1
0
Fork 0
forked from fte/fteqw
fteqw/iqm/iqm.cpp

5843 lines
153 KiB
C++
Raw Normal View History

makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
#define FTEPLUGIN
#define GLQUAKE //this is shit, but ensures index sizes come out the right size
#include "../plugins/plugin.h"
#include "com_mesh.h"
#include "util.h"
#define IQM_UNPACK (1u<<31) //animations will be unpacked into individual frames-as-animations (ie: no more framegroups)
#define IQM_ALLPRIVATE (IQM_UNPACK)
bool noext = false;
bool verbose = false;
bool quiet = false;
struct ejoint
{
const char *name;
int parent;
ejoint() : name(NULL), parent(-1) {}
};
struct triangle { uint vert[3]; triangle() {} triangle(uint v0, uint v1, uint v2) { vert[0] = v0; vert[1] = v1; vert[2] = v2; } };
vector<triangle> triangles, neighbors;
struct mesh { uint name, material; uint firstvert, numverts; uint firsttri, numtris; mesh() : name(0), material(0), firstvert(0), numverts(0), firsttri(0), numtris(0) {} };
vector<mesh> meshes;
struct meshprop
{
uint contents;
uint surfaceflags;
uint body;
uint geomset;
uint geomid;
float mindist;
float maxdist;
meshprop() : contents(0x02000000), surfaceflags(0), body(0), geomset(~0u), geomid(0), mindist(0), maxdist(0) {};
};
vector<meshprop> meshes_fte; //extra crap
uint modelflags; //q1 uses this.
struct event_fte
{
uint anim;
float timestamp; //pose indexes.
uint evcode;
const char *evdata_str;
uint evdata_idx;
};
vector<event_fte> events_fte;
struct anim { uint name; uint firstframe, numframes; float fps; uint flags; anim() : name(0), firstframe(0), numframes(0), fps(0), flags(0) {} };
vector<anim> anims;
struct joint { int group; uint name; int parent; float pos[3], orient[4], scale[3]; joint() : name(0), parent(-1) { memset(pos, 0, sizeof(pos)); memset(orient, 0, sizeof(orient)); memset(scale, 0, sizeof(scale)); } };
vector<joint> joints; //for meshes
struct pose { const char *name; int parent; uint flags; float offset[10], scale[10]; pose() : name(NULL), parent(-1), flags(0) { memset(offset, 0, sizeof(offset)); memset(scale, 0, sizeof(scale)); } };
vector<pose> poses; //aka: animation joints
struct framebounds { Vec3 bbmin, bbmax; double xyradius, radius; framebounds() : bbmin(0, 0, 0), bbmax(0, 0, 0), xyradius(0), radius(0) {} };
vector<framebounds> bounds;
struct transform
{
Vec3 pos;
Quat orient;
Vec3 scale;
transform() {}
transform(const Vec3 &pos, const Quat &orient, const Vec3 &scale = Vec3(1, 1, 1)) : pos(pos), orient(orient), scale(scale) {}
};
struct frame
{
struct framepose
{
int remap;
const char *bonename;
int boneparent;
transform tr;
framepose() : bonename(""),boneparent(-1),tr() {}
framepose(ejoint &j, transform t) : bonename(j.name),boneparent(j.parent),tr(t) {}
};
vector<framepose> pose;
};
vector<frame> frames;
vector<char> stringdata, commentdata;
uint numfverts; //verts generated so far
struct boneoverride
{
const char *name;
bool used;
struct prop
{
const char *rename;
int group;
prop() : rename(NULL), group(-1) {}
} props;
boneoverride() : used(false), props(){}
};
vector<boneoverride> boneoverrides;
struct meshoverride
{
const char *name;
meshprop props;
};
vector<meshoverride> meshoverrides;
struct hitbox
{
int body;
const char *bone;
Vec3 mins, maxs;
};
struct filespec
{
const char *file;
const char *name;
double fps;
uint flags;
int startframe;
int endframe;
meshprop meshprops;
const char *materialprefix;
Quat rotate;
float scale;
Vec3 translate;
bool nomesh;
bool noanim;
vector<event_fte> events;
filespec() { reset(); }
void reset()
{
file = NULL;
name = NULL;
fps = 24;
flags = 0;
startframe = 0;
endframe = -1;
meshprops = meshprop();
materialprefix = NULL;
rotate = Quat(0, 0, 0, 1);
scale = 1;
translate = Vec3(0,0,0);
nomesh = false;
noanim = false;
events.setsize(0);
}
};
struct sharedstring
{
uint offset;
sharedstring() {}
sharedstring(const char *s) : offset(stringdata.length()) { stringdata.put(s, strlen(s)+1); }
};
static inline bool htcmp(const char *x, const sharedstring &s)
{
return htcmp(x, &stringdata[s.offset]);
}
hashtable<sharedstring, uint> stringoffsets;
uint sharestring(const char *s)
{
if(stringdata.empty()) stringoffsets.access("", 0);
return stringoffsets.access(s ? s : "", stringdata.length());
}
struct blendcombo
{
int sorted;
double weights[4];
uchar bones[4];
blendcombo() : sorted(0) {}
void reset() { sorted = 0; }
void addweight(double weight, int bone)
{
if(weight <= 1e-3) return;
loopk(sorted) if(weight > weights[k])
{
for(int l = min(sorted-1, 2); l >= k; l--)
{
weights[l+1] = weights[l];
bones[l+1] = bones[l];
}
weights[k] = weight;
bones[k] = bone;
if(sorted<4) sorted++;
return;
}
if(sorted>=4) return;
weights[sorted] = weight;
bones[sorted] = bone;
sorted++;
}
void finalize()
{
loopj(4-sorted) { weights[sorted+j] = 0; bones[sorted+j] = 0; }
if(sorted <= 0) return;
double total = 0;
loopj(sorted) total += weights[j];
total = 1.0/total;
loopj(sorted) weights[j] *= total;
}
void serialize(uchar *vweights) const
{
int total = 0;
loopk(4) total += (vweights[k] = uchar(0.5 + weights[k]*255));
if(sorted <= 0) return;
while(total > 255)
{
loopk(4) if(vweights[k] > 0 && total > 255) { vweights[k]--; total--; }
}
while(total < 255)
{
loopk(4) if(vweights[k] < 255 && total < 255) { vweights[k]++; total++; }
}
}
bool operator==(const blendcombo &c) { loopi(4) if(weights[i] != c.weights[i] || bones[i] != c.bones[i]) return false; return true; }
bool operator!=(const blendcombo &c) { loopi(4) if(weights[i] != c.weights[i] || bones[i] != c.bones[i]) return true; return false; }
};
vector<Vec4> mpositions;
vector<blendcombo> mblends;
static bool parseindex(char *&c, int &val)
{
while(isspace(*c)) c++;
char *end = NULL;
int rval = strtol(c, &end, 10);
if(c == end) return false;
val = rval;
c = end;
return true;
}
static double parseattrib(char *&c, double ival = 0)
{
while(isspace(*c)) c++;
char *end = NULL;
double val = strtod(c, &end);
if(c == end) val = ival;
else c = end;
return val;
}
static bool maybeparseattrib(char *&c, double &result)
{
while(isspace(*c)) c++;
char *end = NULL;
double val = strtod(c, &end);
if(c == end) return false;
c = end;
result = val;
return true;
}
#if 0
static bool parsename(char *&c, char *buf, int bufsize = sizeof(string))
{
while(isspace(*c)) c++;
char *end;
if(*c == '"')
{
c++;
end = c;
while(*end && *end != '"') end++;
copystring(buf, c, min(int(end-c+1), bufsize));
if(*end == '"') end++;
}
else
{
end = c;
while(*end && !isspace(*end)) end++;
copystring(buf, c, min(int(end-c+1), bufsize));
}
if(c == end) return false;
c = end;
return true;
}
#endif
static char *trimname(char *&c)
{
while(isspace(*c)) c++;
char *start, *end;
if(*c == '"')
{
c++;
start = end = c;
while(*end && *end != '"') end++;
if(*end) { *end = '\0'; end++; }
}
else
{
start = end = c;
while(*end && !isspace(*end)) end++;
if(*end) { *end = '\0'; end++; }
}
c = end;
return start;
}
static Vec4 parseattribs4(char *&c, const Vec4 &ival = Vec4(0, 0, 0, 0))
{
Vec4 val;
loopk(4) val[k] = parseattrib(c, ival[k]);
return val;
}
static Vec3 parseattribs3(char *&c, const Vec3 &ival = Vec3(0, 0, 0))
{
Vec3 val;
loopk(3) val[k] = parseattrib(c, ival[k]);
return val;
}
static blendcombo parseblends(char *&c)
{
blendcombo b;
int index;
while(parseindex(c, index))
{
double weight = parseattrib(c, 0);
b.addweight(weight, index);
}
b.finalize();
return b;
}
struct eanim
{
const char *name;
int startframe, endframe;
double fps;
uint flags;
eanim() : name(NULL), startframe(0), endframe(INT_MAX), fps(0), flags(0) {}
};
struct emesh
{
const char *name, *material;
int firsttri;
bool used;
bool hasexplicits;
meshprop explicits;
emesh() : name(NULL), material(NULL), firsttri(0), used(false), hasexplicits(false) {}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
emesh(const char *name, const char *material, int firsttri = 0) : name(name), material(material), firsttri(firsttri), used(false), hasexplicits(false) {}
};
struct evarray
{
string name;
uint type, format, size;
evarray() : type(IQM_POSITION), format(IQM_FLOAT), size(3) { name[0] = '\0'; }
evarray(uint type, uint format, uint size, const char *initname = "") : type(type), format(format), size(size) { copystring(name, initname); }
};
struct esmoothgroup
{
enum
{
F_USED = 1<<0,
F_UVSMOOTH = 1<<1
};
int key;
float angle;
int flags;
esmoothgroup() : key(-1), angle(-1), flags(0) {}
};
struct etriangle
{
int smoothgroup;
uint vert[3], weld[3];
etriangle()
: smoothgroup(-1)
{
}
etriangle(int v0, int v1, int v2, int smoothgroup = -1)
: smoothgroup(smoothgroup)
{
vert[0] = v0;
vert[1] = v1;
vert[2] = v2;
}
};
vector<Vec4> epositions, etexcoords, etangents, ecolors, ecustom[10];
vector<Vec3> enormals, ebitangents;
vector<blendcombo> eblends;
vector<etriangle> etriangles;
vector<esmoothgroup> esmoothgroups;
vector<int> esmoothindexes;
vector<uchar> esmoothedges;
vector<ejoint> ejoints;
vector<transform> eposes;
vector<Matrix3x4> mjoints;
vector<int> eframes;
vector<eanim> eanims;
vector<emesh> emeshes;
vector<evarray> evarrays;
hashtable<const char *, char *> enames;
const char *getnamekey(const char *name)
{
char **exists = enames.access(name);
if(exists) return *exists;
char *key = newstring(name);
enames[key] = key;
return key;
}
struct weldinfo
{
int tri, vert;
weldinfo *next;
};
void weldvert(const vector<Vec3> &norms, const Vec4 &pos, weldinfo *welds, int &numwelds, unionfind<int> &welder)
{
welder.clear();
int windex = 0;
for(weldinfo *w = welds; w; w = w->next, windex++)
{
etriangle &wt = etriangles[w->tri];
esmoothgroup &wg = esmoothgroups[wt.smoothgroup];
int vindex = windex + 1;
for(weldinfo *v = w->next; v; v = v->next, vindex++)
{
etriangle &vt = etriangles[v->tri];
esmoothgroup &vg = esmoothgroups[vt.smoothgroup];
if(wg.key != vg.key) continue;
if(norms[w->tri].dot(norms[v->tri]) < max(wg.angle, vg.angle)) continue;
if(((wg.flags | vg.flags) & esmoothgroup::F_UVSMOOTH) &&
etexcoords[wt.vert[w->vert]] != etexcoords[vt.vert[v->vert]])
continue;
if(esmoothindexes.length() > max(w->vert, v->vert) && esmoothindexes[w->vert] != esmoothindexes[v->vert])
continue;
if(esmoothedges.length())
{
int w0 = w->vert, w1 = (w->vert+1)%3, w2 = (w->vert+2)%3;
const Vec4 &wp1 = epositions[wt.vert[w1]],
&wp2 = epositions[wt.vert[w2]];
int v0 = v->vert, v1 = (v->vert+1)%3, v2 = (v->vert+2)%3;
const Vec4 &vp1 = epositions[vt.vert[v1]],
&vp2 = epositions[vt.vert[v2]];
int wf = esmoothedges[w->tri], vf = esmoothedges[v->tri];
if((wp1 != vp1 || !(((wf>>w0)|(vf>>v0))&1)) &&
(wp1 != vp2 || !(((wf>>w0)|(vf>>v2))&1)) &&
(wp2 != vp1 || !(((wf>>w2)|(vf>>v0))&1)) &&
(wp2 != vp2 || !(((wf>>w2)|(vf>>v2))&1)))
continue;
}
welder.unite(windex, vindex, -1);
}
}
windex = 0;
for(weldinfo *w = welds; w; w = w->next, windex++)
{
etriangle &wt = etriangles[w->tri];
wt.weld[w->vert] = welder.find(windex, -1, numwelds);
if(wt.weld[w->vert] == uint(numwelds)) numwelds++;
}
}
void smoothverts(bool areaweight = true)
{
if(etriangles.empty()) return;
if(enormals.length())
{
loopv(etriangles)
{
etriangle &t = etriangles[i];
loopk(3) t.weld[k] = t.vert[k];
}
return;
}
if(etexcoords.empty()) loopv(esmoothgroups) esmoothgroups[i].flags &= ~esmoothgroup::F_UVSMOOTH;
if(esmoothedges.length()) while(esmoothedges.length() < etriangles.length()) esmoothedges.add(7);
vector<Vec3> tarea, tnorms;
loopv(etriangles)
{
etriangle &t = etriangles[i];
Vec3 v0(epositions[t.vert[0]]),
v1(epositions[t.vert[1]]),
v2(epositions[t.vert[2]]);
tnorms.add(tarea.add((v2 - v0).cross(v1 - v0)).normalize());
}
int nextalloc = 0;
vector<weldinfo *> allocs;
hashtable<Vec4, weldinfo *> welds(1<<12);
loopv(etriangles)
{
etriangle &t = etriangles[i];
loopk(3)
{
weldinfo **next = &welds.access(epositions[t.vert[k]], NULL);
if(! (nextalloc % 1024)) allocs.add(new weldinfo[1024]);
weldinfo &w = allocs[nextalloc/1024][nextalloc%1024];
nextalloc++;
w.tri = i;
w.vert = k;
w.next = *next;
*next = &w;
}
}
int numwelds = 0;
unionfind<int> welder;
enumerate(welds, Vec4, vpos, weldinfo *, vwelds, weldvert(tnorms, vpos, vwelds, numwelds, welder));
loopv(allocs) delete[] allocs[i];
loopi(numwelds) enormals.add(Vec3(0, 0, 0));
loopv(etriangles)
{
etriangle &t = etriangles[i];
loopk(3) enormals[t.weld[k]]+= areaweight ? tarea[i] : tnorms[i];
}
loopv(enormals) if(enormals[i] != Vec3(0, 0, 0)) enormals[i] = enormals[i].normalize();
}
struct sharedvert
{
int index, weld;
sharedvert() {}
sharedvert(int index, int weld) : index(index), weld(weld) {}
};
static inline bool htcmp(const sharedvert &v, const sharedvert &s)
{
if(epositions[v.index] != epositions[s.index]) return false;
if(etexcoords.length() && etexcoords[v.index] != etexcoords[s.index]) return false;
if(enormals.length() && enormals[v.weld] != enormals[s.weld]) return false;
if(eblends.length() && eblends[v.index] != eblends[s.index]) return false;
if(ecolors.length() && ecolors[v.index] != ecolors[s.index]) return false;
loopi(10) if(ecustom[i].length() && ecustom[i][v.index] != ecustom[i][s.index]) return false;
return true;
}
static inline uint hthash(const sharedvert &v)
{
return hthash(epositions[v.index]);
}
const struct vertexarraytype
{
const char *name;
int code;
} vatypes[] =
{
{ "position", IQM_POSITION },
{ "texcoord", IQM_TEXCOORD },
{ "normal", IQM_NORMAL },
{ "tangent", IQM_TANGENT },
{ "blendindexes", IQM_BLENDINDEXES },
{ "blendweights", IQM_BLENDWEIGHTS },
{ "color", IQM_COLOR },
{ "custom0", IQM_CUSTOM + 0 },
{ "custom1", IQM_CUSTOM + 1 },
{ "custom2", IQM_CUSTOM + 2 },
{ "custom3", IQM_CUSTOM + 3 },
{ "custom4", IQM_CUSTOM + 4 },
{ "custom5", IQM_CUSTOM + 5 },
{ "custom6", IQM_CUSTOM + 6 },
{ "custom7", IQM_CUSTOM + 7 },
{ "custom8", IQM_CUSTOM + 8 },
{ "custom9", IQM_CUSTOM + 9 }
};
int findvertexarraytype(const char *name)
{
loopi(sizeof(vatypes)/sizeof(vatypes[0]))
{
if(!strcasecmp(vatypes[i].name, name))
return vatypes[i].code;
}
return -1;
}
const struct vertexarrayformat
{
const char *name;
int code;
int size;
} vaformats[] =
{
{ "byte", IQM_BYTE, 1 },
{ "ubyte", IQM_UBYTE, 1 },
{ "short", IQM_SHORT, 2 },
{ "ushort", IQM_USHORT, 2 },
{ "int", IQM_INT, 4 },
{ "uint", IQM_UINT, 4 },
{ "half", IQM_HALF, 2 },
{ "float", IQM_FLOAT, 4 },
{ "double", IQM_DOUBLE, 8 }
};
int findvertexarrayformat(const char *name)
{
loopi(sizeof(vaformats)/sizeof(vaformats[0]))
{
if(!strcasecmp(vaformats[i].name, name))
return vaformats[i].code;
}
return -1;
}
struct vertexarray
{
uint type, flags, format, size, offset, count;
vector<uchar> vdata;
vertexarray(uint type, uint format, uint size) : type(type), flags(0), format(format), size(size), offset(0), count(0) {}
int formatsize() const
{
return vaformats[format].size;
}
int bytesize() const
{
return size * vaformats[format].size;
}
};
vector<sharedvert> vmap;
vector<vertexarray> varrays;
vector<uchar> vdata;
struct halfdata
{
ushort val;
halfdata(double d)
{
union
{
ullong i;
double d;
} conv;
conv.d = d;
ushort signbit = ushort((conv.i>>63)&1);
ushort mantissa = ushort((conv.i>>(52-10))&0x3FF);
int exponent = int((conv.i>>52)&0x7FF) - 1023 + 15;
if(exponent <= 0)
{
mantissa |= 0x400;
mantissa >>= min(1-exponent, 10+1);
exponent = 0;
}
else if(exponent >= 0x1F)
{
mantissa = 0;
exponent = 0x1F;
}
val = (signbit<<15) | (ushort(exponent)<<10) | mantissa;
}
};
template<> inline halfdata endianswap<halfdata>(halfdata n) { n.val = endianswap16(n.val); return n; }
template<int TYPE> static inline int remapindex(int i, const sharedvert &v) { return v.index; }
template<> inline int remapindex<IQM_NORMAL>(int i, const sharedvert &v) { return v.weld; }
template<> inline int remapindex<IQM_TANGENT>(int i, const sharedvert &v) { return i; }
template<class T, class U>
static inline void putattrib(T &out, const U &val) { out = T(val); }
template<class T, class U>
static inline void uroundattrib(T &out, const U &val, double scale) { out = T(clamp(0.5 + val*scale, 0.0, scale)); }
template<class T, class U>
static inline void sroundattrib(T &out, const U &val, double scale, double low, double high) { double n = val*scale*0.5; out = T(clamp(n < 0 ? ceil(n - 1) : floor(n), low, high)); }
template<class T, class U>
static inline void scaleattrib(T &out, const U &val) { putattrib(out, val); }
template<class U>
static inline void scaleattrib(char &out, const U &val) { sroundattrib(out, val, 255.0, -128.0, 127.0); }
template<class U>
static inline void scaleattrib(short &out, const U &val) { sroundattrib(out, val, 65535.0, -32768.0, 32767.0); }
template<class U>
static inline void scaleattrib(int &out, const U &val) { sroundattrib(out, val, 4294967295.0, -2147483648.0, 2147483647.0); }
template<class U>
static inline void scaleattrib(uchar &out, const U &val) { uroundattrib(out, val, 255.0); }
template<class U>
static inline void scaleattrib(ushort &out, const U &val) { uroundattrib(out, val, 65535.0); }
template<class U>
static inline void scaleattrib(uint &out, const U &val) { uroundattrib(out, val, 4294967295.0); }
template<int T>
static inline bool normalizedattrib() { return true; }
template<int TYPE, int FMT, class T, class U>
static inline void serializeattrib(const vertexarray &va, T *data, const U &attrib)
{
if(normalizedattrib<TYPE>()) switch(va.size)
{
case 4: scaleattrib(data[3], attrib.w);
case 3: scaleattrib(data[2], attrib.z);
case 2: scaleattrib(data[1], attrib.y);
case 1: scaleattrib(data[0], attrib.x);
}
else switch(va.size)
{
case 4: putattrib(data[3], attrib.w);
case 3: putattrib(data[2], attrib.z);
case 2: putattrib(data[1], attrib.y);
case 1: putattrib(data[0], attrib.x);
}
lilswap(data, va.size);
}
template<int TYPE, int FMT, class T>
static inline void serializeattrib(const vertexarray &va, T *data, const Vec3 &attrib)
{
if(normalizedattrib<TYPE>()) switch(va.size)
{
case 3: scaleattrib(data[2], attrib.z);
case 2: scaleattrib(data[1], attrib.y);
case 1: scaleattrib(data[0], attrib.x);
}
else switch(va.size)
{
case 3: putattrib(data[2], attrib.z);
case 2: putattrib(data[1], attrib.y);
case 1: putattrib(data[0], attrib.x);
}
lilswap(data, va.size);
}
template<int TYPE, int FMT, class T>
static inline void serializeattrib(const vertexarray &va, T *data, const blendcombo &blend)
{
if(TYPE == IQM_BLENDINDEXES)
{
switch(va.size)
{
case 4: putattrib(data[3], blend.bones[3]);
case 3: putattrib(data[2], blend.bones[2]);
case 2: putattrib(data[1], blend.bones[1]);
case 1: putattrib(data[0], blend.bones[0]);
}
}
else if(FMT == IQM_UBYTE)
{
uchar weights[4];
blend.serialize(weights);
switch(va.size)
{
case 4: putattrib(data[3], weights[3]);
case 3: putattrib(data[2], weights[2]);
case 2: putattrib(data[1], weights[1]);
case 1: putattrib(data[0], weights[0]);
}
}
else
{
switch(va.size)
{
case 4: scaleattrib(data[3], blend.weights[3]);
case 3: scaleattrib(data[2], blend.weights[2]);
case 2: scaleattrib(data[1], blend.weights[1]);
case 1: scaleattrib(data[0], blend.weights[0]);
}
}
lilswap(data, va.size);
}
template<int TYPE, class T>
void setupvertexarray(const vector<T> &attribs, uint type, uint fmt, uint size, uint first)
{
const char *name = "";
loopv(evarrays) if(evarrays[i].type == type)
{
evarray &info = evarrays[i];
fmt = info.format;
size = (uint)clamp((int)info.size, 1, 4);
name = info.name;
break;
}
if(type >= IQM_CUSTOM)
{
if(!name[0])
{
defformatstring(customname, "custom%d", type-IQM_CUSTOM);
type = IQM_CUSTOM + sharestring(customname);
}
else type = IQM_CUSTOM + sharestring(name);
}
int k;
for (k = 0; k < varrays.length(); k++)
{
if (varrays[k].type == type && varrays[k].format == fmt && varrays[k].size == size)
break;
}
if (k == varrays.length())
varrays.add(vertexarray(type, fmt, size));
vertexarray &va = varrays[k];
if (va.count != first)
fatal("count != first"); //gaps are a problem.
va.count += vmap.length();
int totalsize = va.bytesize() * vmap.length();
uchar *data = va.vdata.reserve(totalsize);
va.vdata.advance(totalsize);
loopv(vmap)
{
const T &attrib = attribs[remapindex<TYPE>(i, vmap[i])];
switch(va.format)
{
case IQM_BYTE: serializeattrib<TYPE, IQM_BYTE>(va, (char *)data, attrib); break;
case IQM_UBYTE: serializeattrib<TYPE, IQM_UBYTE>(va, (uchar *)data, attrib); break;
case IQM_SHORT: serializeattrib<TYPE, IQM_SHORT>(va, (short *)data, attrib); break;
case IQM_USHORT: serializeattrib<TYPE, IQM_USHORT>(va, (ushort *)data, attrib); break;
case IQM_INT: serializeattrib<TYPE, IQM_INT>(va, (int *)data, attrib); break;
case IQM_UINT: serializeattrib<TYPE, IQM_UINT>(va, (uint *)data, attrib); break;
case IQM_HALF: serializeattrib<TYPE, IQM_HALF>(va, (halfdata *)data, attrib); break;
case IQM_FLOAT: serializeattrib<TYPE, IQM_FLOAT>(va, (float *)data, attrib); break;
case IQM_DOUBLE: serializeattrib<TYPE, IQM_DOUBLE>(va, (double *)data, attrib); break;
}
data += va.bytesize();
}
}
// linear speed vertex cache optimization from Tom Forsyth
#define MAXVCACHE 32
struct triangleinfo
{
bool used;
float score;
uint vert[3];
triangleinfo() {}
triangleinfo(uint v0, uint v1, uint v2)
{
vert[0] = v0;
vert[1] = v1;
vert[2] = v2;
}
};
struct vertexcache : listnode<vertexcache>
{
int index, rank;
float score;
int numuses;
triangleinfo **uses;
vertexcache() : index(-1), rank(-1), score(-1.0f), numuses(0), uses(NULL) {}
void calcscore()
{
if(numuses > 0)
{
score = 2.0f * powf(numuses, -0.5f);
if(rank >= 3) score += powf(1.0f - (rank - 3)/float(MAXVCACHE - 3), 1.5f);
else if(rank >= 0) score += 0.75f;
}
else score = -1.0f;
}
void removeuse(triangleinfo *t)
{
loopi(numuses) if(uses[i] == t)
{
uses[i] = uses[--numuses];
return;
}
}
};
void maketriangles(vector<triangleinfo> &tris, const vector<sharedvert> &mmap)
{
triangleinfo **uses = new triangleinfo *[3*tris.length()];
vertexcache *verts = new vertexcache[mmap.length()];
list<vertexcache> vcache;
loopv(tris)
{
triangleinfo &t = tris[i];
t.used = t.vert[0] == t.vert[1] || t.vert[1] == t.vert[2] || t.vert[2] == t.vert[0];
if(t.used) continue;
loopk(3) verts[t.vert[k]].numuses++;
}
triangleinfo **curuse = uses;
loopvrev(tris)
{
triangleinfo &t = tris[i];
if(t.used) continue;
loopk(3)
{
vertexcache &v = verts[t.vert[k]];
if(!v.uses) { curuse += v.numuses; v.uses = curuse; }
*--v.uses = &t;
}
}
loopv(mmap) verts[i].calcscore();
triangleinfo *besttri = NULL;
float bestscore = -1e16f;
loopv(tris)
{
triangleinfo &t = tris[i];
if(t.used) continue;
t.score = verts[t.vert[0]].score + verts[t.vert[1]].score + verts[t.vert[2]].score;
if(t.score > bestscore) { besttri = &t; bestscore = t.score; }
}
//int reloads = 0;
while(besttri)
{
besttri->used = true;
triangle &t = triangles.add();
loopk(3)
{
vertexcache &v = verts[besttri->vert[k]];
if(v.index < 0) { v.index = vmap.length(); vmap.add(mmap[besttri->vert[k]]); }
t.vert[k] = v.index;
v.removeuse(besttri);
if(v.rank >= 0) vcache.remove(&v)->rank = -1;
// else reloads++;
if(v.numuses <= 0) continue;
vcache.insertfirst(&v);
v.rank = 0;
}
int rank = 0;
for(vertexcache *v = vcache.first(); v != vcache.end(); v = v->next)
{
v->rank = rank++;
v->calcscore();
}
besttri = NULL;
bestscore = -1e16f;
for(vertexcache *v = vcache.first(); v != vcache.end(); v = v->next)
{
loopi(v->numuses)
{
triangleinfo &t = *v->uses[i];
t.score = verts[t.vert[0]].score + verts[t.vert[1]].score + verts[t.vert[2]].score;
if(t.score > bestscore) { besttri = &t; bestscore = t.score; }
}
}
while(vcache.size > MAXVCACHE) vcache.removelast()->rank = -1;
if(!besttri) loopv(tris)
{
triangleinfo &t = tris[i];
if(!t.used && t.score > bestscore) { besttri = &t; bestscore = t.score; }
}
}
// printf("reloads: %d, worst: %d, best: %d\n", reloads, tris.length()*3, mmap.length());
delete[] uses;
delete[] verts;
}
void calctangents(uint priortris, bool areaweight = true)
{
uint numverts = vmap.length();
Vec3 *tangent = new Vec3[2*numverts], *bitangent = tangent+numverts;
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
for (uint i = 0; i < 2*numverts; i++)
tangent[i] = Vec3(0,0,0);
for (int i = priortris; i < triangles.length(); i++)
{
const triangle &t = triangles[i];
sharedvert &i0 = vmap[t.vert[0]],
&i1 = vmap[t.vert[1]],
&i2 = vmap[t.vert[2]];
Vec3 v0(epositions[i0.index]), e1 = Vec3(epositions[i1.index]) - v0, e2 = Vec3(epositions[i2.index]) - v0;
double u1 = etexcoords[i1.index].x - etexcoords[i0.index].x, v1 = etexcoords[i1.index].y - etexcoords[i0.index].y,
u2 = etexcoords[i2.index].x - etexcoords[i0.index].x, v2 = etexcoords[i2.index].y - etexcoords[i0.index].y;
Vec3 u = e2*v1 - e1*v2,
v = e2*u1 - e1*u2;
if(e2.cross(e1).dot(v.cross(u)) < 0)
{
u = -u;
v = -v;
}
if(!areaweight)
{
u = u.normalize();
v = v.normalize();
}
loopj(3)
{
tangent[t.vert[j]] += u;
bitangent[t.vert[j]] += v;
}
}
loopv(vmap)
{
const Vec3 &n = enormals[vmap[i].weld],
&t = tangent[i],
&bt = bitangent[i];
etangents.add(Vec4((t - n*n.dot(t)).normalize(), n.cross(t).dot(bt) < 0 ? -1 : 1));
}
delete[] tangent;
}
struct neighborkey
{
uint e0, e1;
neighborkey() {}
neighborkey(uint i0, uint i1)
{
if(epositions[i0] < epositions[i1]) { e0 = i0; e1 = i1; }
else { e0 = i1; e1 = i0; }
}
uint hash() const { return hthash(epositions[e0]) + hthash(epositions[e1]); }
bool operator==(const neighborkey &n) const
{
return epositions[e0] == epositions[n.e0] && epositions[e1] == epositions[n.e1] &&
(eblends.empty() || (eblends[e0] == eblends[n.e0] && eblends[e1] == eblends[n.e1]));
}
};
static inline uint hthash(const neighborkey &n) { return n.hash(); }
static inline bool htcmp(const neighborkey &x, const neighborkey &y) { return x == y; }
struct neighborval
{
uint tris[2];
neighborval() {}
neighborval(uint i) { tris[0] = i; tris[1] = 0xFFFFFFFFU; }
void add(uint i)
{
if(tris[1] != 0xFFFFFFFFU) tris[0] = tris[1] = 0xFFFFFFFFU;
else if(tris[0] != 0xFFFFFFFFU) tris[1] = i;
}
int opposite(uint i) const
{
return tris[0] == i ? tris[1] : tris[0];
}
};
void makeneighbors(uint priortris)
{
hashtable<neighborkey, neighborval> nhash;
for(int i = priortris; i<triangles.length(); i++)
{
triangle &t = triangles[i];
for(int j = 0, p = 2; j < 3; p = j, j++)
{
neighborkey key(t.vert[p], t.vert[j]);
neighborval *val = nhash.access(key);
if(val) val->add(i);
else nhash[key] = neighborval(i);
}
}
for(int i = priortris; i<triangles.length(); i++)
{
triangle &t = triangles[i];
triangle &n = neighbors.add();
for(int j = 0, p = 2; j < 3; p = j, j++)
n.vert[p] = nhash[neighborkey(t.vert[p], t.vert[j])].opposite(i);
}
}
Quat erotate;
double escale = 1;
Vec3 emeshtrans(0, 0, 0);
Vec3 ejointtrans(0, 0, 0);
double gscale = 1;
Vec3 gmeshtrans(0,0,0);
void printlastmesh(void)
{
if (quiet)
return;
mesh &m = meshes[meshes.length()-1];
meshprop &fm = meshes_fte[meshes.length()-1];
printf(" %smesh %i:\tname=\"%s\",\tmat=\"%s\",\ttri=%i, vert=%i\n", fm.contents?"c":"r", meshes.length()-1,
&stringdata[m.name], &stringdata[m.material], m.numtris, m.numverts);
if (verbose)
{
if (noext)
printf(" writing mesh properties is disabled\n");
else
printf(" c=%#x sf=%#x b=%i gs=%i gi=%i nd=%g fd=%g\n", fm.contents, fm.surfaceflags, fm.body, fm.geomset, fm.geomid, fm.maxdist, fm.mindist);
}
}
void makemeshes(const filespec &spec)
{
if (spec.nomesh)
return;
/*
if (meshes.length())
return;
meshes.setsize(0);
meshes_fte.setsize(0);
triangles.setsize(0);
neighbors.setsize(0);
varrays.setsize(0);
vdata.setsize(0);
*/
int priorverts = numfverts;
int priortris = triangles.length();
hashtable<sharedvert, uint> mshare(1<<12);
vector<sharedvert> mmap;
vector<triangleinfo> tinfo;
if (!noext)
{
loopv(emeshes)
{
if (!emeshes[i].hasexplicits)
emeshes[i].explicits = spec.meshprops;
loopk(meshoverrides.length())
{
if (!strcmp(meshoverrides[k].name, emeshes[i].name))
{
emeshes[i].explicits = meshoverrides[k].props;
for (; k < meshoverrides.length()-1; k++)
meshoverrides[k] = meshoverrides[k+1];
meshoverrides.drop();
break;
}
}
}
}
loopv(emeshes)
{
emesh &em1 = emeshes[i];
if(em1.used) continue;
for(int j = i; j < emeshes.length(); j++)
{
emesh &em = emeshes[j];
if(strcmp(em.name, em1.name) || strcmp(em.material, em1.material) || memcmp(&em.explicits, &em1.explicits, sizeof(em.explicits))) continue;
int lasttri = emeshes.inrange(j+1) ? emeshes[j+1].firsttri : etriangles.length();
for(int k = em.firsttri; k < lasttri; k++)
{
etriangle &et = etriangles[k];
triangleinfo &t = tinfo.add();
loopl(3)
{
sharedvert v(et.vert[l], et.weld[l]);
t.vert[l] = mshare.access(v, mmap.length());
if(!mmap.inrange(t.vert[l])) mmap.add(v);
}
}
em.used = true;
}
if(tinfo.empty()) continue;
mesh &m = meshes.add();
m.name = sharestring(em1.name);
if (spec.materialprefix)
{
char material[512];
formatstring(material, "%s%s", spec.materialprefix, em1.material);
m.material = sharestring(material);
}
else
m.material = sharestring(em1.material);
m.firsttri = triangles.length();
m.firstvert = numfverts+vmap.length();
maketriangles(tinfo, mmap);
m.numtris = triangles.length() - m.firsttri;
m.numverts = numfverts+vmap.length() - m.firstvert;
meshprop &mf = meshes_fte.add();
mf = em1.explicits;
printlastmesh();
mshare.clear();
mmap.setsize(0);
tinfo.setsize(0);
}
numfverts+=vmap.length();
if(triangles.length()) makeneighbors(priortris);
if(escale != 1) loopv(epositions) epositions[i] *= escale;
if(erotate != Quat(0, 0, 0, 1))
{
loopv(epositions) epositions[i].setxyz(erotate.transform(Vec3(epositions[i])));
loopv(enormals) enormals[i] = erotate.transform(enormals[i]);
loopv(etangents) etangents[i].setxyz(erotate.transform(Vec3(etangents[i])));
loopv(ebitangents) ebitangents[i] = erotate.transform(ebitangents[i]);
}
if(emeshtrans != Vec3(0, 0, 0)) loopv(epositions) epositions[i] += emeshtrans;
if(epositions.length()) setupvertexarray<IQM_POSITION>(epositions, IQM_POSITION, IQM_FLOAT, 3, priorverts);
if(etexcoords.length()) setupvertexarray<IQM_TEXCOORD>(etexcoords, IQM_TEXCOORD, IQM_FLOAT, 2, priorverts);
if(enormals.length()) setupvertexarray<IQM_NORMAL>(enormals, IQM_NORMAL, IQM_FLOAT, 3, priorverts);
if(etangents.length())
{
if(ebitangents.length() && enormals.length())
{
loopv(etangents) if(ebitangents.inrange(i) && enormals.inrange(i))
etangents[i].w = enormals[i].cross(Vec3(etangents[i])).dot(ebitangents[i]) < 0 ? -1 : 1;
}
setupvertexarray<IQM_TANGENT>(etangents, IQM_TANGENT, IQM_FLOAT, 4, priorverts);
}
else if(enormals.length() && etexcoords.length())
{
calctangents(priortris);
setupvertexarray<IQM_TANGENT>(etangents, IQM_TANGENT, IQM_FLOAT, 4, priorverts);
}
if(eblends.length())
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if (ejoints.length() > 65535)
setupvertexarray<IQM_BLENDINDEXES>(eblends, IQM_BLENDINDEXES, IQM_UINT, 4, priorverts);
else if (ejoints.length() > 255)
setupvertexarray<IQM_BLENDINDEXES>(eblends, IQM_BLENDINDEXES, IQM_USHORT, 4, priorverts);
else
setupvertexarray<IQM_BLENDINDEXES>(eblends, IQM_BLENDINDEXES, IQM_UBYTE, 4, priorverts);
setupvertexarray<IQM_BLENDWEIGHTS>(eblends, IQM_BLENDWEIGHTS, IQM_UBYTE, 4, priorverts);
}
if(ecolors.length()) setupvertexarray<IQM_COLOR>(ecolors, IQM_COLOR, IQM_UBYTE, 4, priorverts);
loopi(10) if(ecustom[i].length()) setupvertexarray<IQM_CUSTOM>(ecustom[i], IQM_CUSTOM + i, IQM_FLOAT, 4, priorverts);
//make sure we keep this data in a usable form so that we can calc framebounds.
if(epositions.length())
{
Vec4 *o = mpositions.reserve(epositions.length());
mpositions.advance(epositions.length());
loopv(epositions)
o[i] = epositions[i];
}
if(eblends.length())
{
blendcombo *o = mblends.reserve(eblends.length());
mblends.advance(eblends.length());
loopv(eblends)
o[i] = eblends[i];
}
//the generated triangles currently refer to the imported arrays.
//make sure they refer to the final verts
if (priorverts)
for (int i = priortris; i < triangles.length(); i++)
{
triangles[i].vert[0] += priorverts;
triangles[i].vert[1] += priorverts;
triangles[i].vert[2] += priorverts;
}
}
void makebounds(framebounds &bb, Matrix3x4 *invbase, frame &frame)
{
vector<Matrix3x4> buf;
buf.growbuf(joints.length());
buf.setsize(joints.length());
//make sure all final bones have some value, even if its gibberish. should probably ignore verts that depend upon bones not defined in this animation.
//remap<0 means the bone was dropped.
loopv(buf) {buf[i] = Matrix3x4(Quat(0,0,0,1),Vec3(0,0,0));}
loopv(frame.pose) if (frame.pose[i].remap>=0)
{
int bone = frame.pose[i].remap;
int jparent = frame.pose[i].boneparent;
if (jparent >= 0) jparent = frame.pose[jparent].remap;
if(jparent >= 0) buf[bone] = buf[jparent] * Matrix3x4(frame.pose[i].tr.orient, frame.pose[i].tr.pos, frame.pose[i].tr.scale);
else buf[bone] = Matrix3x4(frame.pose[i].tr.orient, frame.pose[i].tr.pos, frame.pose[i].tr.scale);
}
loopv(frame.pose) buf[i] *= invbase[i];
loopv(mpositions)
{
const blendcombo &c = mblends[i];
Matrix3x4 m(Vec4(0, 0, 0, 0), Vec4(0, 0, 0, 0), Vec4(0, 0, 0, 0));
loopk(4) if(c.weights[k] > 0)
m += buf[c.bones[k]] * c.weights[k];
Vec3 p = m.transform(Vec3(mpositions[i]));
if(!i) bb.bbmin = bb.bbmax = p;
else
{
bb.bbmin.x = min(bb.bbmin.x, p.x);
bb.bbmin.y = min(bb.bbmin.y, p.y);
bb.bbmin.z = min(bb.bbmin.z, p.z);
bb.bbmax.x = max(bb.bbmax.x, p.x);
bb.bbmax.y = max(bb.bbmax.y, p.y);
bb.bbmax.z = max(bb.bbmax.z, p.z);
}
double xyradius = p.x*p.x + p.y*p.y;
bb.xyradius = max(bb.xyradius, xyradius);
bb.radius = max(bb.radius, xyradius + p.z*p.z);
}
if(bb.xyradius > 0) bb.xyradius = sqrt(bb.xyradius);
if(bb.radius > 0) bb.radius = sqrt(bb.radius);
}
void makerelativebasepose()
{
int numbasejoints = min(ejoints.length(), eframes.length() ? eframes[0] : eposes.length());
for(int i = numbasejoints-1; i >= 0; i--)
{
ejoint &ej = ejoints[i];
if(ej.parent < 0) continue;
transform &parent = eposes[ej.parent], &child = eposes[i];
child.pos = (-parent.orient).transform(child.pos - parent.pos);
child.orient = (-parent.orient)*child.orient;
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
child.scale = child.scale / parent.scale;
if(child.orient.w > 0) child.orient.flip();
}
}
bool forcejoints = false;
void printlastanim(void)
{
if (quiet)
return;
anim &a = anims[anims.length()-1];
if (a.numframes == 1)
printf(" frame %i:\tname=\"%s\"\tfps=%g, %s\n", anims.length()-1,
&stringdata[a.name], a.fps, (a.flags & IQM_LOOP)?"looped":"clamped");
else
printf(" anim %i:\tname=\"%s\",\tframes=%i, fps=%g, %s\n", anims.length()-1,
&stringdata[a.name], a.numframes, a.fps, (a.flags & IQM_LOOP)?"looped":"clamped");
loopv(events_fte)
{
if (events_fte[i].anim == (uint)anims.length()-1)
printf(" pose %g: %x \"%s\"\n", events_fte[i].timestamp*a.fps, events_fte[i].evcode, events_fte[i].evdata_str);
}
}
void printbones(int parent = -1, size_t ind = 1)
{
char prefix[256];
if (ind >= sizeof(prefix))
ind = sizeof(prefix)-1;
memset(prefix, ' ', ind);
prefix[ind] = 0;
loopv(joints)
{
if (joints[i].parent == parent)
{ //show as 1-based for consistency with quake.
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
conoutf("%sbone %i:\tname=\"%s\"\tparent=%i, group=%i (%f %f %f)", prefix, i+1, &stringdata[joints[i].name], joints[i].parent+1, joints[i].group, joints[i].pos[0], joints[i].pos[1], joints[i].pos[2]);
printbones(i, ind+1);
}
}
}
void printbonelist()
{
loopv(joints)
{
conoutf("bone %i:\tname=\"%s\"\tparent=%i%s, group=%i", i+1, &stringdata[joints[i].name], joints[i].parent+1, joints[i].parent >= i?"(ERROR)":"", joints[i].group);
}
}
int findjoint(const char *name)
{
loopv(joints)
{
if (!strcmp(&stringdata[joints[i].name], name))
return i;
}
return -1;
}
bool floatcmp(float f1, float f2)
{
if (f1 != f2)
return true;
return false;
}
void makeanims(const filespec &spec)
{
if(escale != 1) loopv(eposes) eposes[i].pos *= escale;
if(erotate != Quat(0, 0, 0, 1)) loopv(ejoints)
{
ejoint &ej = ejoints[i];
if(ej.parent < 0) for(int j = i; j < eposes.length(); j += ejoints.length())
{
transform &p = eposes[j];
p.orient = erotate * p.orient;
p.pos = erotate.transform(p.pos);
}
}
int numbasejoints = eframes.length() ? eframes[0] : eposes.length();
if(forcejoints || emeshes.length())
{
bool warned = false;
int *jr = new int[ejoints.length()];
loopv(ejoints)
{
ejoint &ej = ejoints[i];
jr[i] = findjoint(ej.name);
if (jr[i] >= 0)
{
bool rigmismatch = false;
if (warned || forcejoints)
continue;
joint &j = joints[jr[i]];
loopk(3) if (floatcmp(j.pos[k], eposes[i].pos[k] + (ej.parent>=0?0:ejointtrans[k]))) rigmismatch = true;
loopk(4) if (floatcmp(j.orient[k], eposes[i].orient[k])) rigmismatch = true;
loopk(3) if (floatcmp(j.scale[k], eposes[i].scale[k])) rigmismatch = true;
if (rigmismatch)
{
warned = true;
conoutf("warning: rig mismatch (bone %s)", ej.name);
}
continue;
}
jr[i] = joints.length();
joint &j = joints.add();
Matrix3x4 &m = mjoints.add();
const char *name = ej.name;
int group = -1;
loopvk(boneoverrides)
{
if (!strcmp(boneoverrides[k].name, name))
{
boneoverrides[k].used = true;
if (boneoverrides[k].props.rename)
name = boneoverrides[k].props.rename;
if (boneoverrides[k].props.group >= 0)
group = boneoverrides[k].props.group;
break;
}
}
j.name = sharestring(name);
if (ej.parent >= 0)
j.parent = findjoint(ejoints[ej.parent].name);
else
j.parent = -1;
if (group < 0 && j.parent >= 0)
group = joints[j.parent].group;
if (group < 0)
group = 0;
j.group = group;
if(i < numbasejoints)
{
m.invert(Matrix3x4(eposes[i].orient, eposes[i].pos + (ej.parent>=0?Vec3(0,0,0):ejointtrans), eposes[i].scale));
loopk(3) j.pos[k] = eposes[i].pos[k] + (ej.parent>=0?0:ejointtrans[k]);
loopk(4) j.orient[k] = eposes[i].orient[k];
loopk(3) j.scale[k] = eposes[i].scale[k];
}
else m.invert(Matrix3x4(Quat(0, 0, 0, 1), Vec3(0, 0, 0), Vec3(1, 1, 1)));
if(j.parent >= 0) m *= mjoints[j.parent];
}
loopv(eblends)
{
loopk(eblends[i].sorted)
{
int b = eblends[i].bones[k];
if (b >= ejoints.length())
b = 0;
else
b = jr[b];
eblends[i].bones[k] = b;
}
}
delete[] jr;
}
// loopv(spec.events)
// spec.events[i].evdata_idx = 0;
loopv(eanims)
{
eanim &ea = eanims[i];
if (ea.flags & IQM_UNPACK)
{ //some quake mods suck, and are unable to deal with animations
for (int j = ea.startframe, end = eanims.inrange(i+1) ? eanims[i+1].startframe : eframes.length(); j < end && j < ea.endframe; j++)
{
anim &a = anims.add();
char nname[256];
formatstring(nname, "%s%i", ea.name, j+1-ea.startframe);
a.name = sharestring(nname);
a.firstframe = frames.length();
a.numframes = 0;
a.fps = ea.fps;
a.flags = ea.flags&~IQM_ALLPRIVATE;
int offset = eframes[j], range = (eframes.inrange(j+1) ? eframes[j+1] : eposes.length()) - offset;
if(range <= 0) continue;
frame &fr = frames.add();
loopk(min(range, ejoints.length())) fr.pose.add(frame::framepose(ejoints[i], eposes[offset + k]));
loopk(max(ejoints.length() - range, 0)) fr.pose.add(frame::framepose(ejoints[i], transform(Vec3(0, 0, 0), Quat(0, 0, 0, 1), Vec3(1, 1, 1))));
a.numframes++;
printlastanim();
}
}
else
{
anim &a = anims.add();
a.name = sharestring(ea.name);
a.firstframe = frames.length();
a.numframes = 0;
a.fps = ea.fps;
a.flags = ea.flags&~IQM_ALLPRIVATE;
for(int j = ea.startframe, end = eanims.inrange(i+1) ? eanims[i+1].startframe : eframes.length(); j < end && j <= ea.endframe; j++)
{
int offset = eframes[j], range = (eframes.inrange(j+1) ? eframes[j+1] : eposes.length()) - offset;
if(range <= 0) continue;
frame &fr = frames.add();
loopk(min(range, ejoints.length()))
{
if (ejoints[k].parent < 0)
eposes[offset+k].pos += ejointtrans;
fr.pose.add(frame::framepose(ejoints[k], eposes[offset + k]));
}
loopk(max(ejoints.length() - range, 0))
{
if (ejoints[k].parent < 0)
fr.pose.add(frame::framepose(ejoints[k], transform(ejointtrans, Quat(0, 0, 0, 1), Vec3(1, 1, 1))));
else
fr.pose.add(frame::framepose(ejoints[k], transform(Vec3(0, 0, 0), Quat(0, 0, 0, 1), Vec3(1, 1, 1))));
}
a.numframes++;
}
loopvj(spec.events)
{
float p;
if (spec.events[j].anim == ~0u)
{
if (spec.events[j].timestamp < ea.startframe || spec.events[j].timestamp >= ea.startframe + a.numframes)
continue;
p = spec.events[j].timestamp - ea.startframe;
}
else
{
if (spec.events[j].anim != (uint)i)
continue;
else
p = spec.events[j].timestamp;
}
event_fte &ev = events_fte.add(spec.events[j]);
ev.anim = anims.length()-1;
ev.timestamp = p / a.fps;
}
printlastanim();
}
}
// loopv(spec.events) if (!spec.events[i].evdata_idx)
// {
// conoutf("event specifies invalid animation from %s", spec.file);
// }
}
bool resetimporter(const filespec &spec, bool reuse = false)
{
if(reuse)
{
ejoints.setsize(0);
evarrays.setsize(0);
return false;
}
vmap.setsize(0);
epositions.setsize(0);
etexcoords.setsize(0);
enormals.setsize(0);
etangents.setsize(0);
ebitangents.setsize(0);
ecolors.setsize(0);
loopi(10) ecustom[i].setsize(0);
eblends.setsize(0);
etriangles.setsize(0);
esmoothindexes.setsize(0);
esmoothedges.setsize(0);
esmoothgroups.setsize(0);
esmoothgroups.add();
ejoints.setsize(0);
eposes.setsize(0);
eframes.setsize(0);
eanims.setsize(0);
emeshes.setsize(0);
evarrays.setsize(0);
emeshtrans = gmeshtrans+spec.translate;
ejointtrans = spec.translate;
erotate = spec.rotate;
escale = gscale*spec.scale;
return true;
}
bool parseiqe(stream *f)
{
const char *curmesh = getnamekey(""), *curmaterial = getnamekey("");
bool needmesh = true;
int fmoffset = 0;
char buf[512];
if(!f->getline(buf, sizeof(buf))) return false;
if(!strchr(buf, '#') || strstr(buf, "# Inter-Quake Export") != strchr(buf, '#')) return false;
while(f->getline(buf, sizeof(buf)))
{
char *c = buf;
while(isspace(*c)) ++c;
if(isalpha(c[0]) && isalnum(c[1]) && (!c[2] || isspace(c[2]))) switch(*c++)
{
case 'v':
switch(*c++)
{
case 'p': epositions.add(parseattribs4(c, Vec4(0, 0, 0, 1))); continue;
case 't': etexcoords.add(parseattribs4(c)); continue;
case 'n': enormals.add(parseattribs3(c)); continue;
case 'x':
{
Vec4 tangent(parseattribs3(c), 0);
Vec3 bitangent(0, 0, 0);
bitangent.x = parseattrib(c);
if(maybeparseattrib(c, bitangent.y))
{
bitangent.z = parseattrib(c);
ebitangents.add(bitangent);
}
else tangent.w = bitangent.x;
etangents.add(tangent);
continue;
}
case 'b': eblends.add(parseblends(c)); continue;
case 'c': ecolors.add(parseattribs4(c, Vec4(0, 0, 0, 1))); continue;
case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9':
{
int n = c[-1] - '0';
ecustom[n].add(parseattribs4(c));
continue;
}
case 's':
parseindex(c, esmoothindexes.add());
continue;
}
break;
case 'p':
{
transform t;
switch(*c++)
{
case 'q':
{
t.pos = parseattribs3(c);
loopk(3) t.orient[k] = parseattrib(c);
t.orient.restorew();
double w = parseattrib(c, t.orient.w);
if(w != t.orient.w)
{
t.orient.w = w;
t.orient.normalize();
// double x2 = f.orient.x*f.orient.x, y2 = f.orient.y*f.orient.y, z2 = f.orient.z*f.orient.z, w2 = f.orient.w*f.orient.w, s2 = x2 + y2 + z2 + w2;
// f.orient.x = keepsign(f.orient.x, sqrt(max(1.0 - (w2 + y2 + z2) / s2, 0.0)));
// f.orient.y = keepsign(f.orient.y, sqrt(max(1.0 - (w2 + x2 + z2) / s2, 0.0)));
// f.orient.z = keepsign(f.orient.z, sqrt(max(1.0 - (w2 + x2 + y2) / s2, 0.0)));
// f.orient.w = keepsign(f.orient.w, sqrt(max(1.0 - (x2 + y2 + z2) / s2, 0.0)));
}
if(t.orient.w > 0) t.orient.flip();
t.scale = parseattribs3(c, Vec3(1, 1, 1));
eposes.add(t);
continue;
}
case 'm':
{
t.pos = parseattribs3(c);
Matrix3x3 m;
m.a = parseattribs3(c);
m.b = parseattribs3(c);
m.c = parseattribs3(c);
Vec3 mscale(Vec3(m.a.x, m.b.x, m.c.x).magnitude(), Vec3(m.a.y, m.b.y, m.c.y).magnitude(), Vec3(m.a.z, m.b.z, m.c.z).magnitude());
// check determinant for sign of scaling
if(m.determinant() < 0) mscale = -mscale;
m.a /= mscale;
m.b /= mscale;
m.c /= mscale;
t.orient = Quat(m);
if(t.orient.w > 0) t.orient.flip();
t.scale = parseattribs3(c, Vec3(1, 1, 1)) * mscale;
eposes.add(t);
continue;
}
case 'a':
{
t.pos = parseattribs3(c);
Vec3 rot = parseattribs3(c);
t.orient = Quat::fromangles(rot);
t.scale = parseattribs3(c, Vec3(1, 1, 1));
eposes.add(t);
continue;
}
}
break;
}
case 'f':
switch(*c++)
{
case 'a':
{
int i1 = 0, i2 = 0, i3 = 0;
if(!parseindex(c, i1) || !parseindex(c, i2)) continue;
if(needmesh)
{
emeshes.add(emesh(curmesh, curmaterial, etriangles.length()));
needmesh = false;
}
if(i1 < 0) i1 = max(epositions.length() + i1, 0);
if(i2 < 0) i2 = max(epositions.length() + i2, 0);
while(parseindex(c, i3))
{
if(i3 < 0) i3 = max(epositions.length() + i3, 0);
esmoothgroups.last().flags |= esmoothgroup::F_USED;
etriangles.add(etriangle(i1, i2, i3, esmoothgroups.length()-1));
i2 = i3;
}
continue;
}
case 'm':
{
int i1 = 0, i2 = 0, i3 = 0;
if(!parseindex(c, i1) || !parseindex(c, i2)) continue;
if(needmesh)
{
emeshes.add(emesh(curmesh, curmaterial, etriangles.length()));
needmesh = false;
}
i1 = i1 < 0 ? max(epositions.length() + i1, 0) : (fmoffset + i1);
i2 = i2 < 0 ? max(epositions.length() + i2, 0) : (fmoffset + i2);
while(parseindex(c, i3))
{
i3 = i3 < 0 ? max(epositions.length() + i3, 0) : (fmoffset + i3);
esmoothgroups.last().flags |= esmoothgroup::F_USED;
etriangles.add(etriangle(i1, i2, i3, esmoothgroups.length()-1));
i2 = i3;
}
continue;
}
case 's':
{
int i1 = 0, i2 = 0, i3 = 0;
uchar flags = 0;
if(!parseindex(c, i1) || !parseindex(c, i2) || !parseindex(c, i3)) continue;
flags |= clamp(i1, 0, 1);
flags |= clamp(i2, 0, 1)<<1;
flags |= clamp(i3, 0, 1)<<2;
esmoothgroups.last().flags |= esmoothgroup::F_USED;
while(parseindex(c, i3))
{
esmoothedges.add(flags | 4);
flags = 1 | ((flags & 4) >> 1) | (clamp(i3, 0, 1)<<2);
}
esmoothedges.add(flags);
continue;
}
}
break;
}
char *args = c;
while(*args && !isspace(*args)) args++;
if(!strncmp(c, "smoothgroup", max(int(args-c), 11)))
{
if(esmoothgroups.last().flags & esmoothgroup::F_USED) esmoothgroups.dup();
parseindex(args, esmoothgroups.last().key);
}
else if(!strncmp(c, "smoothangle", max(int(args-c), 11)))
{
if(esmoothgroups.last().flags & esmoothgroup::F_USED) esmoothgroups.dup();
double angle = parseattrib(args, 0);
esmoothgroups.last().angle = fabs(cos(clamp(angle, -180.0, 180.0) * M_PI/180));
}
else if(!strncmp(c, "smoothuv", max(int(args-c), 8)))
{
if(esmoothgroups.last().flags & esmoothgroup::F_USED) esmoothgroups.dup();
int val = 1;
if(parseindex(args, val) && val <= 0) esmoothgroups.last().flags &= ~esmoothgroup::F_UVSMOOTH;
else esmoothgroups.last().flags |= esmoothgroup::F_UVSMOOTH;
}
else if(!strncmp(c, "mesh", max(int(args-c), 4)))
{
curmesh = getnamekey(trimname(args));
if(emeshes.empty() || emeshes.last().name != curmesh) needmesh = true;
fmoffset = epositions.length();
#if 0
emesh &m = emeshes.add();
m.firsttri = etriangles.length();
fmoffset = epositions.length();
parsename(args, m.name);
#endif
}
else if(!strncmp(c, "material", max(int(args-c), 8)))
{
curmaterial = getnamekey(trimname(args));
if(emeshes.empty() || emeshes.last().material != curmaterial) needmesh = true;
// if(emeshes.length()) parsename(c, emeshes.last().material);
}
else if(!strncmp(c, "joint", max(int(args-c), 5)))
{
ejoint &j = ejoints.add();
j.name = getnamekey(trimname(args));
parseindex(args, j.parent);
}
else if(!strncmp(c, "vertexarray", max(int(args-c), 11)))
{
evarray &va = evarrays.add();
va.type = findvertexarraytype(trimname(args));
va.format = findvertexarrayformat(trimname(args));
va.size = strtol(args, &args, 10);
copystring(va.name, trimname(args));
}
else if(!strncmp(c, "animation", max(int(args-c), 9)))
{
eanim &a = eanims.add();
a.name = getnamekey(trimname(args));
a.startframe = eframes.length();
if(!eframes.length() || eframes.last() != eposes.length()) eframes.add(eposes.length());
}
else if(!strncmp(c, "frame", max(int(args-c), 5)))
{
if(eanims.length() && eframes.length() && eframes.last() != eposes.length()) eframes.add(eposes.length());
}
else if(!strncmp(c, "framerate", max(int(args-c), 9)))
{
if(eanims.length())
{
double fps = parseattrib(args);
eanims.last().fps = max(fps, 0.0);
}
}
else if(!strncmp(c, "loop", max(int(args-c), 4)))
{
if(eanims.length()) eanims.last().flags |= IQM_LOOP;
}
else if(!strncmp(c, "comment", max(int(args-c), 7)))
{
if(commentdata.length()) break;
for(;;)
{
size_t len = f->read(commentdata.reserve(1024), 1024);
commentdata.advance(len);
if(len < 1024) { commentdata.add('\0'); break; }
}
}
}
return true;
}
bool loadiqe(const char *filename, const filespec &spec)
{
int numfiles = 0;
while(filename)
{
const char *endfile = strchr(filename, ',');
const char *file = endfile ? newstring(filename, endfile-filename) : filename;
stream *f = openfile(file, "r");
if(f)
{
resetimporter(spec, numfiles > 0);
if(parseiqe(f)) numfiles++;
delete f;
}
if(!endfile) break;
delete[] file;
filename = endfile+1;
}
if(!numfiles) return false;
if(eanims.length() == 1)
{
eanim &a = eanims.last();
if(spec.name) a.name = spec.name;
if(spec.fps > 0) a.fps = spec.fps;
a.flags |= spec.flags;
if(spec.endframe >= 0) a.endframe = a.startframe + spec.endframe;
else if(spec.endframe < -1) a.endframe = a.startframe + max(eframes.length() - a.startframe + spec.endframe + 1, 0);
a.startframe += spec.startframe;
}
makeanims(spec);
if(emeshes.length())
{
smoothverts();
makemeshes(spec);
}
return true;
}
struct md5weight
{
int joint;
double bias;
Vec3 pos;
};
struct md5vert
{
double u, v;
uint start, count;
};
struct md5hierarchy
{
const char *name;
int parent, flags, start;
};
vector<md5weight> weightinfo;
vector<md5vert> vertinfo;
void buildmd5verts()
{
loopv(vertinfo)
{
md5vert &v = vertinfo[i];
Vec3 pos(0, 0, 0);
loopk(v.count)
{
md5weight &w = weightinfo[v.start+k];
transform &j = eposes[w.joint];
pos += (j.orient.transform(w.pos) + j.pos)*w.bias;
}
epositions.add(Vec4(pos, 1));
etexcoords.add(Vec4(v.u, v.v, 0, 0));
blendcombo &c = eblends.add();
loopj(v.count)
{
md5weight &w = weightinfo[v.start+j];
c.addweight(w.bias, w.joint);
}
c.finalize();
}
}
void parsemd5mesh(stream *f, char *buf, size_t bufsize)
{
md5weight w;
md5vert v;
etriangle t(0, 0, 0, 0);
int index, firsttri = etriangles.length(), firstvert = vertinfo.length(), firstweight = weightinfo.length(), numtris = 0, numverts = 0, numweights = 0;
emesh m;
while(f->getline(buf, bufsize) && buf[0]!='}')
{
if(strstr(buf, "// meshes:"))
{
char *start = strchr(buf, ':')+1;
if(*start==' ') start++;
char *end = start + strlen(start)-1;
while(end >= start && isspace(*end)) end--;
end[1] = '\0';
m.name = getnamekey(start);
}
else if(strstr(buf, "shader"))
{
char *start = strchr(buf, '"'), *end = start ? strchr(start+1, '"') : NULL;
if(start && end)
{
*end = '\0';
m.material = getnamekey(start+1);
}
}
else if(sscanf(buf, " numverts %d", &numverts)==1)
{
numverts = max(numverts, 0);
if(numverts)
{
vertinfo.reserve(numverts);
vertinfo.advance(numverts);
}
}
else if(sscanf(buf, " numtris %d", &numtris)==1)
{
numtris = max(numtris, 0);
if(numtris)
{
etriangles.reserve(numtris);
etriangles.advance(numtris);
}
m.firsttri = firsttri;
}
else if(sscanf(buf, " numweights %d", &numweights)==1)
{
numweights = max(numweights, 0);
if(numweights)
{
weightinfo.reserve(numweights);
weightinfo.advance(numweights);
}
}
else if(sscanf(buf, " vert %d ( %lf %lf ) %u %u", &index, &v.u, &v.v, &v.start, &v.count)==5)
{
if(index>=0 && index<numverts)
{
v.start += firstweight;
vertinfo[firstvert + index] = v;
}
}
else if(sscanf(buf, " tri %d %u %u %u", &index, &t.vert[0], &t.vert[1], &t.vert[2])==4)
{
if(index>=0 && index<numtris)
{
loopk(3) t.vert[k] += firstvert;
etriangles[firsttri + index] = t;
}
}
else if(sscanf(buf, " weight %d %d %lf ( %lf %lf %lf ) ", &index, &w.joint, &w.bias, &w.pos.x, &w.pos.y, &w.pos.z)==6)
{
if(index>=0 && index<numweights) weightinfo[firstweight + index] = w;
}
}
if(numtris && numverts) emeshes.add(m);
}
bool loadmd5mesh(const char *filename, const filespec &spec)
{
stream *f = openfile(filename, "r");
if(!f) return false;
resetimporter(spec);
esmoothgroups[0].flags |= esmoothgroup::F_UVSMOOTH;
char buf[512];
while(f->getline(buf, sizeof(buf)))
{
int tmp;
if(sscanf(buf, " MD5Version %d", &tmp)==1)
{
if(tmp!=10) { delete f; return false; }
}
else if(sscanf(buf, " numJoints %d", &tmp)==1)
{
if(tmp<1 || (joints.length() && tmp != joints.length())) { delete f; return false; }
}
else if(sscanf(buf, " numMeshes %d", &tmp)==1)
{
if(tmp<1) { delete f; return false; }
}
else if(strstr(buf, "joints {"))
{
ejoint j;
transform p;
while(f->getline(buf, sizeof(buf)) && buf[0]!='}')
{
char *c = buf;
j.name = getnamekey(trimname(c));
if(sscanf(c, " %d ( %lf %lf %lf ) ( %lf %lf %lf )",
&j.parent, &p.pos.x, &p.pos.y, &p.pos.z,
&p.orient.x, &p.orient.y, &p.orient.z)==7)
{
p.orient.restorew();
p.scale = Vec3(1, 1, 1);
ejoints.add(j);
eposes.add(p);
}
}
}
else if(strstr(buf, "mesh {"))
{
parsemd5mesh(f, buf, sizeof(buf));
}
}
delete f;
buildmd5verts();
makeanims(spec);
smoothverts();
makemeshes(spec);
makerelativebasepose();
return true;
}
bool loadmd5anim(const char *filename, const filespec &spec)
{
stream *f = openfile(filename, "r");
if(!f) return false;
resetimporter(spec);
vector<md5hierarchy> hierarchy;
vector<transform> baseframe;
int animdatalen = 0, animframes = 0, frameoffset = eposes.length(), firstframe = eframes.length();
double framerate = 0;
double *animdata = NULL;
char buf[512];
while(f->getline(buf, sizeof(buf)))
{
int tmp;
if(sscanf(buf, " MD5Version %d", &tmp)==1)
{
if(tmp!=10) { delete f; return false; }
}
else if(sscanf(buf, " numJoints %d", &tmp)==1)
{
if(tmp<1) { delete f; return false; }
}
else if(sscanf(buf, " numFrames %d", &animframes)==1)
{
if(animframes<1) { delete f; return false; }
}
else if(sscanf(buf, " frameRate %lf", &framerate)==1);
else if(sscanf(buf, " numAnimatedComponents %d", &animdatalen)==1)
{
if(animdatalen>0) animdata = new double[animdatalen];
}
else if(strstr(buf, "bounds {"))
{
while(f->getline(buf, sizeof(buf)) && buf[0]!='}');
}
else if(strstr(buf, "hierarchy {"))
{
while(f->getline(buf, sizeof(buf)) && buf[0]!='}')
{
char *c = buf;
md5hierarchy h;
h.name = getnamekey(trimname(c));
if(sscanf(c, " %d %d %d", &h.parent, &h.flags, &h.start)==3)
hierarchy.add(h);
}
if(hierarchy.empty()) { delete f; return false; }
loopv(hierarchy)
{
md5hierarchy &h = hierarchy[i];
ejoint &j = ejoints.add();
j.name = h.name;
j.parent = h.parent;
}
}
else if(strstr(buf, "baseframe {"))
{
while(f->getline(buf, sizeof(buf)) && buf[0]!='}')
{
transform j;
if(sscanf(buf, " ( %lf %lf %lf ) ( %lf %lf %lf )", &j.pos.x, &j.pos.y, &j.pos.z, &j.orient.x, &j.orient.y, &j.orient.z)==6)
{
j.orient.restorew();
j.scale = Vec3(1, 1, 1);
baseframe.add(j);
}
}
if(baseframe.length()!=hierarchy.length()) { delete f; return false; }
eposes.reserve(animframes*baseframe.length());
eposes.advance(animframes*baseframe.length());
}
else if(sscanf(buf, " frame %d", &tmp)==1)
{
for(int numdata = 0; f->getline(buf, sizeof(buf)) && buf[0]!='}';)
{
for(char *src = buf, *next = src; numdata < animdatalen; numdata++, src = next)
{
animdata[numdata] = strtod(src, &next);
if(next <= src) break;
}
}
int offset = frameoffset + tmp*baseframe.length();
eframes.add(offset);
loopv(baseframe)
{
md5hierarchy &h = hierarchy[i];
transform j = baseframe[i];
if(h.start < animdatalen && h.flags)
{
double *jdata = &animdata[h.start];
if(h.flags&1) j.pos.x = *jdata++;
if(h.flags&2) j.pos.y = *jdata++;
if(h.flags&4) j.pos.z = *jdata++;
if(h.flags&8) j.orient.x = *jdata++;
if(h.flags&16) j.orient.y = *jdata++;
if(h.flags&32) j.orient.z = *jdata++;
j.orient.restorew();
}
eposes[offset + i] = j;
}
}
}
if(animdata) delete[] animdata;
delete f;
eanim &a = eanims.add();
if(spec.name) a.name = getnamekey(spec.name);
else
{
string name;
copystring(name, filename);
char *end = strrchr(name, '.');
if(end) *end = '\0';
a.name = getnamekey(name);
}
a.startframe = firstframe;
a.fps = spec.fps > 0 ? spec.fps : framerate;
a.flags = spec.flags;
if(spec.endframe >= 0) a.endframe = a.startframe + spec.endframe;
else if(spec.endframe < -1) a.endframe = a.startframe + max(eframes.length() - a.startframe + spec.endframe + 1, 0);
a.startframe += spec.startframe;
makeanims(spec);
return true;
}
namespace smd
{
bool skipcomment(char *&curbuf)
{
while(*curbuf && isspace(*curbuf)) curbuf++;
switch(*curbuf)
{
case '#':
case ';':
case '\r':
case '\n':
case '\0':
return true;
case '/':
if(curbuf[1] == '/') return true;
break;
}
return false;
}
void skipsection(stream *f, char *buf, size_t bufsize)
{
while(f->getline(buf, bufsize))
{
char *curbuf = buf;
if(skipcomment(curbuf)) continue;
if(!strncmp(curbuf, "end", 3)) break;
}
}
void readname(char *&curbuf, char *name, size_t namesize)
{
char *curname = name;
while(*curbuf && isspace(*curbuf)) curbuf++;
bool allowspace = false;
if(*curbuf == '"') { curbuf++; allowspace = true; }
while(*curbuf)
{
char c = *curbuf++;
if(c == '"') break;
if(isspace(c) && !allowspace) break;
if(curname < &name[namesize-1]) *curname++ = c;
}
*curname = '\0';
}
void readnodes(stream *f, char *buf, size_t bufsize)
{
while(f->getline(buf, bufsize))
{
char *curbuf = buf;
if(skipcomment(curbuf)) continue;
if(!strncmp(curbuf, "end", 3)) break;
int id = strtol(curbuf, &curbuf, 10);
string name;
readname(curbuf, name, sizeof(name));
int parent = strtol(curbuf, &curbuf, 10);
if(id < 0 || id > 255 || parent > 255 || !name[0] || (ejoints.inrange(id) && ejoints[id].name)) continue;
ejoint j;
j.name = getnamekey(name);
j.parent = parent;
while(ejoints.length() <= id) ejoints.add();
ejoints[id] = j;
}
}
void readmaterial(char *&curbuf, char *mat, char *name, size_t matsize)
{
char *curmat = mat;
while(*curbuf && isspace(*curbuf)) curbuf++;
char *ext = NULL;
while(*curbuf)
{
char c = *curbuf++;
if(isspace(c)) break;
if(c == '.' && !ext) ext = curmat;
if(curmat < &mat[matsize-1]) *curmat++ = c;
}
*curmat = '\0';
if(!ext) ext = curmat;
memcpy(name, mat, ext - mat);
name[ext - mat] = '\0';
}
void readskeleton(stream *f, char *buf, size_t bufsize)
{
int frame = -1, firstpose = -1;
while(f->getline(buf, bufsize))
{
char *curbuf = buf;
if(skipcomment(curbuf)) continue;
if(sscanf(curbuf, " time %d", &frame) == 1) continue;
else if(!strncmp(curbuf, "end", 3)) break;
else if(frame != 0) continue;
int bone;
Vec3 pos, rot;
if(sscanf(curbuf, " %d %lf %lf %lf %lf %lf %lf", &bone, &pos.x, &pos.y, &pos.z, &rot.x, &rot.y, &rot.z) != 7)
continue;
if(!ejoints.inrange(bone))
continue;
if(firstpose < 0)
{
firstpose = eposes.length();
eposes.reserve(ejoints.length());
eposes.advance(ejoints.length());
}
transform p(pos, Quat::fromangles(rot));
eposes[firstpose + bone] = p;
}
}
void readtriangles(stream *f, char *buf, size_t bufsize)
{
emesh m;
while(f->getline(buf, bufsize))
{
char *curbuf = buf;
if(skipcomment(curbuf)) continue;
if(!strncmp(curbuf, "end", 3)) break;
string name, material;
readmaterial(curbuf, material, name, sizeof(material));
if(!m.name || strcmp(m.name, name))
{
if(m.name && etriangles.length() > m.firsttri) emeshes.add(m);
m.name = getnamekey(name);
m.material = getnamekey(material);
m.firsttri = etriangles.length();
}
Vec4 *pos = epositions.reserve(3) + 2, *tc = etexcoords.reserve(3) + 2;
Vec3 *norm = enormals.reserve(3) + 2;
blendcombo *c = eblends.reserve(3) + 2;
loopi(3)
{
char *curbuf;
do
{
if(!f->getline(buf, bufsize)) goto endsection;
curbuf = buf;
} while(skipcomment(curbuf));
int parent = -1, numlinks = 0, len = 0;
if(sscanf(curbuf, " %d %lf %lf %lf %lf %lf %lf %lf %lf %d%n", &parent, &pos->x, &pos->y, &pos->z, &norm->x, &norm->y, &norm->z, &tc->x, &tc->y, &numlinks, &len) < 9) goto endsection;
curbuf += len;
pos->w = 1;
tc->y = 1 - tc->y;
tc->z = tc->w = 0;
c->reset();
double pweight = 0, tweight = 0;
for(; numlinks > 0; numlinks--)
{
int bone = -1, len = 0;
double weight = 0;
if(sscanf(curbuf, " %d %lf%n", &bone, &weight, &len) < 2) break;
curbuf += len;
tweight += weight;
if(bone == parent) pweight += weight;
else c->addweight(weight, bone);
}
if(tweight < 1) pweight += 1 - tweight;
if(pweight > 0) c->addweight(pweight, parent);
c->finalize();
--pos;
--tc;
--norm;
--c;
}
etriangle &t = etriangles.add();
loopi(3) t.vert[i] = epositions.length() + i;
t.smoothgroup = 0;
epositions.advance(3);
enormals.advance(3);
etexcoords.advance(3);
eblends.advance(3);
}
endsection:
if(m.name && etriangles.length () > m.firsttri) emeshes.add(m);
}
int readframes(stream *f, char *buf, size_t bufsize)
{
int frame = -1, numframes = 0, lastbone = ejoints.length(), frameoffset = eposes.length();
while(f->getline(buf, bufsize))
{
char *curbuf = buf;
if(skipcomment(curbuf)) continue;
int nextframe = -1;
if(sscanf(curbuf, " time %d", &nextframe) == 1)
{
for(; lastbone < ejoints.length(); lastbone++) eposes[frameoffset + frame*ejoints.length() + lastbone] = eposes[frameoffset + lastbone];
if(nextframe >= numframes)
{
eposes.reserve(ejoints.length() * (nextframe + 1 - numframes));
loopi(nextframe - numframes)
{
eframes.add(eposes.length());
eposes.put(&eposes[frameoffset], ejoints.length());
}
eframes.add(eposes.length());
eposes.advance(ejoints.length());
numframes = nextframe + 1;
}
frame = nextframe;
lastbone = 0;
continue;
}
else if(!strncmp(curbuf, "end", 3)) break;
int bone;
Vec3 pos, rot;
if(sscanf(curbuf, " %d %lf %lf %lf %lf %lf %lf", &bone, &pos.x, &pos.y, &pos.z, &rot.x, &rot.y, &rot.z) != 7)
continue;
if(bone < 0 || bone >= ejoints.length())
continue;
for(; lastbone < bone; lastbone++) eposes[frameoffset + frame*ejoints.length() + lastbone] = eposes[frameoffset + lastbone];
lastbone++;
transform p(pos, Quat::fromangles(rot));
eposes[frameoffset + frame*ejoints.length() + bone] = p;
}
for(; lastbone < ejoints.length(); lastbone++) eposes[frameoffset + frame*ejoints.length() + lastbone] = eposes[frameoffset + lastbone];
return numframes;
}
}
bool loadsmd(const char *filename, const filespec &spec)
{
stream *f = openfile(filename, "r");
if(!f) return false;
resetimporter(spec);
char buf[512];
int version = -1, firstframe = eframes.length();
bool hastriangles = false;
while(f->getline(buf, sizeof(buf)))
{
char *curbuf = buf;
if(smd::skipcomment(curbuf)) continue;
if(sscanf(curbuf, " version %d", &version) == 1)
{
if(version != 1) { delete f; return false; }
}
else if(!strncmp(curbuf, "nodes", 5))
smd::readnodes(f, buf, sizeof(buf));
else if(!strncmp(curbuf, "triangles", 9))
{
smd::readtriangles(f, buf, sizeof(buf));
hastriangles = true;
}
else if(!strncmp(curbuf, "skeleton", 8))
smd::readframes(f, buf, sizeof(buf));
else if(!strncmp(curbuf, "vertexanimation", 15))
smd::skipsection(f, buf, sizeof(buf));
}
delete f;
if(hastriangles)
{
eframes.setsize(firstframe);
makeanims(spec);
smoothverts();
makemeshes(spec);
}
else
{
eanim &a = eanims.add();
if(spec.name) a.name = getnamekey(spec.name);
else
{
string name;
const char *shortname = filename;
shortname = strrchr(filename, '/');
if (shortname)
shortname++;
else
shortname = filename;
copystring(name, shortname);
char *end = strrchr(name, '.');
if(end) *end = '\0';
a.name = getnamekey(name);
}
a.startframe = firstframe;
a.fps = spec.fps;
a.flags = spec.flags;
if(spec.endframe >= 0) a.endframe = a.startframe + spec.endframe;
else if(spec.endframe < -1) a.endframe = a.startframe + max(eframes.length() - a.startframe + spec.endframe + 1, 0);
a.startframe += spec.startframe;
makeanims(spec);
}
return true;
}
struct objvert { int attrib[3]; objvert() { attrib[0] = attrib[1] = attrib[2] = -1; } };
static inline uint hthash(const objvert &k) { return k.attrib[0] ^ k.attrib[1] ^ k.attrib[2]; };
static inline bool htcmp(const objvert &x, const objvert &y) { return x.attrib[0] == y.attrib[0] && x.attrib[1] == y.attrib[1] && x.attrib[2] == y.attrib[2]; }
void parseobjvert(char *s, vector<Vec3> &out)
{
Vec3 &v = out.add(Vec3(0, 0, 0));
while(isalpha(*s)) s++;
loopi(3)
{
v[i] = strtod(s, &s);
while(isspace(*s)) s++;
if(!*s) break;
}
}
bool parseobj(stream *f)
{
vector<Vec3> attrib[3];
char buf[512];
hashtable<objvert, int> verthash;
string meshname = "", matname = "";
int curmesh = -1, smooth = 0;
while(f->getline(buf, sizeof(buf)))
{
char *c = buf;
while(isspace(*c)) c++;
switch(*c)
{
case '#': continue;
case 'v':
if(isspace(c[1])) parseobjvert(c, attrib[0]);
else if(c[1]=='t') parseobjvert(c, attrib[1]);
else if(c[1]=='n') parseobjvert(c, attrib[2]);
break;
case 'g':
{
while(isalpha(*c)) c++;
while(isspace(*c)) c++;
char *name = c;
size_t namelen = strlen(name);
while(namelen > 0 && isspace(name[namelen-1])) namelen--;
copystring(meshname, name, min(namelen+1, sizeof(meshname)));
curmesh = -1;
break;
}
case 'u':
{
if(!strncmp(c, "usemtl", 6)) continue;
while(isalpha(*c)) c++;
while(isspace(*c)) c++;
char *name = c;
size_t namelen = strlen(name);
while(namelen > 0 && isspace(name[namelen-1])) namelen--;
copystring(matname, name, min(namelen+1, sizeof(matname)));
curmesh = -1;
break;
}
case 's':
{
if(!isspace(c[1])) continue;
while(isalpha(*c)) c++;
while(isspace(*c)) c++;
int key = strtol(c, &c, 10);
smooth = -1;
loopv(esmoothgroups) if(esmoothgroups[i].key == key) { smooth = i; break; }
if(smooth < 0)
{
smooth = esmoothgroups.length();
esmoothgroups.add().key = key;
}
break;
}
case 'f':
{
if(curmesh < 0)
{
emesh m;
m.name = getnamekey(meshname);
m.material = getnamekey(matname);
m.firsttri = etriangles.length();
curmesh = emeshes.length();
emeshes.add(m);
verthash.clear();
}
int v0 = -1, v1 = -1;
while(isalpha(*c)) c++;
for(;;)
{
while(isspace(*c)) c++;
if(!*c) break;
objvert vkey;
loopi(3)
{
vkey.attrib[i] = strtol(c, &c, 10);
if(vkey.attrib[i] < 0) vkey.attrib[i] = attrib[i].length() + vkey.attrib[i];
else vkey.attrib[i]--;
if(!attrib[i].inrange(vkey.attrib[i])) vkey.attrib[i] = -1;
if(*c!='/') break;
c++;
}
int *index = verthash.access(vkey);
if(!index)
{
index = &verthash[vkey];
*index = epositions.length();
epositions.add(Vec4(vkey.attrib[0] < 0 ? Vec3(0, 0, 0) : attrib[0][vkey.attrib[0]].zxy(), 1));
if(vkey.attrib[2] >= 0) enormals.add(attrib[2][vkey.attrib[2]].zxy());
etexcoords.add(vkey.attrib[1] < 0 ? Vec4(0, 0, 0, 0) : Vec4(attrib[1][vkey.attrib[1]].x, 1-attrib[1][vkey.attrib[1]].y, 0, 0));
}
if(v0 < 0) v0 = *index;
else if(v1 < 0) v1 = *index;
else
{
etriangles.add(etriangle(*index, v1, v0, smooth));
v1 = *index;
}
}
break;
}
}
}
return true;
}
bool loadobj(const char *filename, const filespec &spec)
{
stream *f = openfile(filename, "r");
if(!f) return false;
int numfiles = 0;
while(filename)
{
const char *endfile = strchr(filename, ',');
const char *file = endfile ? newstring(filename, endfile-filename) : filename;
stream *f = openfile(file, "r");
if(f)
{
if(resetimporter(spec, numfiles > 0))
{
esmoothgroups[0].key = 0;
}
if(parseobj(f)) numfiles++;
delete f;
}
if(!endfile) break;
delete[] file;
filename = endfile+1;
}
if(!numfiles) return false;
smoothverts();
makemeshes(spec);
return true;
}
namespace fbx
{
struct token
{
enum { NONE, PROP, NUMBER, STRING, ARRAY, BEGIN, END, LINE };
int type;
union
{
char s[64];
double f;
int i;
};
token() : type(NONE) {}
};
struct tokenizer
{
stream *f;
char *pos;
char buf[4096];
void reset(stream *s) { f = s; pos = buf; buf[0] = '\0'; }
bool parse(token &t)
{
for(;;)
{
while(isspace(*pos)) pos++;
if(!*pos)
{
bool more = f->getline(buf, sizeof(buf));
pos = buf;
if(!more) { buf[0] = '\0'; return false; }
t.type = token::LINE;
return true;
}
size_t slen = 0;
switch(*pos)
{
case ',':
pos++;
continue;
case ';':
pos++;
while(*pos) pos++;
continue;
case '{':
pos++;
t.type = token::BEGIN;
return true;
case '}':
pos++;
t.type = token::END;
return true;
case '"':
pos++;
for(; *pos && *pos != '"'; pos++) if(slen < sizeof(t.s)-1) t.s[slen++] = *pos;
t.s[slen] = '\0';
if(*pos == '"') pos++;
t.type = token::STRING;
return true;
case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': case '.': case '-': case '+':
t.f = strtod(pos, &pos);
t.type = token::NUMBER;
return true;
case '*':
pos++;
t.i = int(strtol(pos, &pos, 10));
t.type = token::ARRAY;
return true;
default:
for(; *pos && !isspace(*pos) && *pos != ':'; pos++) if(slen < sizeof(t.s)-1) t.s[slen++] = *pos;
t.s[slen] = '\0';
if(*pos == ':') pos++;
t.type = token::PROP;
return true;
}
}
return false;
}
bool skipprop()
{
token t;
while(parse(t)) switch(t.type)
{
case token::LINE:
return true;
case token::BEGIN:
while(parse(t)) switch(t.type)
{
case token::PROP:
skipprop();
break;
case token::END:
return true;
}
return true;
case token::END:
return false;
}
return false;
}
bool findbegin()
{
token t;
while(parse(t)) switch(t.type)
{
case token::LINE: return false;
case token::BEGIN: return true;
}
return false;
}
template<class T>
bool readarray(vector<T> &vals, int size = 0)
{
if(!findbegin()) return false;
if(size > 0) vals.reserve(min(size, 1<<16));
token t;
while(parse(t)) switch(t.type)
{
case token::NUMBER: if(size <= 0 || vals.length() < size) vals.add(T(t.f)); break;
case token::END: return true;
}
return false;
}
};
struct node
{
enum { GEOM = 0, MODEL, MATERIAL, LIMB, CLUSTER, SKIN, CURVE, XFORM, ANIMLAYER, ANIMSTACK };
enum { TRANS = 0, ROT, SCALE };
virtual int type() = 0;
virtual ~node() {}
virtual void process() {}
virtual void finish() {}
};
struct namednode : node
{
string name;
namednode() { name[0] = 0; }
};
struct geomnode;
struct modelnode;
struct materialnode;
struct limbnode;
struct clusternode;
struct skinnode;
struct curvenode;
struct xformnode;
struct animlayernode;
struct animstacknode;
struct geomnode : node
{
int mesh, firstvert, lastvert, numverts;
modelnode *model;
vector<int> remap;
vector<blendcombo> blends;
geomnode() : mesh(-1), firstvert(-1), lastvert(-1), numverts(0), model(NULL) {}
int type() { return GEOM; }
void process();
void finish();
};
struct modelnode : namednode
{
materialnode *material;
Vec3 geomtrans, prerot, lcltrans, lclrot, lclscale;
modelnode() : material(NULL), geomtrans(0, 0, 0), prerot(0, 0, 0), lcltrans(0, 0, 0), lclrot(0, 0, 0), lclscale(1, 1, 1) {}
int type() { return MODEL; }
};
struct materialnode : namednode
{
int type() { return MATERIAL; }
};
struct limbnode : namednode
{
limbnode *parent;
int index;
Vec3 trans, rot, prerot, scale;
clusternode *cluster;
limbnode() : parent(NULL), index(-1), trans(0, 0, 0), rot(0, 0, 0), prerot(0, 0, 0), scale(1, 1, 1), cluster(NULL) {}
int type() { return LIMB; }
void process()
{
if(parent) ejoints[index].parent = parent->index;
}
void finish();
};
struct clusternode : node
{
skinnode *skin;
limbnode *limb;
vector<int> indexes;
vector<double> weights, transform, transformlink;
clusternode() : skin(NULL), limb(NULL) {}
int type() { return CLUSTER; }
void process();
};
struct skinnode : node
{
geomnode *geom;
skinnode() : geom(NULL) {}
int type() { return SKIN; }
};
struct curvenode : node
{
vector<double> vals;
int type() { return CURVE; }
bool varies() const
{
loopv(vals) if(vals[i] != vals[0]) return true;
return false;
}
};
struct xformnode : node
{
limbnode *limb;
int xform;
Vec3 val;
curvenode *curves[3];
xformnode() : limb(NULL), xform(-1), val(0, 0, 0) { curves[0] = curves[1] = curves[2] = NULL; }
void setcurve(int i, curvenode *c)
{
if(c->varies()) curves[i] = c;
else if(c->vals.length()) val[i] = c->vals[0];
}
int numframes()
{
int n = 0;
loopi(3) if(curves[i]) { if(!n) n = curves[i]->vals.length(); else if(n != curves[i]->vals.length()) n = -1; }
return n;
}
int type() { return XFORM; }
};
struct animlayernode : namednode
{
vector<xformnode *> xforms;
int numframes()
{
int n = 0;
loopv(xforms)
{
int xn = xforms[i]->numframes();
if(xn) { if(!n) n = xn; else if(n != xn) n = -1; }
}
return n;
}
int type() { return ANIMLAYER; }
};
struct animstacknode : namednode
{
vector<animlayernode *> layers;
double secs;
animstacknode() : secs(0) {}
int numframes()
{
int n;
loopv(layers)
{
int ln = layers[i]->numframes();
if(ln) { if(!n) n = ln; else if(n != ln) n = -1; }
}
return n;
}
int type() { return ANIMSTACK; }
void process();
};
hashtable<double, node *> nodes;
tokenizer p;
void parsegeometry()
{
token t;
if(!p.parse(t)) return;
if(t.type != token::NUMBER) { p.skipprop(); return; }
double id = t.f;
if(!p.findbegin()) return;
vector<double> verts, norms, uvs, colors;
vector<int> polyidxs, uvidxs, coloridxs;
while(p.parse(t)) switch(t.type)
{
case token::END:
goto endgeometry;
case token::PROP:
if(!strcmp(t.s, "Vertices")) p.readarray(verts);
else if(!strcmp(t.s, "PolygonVertexIndex")) p.readarray(polyidxs);
else if(!strcmp(t.s, "LayerElementNormal"))
{
if(p.findbegin())
{
while(p.parse(t)) switch(t.type)
{
case token::PROP:
if(!strcmp(t.s, "Normals")) p.readarray(norms);
else p.skipprop();
break;
case token::END:
goto endnormals;
}
endnormals:;
}
}
else if(!strcmp(t.s, "LayerElementUV"))
{
if(p.findbegin())
{
while(p.parse(t)) switch(t.type)
{
case token::PROP:
if(!strcmp(t.s, "UV")) p.readarray(uvs);
else if(!strcmp(t.s, "UVIndex")) p.readarray(uvidxs);
else p.skipprop();
break;
case token::END:
goto enduvs;
}
enduvs:;
}
}
else if(!strcmp(t.s, "LayerElementColor"))
{
if(p.findbegin())
{
while(p.parse(t)) switch(t.type)
{
case token::PROP:
if(!strcmp(t.s, "Colors")) p.readarray(colors);
else if(!strcmp(t.s, "ColorIndex")) p.readarray(coloridxs);
else p.skipprop();
break;
case token::END:
goto endcolors;
}
endcolors:;
}
}
else p.skipprop();
break;
}
endgeometry:
int poslen = epositions.length();
geomnode *n = new geomnode;
nodes[id] = n;
if(polyidxs.empty()) for(int i = 0; i + 2 < verts.length(); i += 3) epositions.add(Vec4(verts[i], verts[i+1], verts[i+2], 1));
else
{
loopv(polyidxs)
{
int idx = polyidxs[i];
if(idx < 0) idx = -(idx+1);
n->remap.add(idx);
idx *= 3;
epositions.add(Vec4(verts[idx], verts[idx+1], verts[idx+2], 1));
}
}
loopi(epositions.length() - poslen) eblends.add();
emesh m;
m.name = getnamekey("");
m.material = getnamekey("");
m.firsttri = etriangles.length();
for(int i = poslen; i + 2 < epositions.length(); i += 3)
etriangles.add(etriangle(i+1, i, i+2));
emeshes.add(m);
n->mesh = emeshes.length()-1;
n->firstvert = poslen;
n->lastvert = epositions.length();
n->numverts = verts.length()/3;
if(uvidxs.empty())
{
if(polyidxs.length() && uvs.length()/2 == verts.length()/3) loopv(polyidxs)
{
int idx = polyidxs[i];
if(idx < 0) idx = -(idx+1);
idx *= 2;
etexcoords.add(Vec4(uvs[idx], 1-uvs[idx+1], 0, 0));
}
else for(int i = 0; i + 1 < uvs.length(); i += 2) etexcoords.add(Vec4(uvs[i], 1-uvs[i+1], 0, 0));
}
else loopv(uvidxs)
{
int idx = 2*uvidxs[i];
etexcoords.add(Vec4(uvs[idx], 1-uvs[idx+1], 0, 0));
}
if(polyidxs.length() && norms.length() == verts.length()) loopv(polyidxs)
{
int idx = polyidxs[i];
if(idx < 0) idx = -(idx+1);
idx *= 3;
enormals.add(Vec3(norms[idx], norms[idx+1], norms[idx+2]));
}
else for(int i = 0; i + 2 < norms.length(); i += 3) enormals.add(Vec3(norms[i], norms[i+1], norms[i+2]));
if(coloridxs.empty())
{
if(polyidxs.length() && colors.length()/4 == verts.length()/3) loopv(polyidxs)
{
int idx = polyidxs[i];
if(idx < 0) idx = -(idx+1);
idx *= 4;
ecolors.add(Vec4(colors[idx], colors[idx+1], colors[idx+2], colors[idx+3]));
}
else for(int i = 0; i + 3 < colors.length(); i += 4) ecolors.add(Vec4(colors[i], colors[i+1], colors[i+2], colors[i+3]));
}
else loopv(coloridxs)
{
int idx = 4*coloridxs[i];
ecolors.add(Vec4(colors[idx], colors[idx+1], colors[idx+2], colors[idx+3]));
}
}
void parsemodel()
{
token id, name, type, t;
if(!p.parse(id) || !p.parse(name) || !p.parse(type)) return;
if(id.type != token::NUMBER || type.type != token::STRING || name.type != token::STRING) { p.skipprop(); return; }
char *str = name.s;
if(strstr(str, "Model::") == str) str += strlen("Model::");
if(!strcmp(type.s, "Mesh"))
{
modelnode *n = new modelnode;
copystring(n->name, str);
nodes[id.f] = n;
if(!p.findbegin()) return;
while(p.parse(t)) switch(t.type)
{
case token::END: return;
case token::PROP:
if(!strcmp(t.s, "Properties70"))
{
if(!p.findbegin()) return;
while(p.parse(t)) switch(t.type)
{
case token::END: goto endmeshprops;
case token::PROP:
if(!strcmp(t.s, "P"))
{
if(!p.parse(t)) return;
if(t.type == token::STRING)
{
if(!strcmp(t.s, "PreRotation"))
{
loopi(3) if(!p.parse(t)) return;
loopi(3) { if(!p.parse(t)) return; if(t.type != token::NUMBER) break; n->prerot[i] = t.f; }
}
else if(!strcmp(t.s, "GeometricTranslation"))
{
loopi(3) if(!p.parse(t)) return;
loopi(3) { if(!p.parse(t)) return; if(t.type != token::NUMBER) break; n->geomtrans[i] = t.f; }
}
else if(!strcmp(t.s, "Lcl Translation"))
{
loopi(3) if(!p.parse(t)) return;
loopi(3) { if(!p.parse(t)) return; if(t.type != token::NUMBER) break; n->lcltrans[i] = t.f; }
}
else if(!strcmp(t.s, "Lcl Rotation"))
{
loopi(3) if(!p.parse(t)) return;
loopi(3) { if(!p.parse(t)) return; if(t.type != token::NUMBER) break; n->lclrot[i] = t.f; }
}
else if(!strcmp(t.s, "Lcl Scaling"))
{
loopi(3) if(!p.parse(t)) return;
loopi(3) { if(!p.parse(t)) return; if(t.type != token::NUMBER) break; n->lclscale[i] = t.f; }
}
}
}
p.skipprop();
break;
}
endmeshprops:;
}
p.skipprop();
break;
}
}
else if(!strcmp(type.s, "LimbNode"))
{
limbnode *n = new limbnode;
copystring(n->name, str);
n->index = ejoints.length();
nodes[id.f] = n;
ejoint &j = ejoints.add();
j.name = getnamekey(str);
j.parent = -1;
eposes.add(transform(Vec3(0, 0, 0), Quat(0, 0, 0, 1)));
if(!p.findbegin()) return;
while(p.parse(t)) switch(t.type)
{
case token::END:
{
transform &x = eposes[n->index];
x.pos = n->trans;
x.orient = Quat::fromdegrees(n->rot);
x.scale = n->scale;
}
return;
case token::PROP:
if(!strcmp(t.s, "Properties70"))
{
if(!p.findbegin()) return;
while(p.parse(t)) switch(t.type)
{
case token::END: goto endlimbprops;
case token::PROP:
if(!strcmp(t.s, "P"))
{
if(!p.parse(t)) return;
if(t.type == token::STRING)
{
if(!strcmp(t.s, "PreRotation"))
{
loopi(3) if(!p.parse(t)) return;
loopi(3) { if(!p.parse(t)) return; if(t.type != token::NUMBER) break; n->prerot[i] = t.f; }
}
else if(!strcmp(t.s, "Lcl Translation"))
{
loopi(3) if(!p.parse(t)) return;
loopi(3) { if(!p.parse(t)) return; if(t.type != token::NUMBER) break; n->trans[i] = t.f; }
}
else if(!strcmp(t.s, "Lcl Rotation"))
{
loopi(3) if(!p.parse(t)) return;
loopi(3) { if(!p.parse(t)) return; if(t.type != token::NUMBER) break; n->rot[i] = t.f; }
}
else if(!strcmp(t.s, "Lcl Scaling"))
{
loopi(3) if(!p.parse(t)) return;
loopi(3) { if(!p.parse(t)) return; if(t.type != token::NUMBER) break; n->scale[i] = t.f; }
}
}
}
p.skipprop();
break;
}
endlimbprops:;
}
p.skipprop();
break;
}
}
p.skipprop();
}
void parsematerial()
{
token id, name;
if(!p.parse(id) || !p.parse(name)) return;
if(id.type == token::NUMBER)
{
if(name.type == token::STRING)
{
char *str = name.s;
if(strstr(str, "Material::") == str) str += strlen("Material::");
materialnode *n = new materialnode;
copystring(n->name, str);
nodes[id.f] = n;
}
}
p.skipprop();
}
void parsedeformer()
{
token id, name, type, t;
if(!p.parse(id) || !p.parse(name) || !p.parse(type)) return;
if(id.type != token::NUMBER || type.type != token::STRING || name.type != token::STRING)
{
p.skipprop();
return;
}
if(!strcmp(type.s, "Skin"))
{
skinnode *n = new skinnode;
nodes[id.f] = n;
}
else if(!strcmp(type.s, "Cluster"))
{
if(!p.findbegin()) return;
clusternode *n = new clusternode;
nodes[id.f] = n;
while(p.parse(t)) switch(t.type)
{
case token::END: return;
case token::PROP:
if(!strcmp(t.s, "Indexes")) p.readarray(n->indexes);
else if(!strcmp(t.s, "Weights")) p.readarray(n->weights);
else if(!strcmp(t.s, "Transform")) p.readarray(n->transform);
else if(!strcmp(t.s, "TransformLink")) p.readarray(n->transformlink);
else p.skipprop();
break;
}
return;
}
p.skipprop();
}
void parsecurve()
{
token id, t;
if(!p.parse(id)) return;
if(id.type != token::NUMBER) { p.skipprop(); return; }
curvenode *n = new curvenode;
nodes[id.f] = n;
while(p.parse(t)) switch(t.type)
{
case token::END: return;
case token::PROP:
if(!strcmp(t.s, "KeyValueFloat")) p.readarray(n->vals);
else p.skipprop();
break;
}
}
void parsexform()
{
token id, t;
if(!p.parse(id)) return;
if(id.type != token::NUMBER) { p.skipprop(); return; }
if(!p.findbegin()) return;
xformnode *n = new xformnode;
nodes[id.f] = n;
while(p.parse(t)) switch(t.type)
{
case token::END: return;
case token::PROP:
if(!strcmp(t.s, "Properties70"))
{
if(!p.findbegin()) return;
while(p.parse(t)) switch(t.type)
{
case token::END: goto endprops;
case token::PROP:
if(!strcmp(t.s, "P"))
{
token name, type, val;
if(!p.parse(name) || !p.parse(type) || !p.parse(t) || !p.parse(t)) return;
if(name.type == token::STRING)
{
if(!strcmp(name.s, "d|X")) { if(p.parse(val) && val.type == token::NUMBER) n->val.x = val.f; }
else if(!strcmp(name.s, "d|Y")) { if(p.parse(val) && val.type == token::NUMBER) n->val.y = val.f; }
else if(!strcmp(name.s, "d|Z")) { if(p.parse(val) && val.type == token::NUMBER) n->val.z = val.f; }
}
}
p.skipprop();
break;
}
endprops:;
}
else p.skipprop();
break;
}
}
void parseanimlayer()
{
token id, name;
if(!p.parse(id) || !p.parse(name)) return;
if(id.type != token::NUMBER || name.type != token::STRING) { p.skipprop(); return; }
char *str = name.s;
if(strstr(str, "AnimLayer::") == str) str += strlen("AnimLayer::");
animlayernode *n = new animlayernode;
copystring(n->name, str);
nodes[id.f] = n;
p.skipprop();
}
#define FBX_SEC 46186158000.0
void parseanimstack()
{
token id, name, t;
if(!p.parse(id) || !p.parse(name)) return;
if(id.type != token::NUMBER || name.type != token::STRING) { p.skipprop(); return; }
char *str = name.s;
if(strstr(str, "AnimStack::") == str) str += strlen("AnimStack::");
animstacknode *n = new animstacknode;
copystring(n->name, str);
nodes[id.f] = n;
if(!p.findbegin()) return;
while(p.parse(t)) switch(t.type)
{
case token::END: return;
case token::PROP:
if(!strcmp(t.s, "Properties70"))
{
if(!p.findbegin()) return;
while(p.parse(t)) switch(t.type)
{
case token::END: goto endprops;
case token::PROP:
if(!strcmp(t.s, "P"))
{
token name, type, val;
if(!p.parse(name) || !p.parse(type) || !p.parse(t) || !p.parse(t)) return;
if(name.type == token::STRING)
{
if(!strcmp(name.s, "LocalStop")) { if(p.parse(val) && val.type == token::NUMBER) n->secs = val.f / FBX_SEC; }
}
}
p.skipprop();
break;
}
endprops:;
}
else p.skipprop();
break;
}
}
void parseobjects()
{
if(!p.findbegin()) return;
token t;
while(p.parse(t)) switch(t.type)
{
case token::END:
return;
case token::PROP:
if(!strcmp(t.s, "Geometry")) parsegeometry();
else if(!strcmp(t.s, "Model")) parsemodel();
else if(!strcmp(t.s, "Material")) parsematerial();
else if(!strcmp(t.s, "Deformer")) parsedeformer();
else if(!strcmp(t.s, "AnimationCurve")) parsecurve();
else if(!strcmp(t.s, "AnimationCurveNode")) parsexform();
else if(!strcmp(t.s, "AnimationLayer")) parseanimlayer();
else if(!strcmp(t.s, "AnimationStack")) parseanimstack();
else p.skipprop();
break;
}
}
void parseconnection()
{
token type, from, to, prop;
if(!p.parse(type) || !p.parse(from) || !p.parse(to)) return;
if(type.type == token::STRING && from.type == token::NUMBER && to.type == token::NUMBER)
{
node *nf = nodes.find(from.f, NULL), *nt = nodes.find(to.f, NULL);
if(!strcmp(type.s, "OO") && nf && nt)
{
if(nf->type() == node::GEOM && nt->type() == node::MODEL)
((geomnode *)nf)->model = (modelnode *)nt;
else if(nf->type() == node::MATERIAL && nt->type() == node::MODEL)
((modelnode *)nt)->material = (materialnode *)nf;
else if(nf->type() == node::LIMB && nt->type() == node::LIMB)
((limbnode *)nf)->parent = (limbnode *)nt;
else if(nf->type() == node::CLUSTER && nt->type() == node::SKIN)
((clusternode *)nf)->skin = (skinnode *)nt;
else if(nf->type() == node::SKIN && nt->type() == node::GEOM)
((skinnode *)nf)->geom = (geomnode *)nt;
else if(nf->type() == node::LIMB && nt->type() == node::CLUSTER)
{
((clusternode *)nt)->limb = (limbnode *)nf;
((limbnode *)nf)->cluster = (clusternode *)nt;
}
else if(nf->type() == node::ANIMLAYER && nt->type() == node::ANIMSTACK)
((animstacknode *)nt)->layers.add((animlayernode *)nf);
else if(nf->type() == node::XFORM && nt->type() == node::ANIMLAYER)
((animlayernode *)nt)->xforms.add((xformnode *)nf);
}
else if(!strcmp(type.s, "OP") && nf && nt && p.parse(prop) && prop.type == token::STRING)
{
if(nf->type() == node::CURVE && nt->type() == node::XFORM)
{
if(!strcmp(prop.s, "d|X")) ((xformnode *)nt)->setcurve(0, (curvenode *)nf);
else if(!strcmp(prop.s, "d|Y")) ((xformnode *)nt)->setcurve(1, (curvenode *)nf);
else if(!strcmp(prop.s, "d|Z")) ((xformnode *)nt)->setcurve(2, (curvenode *)nf);
}
else if(nf->type() == node::XFORM && nt->type() == node::LIMB)
{
((xformnode *)nf)->limb = (limbnode *)nt;
if(!strcmp(prop.s, "Lcl Translation")) ((xformnode *)nf)->xform = xformnode::TRANS;
else if(!strcmp(prop.s, "Lcl Rotation")) ((xformnode *)nf)->xform = xformnode::ROT;
else if(!strcmp(prop.s, "Lcl Scaling")) ((xformnode *)nf)->xform = xformnode::SCALE;
}
}
}
p.skipprop();
}
void parseconnections()
{
if(!p.findbegin()) return;
token t;
while(p.parse(t)) switch(t.type)
{
case token::END:
return;
case token::PROP:
if(!strcmp(t.s, "C")) parseconnection();
else p.skipprop();
break;
}
}
void geomnode::process()
{
if(model)
{
emeshes[mesh].name = getnamekey(model->name);
if(model->material) emeshes[mesh].material = getnamekey(model->material->name);
if(model->geomtrans != Vec3(0, 0, 0)) for(int i = firstvert; i < lastvert; i++) epositions[i] += model->geomtrans;
if(model->lclscale != Vec3(1, 1, 1))
{
for(int i = firstvert; i < lastvert; i++)
{
epositions[i].setxyz(model->lclscale * Vec3(epositions[i]));
}
}
if(model->lclrot != Vec3(0, 0, 0))
{
Quat lclquat = Quat::fromdegrees(model->lclrot);
for(int i = firstvert; i < lastvert; i++)
{
epositions[i].setxyz(lclquat.transform(Vec3(epositions[i])));
enormals[i] = lclquat.transform(enormals[i]);
}
}
if(model->prerot != Vec3(0, 0, 0))
{
Quat prequat = Quat::fromdegrees(model->prerot);
for(int i = firstvert; i < lastvert; i++)
{
epositions[i].setxyz(prequat.transform(Vec3(epositions[i])));
enormals[i] = prequat.transform(enormals[i]);
}
}
if(model->lcltrans != Vec3(0, 0, 0)) for(int i = firstvert; i < lastvert; i++) epositions[i] += model->lcltrans;
}
}
void clusternode::process()
{
if(!limb || limb->index > 255 || !skin || !skin->geom || indexes.length() != weights.length()) return;
geomnode *g = skin->geom;
if(g->blends.empty()) loopi(g->numverts) g->blends.add();
loopv(indexes)
{
int idx = indexes[i];
double weight = weights[i];
g->blends[idx].addweight(weight, limb->index);
}
}
void animstacknode::process()
{
if(layers.empty()) return;
animlayernode *l = layers[0];
int numframes = l->numframes();
if(numframes < 0) return;
eanim &a = eanims.add();
a.name = getnamekey(name);
a.startframe = eframes.length();
a.fps = secs > 0 ? numframes/secs : 0;
transform *poses = eposes.reserve(numframes*ejoints.length());
loopj(numframes)
{
eframes.add(eposes.length());
eposes.put(eposes.getbuf(), ejoints.length());
}
loopv(l->xforms)
{
xformnode &x = *l->xforms[i];
if(!x.limb) continue;
transform *dst = &poses[x.limb->index];
loopj(numframes)
{
Vec3 val = x.val;
loopk(3) if(x.curves[k]) val[k] = x.curves[k]->vals[j];
switch(x.xform)
{
case xformnode::TRANS: dst->pos = val; break;
case xformnode::ROT: dst->orient = Quat::fromdegrees(val); break;
case xformnode::SCALE: dst->scale = val; break;
}
dst += ejoints.length();
}
}
#if 0
loopv(eposes)
{
transform &t = eposes[i];
Matrix3x3 m(t.orient, t.scale);
Vec3 mscale(Vec3(m.a.x, m.b.x, m.c.x).magnitude(), Vec3(m.a.y, m.b.y, m.c.y).magnitude(), Vec3(m.a.z, m.b.z, m.c.z).magnitude());
if(m.determinant() < 0) mscale = -mscale;
m.a /= mscale;
m.b /= mscale;
m.c /= mscale;
Quat morient(m); if(morient.w > 0) morient.flip();
t.orient = morient;
t.scale = mscale;
}
#endif
}
void geomnode::finish()
{
if(blends.empty()) return;
loopv(blends) blends[i].finalize();
while(eblends.length() < lastvert) eblends.add();
if(remap.length()) loopv(remap) eblends[firstvert + i] = blends[remap[i]];
else loopv(blends) eblends[firstvert + i] = blends[i];
}
void limbnode::finish()
{
if(prerot == Vec3(0, 0, 0)) return;
Quat prequat = Quat::fromdegrees(prerot);
for(int i = index; i < eposes.length(); i += ejoints.length())
eposes[i].orient = prequat * eposes[i].orient;
}
bool checkversion(stream *f)
{
return f->getline(p.buf, sizeof(p.buf)) && strstr(p.buf, "FBX 7");
}
void parse(stream *f)
{
p.reset(f);
token t;
while(p.parse(t)) switch(t.type)
{
case token::PROP:
if(!strcmp(t.s, "Objects")) parseobjects();
else if(!strcmp(t.s, "Connections")) parseconnections();
else p.skipprop();
break;
}
enumerate(nodes, double, id, node *, n, { (void)id; n->process(); });
enumerate(nodes, double, id, node *, n, { (void)id; n->finish(); });
enumerate(nodes, double, id, node *, n, { (void)id; delete n; });
nodes.clear();
}
}
bool loadfbx(const char *filename, const filespec &spec)
{
int numfiles = 0;
while(filename)
{
const char *endfile = strchr(filename, ',');
const char *file = endfile ? newstring(filename, endfile-filename) : filename;
stream *f = openfile(file, "r");
if(f)
{
if(fbx::checkversion(f))
{
resetimporter(spec, numfiles > 0);
numfiles++;
fbx::parse(f);
}
delete f;
}
if(!endfile) break;
delete[] file;
filename = endfile+1;
}
if(!numfiles) return false;
if(eanims.length() == 1)
{
eanim &a = eanims.last();
if(spec.name) a.name = spec.name;
if(spec.fps > 0) a.fps = spec.fps;
a.flags |= spec.flags;
if(spec.endframe >= 0) a.endframe = a.startframe + spec.endframe;
else if(spec.endframe < -1) a.endframe = a.startframe + max(eframes.length() - a.startframe + spec.endframe + 1, 0);
a.startframe += spec.startframe;
}
erotate *= Quat(M_PI/2, Vec3(1, 0, 0));
makeanims(spec);
if(emeshes.length())
{
smoothverts();
makemeshes(spec);
}
return true;
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
namespace fte
{
static plugfsfuncs_t cppfsfuncs =
{
.OpenVFS = [](const char *filename, const char *mode, enum fs_relative relativeto)
{
stream *f = openfile(filename, "rb");
if (!f)
return (vfsfile_t*)nullptr;
struct cppfile_t : public vfsfile_t
{
static int ReadBytes (struct vfsfile_s *file, void *buffer, int bytestoread)
{
auto c = static_cast<cppfile_t*>(file);
return c->f->read(buffer, bytestoread);
}
static qboolean Seek (struct vfsfile_s *file, qofs_t pos)
{
auto c = static_cast<cppfile_t*>(file);
return c->f->seek(pos)?qtrue:qfalse;
}
static qofs_t Tell (struct vfsfile_s *file)
{
auto c = static_cast<cppfile_t*>(file);
return c->f->tell();
}
static qofs_t GetLen (struct vfsfile_s *file)
{
auto c = static_cast<cppfile_t*>(file);
return c->f->size();
}
static qboolean Close (struct vfsfile_s *file)
{
auto c = static_cast<cppfile_t*>(file);
c->f->close();
delete c;
return qtrue;
}
cppfile_t(stream *sourcefile):f(sourcefile)
{
vfsfile_t::ReadBytes = ReadBytes;
vfsfile_t::Seek = Seek;
vfsfile_t::Tell = Tell;
vfsfile_t::GetLen = GetLen;
vfsfile_t::Close = Close;
}
stream *f;
};
cppfile_t *c = new cppfile_t(f);
return static_cast<vfsfile_t*>(c);
},
};
static plugmodfuncs_t cppmodfuncs =
{
.version = MODPLUGFUNCS_VERSION,
.RegisterModelFormatText = [](const char *formatname, char *magictext, qboolean (QDECL *load) (struct model_s *mod, void *buffer, size_t fsize))
{ //called explicitly because we're lame.
return 0;
},
.RegisterModelFormatMagic = [](const char *formatname, unsigned int magic, qboolean (QDECL *load) (struct model_s *mod, void *buffer, size_t fsize))
{ //called explicitly because we're lame.
return 0;
},
.ZG_Malloc = [](zonegroup_t *ctx, size_t size)
{
/*leak the memory, because we're lazy*/
void *ret = malloc(size);
memset(ret, 0, size);
return ret;
},
.ConcatTransforms = [](const float in1[3][4], const float in2[3][4], float out[3][4])
{
out[0][0] = in1[0][0] * in2[0][0] + in1[0][1] * in2[1][0] +
in1[0][2] * in2[2][0];
out[0][1] = in1[0][0] * in2[0][1] + in1[0][1] * in2[1][1] +
in1[0][2] * in2[2][1];
out[0][2] = in1[0][0] * in2[0][2] + in1[0][1] * in2[1][2] +
in1[0][2] * in2[2][2];
out[0][3] = in1[0][0] * in2[0][3] + in1[0][1] * in2[1][3] +
in1[0][2] * in2[2][3] + in1[0][3];
out[1][0] = in1[1][0] * in2[0][0] + in1[1][1] * in2[1][0] +
in1[1][2] * in2[2][0];
out[1][1] = in1[1][0] * in2[0][1] + in1[1][1] * in2[1][1] +
in1[1][2] * in2[2][1];
out[1][2] = in1[1][0] * in2[0][2] + in1[1][1] * in2[1][2] +
in1[1][2] * in2[2][2];
out[1][3] = in1[1][0] * in2[0][3] + in1[1][1] * in2[1][3] +
in1[1][2] * in2[2][3] + in1[1][3];
out[2][0] = in1[2][0] * in2[0][0] + in1[2][1] * in2[1][0] +
in1[2][2] * in2[2][0];
out[2][1] = in1[2][0] * in2[0][1] + in1[2][1] * in2[1][1] +
in1[2][2] * in2[2][1];
out[2][2] = in1[2][0] * in2[0][2] + in1[2][1] * in2[1][2] +
in1[2][2] * in2[2][2];
out[2][3] = in1[2][0] * in2[0][3] + in1[2][1] * in2[1][3] +
in1[2][2] * in2[2][3] + in1[2][3];
},
.GenMatrixPosQuat4Scale = [](const vec3_t pos, const vec4_t quat, const vec3_t scale, float result[12])
{
Matrix3x4 m = Matrix3x4(Quat(quat), Vec3(pos), Vec3(scale));
result[0] = m.a.x;
result[1] = m.b.x;
result[2] = m.c.x;
result[3] = m.a.w;
result[4] = m.a.y;
result[5] = m.b.y;
result[6] = m.c.y;
result[7] = m.b.w;
result[8] = m.a.z;
result[9] = m.b.z;
result[10] = m.c.z;
result[11] = m.c.w;
},
.GetTexture = [](const char *identifier, const char *subpath, unsigned int flags, void *fallbackdata, void *fallbackpalette, int fallbackwidth, int fallbackheight, uploadfmt_t fallbackfmt)
{
image_t *img = (image_t*)cppmodfuncs.ZG_Malloc(NULL, sizeof(*img)+strlen(identifier)+1);
img->ident = (char*)(img+1);
strcpy(img->ident, identifier);
img->flags = flags;
return img;
},
.AccumulateTextureVectors = [](vecV_t *const vc, vec2_t *const tc, vec3_t *nv, vec3_t *sv, vec3_t *tv, const index_t *idx, int numidx, qboolean calcnorms)
{ //once per surface that shares the set of verts
},
.NormaliseTextureVectors = [](vec3_t *n, vec3_t *s, vec3_t *t, int v, qboolean calcnorms)
{ //once per shared set of verts.
},
};
static plugcorefuncs_t cppplugfuncs =
{
.GetEngineInterface = [](const char *interfacename, size_t structsize)
{
void *ret = nullptr;
if (!strcmp(interfacename, plugfsfuncs_name))
ret = &cppfsfuncs;
if (!strcmp(interfacename, plugmodfuncs_name))
ret = &cppmodfuncs;
return ret;
},
};
static plugcvarfuncs_t cppcvarfuncs =
{
.GetNVFDG = [](const char *name, const char *defaultval, unsigned int flags, const char *description, const char *groupname)
{ //could maybe fill with environment settings perhaps? yuck.
auto v = new cvar_t();
v->name = strdup(name);
v->string = strdup(defaultval);
v->value = atof(v->string);
v->ival = atoi(v->string);
return v;
},
};
extern "C"
{ //our plugin-style stuff has a few external dependancies not provided via pointers...
void ImgTool_SetupPalette(void);
qboolean QDECL Mod_LoadGLTFModel (struct model_s *mod, void *buffer, size_t fsize);
qboolean QDECL Mod_LoadGLBModel (struct model_s *mod, void *buffer, size_t fsize);
qboolean Plug_GLTF_Init(void);
plugcorefuncs_t *plugfuncs = &cppplugfuncs;
plugcmdfuncs_t *cmdfuncs;
plugcvarfuncs_t *cvarfuncs = &cppcvarfuncs;
void Q_strlcpy(char *d, const char *s, int n)
{
int i;
n--;
if (n < 0)
return; //this could be an error
for (i=0; *s; i++)
{
if (i == n)
break;
*d++ = *s++;
}
*d='\0';
}
void Q_strlcat(char *d, const char *s, int n)
{
if (n)
{
int dlen = strlen(d);
int slen = strlen(s)+1;
if (slen > (n-1)-dlen)
slen = (n-1)-dlen;
memcpy(d+dlen, s, slen);
d[n - 1] = 0;
}
}
}
transform ftetransform(float bm[12], bool invert)
{
Matrix3x3 m(Vec3(bm[0], bm[1], bm[2]), Vec3(bm[4], bm[5], bm[6]), Vec3(bm[8], bm[9], bm[10]));
m.transpose();
Vec3 pos(bm[3], bm[7], bm[11]);
transform t;
Vec3 mscale(Vec3(m.a.x, m.b.x, m.c.x).magnitude(), Vec3(m.a.y, m.b.y, m.c.y).magnitude(), Vec3(m.a.z, m.b.z, m.c.z).magnitude());
// check determinant for sign of scaling
if(Matrix3x3(m).determinant() < 0) mscale = -mscale;
m.a /= mscale;
m.b /= mscale;
m.c /= mscale;
t.orient = Quat(m);
if(t.orient.w > 0) t.orient.flip();
t.scale = mscale;
if (invert)
{
// invert the translate
t.pos[0] = -(pos[0] * m.a[0] + pos[1] * m.a[1] + pos[2] * m.a[2]);
t.pos[1] = -(pos[0] * m.b[0] + pos[1] * m.b[1] + pos[2] * m.b[2]);
t.pos[2] = -(pos[0] * m.c[0] + pos[1] * m.c[1] + pos[2] * m.c[2]);
}
else
t.pos = pos;
return t;
}
bool loadfte(model_t *mod, const filespec &spec)
{ //import from fte's structs and convert to iqmtool's c++isms
if (mod->type != mod_alias)
return false; //err...
galiasinfo_t *surf = (galiasinfo_t*)mod->meshinfo;
if (surf)
{
resetimporter(spec);
if (surf->baseframeofs)
{
for (int b = 0; b < surf->numbones; b++)
{
transform p(ftetransform(surf->ofsbones[b].inverse, true));
//spit out the joint info
ejoint &j = ejoints.add();
j.name = surf->ofsbones[b].name;
j.parent = surf->ofsbones[b].parent;
//and the base pose
eposes.add(p);
}
makerelativebasepose();
#if 1 //import the animations
for (int animidx = 0; animidx < surf->numanimations; animidx++)
{
auto &anim = surf->ofsanimations[animidx];
int firstframe = eframes.length();
vector<float[12]> bonebuf;
for (int f = 0; f < anim.numposes; f++)
{
skeltype_t sk = anim.skeltype;
float time = f/anim.rate;
float *bonedata;
if (anim.GetRawBones)
bonedata = anim.GetRawBones(surf, &anim, time, bonebuf.reserve(surf->numbones)[0], surf->numbones);
else if (anim.boneofs)
bonedata = (float*)anim.boneofs;
else
bonedata = (float*)surf->baseframeofs, sk = SKEL_ABSOLUTE; //abs...
if (sk == SKEL_RELATIVE)
;
else
printf("Unusable skeletal type for import - %i\n", (int)sk);
eframes.add(eposes.length());
for (int b = 0; b < surf->numbones; b++, bonedata += 12)
eposes.add(ftetransform(bonedata, false));
}
eanim &a = eanims.add();
if(spec.name) a.name = getnamekey(spec.name);
else
{
string name;
copystring(name, mod->name);
char *end = strrchr(name, '.');
if(end) *end = '\0';
a.name = getnamekey(name);
}
a.startframe = firstframe;
a.fps = anim.rate;
a.flags = anim.loop?IQM_LOOP:0;
a.endframe = eframes.length();
}
#endif
}
for(; surf; surf = surf->nextsurf)
{
if (surf->shares_bones != 0)
continue;
if (surf->numindexes)
{
unsigned int firstvert=epositions.length();
for (int v = 0; v < surf->numverts; v++)
{
Vec3 pos(surf->ofs_skel_xyz[v][0], surf->ofs_skel_xyz[v][1], surf->ofs_skel_xyz[v][2]);
Vec3 norm;
if (surf->ofs_skel_norm)
norm = Vec3(surf->ofs_skel_norm[v][0], surf->ofs_skel_norm[v][1], surf->ofs_skel_norm[v][2]);
else
norm = Vec3(0,0,0);
etexcoords.add(Vec4(surf->ofs_st_array[v][0], surf->ofs_st_array[v][1], 0, 0));
if (surf->ofs_rgbaf)
ecolors.add(Vec4(surf->ofs_rgbaf[v][0], surf->ofs_rgbaf[v][1], surf->ofs_rgbaf[v][2], surf->ofs_rgbaf[v][3]));
else if (surf->ofs_rgbaub)
ecolors.add(Vec4(surf->ofs_rgbaub[v][0]/255.0, surf->ofs_rgbaub[v][1]/255.0, surf->ofs_rgbaub[v][2]/255.0, surf->ofs_rgbaub[v][3]/255.0));
// if (surf->ofs_skel_svect)
// etangents.add (Vec4(surf->ofs_skel_svect[v][0], surf->ofs_skel_svect[v][1], surf->ofs_skel_svect[v][2], 0));
// if (surf->ofs_skel_tvect)
// ebitangents.add(Vec3(surf->ofs_skel_tvect[v][0], surf->ofs_skel_tvect[v][1], surf->ofs_skel_tvect[v][2]));
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if (surf->shares_bones == 0 && surf->ofs_skel_weight && surf->ofs_skel_idx)
{
blendcombo b = {};
//Vec3 newpos(0,0,0);
//Vec3 newnorm(0,0,0);
for (size_t w = 0; w < 4; w++)
{
//newpos += bonerepositions[surf->ofs_skel_idx[v][w]].transform(pos) * surf->ofs_skel_weight[v][w];
//newnorm += bonerepositions[surf->ofs_skel_idx[v][w]].transform3(norm) * surf->ofs_skel_weight[v][w];
if (surf->ofs_skel_weight[v][w] > 0)
b.addweight(surf->ofs_skel_weight[v][w], surf->ofs_skel_idx[v][w]);
}
b.finalize();
eblends.add(b);
//pos = newpos;
//norm = newnorm;
}
epositions.add(Vec4(pos));
enormals.add(norm);
}
//iqms don't support skins/skingroups themselves.
//we have only surface name and texture(aka material) name.
//so use the diffuse texture's name where we can
// a) its already processed properly so no ''path/model.gltf/sectionthatdoesntevenexistondisk' locations.
// b) its more likely to show something without needing to synthesize shaders/textures.
//we should probably cvar this.
const char *materialname;
if (surf->numskins && surf->ofsskins[0].numframes && surf->ofsskins[0].frame[0].texnums.base && *surf->ofsskins[0].frame[0].texnums.base->ident != '$')
materialname = surf->ofsskins[0].frame[0].texnums.base->ident;
else if (surf->numskins)
materialname = surf->ofsskins[0].name;
else
materialname = surf->surfacename;
emesh mesh(surf->surfacename, materialname, etriangles.length());
//add in some extra surface properties.
mesh.hasexplicits = true;
mesh.explicits.contents = surf->contents;
mesh.explicits.surfaceflags = surf->csurface.flags;
mesh.explicits.body = surf->surfaceid;
mesh.explicits.geomset = surf->geomset;
mesh.explicits.geomid = surf->geomid;
mesh.explicits.mindist = surf->mindist;
mesh.explicits.maxdist = surf->maxdist;
emeshes.add(mesh);
for (int idx = 0; idx+2 < surf->numindexes; idx+=3)
etriangles.add(etriangle(surf->ofs_indexes[idx+0]+firstvert, surf->ofs_indexes[idx+1]+firstvert, surf->ofs_indexes[idx+2]+firstvert));
}
}
makeanims(spec);
if (emeshes.length())
{
smoothverts();
makemeshes(spec);
}
return true;
}
return false;
}
bool loadglb(const char *filename, const filespec &spec)
{
bool ret = false;
model_t mod={};
stream *f = openfile(filename, "rb");
Q_strlcpy(mod.name, filename, sizeof(mod.name));
if (f)
{
size_t sz = f->size();
auto filebuf = new char[sz];
if (sz == f->read(filebuf, sz))
{
if (Plug_GLTF_Init())
if (Mod_LoadGLBModel(&mod, filebuf, sz))
ret = loadfte(&mod, spec);
}
delete[] filebuf;
delete f;
}
return ret;
}
bool loadgltf(const char *filename, const filespec &spec)
{
bool ret = false;
model_t mod={};
stream *f = openfile(filename, "rb");
Q_strlcpy(mod.name, filename, sizeof(mod.name));
if (f)
{
size_t sz = f->size();
auto filebuf = new char[sz];
if (sz == f->read(filebuf, sz))
{
if (Plug_GLTF_Init())
if (Mod_LoadGLTFModel(&mod, filebuf, sz))
ret = loadfte(&mod, spec);
}
delete[] filebuf;
delete f;
}
return ret;
}
}
void genhitboxes(vector<hitbox> &hitboxes)
{
//for half-life weenies that are too lazy to define their own hitmeshes
if (!hitboxes.length())
return;
filespec inspec;
inspec.reset();
resetimporter(inspec);
loopv(hitboxes)
{
hitbox &hb = hitboxes[i];
int bone = -1;
for (bone = 0; bone < joints.length(); bone++)
if (!strcasecmp(hb.bone, &stringdata[joints[bone].name]))
break;
if (bone == joints.length())
{
fatal("error: hitbox attached to invalid bone %s", hb.bone);
continue; //this hitbox is invalid
}
emesh &m = emeshes.add();
int firstvert = epositions.length();
m.firsttri = etriangles.length();
m.material = "textures/common/hitmesh"; //to be vaugely compatible with q3map2's default shader names
string tmp;
formatstring(tmp, "hitbox%i", hitboxes[i].body);
m.name = newstring(tmp);
m.hasexplicits = true;
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
m.explicits = {};
m.explicits.contents = 0x02000000;
m.explicits.surfaceflags = 0x80;
m.explicits.body = hitboxes[i].body;
m.explicits.geomset = ~0u;
//spit out some verts
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
Matrix3x4 bm(mjoints[bone]);
bm.invert();
for (int j = 0; j < 8; j++)
{
Vec3 p = Vec3((j&1)?hb.mins[0]:hb.maxs[0], (j&2)?hb.mins[1]:hb.maxs[1], (j&4)?hb.mins[2]:hb.maxs[2]);
p = bm.transform(p);
epositions.add(Vec4(p, 0));
enormals.add(p);
etexcoords.add(Vec4(0,0,0,0));
eblends.add(blendcombo()).addweight(1, bone);
}
//and some triangles for them
etriangles.add(etriangle(firstvert+2, firstvert+1, firstvert+0));
etriangles.add(etriangle(firstvert+2, firstvert+3, firstvert+1));
etriangles.add(etriangle(firstvert+4, firstvert+5, firstvert+6));
etriangles.add(etriangle(firstvert+5, firstvert+7, firstvert+6));
etriangles.add(etriangle(firstvert+0, firstvert+1, firstvert+4));
etriangles.add(etriangle(firstvert+1, firstvert+5, firstvert+4));
etriangles.add(etriangle(firstvert+6, firstvert+3, firstvert+2));
etriangles.add(etriangle(firstvert+6, firstvert+7, firstvert+3));
etriangles.add(etriangle(firstvert+4, firstvert+2, firstvert+0));
etriangles.add(etriangle(firstvert+4, firstvert+6, firstvert+2));
etriangles.add(etriangle(firstvert+1, firstvert+3, firstvert+5));
etriangles.add(etriangle(firstvert+3, firstvert+7, firstvert+5));
}
smoothverts();
makemeshes(inspec);
}
int framesize = 0;
vector<ushort> animdata;
#define QUANTIZE(offset, base, scale) ushort(0.5f + (float(offset) - base) / scale)
static int jsort(const void *va, const void *vb)
{
joint &a = joints[*(int*)va];
joint &b = joints[*(int*)vb];
if (a.group == b.group)
{
if (*(int*)va > *(int*)vb)
return 1;
else
return -1;
}
else if (a.group < b.group)
return -1;
else
return 1;
}
void calcanimdata()
{
hashtable<const char *, bool> bonewarnings;
//reorder the joints according to their groups, including a lookup so we can fix up other mappings
int *jointremap = new int[joints.length()];
int *jointremapinv = new int[joints.length()];
loopv(joints) jointremap[i] = i;
qsort(jointremap, joints.length(), sizeof(int), jsort);
vector<joint> oj;
joints.swap(oj);
bool dodgyorder = false;
loopv(oj)
jointremapinv[jointremap[i]] = i;
loopv(oj)
{
joint &j = joints.add(oj[jointremap[i]]);
if (j.parent >= 0)
{
j.parent = jointremapinv[j.parent];
if (j.parent >= i)
dodgyorder = true;
}
}
if (dodgyorder)
{
printbonelist();
fatal("Bone group reordering resulted in invalid order");
}
//try and ensure that the animation bone order matches the mesh bones
loopv(joints)
{
pose &j = poses.add();
j.name = &stringdata[joints[i].name];
j.parent = joints[i].parent;
loopk(10) { j.offset[k] = 1e16f; j.scale[k] = -1e16f; }
}
loopv(frames)
{
frame &fr = frames[i];
loopl(fr.pose.length())
{
frame::framepose &p = fr.pose[l];
p.remap = -1;
loopvk(poses)
{
// if (poses[k].parent == p.boneparent)
if (!strcmp(poses[k].name, p.bonename))
{
if (poses[k].parent == -1 || p.boneparent == -1)
{
if (poses[k].parent != -1 || p.boneparent != -1)
fatal("Error: bone %s has inconsistent parents\n", p.bonename);
}
else if (strcmp(poses[poses[k].parent].name, fr.pose[p.boneparent].bonename))
fatal("Error: bone %s has inconsistent parents (%s vs %s)\n", p.bonename, poses[poses[k].parent].name, fr.pose[p.boneparent].bonename);
p.remap = k;
break;
}
}
if(p.remap < 0)
{
//if we have a mesh, then any extra bones are surplus to requirements.
//otherwise we play safe and keep all (which is kinda awkward, because there's no way to name them in the output iqm).
if (!joints.empty())
{
if (!bonewarnings.find(p.bonename, false))
{
const char *a = "UNKNOWN";
loopvj(anims)
{
if ((uint)i >= anims[j].firstframe && (uint)i < anims[j].firstframe+anims[j].numframes)
{
a = &stringdata[anims[j].name];
break;
}
}
bonewarnings.access(p.bonename, true);
if (p.boneparent >= 0)
conoutf("warning: ignoring bone %s (parent %s) (surplus in %s)", p.bonename, fr.pose[p.boneparent].bonename, a);
else
conoutf("warning: ignoring bone %s (root) (surplus in %s)", p.bonename, a);
}
continue;
}
if (p.boneparent >= 0)
conoutf("bone %s (%s)", p.bonename, poses[p.boneparent].name);
else
conoutf("bone %s", p.bonename);
p.remap = poses.length();
pose &j = poses.add();
j.name = p.bonename;
j.parent = -1;
loopk(p.remap)
{
if (!strcmp(poses[k].name, fr.pose[p.boneparent].bonename))
{
j.parent = k;
break;
}
}
loopk(10) { j.offset[k] = 1e16f; j.scale[k] = -1e16f; }
}
pose &j = poses[p.remap];
transform &f = p.tr;
loopk(3)
{
j.offset[k] = min(j.offset[k], float(f.pos[k]));
j.scale[k] = max(j.scale[k], float(f.pos[k]));
}
loopk(4)
{
j.offset[3+k] = min(j.offset[3+k], float(f.orient[k]));
j.scale[3+k] = max(j.scale[3+k], float(f.orient[k]));
}
loopk(3)
{
j.offset[7+k] = min(j.offset[7+k], float(f.scale[k]));
j.scale[7+k] = max(j.scale[7+k], float(f.scale[k]));
}
}
}
loopv(poses)
{
pose &j = poses[i];
loopk(10)
{
j.scale[k] -= j.offset[k];
if(j.scale[k] >= 1e-10f) { framesize++; j.scale[k] /= 0xFFFF; j.flags |= 1<<k; }
else j.scale[k] = 0.0f;
}
}
#if 0
int runlength = 0, blocksize = 0, blocks = 0;
#define FLUSHVAL(val) \
if(!blocksize || (animdata.last() == val ? runlength >= 0xFF : runlength || blocksize > 0xFF)) \
{ \
animdata.add(0); \
animdata.add(val); \
blocksize = 1; \
runlength = 0; \
blocks++; \
} \
else if(animdata.last() == val) \
{ \
animdata[animdata.length()-blocksize-1] += 0x10; \
runlength++; \
} \
else \
{ \
animdata[animdata.length()-blocksize-1]++; \
animdata.add(val); \
blocksize++; \
}
loopv(joints)
{
joint &j = joints[i];
loopk(3) if(j.flags & (0x01<<k))
{
for(int l = i; l < frames.length(); l += poses.length())
{
transform &f = frames[l];
ushort val = QUANTIZE(f.pos[k], j.offset[k], j.scale[k]);
FLUSHVAL(val);
}
}
loopk(4) if(j.flags & (0x08<<k))
{
for(int l = i; l < frames.length(); l += poses.length())
{
transform &f = frames[l];
ushort val = QUANTIZE(f.orient[k], j.offset[3+k], j.scale[3+k]);
FLUSHVAL(val);
}
}
loopk(3) if(j.flags & (0x80<<k))
{
for(int l = i; l < frames.length(); l += poses.length())
{
transform &f = frames[l];
ushort val = QUANTIZE(f.scale[k], j.offset[7+k], j.scale[7+k]);
FLUSHVAL(val);
}
}
}
printf("%d frames of size %d/%d compressed from %d/%d to %d in %d blocks", frames.length()/poses.length(), framesize, poses.length()*9, framesize*frames.length()/poses.length(), poses.length()*9*frames.length()/poses.length(), animdata.length(), blocks);
#else
transform *tr = new transform[poses.length()];
char *def = new char[poses.length()];
loopvk(poses) {def[k]=0;}
loopv(frames)
{
frame &fr = frames[i];
loopvk(poses) {tr[k] = transform(Vec3(0,0,0),Quat(0,0,0,1)); def[k]|=1;}
loopvk(fr.pose) if (fr.pose[k].remap>=0) {tr[fr.pose[k].remap] = fr.pose[k].tr; def[fr.pose[k].remap] &= ~1;}
loopvk(poses)
{
if (def[k] == 1)
{ //if this bone didn't have any data and is still in an identity pose, warn about it.
def[k] |= 2;
const char *a = "UNKNOWN";
loopvj(anims)
{
if ((uint)i >= anims[j].firstframe && (uint)i < anims[j].firstframe+anims[j].numframes)
{
a = &stringdata[anims[j].name];
break;
}
}
conoutf("warning: bone %s defaulted (missing in %s)", poses[k].name, a);
}
pose &j = poses[k];
transform &f = tr[k];
loopk(3) if(j.flags & (0x01<<k)) animdata.add(QUANTIZE(f.pos[k], j.offset[k], j.scale[k]));
loopk(4) if(j.flags & (0x08<<k)) animdata.add(QUANTIZE(f.orient[k], j.offset[3+k], j.scale[3+k]));
loopk(3) if(j.flags & (0x80<<k)) animdata.add(QUANTIZE(f.scale[k], j.offset[7+k], j.scale[7+k]));
}
}
delete[] tr;
#endif
//combine the arrays into a single vertex data lump.
loopv(varrays)
{
vertexarray &va = varrays[i];
va.offset = vdata.length();
//align it, if needed
uint align = max(va.formatsize(), 8);
if(va.offset%align) { uint pad = align - va.offset%align; va.offset += pad; loopi(pad) vdata.add(0); }
//splurge the data out.
uchar *final = vdata.reserve(va.bytesize() * va.count);
uchar *src = &va.vdata[0];
if (va.type == IQM_BLENDINDEXES && (va.format == IQM_UBYTE || va.format == IQM_BYTE))
loopk(va.size*va.count) final[k] = jointremapinv[src[k]];
else if (va.type == IQM_BLENDINDEXES && (va.format == IQM_USHORT || va.format == IQM_SHORT))
loopk(va.size*va.count) ((ushort*)final)[k] = jointremapinv[((ushort*)src)[k]];
else if (va.type == IQM_BLENDINDEXES && (va.format == IQM_UINT || va.format == IQM_INT))
loopk(va.size*va.count) ((uint*)final)[k] = jointremapinv[((uint*)src)[k]];
else
memcpy(final, &va.vdata[0], va.bytesize() * va.count);
vdata.advance(va.bytesize() * va.count);
va.vdata.setsize(0); //no longer needed.
}
while(vdata.length()%4) vdata.add(0);
while(stringdata.length()%4) stringdata.add('\0');
while(commentdata.length()%4) commentdata.add('\0');
while(animdata.length()%2) animdata.add(0);
delete[] jointremap;
delete[] jointremapinv;
if (joints.length()) loopv(frames) makebounds(bounds.add(), mjoints.getbuf(), frames[i]);
}
bool writeiqm(const char *filename)
{
vector<iqmextension> extensions;
stream *f = openfile(filename, "wb");
if(!f) return false;
iqmheader hdr;
memset(&hdr, 0, sizeof(hdr));
copystring(hdr.magic, IQM_MAGIC, sizeof(hdr.magic));
hdr.version = IQM_VERSION;
hdr.filesize = sizeof(hdr);
hdr.flags = modelflags;
iqmextension *ext_meshes_fte = NULL;
if (meshes_fte.length())
{
ext_meshes_fte = &extensions.add();
ext_meshes_fte->name = sharestring("FTE_MESH");
}
iqmextension *ext_events_fte = NULL;
if (events_fte.length())
{
ext_events_fte = &extensions.add();
ext_events_fte->name = sharestring("FTE_EVENT");
loopv(events_fte)
{
event_fte &ev = events_fte[i];
ev.evdata_idx = sharestring(ev.evdata_str);
}
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if(stringdata.length()) hdr.ofs_text = hdr.filesize, hdr.num_text = stringdata.length(), hdr.filesize += hdr.num_text;
hdr.num_meshes = meshes.length(); if(meshes.length()) hdr.ofs_meshes = hdr.filesize; hdr.filesize += meshes.length() * sizeof(iqmmesh);
uint voffset = hdr.filesize + varrays.length() * sizeof(iqmvertexarray);
hdr.num_vertexarrays = varrays.length(); if(varrays.length()) hdr.ofs_vertexarrays = hdr.filesize; hdr.filesize += varrays.length() * sizeof(iqmvertexarray);
uint valign = (8 - (hdr.filesize%8))%8;
voffset += valign;
hdr.filesize += valign + vdata.length();
hdr.num_vertexes = numfverts;
hdr.num_triangles = triangles.length(); if(triangles.length()) hdr.ofs_triangles = hdr.filesize; hdr.filesize += triangles.length() * sizeof(iqmtriangle);
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if(neighbors.length()) hdr.ofs_adjacency = hdr.filesize, hdr.filesize += neighbors.length() * sizeof(iqmtriangle);
hdr.num_joints = joints.length(); if(joints.length()) hdr.ofs_joints = hdr.filesize; hdr.filesize += joints.length() * sizeof(iqmjoint);
hdr.num_poses = poses.length(); if(poses.length()) hdr.ofs_poses = hdr.filesize; hdr.filesize += poses.length() * sizeof(iqmpose);
hdr.num_anims = anims.length(); if(anims.length()) hdr.ofs_anims = hdr.filesize; hdr.filesize += anims.length() * sizeof(iqmanim);
hdr.num_frames = frames.length(); hdr.num_framechannels = framesize;
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if(animdata.length()) hdr.ofs_frames = hdr.filesize, hdr.filesize += animdata.length() * sizeof(ushort);
if(bounds.length()) hdr.ofs_bounds = hdr.filesize, hdr.filesize += bounds.length() * sizeof(float[8]);
if(commentdata.length()) hdr.ofs_comment = hdr.filesize, hdr.num_comment = commentdata.length(), hdr.filesize += hdr.num_comment;
if (extensions.length()) hdr.ofs_extensions = hdr.filesize, hdr.num_extensions = extensions.length(), hdr.filesize += sizeof(iqmextension) * hdr.num_extensions;
if (ext_meshes_fte) ext_meshes_fte->ofs_data = hdr.filesize, ext_meshes_fte->num_data = meshes_fte.length()*sizeof(iqmext_fte_mesh), hdr.filesize += ext_meshes_fte->num_data;
if (ext_events_fte) ext_events_fte->ofs_data = hdr.filesize, ext_events_fte->num_data = events_fte.length()*sizeof(iqmext_fte_events), hdr.filesize += ext_events_fte->num_data;
lilswap(&hdr.version, (sizeof(hdr) - sizeof(hdr.magic))/sizeof(uint));
f->write(&hdr, sizeof(hdr));
if(stringdata.length()) f->write(stringdata.getbuf(), stringdata.length());
loopv(meshes)
{
mesh &m = meshes[i];
f->putlil(m.name);
f->putlil(m.material);
f->putlil(m.firstvert);
f->putlil(m.numverts);
f->putlil(m.firsttri);
f->putlil(m.numtris);
}
loopv(varrays)
{
vertexarray &v = varrays[i];
f->putlil(v.type);
f->putlil(v.flags);
f->putlil(v.format);
f->putlil(v.size);
f->putlil(voffset + v.offset);
}
loopi(valign) f->putchar(0);
f->write(vdata.getbuf(), vdata.length());
loopv(triangles)
{
triangle &t = triangles[i];
loopk(3) f->putlil(t.vert[k]);
}
loopv(neighbors)
{
triangle &t = neighbors[i];
loopk(3) f->putlil(t.vert[k]);
}
loopv(joints)
{
joint &j = joints[i];
f->putlil(j.name);
f->putlil(j.parent);
loopk(3) f->putlil(float(j.pos[k]));
loopk(4) f->putlil(float(j.orient[k]));
loopk(3) f->putlil(float(j.scale[k]));
}
loopv(poses)
{
pose &p = poses[i];
f->putlil(p.parent);
f->putlil(p.flags);
loopk(10) f->putlil(p.offset[k]);
loopk(10) f->putlil(p.scale[k]);
}
loopv(anims)
{
anim &a = anims[i];
f->putlil(a.name);
f->putlil(a.firstframe);
f->putlil(a.numframes);
f->putlil(a.fps);
f->putlil(a.flags);
}
loopv(animdata) f->putlil(animdata[i]);
loopv(bounds)
{
framebounds &b = bounds[i];
loopk(3) f->putlil(float(b.bbmin[k]));
loopk(3) f->putlil(float(b.bbmax[k]));
f->putlil(float(b.xyradius));
f->putlil(float(b.radius));
}
if(commentdata.length()) f->write(commentdata.getbuf(), commentdata.length());
loopv (extensions)
{
iqmextension &ext = extensions[i];
f->putlil(ext.name);
f->putlil(ext.num_data);
f->putlil(ext.ofs_data);
if (i == extensions.length()-1)
f->putlil(0);
else
f->putlil((uint)(hdr.ofs_extensions + (i+1)*sizeof(ext)));
}
if (ext_meshes_fte) loopv(meshes_fte)
{
meshprop &mf = meshes_fte[i];
f->putlil(mf.contents);
f->putlil(mf.surfaceflags);
f->putlil(mf.body);
f->putlil(mf.geomset);
f->putlil(mf.geomid);
f->putlil(mf.mindist);
f->putlil(mf.maxdist);
}
if (ext_events_fte) loopv(events_fte)
{
event_fte &ev = events_fte[i];
f->putlil(ev.anim);
f->putlil(ev.timestamp);
f->putlil(ev.evcode);
f->putlil(ev.evdata_idx);
}
delete f;
return true;
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
static uchar qmdl_bestnorm(Vec3 &v)
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
#define NUMVERTEXNORMALS 162
static float r_avertexnormals[NUMVERTEXNORMALS][3] = {
#include "anorms.h"
};
uchar best = 0;
float bestdot = -FLT_MAX, dot;
for (size_t i = 0; i < countof(r_avertexnormals); i++)
{
dot = DotProduct(v, r_avertexnormals[i]);
if (dot > bestdot)
{
bestdot = dot;
best = i;
}
}
return best;
}
struct qmdl_vertex_t
{
unsigned char v[3];
unsigned char normalIndex;
};
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
static bool writemdl(const char *filename, bool md16)
{
if (meshes.length() != 1)
{
conoutf("warning: mdl output requires exactly one mesh");
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if (meshes.length() < 0)
return false; //must have ONE mesh only.
else
conoutf("using first...");
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
auto mesh = meshes[0]; //should probably favour the mesh with the most verts or something.
vertexarray *texcoords = NULL;
vertexarray *vertcoords = NULL;
vertexarray *vertnorm = NULL;
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
vertexarray *vertbones = NULL;
vertexarray *vertweights = NULL;
uint skinwidth = 0;
uint skinheight = 0;
Vec3 offset={0,0,0};
Vec3 scale={1,1,1};
uint numskins = 0;
vector<uchar> skindata;
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
unsigned char *paletteddata;
loopv(varrays)
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if(varrays[i].type == IQM_TEXCOORD && varrays[i].format == IQM_FLOAT && varrays[i].size == 2)
texcoords = &varrays[i];
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if(varrays[i].type == IQM_POSITION && varrays[i].format == IQM_FLOAT && varrays[i].size == 3)
vertcoords = &varrays[i];
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if(varrays[i].type == IQM_NORMAL && varrays[i].format == IQM_FLOAT && varrays[i].size == 3)
vertnorm = &varrays[i];
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if(varrays[i].type == IQM_BLENDINDEXES && varrays[i].format == IQM_UBYTE && varrays[i].size == 4)
vertbones = &varrays[i];
if(varrays[i].type == IQM_BLENDWEIGHTS && varrays[i].format == IQM_UBYTE && varrays[i].size == 4)
vertweights = &varrays[i];
}
if (!texcoords)
{
conoutf("warning: mdl output requires a float texcoord array");
return false; //must have some vertex coords...
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if (!vertcoords)
{
conoutf("warning: mdl output requires a suitable vertex positions array...");
return false; //must have some vertex coords...
}
if (!vertnorm)
{
conoutf("warning: mdl output requires a suitable vertex normals array...");
return false; //must have some vertex coords...
}
float *tcdata = (float*)texcoords->vdata.getbuf();
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
//the actual mdl limit is really annoying to calculate.
if (mesh.numverts >= 1024)
conoutf("Writing mdl %s with %u verts exceeds regular limit of %u", filename, mesh.numverts, 1024);
//read the skin...
size_t filesize=0;
qbyte *filedata = NULL;
auto s = openfile(&stringdata[mesh.material], "rb");
if (s)
{
filesize = s->size();
filedata = (qbyte*)malloc(filesize);
s->read(filedata, filesize);
delete s;
}
//decode it...
fte::ImgTool_SetupPalette();
struct pendingtextureinfo *tex = NULL;
if (filedata)
tex = Image_LoadMipsFromMemory(IF_NOMIPMAP, &stringdata[mesh.material], &stringdata[mesh.material], filedata, filesize);
else
conoutf("could not open file %s", &stringdata[mesh.material]);
if (tex)
{ //okay, we have a valid image!
#if 1
//downsize it to work around glquake's limitations. square textures will generally end up 256*256 instead of 512*512 due to that stupid 480 height limit
int newwidth = tex->mip[0].width;
int newheight = tex->mip[0].height;
auto npotup = [](unsigned val)
{ //convert to npot, rounding up.
unsigned scaled = 1;
while(scaled < val)
scaled<<=1;
return scaled;
};
while (newwidth*newheight > 640*480/*GL_Upload8 limit*/ || npotup(newwidth)*npotup(newheight) > 1024*512/*GL_Upload32 limit, may be higher thanks to gl_max_size or gl_picmip but really that sucks*/ || newheight > 480/*Mod_LoadAliasModel limit -- weird MAX_LBM_HEIGHT check*/)
{
newwidth >>= 1;
newheight >>= 1;
}
auto resized = (tex->mip[0].width == newwidth&&tex->mip[0].height == newheight)?NULL:Image_ResampleTexture(tex->encoding, tex->mip[0].data, tex->mip[0].width, tex->mip[0].height, NULL, newwidth, newheight);
if (resized)
{
tex->mip[0].data = resized;
tex->mip[0].datasize = 0; /*o.O*/
tex->mip[0].width = newwidth;
tex->mip[0].height = newheight;
}
#endif
//palettize it, to match the q1 palette.
qboolean allowedformats[PTI_MAX] = {};
allowedformats[PTI_P8]=qtrue;
//FIXME: add hexen2's alpha stuff?
conoutf("Palettizing \"%s\" (%u*%u)", &stringdata[mesh.material], tex->mip[0].width, tex->mip[0].height);
Image_ChangeFormat(tex, allowedformats, PTI_INVALID, "foo");
skinwidth = tex->mip[0].width;
skinheight = tex->mip[0].height;
}
else
{ //texture coords are ints. if we don't have a large enough texture then we don't have much texture coord precision either, so use something reasonable.
skinwidth = 128;
skinheight = 128;
}
paletteddata = skindata.reserve(skinwidth*skinheight);
if (tex)
{
memcpy(paletteddata, tex->mip[0].data, skinwidth*skinheight);
*paletteddata^=1; //try to work around glquake's flood fill...
}
else
{ //fill with some sort of grey.
memset(paletteddata, 8, skinwidth*skinheight);
paletteddata[0] = 7; //work around flood filling...
}
skindata.advance(skinwidth*skinheight);
numskins++;
//we're going to need the transformed pose data, without any bone weights getting in the way.
vector<Vec3> vpos, vnorm;
Vec3 min={FLT_MAX,FLT_MAX,FLT_MAX}, max={-FLT_MAX,-FLT_MAX,-FLT_MAX};
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if (!anims.length())
{
Vec3 *outv = vpos.reserve(mesh.numverts);
Vec3 *outn = vnorm.reserve(mesh.numverts);
auto invert = (float*)vertcoords->vdata.getbuf();
auto innorm = (float*)vertnorm->vdata.getbuf();
// auto inbones = (uchar*)vertbones->vdata.getbuf();
// auto inweights = (uchar*)vertweights->vdata.getbuf();
//FIXME: generate bone matricies from base pose? or just use the vertex data as-is...
for (uint i = mesh.firstvert; i < mesh.firstvert+mesh.numverts; i++, outv++, outn++)
{
//FIXME: generate vert's matrix
//transform each vert
*outv = Vec3(invert[i*3+0], invert[i*3+1], invert[i*3+2]);
//bound it to find the model's extents
for (uint c = 0; c < 3; c++)
{
if (min.v[c] > outv->v[c])
min.v[c] = outv->v[c];
if (max.v[c] < outv->v[c])
max.v[c] = outv->v[c];
}
*outn = Vec3(innorm[i*3+0], innorm[i*3+1], innorm[i*3+2]);
}
vpos.advance(mesh.numverts);
vnorm.advance(mesh.numverts);
}
else loopv(anims)
{
anim &a = anims[i];
Vec3 *outv = vpos.reserve(mesh.numverts*a.numframes);
Vec3 *outn = vnorm.reserve(mesh.numverts*a.numframes);
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
// Matrix3x4 bonepose[joints.length()];
vector<Matrix3x4> bonepose;
bonepose.reserve(joints.length());
auto invert = (float*)vertcoords->vdata.getbuf();
auto innorm = (float*)vertnorm->vdata.getbuf();
auto inbones = vertbones?(uchar*)vertbones->vdata.getbuf():NULL;
auto inweights = vertweights?(uchar*)vertweights->vdata.getbuf():NULL;
if (!inbones || !inweights)
printf("no bone indexes\n");
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
for (uint j = 0; j < a.numframes; j++)
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
//build absolute poses.
frame &fr = frames[a.firstframe+j];
for (int b = 0; b < fr.pose.length(); b++)
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
auto &frpose = fr.pose[b];
bonepose[b] = Matrix3x4(Quat(frpose.tr.orient), Vec3(frpose.tr.pos), Vec3(frpose.tr.scale));
if (frpose.boneparent >= 0)
bonepose[b] = bonepose[frpose.boneparent] * bonepose[b];
}
//done with parents... now we want to invert them
for (int b = 0; b < fr.pose.length(); b++)
{
Matrix3x4 invbind = Matrix3x4(mjoints[b]);
//invbind.invert();
bonepose[b] = bonepose[b] * invbind;
}
for (uint i = mesh.firstvert; i < mesh.numverts; i++, outv++, outn++)
{
//generate per-vert matrix...
Matrix3x4 blend;
blend *= 0;
if (inweights && inbones)
{
if (inweights[i*4+0]) blend += bonepose[inbones[i*4+0]] * (inweights[i*4+0]/255.0);
if (inweights[i*4+1]) blend += bonepose[inbones[i*4+1]] * (inweights[i*4+1]/255.0);
if (inweights[i*4+2]) blend += bonepose[inbones[i*4+2]] * (inweights[i*4+2]/255.0);
if (inweights[i*4+3]) blend += bonepose[inbones[i*4+3]] * (inweights[i*4+3]/255.0);
}
//transform each vert
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
*outv = blend.transform(Vec3(invert[i*3+0], invert[i*3+1], invert[i*3+2]));
//bound it to find the model's extents
for (uint c = 0; c < 3; c++)
{
if (min.v[c] > outv->v[c])
min.v[c] = outv->v[c];
if (max.v[c] < outv->v[c])
max.v[c] = outv->v[c];
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
*outn = blend.transform3(Vec3(innorm[i*3+0], innorm[i*3+1], innorm[i*3+2]));
}
vpos.advance(mesh.numverts);
vnorm.advance(mesh.numverts);
}
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
offset = min;
scale = (max-min)/255; //ignore low order info here
stream *f = openfile(filename, "wb");
if(!f) return false;
if (md16)
f->putlil((uint)(('M'<<0)|('D'<<8)|('1'<<16)|('6'<<24)));
else
f->putlil((uint)(('I'<<0)|('D'<<8)|('P'<<16)|('O'<<24)));
f->putlil((uint)6); //version
f->putlil((float)scale[0]);
f->putlil((float)scale[1]);
f->putlil((float)scale[2]);
f->putlil((float)offset[0]);
f->putlil((float)offset[1]);
f->putlil((float)offset[2]);
f->putlil(0.f); //radius
f->putlil(0.f); //eyeposx, never used afaik
f->putlil(0.f); //eyeposy
f->putlil(0.f); //eyeposz
f->putlil((uint)numskins);
f->putlil((uint)skinwidth);
f->putlil((uint)skinheight);
f->putlil((uint)mesh.numverts);
f->putlil((uint)mesh.numtris);
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if (!anims.length())
f->putlil((uint)1); //numanims
else
f->putlil((uint)anims.length()); //numanims
f->putlil((uint)0); //synctype
f->putlil((uint)modelflags); //flags
f->putlil(0.f); //size
//skins
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
for (uint i = 0; i < numskins; i++)
{
f->putlil((uint)0); //ALIAS_SKIN_SINGLE
f->write(skindata.getbuf()+i*skinwidth*skinheight, skinwidth*skinheight);
}
//texcoords
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
for (uint i = mesh.firstvert; i < mesh.firstvert+mesh.numverts; i++)
{
float s = (tcdata[i*2+0])*skinwidth;
float t = (tcdata[i*2+1])*skinheight;
s -= 0.5; //glquake has some annoying half-texel offset thing.
t -= 0.5;
s = bound(0, s, skinwidth);
t = bound(0, t, skinwidth);
f->putlil((uint)(0?32:0)); //onseam. no verts are ever onseam for us, as we don't do that nonsense here.
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
f->putlil((int)s); //mdl texcoords are ints, in texels. which sucks, but what can you do...
f->putlil((int)t);
}
//tris
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
for (uint i = mesh.firsttri; i < mesh.firsttri+mesh.numtris; i++)
{
f->putlil((uint)1); //faces front. All are effectively front-facing for us. This avoids annoying tc additions.
f->putlil((uint)triangles[i].vert[0]);
f->putlil((uint)triangles[i].vert[1]);
f->putlil((uint)triangles[i].vert[2]);
}
//animations
vector<qmdl_vertex_t> high, low;
size_t voffset = 0;
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if (!anims.length())
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
qmdl_vertex_t *th=high.reserve(mesh.numverts),*tl=low.reserve(mesh.numverts);
for (uint i = mesh.firstvert; i < mesh.numverts; i++, th++, tl++)
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
int l;
for (uint c = 0; c < 3; c++)
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
l = (((vpos[voffset][c]-offset[c])*256) / scale[c]);
if (l<0) l = 0;
if (l > 0xff00) l = 0xff00; //0xffff would exceed the bounds values, so don't use it.
th->v[c] = l>>8;
tl->v[c] = l&0xff;
}
tl->normalIndex = th->normalIndex = qmdl_bestnorm(vnorm[voffset]);
voffset++;
}
high.advance(mesh.numverts);
low.advance(mesh.numverts);
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
voffset = 0;
f->putlil((uint)0); //single-pose type
char name[16]="base";
qmdl_vertex_t min={{255,255,255}}, max={{0,0,0}};
for (uint k = 0; k < mesh.numverts; k++)
{
for (uint c = 0; c < 3; c++)
{
if (min.v[c] > high[voffset+k].v[c])
min.v[c] = high[voffset+k].v[c];
if (max.v[c] < high[voffset+k].v[c])
max.v[c] = high[voffset+k].v[c];
}
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
f->put(min);
f->put(max);
name[countof(name)-1] = 0;
for (uint k = 0; k < countof(name); k++)
f->put(name[k]);
f->write(&high[voffset], sizeof(qmdl_vertex_t)*mesh.numverts);
if (md16)
f->write(&low[voffset], sizeof(qmdl_vertex_t)*mesh.numverts);
voffset += mesh.numverts;
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
else
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
loopv(anims)
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
anim &a = anims[i];
for (uint j = 0; j < a.numframes; j++)
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
qmdl_vertex_t *th=high.reserve(mesh.numverts),*tl=low.reserve(mesh.numverts);
for (uint i = mesh.firstvert; i < mesh.numverts; i++, th++, tl++)
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
int l;
for (uint c = 0; c < 3; c++)
{
l = (((vpos[voffset][c]-offset[c])*256) / scale[c]);
if (l<0) l = 0;
if (l > 0xff00) l = 0xff00; //0xffff would exceed the bounds values, so don't use it.
th->v[c] = l>>8;
tl->v[c] = l&0xff;
}
tl->normalIndex = th->normalIndex = qmdl_bestnorm(vnorm[voffset]);
voffset++;
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
high.advance(mesh.numverts);
low.advance(mesh.numverts);
}
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
voffset = 0;
loopv(anims)
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
anim &a = anims[i];
if (a.numframes == 1)
f->putlil((uint)0); //single-pose type
else
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
f->putlil((uint)1); //anim type
f->putlil((uint)a.numframes);
qmdl_vertex_t min={{255,255,255}}, max={{0,0,0}};
for (uint k = 0; k < mesh.numverts*a.numframes; k++)
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
for (uint c = 0; c < 3; c++)
{
if (min.v[c] > high[voffset+k].v[c])
min.v[c] = high[voffset+k].v[c];
if (max.v[c] < high[voffset+k].v[c])
max.v[c] = high[voffset+k].v[c];
}
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
f->put(min);
f->put(max);
for (uint j = 0; j < a.numframes; j++)
f->putlil(1.0f/a.fps); //intervals. we use the same value for each
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
for (uint j = 0; j < a.numframes; j++)
{
char name[16]={0};
qmdl_vertex_t min={{255,255,255}}, max={{0,0,0}};
for (uint k = 0; k < mesh.numverts; k++)
{
for (uint c = 0; c < 3; c++)
{
if (min.v[c] > high[voffset+k].v[c])
min.v[c] = high[voffset+k].v[c];
if (max.v[c] < high[voffset+k].v[c])
max.v[c] = high[voffset+k].v[c];
}
}
f->put(min);
f->put(max);
strncpy(name, &stringdata[a.name], sizeof(name));
name[countof(name)-1] = 0;
for (uint k = 0; k < countof(name); k++)
f->put(name[k]);
f->write(&high[voffset], sizeof(qmdl_vertex_t)*mesh.numverts);
if (md16)
f->write(&low[voffset], sizeof(qmdl_vertex_t)*mesh.numverts);
voffset += mesh.numverts;
}
}
}
delete f;
return true;
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
static bool writeqmdl(const char *filename)
{
return writemdl(filename, false);
}
static bool writemd16(const char *filename)
{
return writemdl(filename, true);
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
static bool writemd3(const char *filename)
{
fprintf(stderr, "writemd3 is not implemented yet\n");
return false;
}
void help(bool exitstatus = EXIT_SUCCESS)
{
fprintf(exitstatus != EXIT_SUCCESS ? stderr : stdout,
"-- FTE's Fork of Lee Salzman's iqm exporter --\n"
"Usage:\n"
"\n"
"./iqm cmdfile.cmd\n"
"./iqm [options] output.iqm mesh.iqe anim1.iqe ... animN.iqe\n"
"./iqm [options] output.iqm mesh.md5mesh anim1.md5anim ... animN.md5anim\n"
"./iqm [options] output.iqm mesh.smd anim1.smd ... animN.smd\n"
"./iqm [options] output.iqm mesh.fbx anim1.fbx ... animN.fbx\n"
"./iqm [options] output.iqm mesh.obj\n"
"\n"
"For certain formats, IQE, OBJ, and FBX, it is possible to combine multiple mesh\n"
"files of the exact same vertex layout and skeleton by supplying them as\n"
"\"mesh1.iqe,mesh2.iqe,mesh3.iqe\", that is, a comma-separated list of the mesh\n"
"files (with no spaces) in place of the usual mesh filename.\n"
"\n"
"Options can be any of the following command-line switches:\n"
"\n"
" -s N\n"
" --scale N\n"
" Sets the output scale to N (float).\n"
"\n"
" --meshtrans Z\n"
" --meshtrans X,Y,Z\n"
" Translates a mesh by X,Y,Z (floats). This does not affect the skeleton.\n"
"\n"
" -j\n"
" --forcejoints\n"
" Forces the exporting of joint information in animation files without\n"
" meshes.\n"
"\n"
" -q\n"
" Quiet. Only display warnings or errors.\n"
"\n"
" -v\n"
" Verbose. Print lots of extra info.\n"
"\n"
" -n\n"
" No Extensions. Disables the use of fte-specific iqm extensions.\n"
"\n"
"Each animation file can be preceded by any combination of the following command-\n"
"line switches:\n"
"\n"
" --name A\n"
" Sets the name of the animation to A.\n"
" --fps N\n"
" Sets the FPS of the animation to N (float).\n"
" --loop\n"
" Sets the loop flag for the animation.\n"
" --start N\n"
" Sets the first frame of the animation to N (integer).\n"
" --end N\n"
" Sets the last frame of the animation to N (integer).\n"
"\n"
"You can supply either a mesh file, animation files, or both.\n"
"Note that if an input mesh file is supplied, it must come before the animation\n"
"files in the file list.\n"
"The output IQM file will contain the supplied mesh and any supplied animations.\n"
"If no mesh is provided,the IQM file will simply contain the supplied animations.\n"
);
exit(exitstatus);
}
struct bitnames
{
const char *std;
const char *name;
unsigned int bits;
};
//chops up the input string, returning subsections of it, like strtok_r, except with more specific separators.
char *mystrtok(char **ctx)
{
char *ret = NULL;
char *p = *ctx;
//skip whitespace
while (*p == ' ' || *p == '\t' || *p == '\r' || *p == '\n')
p++;
if (!*p)
return NULL; //eof
if (*p == '\"')
{
ret = ++p;
while (*p && *p != '\"')
p++;
if (*p)
*p++ = '\0';
*ctx = p;
}
else
{
ret = p;
//we're screwed if we reach a quote without trailing whitespace, blame the user in that case.
while (*p && *p != ' ' && *p != '\t' && *p != '\r' && *p != '\n')
p++;
if (*p)
*p++ = '\0';
*ctx = p;
}
return ret;
}
unsigned int parsebits(bitnames *names, char **line)
{
unsigned int bits = 0;
char *comma;
for (char *value = mystrtok(line); value; value = comma)
{
comma = strchr(value, ',');
if (comma)
*comma++ = 0;
char *end;
strtoul(value, &end, 0);
if (end && !*end)
bits |= strtoul(value, NULL, 0);
else
{
size_t i;
char *std = value;
value = strchr(value, '_');
if (value)
*value++ = 0;
else
{
value = std;
std += strlen(std);
}
for (i = 0; names[i].name; i++)
{
if (!*std || !strcasecmp(names[i].std, std))
{
if (!strcasecmp(names[i].name, value))
{
bits |= names[i].bits;
break;
}
}
}
if (!names[i].name)
{ //stuff with no specific standard, mostly for consistency
if (!*names[i].std)
{
if (!strcasecmp(names[i].name, value))
{
bits |= names[i].bits;
break;
}
}
if (!names[i].name)
fatal("Unknown bit name: %s\n", value);
}
}
}
return bits;
}
bool parsebonefield(const char *tok, char **line, boneoverride::prop &spec, bool defaults)
{
if (!strcasecmp(tok, "rename"))
spec.rename = newstring(mystrtok(line));
else if (!strcasecmp(tok, "group"))
spec.group = atoi(mystrtok(line));
else
return false;
return true;
}
bool parsemeshfield(const char *tok, char **line, meshprop &spec, bool defaults)
{
if (!strcasecmp(tok, "contents"))
{
//these should be (mostly) compatible with q2+q3
bitnames contentnames[] = {
{"", "empty", 0x00000000},
{"", "solid", 0x00000001},
{"", "lava", 0x00000008},
{"", "slime", 0x00000010},
{"", "water", 0x00000020},
{"", "fluid", 0x00000038},
{"fte", "ladder", 0x00004000},
{"", "playerclip", 0x00010000},
{"", "monsterclip", 0x00010000},
{"", "body", 0x02000000},
{"", "corpse", 0x04000000},
{"q2", "ladder", 0x20000000},
{"fte", "sky", 0x80000000},{"q3", "nodrop", 0x80000000},
{NULL}
};
spec.contents = parsebits(contentnames, line);
}
else if (!strcasecmp(tok, "surfaceflags"))
{
bitnames surfaceflagnames[] = {
{"fte", "nodraw", 0x00000080},{"q3", "nodraw", 0x00000080},
{NULL}
};
spec.surfaceflags = parsebits(surfaceflagnames, line);
}
else if (!strcasecmp(tok, "body"))
spec.body = strtoul(mystrtok(line), NULL, 0);
else if (!strcasecmp(tok, "geomset"))
{
spec.geomset = strtoul(mystrtok(line), NULL, 0);
spec.geomid = strtoul(mystrtok(line), NULL, 0);
}
else if (!strcasecmp(tok, "lodrange"))
{
spec.mindist = atof(mystrtok(line));
spec.maxdist = atof(mystrtok(line));
}
else
return false;
return true;
}
bool parseanimfield(const char *tok, char **line, filespec &spec, bool defaults)
{
if (!strcasecmp(tok, "name") && !defaults)
spec.name = newstring(mystrtok(line));
else if (!strcasecmp(tok, "fps"))
spec.fps = atof(mystrtok(line));
else if (!strcasecmp(tok, "loop"))
spec.flags |= IQM_LOOP;
else if (!strcasecmp(tok, "clamp"))
spec.flags &= ~IQM_LOOP;
else if (!strcasecmp(tok, "unpack"))
spec.flags |= IQM_UNPACK;
else if (!strcasecmp(tok, "pack"))
spec.flags &= ~IQM_UNPACK;
else if (!strcasecmp(tok, "nomesh"))
spec.nomesh = !!strtoul(mystrtok(line), NULL, 0);
else if (!strcasecmp(tok, "noanim"))
spec.noanim = !!strtoul(mystrtok(line), NULL, 0);
else if (!strcasecmp(tok, "materialprefix"))
spec.materialprefix = newstring(mystrtok(line));
else if(!strcasecmp(tok, "start"))
spec.startframe = max(atoi(mystrtok(line)), 0);
else if(!strcasecmp(tok, "end"))
spec.endframe = atoi(mystrtok(line));
else if (!strcasecmp(tok, "rotate"))
{
Vec3 ang;
ang.x = atof(mystrtok(line))*-M_PI/180;
ang.z = atof(mystrtok(line))*-M_PI/180;
ang.y = atof(mystrtok(line))*-M_PI/180;
spec.rotate = Quat::fromangles(ang);
}
else if (!strcasecmp(tok, "scale"))
spec.scale = atof(mystrtok(line));
else if (!strcasecmp(tok, "origin"))
{
spec.translate.x = atof(mystrtok(line));
spec.translate.y = atof(mystrtok(line));
spec.translate.z = atof(mystrtok(line));
}
else if (!strcasecmp(tok, "event"))
{
const char *poseidx = mystrtok(line);
char *dot;
if (!strcmp(poseidx, "reset"))
{
spec.events.setsize(0);
return true;
}
event_fte &ev = spec.events.add();
ev.anim = strtod(poseidx, &dot);
if (*dot == ':')
ev.timestamp = strtoul(dot+1, NULL, 0);
else
{
ev.timestamp = strtod(poseidx, &dot);
ev.anim = ~0u; //fix up according to poses...
}
ev.evcode = atoi(mystrtok(line));
ev.evdata_str = newstring(mystrtok(line));
}
else if (parsemeshfield(tok, line, spec.meshprops, defaults))
;
else
return false;
return true;
}
struct
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
const char *extname;
bool (*write)(const char *filename);
const char *cmdname;
const char *altcmdname;
} outputtypes[] =
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
{".vvm", writeiqm, "output_vvm"},
{".iqm", writeiqm, "output_iqm"},
{".mdl", writeqmdl, "output_qmdl"},
{".md16", writemd16, "output_md16"},
{".md3", writemd3, "output_md3"},
};
void parsecommands(char *filename, const char *outfiles[countof(outputtypes)], vector<filespec> &infiles, vector<hitbox> &hitboxes)
{
filespec defaultspec;
defaultspec.reset();
if (!quiet)
conoutf("execing %s", filename);
stream *f = openfile(filename, "rt");
if(!f)
{
fatal("Couldn't open command-file \"%s\"\n", filename);
return;
}
char buf[2048];
while(f->getline(buf, sizeof(buf)))
{
const char *tok;
char *line = buf;
tok = mystrtok(&line);
if (tok && *tok == '$')
tok++;
if (!tok)
continue;
else if (*tok == '#' || !strncasecmp(tok, "//", 2))
{ //comments
while (mystrtok(&line))
;
}
// else if (!strcasecmp(tok, "outputdir"))
// (void)mystrtok(&line);
else if (!strcasecmp(tok, "hitbox") || !strcasecmp(tok, "hbox"))
{
hitbox &hb = hitboxes.add();
hb.body = strtoul(mystrtok(&line), NULL, 0);
hb.bone = newstring(mystrtok(&line));
for (int i = 0; i < 3; i++)
hb.mins[i] = atof(mystrtok(&line));
for (int i = 0; i < 3; i++)
hb.maxs[i] = atof(mystrtok(&line));
}
else if (!strcasecmp(tok, "exec"))
parsecommands(mystrtok(&line), outfiles, infiles, hitboxes);
else if (!strcasecmp(tok, "modelflags"))
{
bitnames modelflagnames[] = {
{"q1", "rocket", 0x01},
{"q1", "grenade", 0x02},
{"q1", "gib", 0x04},
{"q1", "rotate", 0x08},
{"q1", "tracer1", 0x10},
{"q1", "zomgib", 0x20},
{"q1", "tracer2", 0x40},
{"q1", "tracer3", 0x80},
{NULL}
};
modelflags = parsebits(modelflagnames, &line);
}
else if (parseanimfield(tok, &line, defaultspec, true))
;
else if (!strcasecmp(tok, "mesh"))
{
meshoverride &mo = meshoverrides.add();
mo.name = newstring(mystrtok(&line));
mo.props = defaultspec.meshprops;
while(( tok = mystrtok(&line)))
{
//fixme: should probably separate this out.
if (parsemeshfield(tok, &line, mo.props, false))
;
else
{
printf("unknown mesh token \"%s\"\n", tok);
break;
}
}
}
else if (!strcasecmp(tok, "bone"))
{
boneoverride &mo = boneoverrides.add();
mo.name = newstring(mystrtok(&line));
mo.props = boneoverride::prop();
while(( tok = mystrtok(&line)))
{
//fixme: should probably separate this out.
if (parsebonefield(tok, &line, mo.props, false))
;
else
{
printf("unknown mesh token \"%s\"\n", tok);
break;
}
}
}
else if (!strcasecmp(tok, "import") || !strcasecmp(tok, "model") || !strcasecmp(tok, "scene") || !strcasecmp(tok, "animation"))
{
filespec inspec = defaultspec;
//first token is always the filename(s)
inspec.file = newstring(mystrtok(&line));
while(( tok = mystrtok(&line)))
{
if (parseanimfield(tok, &line, inspec, false))
;
else
{
printf("unknown scene token \"%s\"\n", tok);
break;
}
}
infiles.add(inspec);
}
else
{
size_t n, j;
if (!strcasecmp(tok, "output"))
tok = "output_iqm";
for (n = 0; n < countof(outputtypes); n++)
{
if (!strcasecmp(tok, outputtypes[n].cmdname))
{
outfiles[n] = newstring(mystrtok(&line));
for (j = 0; j < countof(outputtypes); j++)
{
if (n!=j && outfiles[j] && !strcasecmp(outfiles[n], outfiles[j]))
{
printf("cancelling %s\n", outputtypes[j].cmdname);
outfiles[j] = NULL;
break;
}
}
tok = "";
break;
}
}
if (*tok)
{
printf("unsupported command \"%s\"\n", tok);
continue;
}
}
if ((tok=mystrtok(&line)))
if (*tok)
printf("unexpected junk at end-of-line \"%s\" \"%s\"\n", buf, tok);
}
delete f;
}
int main(int argc, char **argv)
{
if(argc <= 1) help(EXIT_FAILURE);
vector<filespec> infiles;
vector<hitbox> hitboxes;
filespec inspec;
const char *outfiles[countof(outputtypes)] = {};
for(int i = 1; i < argc; i++)
{
if(argv[i][0] == '-')
{
if(argv[i][1] == '-')
{
if(!strcasecmp(&argv[i][2], "cmd")) { if(i + 1 < argc) parsecommands(argv[++i], outfiles, infiles, hitboxes); }
else if(!strcasecmp(&argv[i][2], "noext")) noext = true;
else if(!strcasecmp(&argv[i][2], "fps")) { if(i + 1 < argc) inspec.fps = atof(argv[++i]); }
else if(!strcasecmp(&argv[i][2], "name")) { if(i + 1 < argc) inspec.name = argv[++i]; }
else if(!strcasecmp(&argv[i][2], "loop")) { inspec.flags |= IQM_LOOP; }
else if(!strcasecmp(&argv[i][2], "start")) { if(i + 1 < argc) inspec.startframe = max(atoi(argv[++i]), 0); }
else if(!strcasecmp(&argv[i][2], "end")) { if(i + 1 < argc) inspec.endframe = atoi(argv[++i]); }
else if(!strcasecmp(&argv[i][2], "scale")) { if(i + 1 < argc) inspec.scale = clamp(atof(argv[++i]), 1e-8, 1e8); }
else if(!strcasecmp(&argv[i][2], "help")) help();
else if(!strcasecmp(&argv[i][2], "forcejoints")) forcejoints = true;
else if(!strcasecmp(&argv[i][2], "meshtrans"))
{
if(i + 1 < argc) switch(sscanf(argv[++i], "%lf , %lf , %lf", &gmeshtrans.x, &gmeshtrans.y, &gmeshtrans.z))
{
case 1: gmeshtrans = Vec3(0, 0, gmeshtrans.x); break;
}
}
}
else switch(argv[i][1])
{
case 'h':
help();
break;
case 's':
if(i + 1 < argc) gscale = clamp(atof(argv[++i]), 1e-8, 1e8);
break;
case 'j':
forcejoints = true;
break;
case 'v':
verbose = true;
break;
case 'q':
quiet = true;
break;
case 'n':
noext = true;
break;
}
}
else
{
const char *type = strrchr(argv[i], '.');
if (type && (!strcasecmp(type, ".cmd")||!strcasecmp(type, ".cfg")||!strcasecmp(type, ".txt")||!strcasecmp(type, ".qc"))) //.qc to humour halflife fanboys
parsecommands(argv[i], outfiles, infiles, hitboxes);
else
{
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
size_t j;
for (j = 0; j < countof(outfiles); j++)
if (outfiles[j])
break;
if(j == countof(outfiles))
{
for (j = countof(outfiles); j --> 0; )
if (type && !strcasecmp(type, outputtypes[j].extname))
break;
outfiles[j] = argv[i]; //first arg is the output name, if its not an export script thingie.
}
else
{
infiles.add(inspec).file = argv[i];
inspec.reset();
}
}
}
}
size_t n;
for (n = 0; n < countof(outputtypes) && !outfiles[n]; n++);
if(n == countof(outfiles)) fatal("no output file specified");
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
if(infiles.empty())
{
if (outfiles[0])
{
inspec.reset();
infiles.add(inspec).file = outfiles[0];
outfiles[0] = NULL;
}
else
fatal("no input files specified");
}
if(gscale != 1) printf("scale: %f\n", escale);
if(gmeshtrans != Vec3(0, 0, 0)) printf("mesh translate: %f, %f, %f\n", gmeshtrans.x, gmeshtrans.y, gmeshtrans.z);
loopv(infiles)
{
const filespec &inspec = infiles[i];
const char *infile = inspec.file, *type = strrchr(infile, '.');
if (verbose)
conoutf("importing %s", infile);
if(!type) fatal("no file type: %s", infile);
else if(!strcasecmp(type, ".md5mesh"))
{
if(!loadmd5mesh(infile, inspec)) fatal("failed reading: %s", infile);
}
else if(!strcasecmp(type, ".md5anim"))
{
if(!loadmd5anim(infile, inspec)) fatal("failed reading: %s", infile);
}
else if(!strcasecmp(type, ".iqe"))
{
if(!loadiqe(infile, inspec)) fatal("failed reading: %s", infile);
}
else if(!strcasecmp(type, ".smd"))
{
if(!loadsmd(infile, inspec)) fatal("failed reading: %s", infile);
}
else if(!strcasecmp(type, ".fbx"))
{
if(!loadfbx(infile, inspec)) fatal("failed reading: %s", infile);
}
else if(!strcasecmp(type, ".obj"))
{
if(!loadobj(infile, inspec)) fatal("failed reading: %s", infile);
}
makefile: attempt to fix freetype when not using makelibs (should make it slightly easier for people to compile with msys2 without needing to resort to cmake). emenu: clean up hexen2's maplist options slightly. emenu: modelviewer should now be slightly more friendly (click+wasd to move around). particles: fix up randomised s coords. csqc: try to fix issue with applycustomskin not refcounting properly. client: [s_]precache and (new) mod_precache cvars can be set to 2 to precache the resources after load, for faster loading at the expense of some early stutter, without risking later mid-game stuttering. gltf: add support for morphweights in a cpu-fallback path. don't expect good performance on surfaces with morphtargets for now. gtlf: add some support for gltf1 files. far from perfect. shaders: gltf1 semantics handling shaders: const correctness iqmtool: fix up mdl skin export. iqmtool: integrate the engine's gltf2 loader. works with animated models, but unanimated ones suffer from basepose-different-from-bindpose issues. q3bsp: hopefully fixed bih traces. still disabled for now. qc: change default value of pr_gc_threaded to 1. qcext: add the '__deprecated' keyword to various symbols in fteextensions.qc, now that fteqcc supports it. ssqc: spit out a more readable error for WriteByte(MSG_CSQC,...) outside of SendEntity. ssqc: add registercommand builtin, for consistency with menuqc and csqc (though only one can register any single command). sv: report userinfo/serverinfo sizes (some clients still have arbitrary limits, plus its nice to see how abusive things are) sv: try to optimise sv_cullentities_trace a little. movechain: relink moved ents. csqc: add spriteframe builtin, for freecs to use instead of more ugly less reliable hacks. menuqc: fopen("tls://host:port", FILE_STREAM) should now open a tls stream. tcp:// should also work. git-svn-id: https://svn.code.sf.net/p/fteqw/code/trunk@5704 fc73d0e0-1445-4013-8a0c-d673dee63da5
2020-06-13 00:05:47 +00:00
else if(!strcasecmp(type, ".glb"))
{
if(!fte::loadglb(infile, inspec)) fatal("failed reading: %s", infile);
}
else if(!strcasecmp(type, ".gltf"))
{
if(!fte::loadgltf(infile, inspec)) fatal("failed reading: %s", infile);
}
else fatal("unknown file type: %s", type);
}
genhitboxes(hitboxes);
loopv(boneoverrides) if (!boneoverrides[i].used)
conoutf("warning: bone \"%s\" overriden, but not present", boneoverrides[i].name);
if (noext && meshoverrides.length())
conoutf("warning: mesh overrides used, but iqm extensions disabled");
else loopv(meshoverrides)
conoutf("warning: mesh \"%s\" overriden, but not present", meshoverrides[i].name);
calcanimdata();
conoutf("bone list:");
printbones();
// printbonelist();
if (!quiet)
conoutf("");
for (size_t n = 0; n < countof(outputtypes); n++)
{
if (outfiles[n] != NULL)
{
if(outputtypes[n].write(outfiles[n]))
{
if (!quiet)
conoutf("exported: %s", outfiles[n]);
}
else fatal("failed writing: %s", outfiles[n]);
}
}
return EXIT_SUCCESS;
}