da09f7ba44
git-svn-id: https://svn.code.sf.net/p/fteqw/code/branches/wip@3656 fc73d0e0-1445-4013-8a0c-d673dee63da5
2517 lines
70 KiB
C
2517 lines
70 KiB
C
#include "quakedef.h"
|
|
#ifdef D3DQUAKE
|
|
#include "shader.h"
|
|
#include <d3d9.h>
|
|
|
|
#ifndef GLQUAKE
|
|
/*shaders have a few GL_FOO constants in them. they shouldn't, but they do.*/
|
|
#include <GL/gl.h>
|
|
#include "glsupp.h"
|
|
#endif
|
|
|
|
extern LPDIRECT3DDEVICE9 pD3DDev9;
|
|
|
|
//#define d3dcheck(foo) foo
|
|
#define d3dcheck(foo) do{HRESULT err = foo; if (FAILED(err)) Sys_Error("D3D reported error on backend line %i - error 0x%x\n", __LINE__, err);} while(0)
|
|
|
|
#define MAX_TMUS 4
|
|
|
|
/*========================================== tables for deforms =====================================*/
|
|
#define frand() (rand()*(1.0/RAND_MAX))
|
|
#define FTABLE_SIZE 1024
|
|
#define FTABLE_CLAMP(x) (((int)((x)*FTABLE_SIZE) & (FTABLE_SIZE-1)))
|
|
#define FTABLE_EVALUATE(table,x) (table ? table[FTABLE_CLAMP(x)] : frand()*((x)-floor(x)))
|
|
|
|
static float r_sintable[FTABLE_SIZE];
|
|
static float r_triangletable[FTABLE_SIZE];
|
|
static float r_squaretable[FTABLE_SIZE];
|
|
static float r_sawtoothtable[FTABLE_SIZE];
|
|
static float r_inversesawtoothtable[FTABLE_SIZE];
|
|
|
|
static float *FTableForFunc ( unsigned int func )
|
|
{
|
|
switch (func)
|
|
{
|
|
case SHADER_FUNC_SIN:
|
|
return r_sintable;
|
|
|
|
case SHADER_FUNC_TRIANGLE:
|
|
return r_triangletable;
|
|
|
|
case SHADER_FUNC_SQUARE:
|
|
return r_squaretable;
|
|
|
|
case SHADER_FUNC_SAWTOOTH:
|
|
return r_sawtoothtable;
|
|
|
|
case SHADER_FUNC_INVERSESAWTOOTH:
|
|
return r_inversesawtoothtable;
|
|
}
|
|
|
|
//bad values allow us to crash (so I can debug em)
|
|
return NULL;
|
|
}
|
|
|
|
static void FTable_Init(void)
|
|
{
|
|
unsigned int i;
|
|
double t;
|
|
for (i = 0; i < FTABLE_SIZE; i++)
|
|
{
|
|
t = (double)i / (double)FTABLE_SIZE;
|
|
|
|
r_sintable[i] = sin(t * 2*M_PI);
|
|
|
|
if (t < 0.25)
|
|
r_triangletable[i] = t * 4.0;
|
|
else if (t < 0.75)
|
|
r_triangletable[i] = 2 - 4.0 * t;
|
|
else
|
|
r_triangletable[i] = (t - 0.75) * 4.0 - 1.0;
|
|
|
|
if (t < 0.5)
|
|
r_squaretable[i] = 1.0f;
|
|
else
|
|
r_squaretable[i] = -1.0f;
|
|
|
|
r_sawtoothtable[i] = t;
|
|
r_inversesawtoothtable[i] = 1.0 - t;
|
|
}
|
|
}
|
|
|
|
typedef vec3_t mat3_t[3];
|
|
static mat3_t axisDefault={{1, 0, 0},
|
|
{0, 1, 0},
|
|
{0, 0, 1}};
|
|
|
|
static void Matrix3_Transpose (mat3_t in, mat3_t out)
|
|
{
|
|
out[0][0] = in[0][0];
|
|
out[1][1] = in[1][1];
|
|
out[2][2] = in[2][2];
|
|
|
|
out[0][1] = in[1][0];
|
|
out[0][2] = in[2][0];
|
|
out[1][0] = in[0][1];
|
|
out[1][2] = in[2][1];
|
|
out[2][0] = in[0][2];
|
|
out[2][1] = in[1][2];
|
|
}
|
|
static void Matrix3_Multiply_Vec3 (const mat3_t a, const vec3_t b, vec3_t product)
|
|
{
|
|
product[0] = a[0][0]*b[0] + a[0][1]*b[1] + a[0][2]*b[2];
|
|
product[1] = a[1][0]*b[0] + a[1][1]*b[1] + a[1][2]*b[2];
|
|
product[2] = a[2][0]*b[0] + a[2][1]*b[1] + a[2][2]*b[2];
|
|
}
|
|
|
|
static int Matrix3_Compare(const mat3_t in, const mat3_t out)
|
|
{
|
|
return !memcmp(in, out, sizeof(mat3_t));
|
|
}
|
|
|
|
/*================================================*/
|
|
|
|
typedef struct
|
|
{
|
|
backendmode_t mode;
|
|
unsigned int flags;
|
|
|
|
float curtime;
|
|
const entity_t *curentity;
|
|
shader_t *curshader;
|
|
texnums_t *curtexnums;
|
|
texid_t curlightmap;
|
|
texid_t curdeluxmap;
|
|
int curvertdecl;
|
|
unsigned int shaderbits;
|
|
unsigned int curcull;
|
|
unsigned int lastpasscount;
|
|
|
|
texid_t curtex[MAX_TMUS];
|
|
unsigned int tmuflags[MAX_TMUS];
|
|
|
|
mesh_t **meshlist;
|
|
unsigned int nummeshes;
|
|
|
|
D3DCOLOR passcolour;
|
|
qboolean passsinglecolour;
|
|
|
|
/*FIXME: we shouldn't lock these so much - we need to cache which batches have been submitted and set up streams separately from the vertex data*/
|
|
IDirect3DVertexBuffer9 *dynxyz_buff;
|
|
unsigned int dynxyz_offs;
|
|
unsigned int dynxyz_size;
|
|
|
|
IDirect3DVertexBuffer9 *dynst_buff[MAX_TMUS];
|
|
unsigned int dynst_offs[MAX_TMUS];
|
|
unsigned int dynst_size;
|
|
|
|
IDirect3DVertexBuffer9 *dyncol_buff;
|
|
unsigned int dyncol_offs;
|
|
unsigned int dyncol_size;
|
|
|
|
IDirect3DIndexBuffer9 *dynidx_buff;
|
|
unsigned int dynidx_offs;
|
|
unsigned int dynidx_size;
|
|
|
|
unsigned int wbatch;
|
|
unsigned int maxwbatches;
|
|
batch_t *wbatches;
|
|
} d3dbackend_t;
|
|
|
|
#define DYNVBUFFSIZE 65536
|
|
#define DYNIBUFFSIZE 65536
|
|
|
|
static d3dbackend_t shaderstate;
|
|
|
|
extern int be_maxpasses;
|
|
|
|
enum
|
|
{
|
|
D3D_VDEC_COL4B = 1<<0,
|
|
//D3D_VDEC_NORMS = 1<<1,
|
|
D3D_VDEC_ST0 = 1<<1,
|
|
D3D_VDEC_ST1 = 1<<2,
|
|
D3D_VDEC_ST2 = 1<<3,
|
|
D3D_VDEC_ST3 = 1<<4,
|
|
D3D_VDEC_MAX = 1<<5
|
|
};
|
|
IDirect3DVertexDeclaration9 *vertexdecls[D3D_VDEC_MAX];
|
|
|
|
|
|
void BE_D3D_Reset(qboolean before)
|
|
{
|
|
int i, tmu;
|
|
if (before)
|
|
{
|
|
IDirect3DDevice9_SetVertexDeclaration(pD3DDev9, NULL);
|
|
shaderstate.curvertdecl = 0;
|
|
for (i = 0; i < 5+MAX_TMUS; i++)
|
|
IDirect3DDevice9_SetStreamSource(pD3DDev9, i, NULL, 0, 0);
|
|
IDirect3DDevice9_SetIndices(pD3DDev9, NULL);
|
|
|
|
if (shaderstate.dynxyz_buff)
|
|
IDirect3DVertexBuffer9_Release(shaderstate.dynxyz_buff);
|
|
shaderstate.dynxyz_buff = NULL;
|
|
for (tmu = 0; tmu < MAX_TMUS; tmu++)
|
|
{
|
|
if (shaderstate.dynst_buff[tmu])
|
|
IDirect3DVertexBuffer9_Release(shaderstate.dynst_buff[tmu]);
|
|
shaderstate.dynst_buff[tmu] = NULL;
|
|
}
|
|
if (shaderstate.dyncol_buff)
|
|
IDirect3DVertexBuffer9_Release(shaderstate.dyncol_buff);
|
|
shaderstate.dyncol_buff = NULL;
|
|
if (shaderstate.dynidx_buff)
|
|
IDirect3DIndexBuffer9_Release(shaderstate.dynidx_buff);
|
|
shaderstate.dynidx_buff = NULL;
|
|
|
|
for (i = 0; i < D3D_VDEC_MAX; i++)
|
|
{
|
|
if (vertexdecls[i])
|
|
IDirect3DVertexDeclaration9_Release(vertexdecls[i]);
|
|
vertexdecls[i] = NULL;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
D3DVERTEXELEMENT9 decl[8], declend=D3DDECL_END();
|
|
int elements;
|
|
|
|
for (i = 0; i < D3D_VDEC_MAX; i++)
|
|
{
|
|
elements = 0;
|
|
decl[elements].Stream = 0;
|
|
decl[elements].Offset = 0;
|
|
decl[elements].Type = D3DDECLTYPE_FLOAT3;
|
|
decl[elements].Method = D3DDECLMETHOD_DEFAULT;
|
|
decl[elements].Usage = D3DDECLUSAGE_POSITION;
|
|
decl[elements].UsageIndex = 0;
|
|
elements++;
|
|
|
|
if (i & D3D_VDEC_COL4B)
|
|
{
|
|
decl[elements].Stream = 1;
|
|
decl[elements].Offset = 0;
|
|
decl[elements].Type = D3DDECLTYPE_D3DCOLOR;
|
|
decl[elements].Method = D3DDECLMETHOD_DEFAULT;
|
|
decl[elements].Usage = D3DDECLUSAGE_COLOR;
|
|
decl[elements].UsageIndex = 0;
|
|
elements++;
|
|
}
|
|
|
|
/* if (i & D3D_VDEC_NORMS)
|
|
{
|
|
decl[elements].Stream = 2;
|
|
decl[elements].Offset = 0;
|
|
decl[elements].Type = D3DDECLTYPE_FLOAT2;
|
|
decl[elements].Method = D3DDECLMETHOD_DEFAULT;
|
|
decl[elements].Usage = D3DDECLUSAGE_TEXCOORD;
|
|
decl[elements].UsageIndex = 1;
|
|
elements++;
|
|
|
|
decl[elements].Stream = 3;
|
|
decl[elements].Offset = 0;
|
|
decl[elements].Type = D3DDECLTYPE_FLOAT2;
|
|
decl[elements].Method = D3DDECLMETHOD_DEFAULT;
|
|
decl[elements].Usage = D3DDECLUSAGE_TEXCOORD;
|
|
decl[elements].UsageIndex = 1;
|
|
elements++;
|
|
|
|
decl[elements].Stream = 4;
|
|
decl[elements].Offset = 0;
|
|
decl[elements].Type = D3DDECLTYPE_FLOAT2;
|
|
decl[elements].Method = D3DDECLMETHOD_DEFAULT;
|
|
decl[elements].Usage = D3DDECLUSAGE_TEXCOORD;
|
|
decl[elements].UsageIndex = 1;
|
|
elements++;
|
|
}
|
|
*/
|
|
for (tmu = 0; tmu < MAX_TMUS; tmu++)
|
|
{
|
|
if (i & (D3D_VDEC_ST0<<tmu))
|
|
{
|
|
decl[elements].Stream = 5+tmu;
|
|
decl[elements].Offset = 0;
|
|
decl[elements].Type = D3DDECLTYPE_FLOAT2;
|
|
decl[elements].Method = D3DDECLMETHOD_DEFAULT;
|
|
decl[elements].Usage = D3DDECLUSAGE_TEXCOORD;
|
|
decl[elements].UsageIndex = tmu;
|
|
elements++;
|
|
}
|
|
}
|
|
|
|
decl[elements] = declend;
|
|
elements++;
|
|
|
|
IDirect3DDevice9_CreateVertexDeclaration(pD3DDev9, decl, &vertexdecls[i]);
|
|
}
|
|
|
|
IDirect3DDevice9_CreateVertexBuffer(pD3DDev9, shaderstate.dynxyz_size, D3DUSAGE_DYNAMIC|D3DUSAGE_WRITEONLY, 0, D3DPOOL_DEFAULT, &shaderstate.dynxyz_buff, NULL);
|
|
for (tmu = 0; tmu < MAX_TMUS; tmu++)
|
|
IDirect3DDevice9_CreateVertexBuffer(pD3DDev9, shaderstate.dynst_size, D3DUSAGE_DYNAMIC|D3DUSAGE_WRITEONLY, 0, D3DPOOL_DEFAULT, &shaderstate.dynst_buff[tmu], NULL);
|
|
IDirect3DDevice9_CreateVertexBuffer(pD3DDev9, shaderstate.dyncol_size, D3DUSAGE_DYNAMIC|D3DUSAGE_WRITEONLY, 0, D3DPOOL_DEFAULT, &shaderstate.dyncol_buff, NULL);
|
|
IDirect3DDevice9_CreateIndexBuffer(pD3DDev9, shaderstate.dynidx_size, D3DUSAGE_DYNAMIC|D3DUSAGE_WRITEONLY, D3DFMT_QINDEX, D3DPOOL_DEFAULT, &shaderstate.dynidx_buff, NULL);
|
|
}
|
|
}
|
|
|
|
static void BE_ApplyTMUState(unsigned int tu, unsigned int flags)
|
|
{
|
|
if ((flags ^ shaderstate.tmuflags[tu]) & SHADER_PASS_CLAMP)
|
|
{
|
|
shaderstate.tmuflags[tu] ^= SHADER_PASS_CLAMP;
|
|
if (flags & SHADER_PASS_CLAMP)
|
|
{
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_ADDRESSU, D3DTADDRESS_CLAMP);
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_ADDRESSV, D3DTADDRESS_CLAMP);
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_ADDRESSW, D3DTADDRESS_CLAMP);
|
|
}
|
|
else
|
|
{
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_ADDRESSU, D3DTADDRESS_WRAP);
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_ADDRESSV, D3DTADDRESS_WRAP);
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_ADDRESSW, D3DTADDRESS_WRAP);
|
|
}
|
|
}
|
|
if ((flags ^ shaderstate.tmuflags[tu]) & SHADER_PASS_NOMIPMAP)
|
|
{
|
|
shaderstate.tmuflags[tu] ^= SHADER_PASS_NOMIPMAP;
|
|
/*lightmaps don't use mipmaps*/
|
|
if (flags & SHADER_PASS_NOMIPMAP)
|
|
{
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_MINFILTER, D3DTEXF_LINEAR);
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_MIPFILTER, D3DTEXF_NONE);
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_MAGFILTER, D3DTEXF_LINEAR);
|
|
}
|
|
else
|
|
{
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_MINFILTER, D3DTEXF_LINEAR);
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_MIPFILTER, D3DTEXF_LINEAR);
|
|
IDirect3DDevice9_SetSamplerState(pD3DDev9, tu, D3DSAMP_MAGFILTER, D3DTEXF_LINEAR);
|
|
}
|
|
}
|
|
}
|
|
|
|
static void D3DBE_ApplyShaderBits(unsigned int bits)
|
|
{
|
|
unsigned int delta;
|
|
delta = bits ^ shaderstate.shaderbits;
|
|
if (!delta)
|
|
return;
|
|
shaderstate.shaderbits = bits;
|
|
|
|
if (delta & SBITS_BLEND_BITS)
|
|
{
|
|
if (bits & SBITS_BLEND_BITS)
|
|
{
|
|
D3DBLEND src;
|
|
D3DBLEND dst;
|
|
|
|
switch(bits & SBITS_SRCBLEND_BITS)
|
|
{
|
|
case SBITS_SRCBLEND_ZERO: src = D3DBLEND_ZERO; break;
|
|
case SBITS_SRCBLEND_ONE: src = D3DBLEND_ONE; break;
|
|
case SBITS_SRCBLEND_DST_COLOR: src = D3DBLEND_DESTCOLOR; break;
|
|
case SBITS_SRCBLEND_ONE_MINUS_DST_COLOR: src = D3DBLEND_INVDESTCOLOR; break;
|
|
case SBITS_SRCBLEND_SRC_ALPHA: src = D3DBLEND_SRCALPHA; break;
|
|
case SBITS_SRCBLEND_ONE_MINUS_SRC_ALPHA: src = D3DBLEND_INVSRCALPHA; break;
|
|
case SBITS_SRCBLEND_DST_ALPHA: src = D3DBLEND_DESTALPHA; break;
|
|
case SBITS_SRCBLEND_ONE_MINUS_DST_ALPHA: src = D3DBLEND_INVDESTALPHA; break;
|
|
case SBITS_SRCBLEND_ALPHA_SATURATE: src = D3DBLEND_SRCALPHASAT; break;
|
|
default: Sys_Error("Bad shader blend src\n"); return;
|
|
}
|
|
switch(bits & SBITS_DSTBLEND_BITS)
|
|
{
|
|
case SBITS_DSTBLEND_ZERO: dst = D3DBLEND_ZERO; break;
|
|
case SBITS_DSTBLEND_ONE: dst = D3DBLEND_ONE; break;
|
|
case SBITS_DSTBLEND_SRC_ALPHA: dst = D3DBLEND_SRCALPHA; break;
|
|
case SBITS_DSTBLEND_ONE_MINUS_SRC_ALPHA: dst = D3DBLEND_INVSRCALPHA; break;
|
|
case SBITS_DSTBLEND_DST_ALPHA: dst = D3DBLEND_DESTALPHA; break;
|
|
case SBITS_DSTBLEND_ONE_MINUS_DST_ALPHA: dst = D3DBLEND_INVDESTALPHA; break;
|
|
case SBITS_DSTBLEND_SRC_COLOR: dst = D3DBLEND_SRCCOLOR; break;
|
|
case SBITS_DSTBLEND_ONE_MINUS_SRC_COLOR: dst = D3DBLEND_INVSRCCOLOR; break;
|
|
default: Sys_Error("Bad shader blend dst\n"); return;
|
|
}
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHABLENDENABLE, TRUE);
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_SRCBLEND, src);
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_DESTBLEND, dst);
|
|
}
|
|
else
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHABLENDENABLE, FALSE);
|
|
}
|
|
|
|
if (delta & SBITS_ATEST_BITS)
|
|
{
|
|
switch(bits & SBITS_ATEST_BITS)
|
|
{
|
|
case SBITS_ATEST_NONE:
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHATESTENABLE, FALSE);
|
|
// IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHAREF, 0);
|
|
// IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHAFUNC, 0);
|
|
break;
|
|
case SBITS_ATEST_GT0:
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHATESTENABLE, TRUE);
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHAREF, 0);
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHAFUNC, D3DCMP_GREATER);
|
|
break;
|
|
case SBITS_ATEST_LT128:
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHATESTENABLE, TRUE);
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHAREF, 128);
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHAFUNC, D3DCMP_LESS);
|
|
break;
|
|
case SBITS_ATEST_GE128:
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHATESTENABLE, TRUE);
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHAREF, 128);
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ALPHAFUNC, D3DCMP_GREATEREQUAL);
|
|
break;
|
|
}
|
|
}
|
|
|
|
if (delta & SBITS_MISC_DEPTHWRITE)
|
|
{
|
|
if (bits & SBITS_MISC_DEPTHWRITE)
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ZWRITEENABLE, TRUE);
|
|
else
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ZWRITEENABLE, FALSE);
|
|
}
|
|
|
|
if(delta & SBITS_MISC_NODEPTHTEST)
|
|
{
|
|
if(bits & SBITS_MISC_NODEPTHTEST)
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ZENABLE, FALSE);
|
|
else
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ZENABLE, TRUE);
|
|
}
|
|
|
|
if (delta & (SBITS_MISC_DEPTHEQUALONLY|SBITS_MISC_DEPTHCLOSERONLY))
|
|
{
|
|
switch(bits & (SBITS_MISC_DEPTHEQUALONLY|SBITS_MISC_DEPTHCLOSERONLY))
|
|
{
|
|
default:
|
|
case 0:
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ZFUNC, D3DCMP_LESSEQUAL);
|
|
break;
|
|
case SBITS_MISC_DEPTHEQUALONLY:
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ZFUNC, D3DCMP_EQUAL);
|
|
break;
|
|
case SBITS_MISC_DEPTHCLOSERONLY:
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_ZFUNC, D3DCMP_LESS);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
void BE_Init(void)
|
|
{
|
|
unsigned int i;
|
|
be_maxpasses = MAX_TMUS;
|
|
memset(&shaderstate, 0, sizeof(shaderstate));
|
|
shaderstate.curvertdecl = -1;
|
|
|
|
FTable_Init();
|
|
|
|
shaderstate.dynxyz_size = sizeof(vecV_t) * DYNVBUFFSIZE;
|
|
shaderstate.dyncol_size = sizeof(byte_vec4_t) * DYNVBUFFSIZE;
|
|
shaderstate.dynst_size = sizeof(vec2_t) * DYNVBUFFSIZE;
|
|
shaderstate.dynidx_size = sizeof(index_t) * DYNIBUFFSIZE;
|
|
|
|
BE_D3D_Reset(false);
|
|
|
|
/*force all state to change, thus setting a known state*/
|
|
shaderstate.shaderbits = ~0;
|
|
D3DBE_ApplyShaderBits(0);
|
|
|
|
for (i = 0; i < MAX_TMUS; i++)
|
|
{
|
|
shaderstate.tmuflags[i] = ~0;
|
|
BE_ApplyTMUState(i, 0);
|
|
}
|
|
}
|
|
|
|
static void allocvertexbuffer(IDirect3DVertexBuffer9 *buff, unsigned int bmaxsize, unsigned int *offset, void **data, unsigned int bytes)
|
|
{
|
|
unsigned int boff;
|
|
if (*offset + bytes > bmaxsize)
|
|
{
|
|
boff = 0;
|
|
*offset = bytes;
|
|
}
|
|
else
|
|
{
|
|
boff = *offset;
|
|
*offset += bytes;
|
|
}
|
|
d3dcheck(IDirect3DVertexBuffer9_Lock(buff, boff, bytes, data, boff?D3DLOCK_NOOVERWRITE:D3DLOCK_DISCARD));
|
|
}
|
|
|
|
static unsigned int allocindexbuffer(void **dest, unsigned int entries)
|
|
{
|
|
unsigned int bytes = entries*sizeof(index_t);
|
|
unsigned int offset;
|
|
|
|
if (shaderstate.dynidx_offs + bytes > DYNIBUFFSIZE)
|
|
{
|
|
offset = 0;
|
|
shaderstate.dynidx_offs = 0;
|
|
}
|
|
else
|
|
{
|
|
offset = shaderstate.dynidx_offs;
|
|
shaderstate.dynidx_offs += bytes;
|
|
}
|
|
|
|
d3dcheck(IDirect3DIndexBuffer9_Lock(shaderstate.dynidx_buff, offset, entries, dest, offset?D3DLOCK_NOOVERWRITE:D3DLOCK_DISCARD));
|
|
return offset/sizeof(index_t);
|
|
}
|
|
|
|
static void BindTexture(unsigned int tu, void *id)
|
|
{
|
|
if (shaderstate.curtex[tu].ptr != id)
|
|
{
|
|
shaderstate.curtex[tu].ptr = id;
|
|
IDirect3DDevice9_SetTexture (pD3DDev9, tu, id);
|
|
}
|
|
}
|
|
|
|
static void SelectPassTexture(unsigned int tu, shaderpass_t *pass)
|
|
{
|
|
switch(pass->texgen)
|
|
{
|
|
default:
|
|
case T_GEN_DIFFUSE:
|
|
BindTexture(tu, shaderstate.curtexnums->base.ptr);
|
|
break;
|
|
case T_GEN_NORMALMAP:
|
|
BindTexture( tu, shaderstate.curtexnums->bump.ptr);
|
|
break;
|
|
case T_GEN_SPECULAR:
|
|
BindTexture(tu, shaderstate.curtexnums->specular.ptr);
|
|
break;
|
|
case T_GEN_UPPEROVERLAY:
|
|
BindTexture(tu, shaderstate.curtexnums->upperoverlay.ptr);
|
|
break;
|
|
case T_GEN_LOWEROVERLAY:
|
|
BindTexture(tu, shaderstate.curtexnums->loweroverlay.ptr);
|
|
break;
|
|
case T_GEN_FULLBRIGHT:
|
|
BindTexture(tu, shaderstate.curtexnums->fullbright.ptr);
|
|
break;
|
|
case T_GEN_ANIMMAP:
|
|
BindTexture(tu, pass->anim_frames[(int)(pass->anim_fps * shaderstate.curtime) % pass->anim_numframes].ptr);
|
|
break;
|
|
case T_GEN_SINGLEMAP:
|
|
BindTexture(tu, pass->anim_frames[0].ptr);
|
|
break;
|
|
case T_GEN_DELUXMAP:
|
|
BindTexture(tu, shaderstate.curdeluxmap.ptr);
|
|
break;
|
|
case T_GEN_LIGHTMAP:
|
|
BindTexture(tu, shaderstate.curlightmap.ptr);
|
|
break;
|
|
|
|
/*case T_GEN_CURRENTRENDER:
|
|
FIXME: no code to grab the current screen and convert to a texture
|
|
break;*/
|
|
case T_GEN_VIDEOMAP:
|
|
BindTexture(tu, Media_UpdateForShader(pass->cin).ptr);
|
|
break;
|
|
}
|
|
|
|
BE_ApplyTMUState(tu, pass->flags);
|
|
|
|
switch (pass->blendmode)
|
|
{
|
|
case GL_DOT3_RGB_ARB:
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLORARG1, D3DTA_TEXTURE);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLORARG2, D3DTA_CURRENT);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLOROP, D3DTOP_DOTPRODUCT3);
|
|
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG1, D3DTA_TEXTURE);
|
|
// IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG2, D3DTA_DIFFUSE);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAOP, D3DTOP_SELECTARG1);
|
|
break;
|
|
case GL_REPLACE:
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLORARG1, D3DTA_TEXTURE);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLOROP, D3DTOP_SELECTARG1);
|
|
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG1, D3DTA_TEXTURE);
|
|
// IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG2, D3DTA_DIFFUSE);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAOP, D3DTOP_SELECTARG1);
|
|
break;
|
|
case GL_ADD:
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLORARG1, D3DTA_TEXTURE);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLORARG2, D3DTA_CURRENT);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLOROP, D3DTOP_BLENDTEXTUREALPHA);
|
|
shaderstate.passcolour &= 0xff000000;
|
|
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG1, D3DTA_TEXTURE);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG2, D3DTA_DIFFUSE);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAOP, D3DTOP_MODULATE);
|
|
break;
|
|
case GL_DECAL:
|
|
if (!tu)
|
|
goto forcemod;
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLORARG1, D3DTA_TEXTURE);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLORARG2, D3DTA_CURRENT);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLOROP, D3DTOP_BLENDTEXTUREALPHA);
|
|
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG1, D3DTA_TEXTURE);
|
|
// IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG2, D3DTA_CURRENT);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAOP, D3DTOP_SELECTARG1);
|
|
break;
|
|
default:
|
|
case GL_MODULATE:
|
|
forcemod:
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLORARG1, D3DTA_TEXTURE);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLORARG2, D3DTA_CURRENT);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLOROP, D3DTOP_MODULATE);
|
|
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG1, D3DTA_TEXTURE);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG2, D3DTA_CURRENT);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAOP, D3DTOP_MODULATE);
|
|
break;
|
|
}
|
|
|
|
if (tu == 0)
|
|
{
|
|
if (shaderstate.passsinglecolour)
|
|
{
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLORARG2, D3DTA_CONSTANT);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG2, D3DTA_CONSTANT);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_CONSTANT, shaderstate.passcolour);
|
|
}
|
|
else
|
|
{
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_COLORARG2, D3DTA_DIFFUSE);
|
|
IDirect3DDevice9_SetTextureStageState(pD3DDev9, tu, D3DTSS_ALPHAARG2, D3DTA_DIFFUSE);
|
|
}
|
|
}
|
|
}
|
|
|
|
static void colourgenbyte(const shaderpass_t *pass, int cnt, const byte_vec4_t *src, byte_vec4_t *dst, const mesh_t *mesh)
|
|
{
|
|
D3DCOLOR block;
|
|
switch (pass->rgbgen)
|
|
{
|
|
case RGB_GEN_ENTITY:
|
|
block = D3DCOLOR_COLORVALUE(shaderstate.curentity->shaderRGBAf[0], shaderstate.curentity->shaderRGBAf[1], shaderstate.curentity->shaderRGBAf[2], shaderstate.curentity->shaderRGBAf[3]);
|
|
while((cnt)--)
|
|
{
|
|
((D3DCOLOR*)dst)[cnt] = block;
|
|
}
|
|
break;
|
|
case RGB_GEN_ONE_MINUS_ENTITY:
|
|
block = D3DCOLOR_COLORVALUE(1-shaderstate.curentity->shaderRGBAf[0], 1-shaderstate.curentity->shaderRGBAf[1], 1-shaderstate.curentity->shaderRGBAf[2], 1-shaderstate.curentity->shaderRGBAf[3]);
|
|
while((cnt)--)
|
|
{
|
|
((D3DCOLOR*)dst)[cnt] = block;
|
|
}
|
|
break;
|
|
case RGB_GEN_VERTEX:
|
|
case RGB_GEN_EXACT_VERTEX:
|
|
if (!src)
|
|
{
|
|
while((cnt)--)
|
|
{
|
|
dst[cnt][0] = 255;//shaderstate.identitylighting;
|
|
dst[cnt][1] = 255;//shaderstate.identitylighting;
|
|
dst[cnt][2] = 255;//shaderstate.identitylighting;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
while((cnt)--)
|
|
{
|
|
qbyte r, g, b;
|
|
r=src[cnt][0];
|
|
g=src[cnt][1];
|
|
b=src[cnt][2];
|
|
dst[cnt][0] = b;
|
|
dst[cnt][1] = g;
|
|
dst[cnt][2] = r;
|
|
}
|
|
}
|
|
break;
|
|
case RGB_GEN_ONE_MINUS_VERTEX:
|
|
while((cnt)--)
|
|
{
|
|
qbyte r, g, b;
|
|
r=255-src[cnt][0];
|
|
g=255-src[cnt][1];
|
|
b=255-src[cnt][2];
|
|
dst[cnt][0] = b;
|
|
dst[cnt][1] = g;
|
|
dst[cnt][2] = r;
|
|
}
|
|
break;
|
|
case RGB_GEN_IDENTITY_LIGHTING:
|
|
//compensate for overbrights
|
|
while((cnt)--)
|
|
{
|
|
dst[cnt][0] = 255;//shaderstate.identitylighting;
|
|
dst[cnt][1] = 255;//shaderstate.identitylighting;
|
|
dst[cnt][2] = 255;//shaderstate.identitylighting;
|
|
}
|
|
break;
|
|
default:
|
|
case RGB_GEN_IDENTITY:
|
|
block = D3DCOLOR_RGBA(255, 255, 255, 255);
|
|
while((cnt)--)
|
|
{
|
|
((D3DCOLOR*)dst)[cnt] = block;
|
|
}
|
|
break;
|
|
case RGB_GEN_CONST:
|
|
block = D3DCOLOR_COLORVALUE(pass->rgbgen_func.args[0], pass->rgbgen_func.args[1], pass->rgbgen_func.args[2], 1);
|
|
while((cnt)--)
|
|
{
|
|
((D3DCOLOR*)dst)[cnt] = block;
|
|
}
|
|
break;
|
|
case RGB_GEN_LIGHTING_DIFFUSE:
|
|
//collect lighting details for mobile entities
|
|
if (!mesh->normals_array)
|
|
{
|
|
block = D3DCOLOR_RGBA(255, 255, 255, 255);
|
|
while((cnt)--)
|
|
{
|
|
((D3DCOLOR*)dst)[cnt] = block;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
R_LightArraysByte_BGR(mesh->xyz_array, dst, cnt, mesh->normals_array);
|
|
}
|
|
break;
|
|
case RGB_GEN_WAVE:
|
|
{
|
|
float *table;
|
|
float c;
|
|
|
|
table = FTableForFunc(pass->rgbgen_func.type);
|
|
c = pass->rgbgen_func.args[2] + shaderstate.curtime * pass->rgbgen_func.args[3];
|
|
c = FTABLE_EVALUATE(table, c) * pass->rgbgen_func.args[1] + pass->rgbgen_func.args[0];
|
|
c = bound(0.0f, c, 1.0f);
|
|
block = D3DCOLOR_COLORVALUE(c, c, c, 1);
|
|
|
|
while((cnt)--)
|
|
{
|
|
((D3DCOLOR*)dst)[cnt] = block;
|
|
}
|
|
}
|
|
break;
|
|
|
|
case RGB_GEN_TOPCOLOR:
|
|
case RGB_GEN_BOTTOMCOLOR:
|
|
#pragma message("fix 24bit player colours")
|
|
block = D3DCOLOR_RGBA(255, 255, 255, 255);
|
|
while((cnt)--)
|
|
{
|
|
((D3DCOLOR*)dst)[cnt] = block;
|
|
}
|
|
// Con_Printf("RGB_GEN %i not supported\n", pass->rgbgen);
|
|
break;
|
|
}
|
|
}
|
|
|
|
static void alphagenbyte(const shaderpass_t *pass, int cnt, const byte_vec4_t *src, byte_vec4_t *dst, const mesh_t *mesh)
|
|
{
|
|
/*FIXME: Skip this if the rgbgen did it*/
|
|
float *table;
|
|
unsigned char t;
|
|
float f;
|
|
vec3_t v1, v2;
|
|
|
|
switch (pass->alphagen)
|
|
{
|
|
default:
|
|
case ALPHA_GEN_IDENTITY:
|
|
while(cnt--)
|
|
dst[cnt][3] = 255;
|
|
break;
|
|
|
|
case ALPHA_GEN_CONST:
|
|
t = pass->alphagen_func.args[0]*255;
|
|
while(cnt--)
|
|
dst[cnt][3] = t;
|
|
break;
|
|
|
|
case ALPHA_GEN_WAVE:
|
|
table = FTableForFunc(pass->alphagen_func.type);
|
|
f = pass->alphagen_func.args[2] + shaderstate.curtime * pass->alphagen_func.args[3];
|
|
f = FTABLE_EVALUATE(table, f) * pass->alphagen_func.args[1] + pass->alphagen_func.args[0];
|
|
t = bound(0.0f, f, 1.0f)*255;
|
|
while(cnt--)
|
|
dst[cnt][3] = t;
|
|
break;
|
|
|
|
case ALPHA_GEN_PORTAL:
|
|
//FIXME: should this be per-vert?
|
|
VectorAdd(mesh->xyz_array[0], shaderstate.curentity->origin, v1);
|
|
VectorSubtract(r_origin, v1, v2);
|
|
f = VectorLength(v2) * (1.0 / 255.0);
|
|
t = bound(0.0f, f, 1.0f)*255;
|
|
|
|
while(cnt--)
|
|
dst[cnt][3] = t;
|
|
break;
|
|
|
|
case ALPHA_GEN_VERTEX:
|
|
if (!src)
|
|
{
|
|
while(cnt--)
|
|
{
|
|
dst[cnt][3] = 255;
|
|
}
|
|
break;
|
|
}
|
|
|
|
while(cnt--)
|
|
{
|
|
dst[cnt][3] = src[cnt][3];
|
|
}
|
|
break;
|
|
|
|
case ALPHA_GEN_ENTITY:
|
|
t = bound(0, shaderstate.curentity->shaderRGBAf[3], 1)*255;
|
|
while(cnt--)
|
|
{
|
|
dst[cnt][3] = t;
|
|
}
|
|
break;
|
|
|
|
case ALPHA_GEN_SPECULAR:
|
|
{
|
|
int i;
|
|
VectorSubtract(r_origin, shaderstate.curentity->origin, v1);
|
|
|
|
if (!Matrix3_Compare(shaderstate.curentity->axis, axisDefault))
|
|
{
|
|
Matrix3_Multiply_Vec3(shaderstate.curentity->axis, v2, v2);
|
|
}
|
|
else
|
|
{
|
|
VectorCopy(v1, v2);
|
|
}
|
|
|
|
for (i = 0; i < cnt; i++)
|
|
{
|
|
VectorSubtract(v2, mesh->xyz_array[i], v1);
|
|
f = DotProduct(v1, mesh->normals_array[i] ) * Q_rsqrt(DotProduct(v1,v1));
|
|
f = f * f * f * f * f;
|
|
dst[i][3] = bound (0.0f, (int)(f*255), 255);
|
|
}
|
|
}
|
|
break;
|
|
}
|
|
}
|
|
|
|
static unsigned int BE_GenerateColourMods(unsigned int vertcount, const shaderpass_t *pass)
|
|
{
|
|
unsigned int ret = 0;
|
|
unsigned char *map;
|
|
const mesh_t *m;
|
|
unsigned int mno;
|
|
|
|
m = shaderstate.meshlist[0];
|
|
|
|
if (pass->flags & SHADER_PASS_NOCOLORARRAY)
|
|
{
|
|
shaderstate.passsinglecolour = true;
|
|
shaderstate.passcolour = D3DCOLOR_RGBA(255,255,255,255);
|
|
colourgenbyte(pass, 1, (byte_vec4_t*)&shaderstate.passcolour, (byte_vec4_t*)&shaderstate.passcolour, m);
|
|
alphagenbyte(pass, 1, (byte_vec4_t*)&shaderstate.passcolour, (byte_vec4_t*)&shaderstate.passcolour, m);
|
|
/*FIXME: just because there's no rgba set, there's no reason to assume it should be a single colour (unshaded ents)*/
|
|
d3dcheck(IDirect3DDevice9_SetStreamSource(pD3DDev9, 1, NULL, 0, 0));
|
|
}
|
|
else
|
|
{
|
|
unsigned int v;
|
|
int c;
|
|
float *src;
|
|
|
|
shaderstate.passsinglecolour = false;
|
|
|
|
ret |= D3D_VDEC_COL4B;
|
|
allocvertexbuffer(shaderstate.dyncol_buff, shaderstate.dyncol_size, &shaderstate.dyncol_offs, (void**)&map, vertcount*sizeof(D3DCOLOR));
|
|
if (m->colors4b_array)
|
|
{
|
|
for (vertcount = 0, mno = 0; mno < shaderstate.nummeshes; mno++)
|
|
{
|
|
m = shaderstate.meshlist[mno];
|
|
colourgenbyte(pass, m->numvertexes, (byte_vec4_t*)m->colors4b_array, (byte_vec4_t*)map, m);
|
|
alphagenbyte(pass, m->numvertexes, (byte_vec4_t*)m->colors4b_array, (byte_vec4_t*)map, m);
|
|
map += m->numvertexes*4;
|
|
vertcount += m->numvertexes;
|
|
}
|
|
}
|
|
else if (m->colors4f_array &&
|
|
((pass->rgbgen == RGB_GEN_VERTEX) ||
|
|
(pass->rgbgen == RGB_GEN_EXACT_VERTEX) ||
|
|
(pass->rgbgen == RGB_GEN_ONE_MINUS_VERTEX) ||
|
|
(pass->alphagen == ALPHA_GEN_VERTEX)))
|
|
{
|
|
for (vertcount = 0, mno = 0; mno < shaderstate.nummeshes; mno++)
|
|
{
|
|
m = shaderstate.meshlist[mno];
|
|
src = m->colors4f_array[0];
|
|
for (v = 0; v < m->numvertexes; v++)
|
|
{
|
|
c = src[0]*255;
|
|
map[0] = bound(0, c, 255);
|
|
c = src[1]*255;
|
|
map[1] = bound(0, c, 255);
|
|
c = src[2]*255;
|
|
map[2] = bound(0, c, 255);
|
|
c = src[3]*255;
|
|
map[3] = bound(0, c, 255);
|
|
map += 4;
|
|
src += 4;
|
|
}
|
|
vertcount += m->numvertexes;
|
|
}
|
|
map -= vertcount*4;
|
|
/*FIXME: m is wrong. its the last ent only*/
|
|
colourgenbyte(pass, vertcount, (byte_vec4_t*)map, (byte_vec4_t*)map, m);
|
|
alphagenbyte(pass, vertcount, (byte_vec4_t*)map, (byte_vec4_t*)map, m);
|
|
}
|
|
else
|
|
{
|
|
for (vertcount = 0, mno = 0; mno < shaderstate.nummeshes; mno++)
|
|
{
|
|
m = shaderstate.meshlist[mno];
|
|
colourgenbyte(pass, m->numvertexes, NULL, (byte_vec4_t*)map, m);
|
|
alphagenbyte(pass, m->numvertexes, NULL, (byte_vec4_t*)map, m);
|
|
map += m->numvertexes*4;
|
|
vertcount += m->numvertexes;
|
|
}
|
|
}
|
|
d3dcheck(IDirect3DVertexBuffer9_Unlock(shaderstate.dyncol_buff));
|
|
d3dcheck(IDirect3DDevice9_SetStreamSource(pD3DDev9, 1, shaderstate.dyncol_buff, shaderstate.dyncol_offs - vertcount*sizeof(D3DCOLOR), sizeof(D3DCOLOR)));
|
|
}
|
|
return ret;
|
|
}
|
|
/*********************************************************************************************************/
|
|
/*========================================== texture coord generation =====================================*/
|
|
|
|
static void tcgen_environment(float *st, unsigned int numverts, float *xyz, float *normal)
|
|
{
|
|
int i;
|
|
vec3_t viewer, reflected;
|
|
float d;
|
|
|
|
vec3_t rorg;
|
|
|
|
RotateLightVector(shaderstate.curentity->axis, shaderstate.curentity->origin, r_origin, rorg);
|
|
|
|
for (i = 0 ; i < numverts ; i++, xyz += 3, normal += 3, st += 2 )
|
|
{
|
|
VectorSubtract (rorg, xyz, viewer);
|
|
VectorNormalizeFast (viewer);
|
|
|
|
d = DotProduct (normal, viewer);
|
|
|
|
reflected[0] = normal[0]*2*d - viewer[0];
|
|
reflected[1] = normal[1]*2*d - viewer[1];
|
|
reflected[2] = normal[2]*2*d - viewer[2];
|
|
|
|
st[0] = 0.5 + reflected[1] * 0.5;
|
|
st[1] = 0.5 - reflected[2] * 0.5;
|
|
}
|
|
}
|
|
|
|
static float *tcgen(const shaderpass_t *pass, int cnt, float *dst, const mesh_t *mesh)
|
|
{
|
|
int i;
|
|
vecV_t *src;
|
|
switch (pass->tcgen)
|
|
{
|
|
default:
|
|
case TC_GEN_BASE:
|
|
return (float*)mesh->st_array;
|
|
case TC_GEN_LIGHTMAP:
|
|
return (float*)mesh->lmst_array;
|
|
case TC_GEN_NORMAL:
|
|
return (float*)mesh->normals_array;
|
|
case TC_GEN_SVECTOR:
|
|
return (float*)mesh->snormals_array;
|
|
case TC_GEN_TVECTOR:
|
|
return (float*)mesh->tnormals_array;
|
|
case TC_GEN_ENVIRONMENT:
|
|
tcgen_environment(dst, cnt, (float*)mesh->xyz_array, (float*)mesh->normals_array);
|
|
return dst;
|
|
|
|
case TC_GEN_DOTPRODUCT:
|
|
return dst;//mesh->st_array[0];
|
|
case TC_GEN_VECTOR:
|
|
src = mesh->xyz_array;
|
|
for (i = 0; i < cnt; i++, dst += 2)
|
|
{
|
|
static vec3_t tc_gen_s = { 1.0f, 0.0f, 0.0f };
|
|
static vec3_t tc_gen_t = { 0.0f, 1.0f, 0.0f };
|
|
|
|
dst[0] = DotProduct(tc_gen_s, src[i]);
|
|
dst[1] = DotProduct(tc_gen_t, src[i]);
|
|
}
|
|
return dst;
|
|
}
|
|
}
|
|
|
|
/*src and dst can be the same address when tcmods are chained*/
|
|
static void tcmod(const tcmod_t *tcmod, int cnt, const float *src, float *dst, const mesh_t *mesh)
|
|
{
|
|
float *table;
|
|
float t1, t2;
|
|
float cost, sint;
|
|
int j;
|
|
#define R_FastSin(x) sin((x)*(2*M_PI))
|
|
switch (tcmod->type)
|
|
{
|
|
case SHADER_TCMOD_ROTATE:
|
|
cost = tcmod->args[0] * shaderstate.curtime;
|
|
sint = R_FastSin(cost);
|
|
cost = R_FastSin(cost + 0.25);
|
|
|
|
for (j = 0; j < cnt; j++, dst+=2,src+=2)
|
|
{
|
|
t1 = cost * (src[0] - 0.5f) - sint * (src[1] - 0.5f) + 0.5f;
|
|
t2 = cost * (src[1] - 0.5f) + sint * (src[0] - 0.5f) + 0.5f;
|
|
dst[0] = t1;
|
|
dst[1] = t2;
|
|
}
|
|
break;
|
|
|
|
case SHADER_TCMOD_SCALE:
|
|
t1 = tcmod->args[0];
|
|
t2 = tcmod->args[1];
|
|
|
|
for (j = 0; j < cnt; j++, dst+=2,src+=2)
|
|
{
|
|
dst[0] = src[0] * t1;
|
|
dst[1] = src[1] * t2;
|
|
}
|
|
break;
|
|
|
|
case SHADER_TCMOD_TURB:
|
|
t1 = tcmod->args[2] + shaderstate.curtime * tcmod->args[3];
|
|
t2 = tcmod->args[1];
|
|
|
|
for (j = 0; j < cnt; j++, dst+=2,src+=2)
|
|
{
|
|
dst[0] = src[0] + R_FastSin (src[0]*t2+t1) * t2;
|
|
dst[1] = src[1] + R_FastSin (src[1]*t2+t1) * t2;
|
|
}
|
|
break;
|
|
|
|
case SHADER_TCMOD_STRETCH:
|
|
table = FTableForFunc(tcmod->args[0]);
|
|
t2 = tcmod->args[3] + shaderstate.curtime * tcmod->args[4];
|
|
t1 = FTABLE_EVALUATE(table, t2) * tcmod->args[2] + tcmod->args[1];
|
|
t1 = t1 ? 1.0f / t1 : 1.0f;
|
|
t2 = 0.5f - 0.5f * t1;
|
|
for (j = 0; j < cnt; j++, dst+=2,src+=2)
|
|
{
|
|
dst[0] = src[0] * t1 + t2;
|
|
dst[1] = src[1] * t1 + t2;
|
|
}
|
|
break;
|
|
|
|
case SHADER_TCMOD_SCROLL:
|
|
t1 = tcmod->args[0] * shaderstate.curtime;
|
|
t2 = tcmod->args[1] * shaderstate.curtime;
|
|
|
|
for (j = 0; j < cnt; j++, dst += 2, src+=2)
|
|
{
|
|
dst[0] = src[0] + t1;
|
|
dst[1] = src[1] + t2;
|
|
}
|
|
break;
|
|
|
|
case SHADER_TCMOD_TRANSFORM:
|
|
for (j = 0; j < cnt; j++, dst+=2, src+=2)
|
|
{
|
|
t1 = src[0];
|
|
t2 = src[1];
|
|
dst[0] = t1 * tcmod->args[0] + t2 * tcmod->args[2] + tcmod->args[4];
|
|
dst[1] = t2 * tcmod->args[1] + t1 * tcmod->args[3] + tcmod->args[5];
|
|
}
|
|
break;
|
|
|
|
default:
|
|
break;
|
|
}
|
|
}
|
|
|
|
static void GenerateTCMods(const shaderpass_t *pass, float *dest)
|
|
{
|
|
mesh_t *mesh;
|
|
unsigned int fvertex = 0, mno;
|
|
int i;
|
|
float *src;
|
|
for (mno = 0; mno < shaderstate.nummeshes; mno++)
|
|
{
|
|
mesh = shaderstate.meshlist[mno];
|
|
src = tcgen(pass, mesh->numvertexes, dest, mesh);
|
|
//tcgen might return unmodified info
|
|
if (pass->numtcmods)
|
|
{
|
|
tcmod(&pass->tcmods[0], mesh->numvertexes, src, dest, mesh);
|
|
for (i = 1; i < pass->numtcmods; i++)
|
|
{
|
|
tcmod(&pass->tcmods[i], mesh->numvertexes, dest, dest, mesh);
|
|
}
|
|
}
|
|
else if (src != dest)
|
|
{
|
|
memcpy(dest, src, 8*mesh->numvertexes);
|
|
}
|
|
dest += mesh->numvertexes*2;
|
|
}
|
|
}
|
|
|
|
//end texture coords
|
|
/*******************************************************************************************************************/
|
|
|
|
static void deformgen(const deformv_t *deformv, int cnt, const vecV_t *src, vecV_t *dst, const mesh_t *mesh)
|
|
{
|
|
float *table;
|
|
int j, k;
|
|
float args[4];
|
|
float deflect;
|
|
switch (deformv->type)
|
|
{
|
|
default:
|
|
case DEFORMV_NONE:
|
|
if (src != (const avec4_t*)dst)
|
|
memcpy(dst, src, sizeof(*src)*cnt);
|
|
break;
|
|
|
|
case DEFORMV_WAVE:
|
|
if (!mesh->normals_array)
|
|
{
|
|
if (src != (const avec4_t*)dst)
|
|
memcpy(dst, src, sizeof(*src)*cnt);
|
|
return;
|
|
}
|
|
args[0] = deformv->func.args[0];
|
|
args[1] = deformv->func.args[1];
|
|
args[3] = deformv->func.args[2] + deformv->func.args[3] * shaderstate.curtime;
|
|
table = FTableForFunc(deformv->func.type);
|
|
|
|
for ( j = 0; j < cnt; j++ )
|
|
{
|
|
deflect = deformv->args[0] * (src[j][0]+src[j][1]+src[j][2]) + args[3];
|
|
deflect = FTABLE_EVALUATE(table, deflect) * args[1] + args[0];
|
|
|
|
// Deflect vertex along its normal by wave amount
|
|
VectorMA(src[j], deflect, mesh->normals_array[j], dst[j]);
|
|
}
|
|
break;
|
|
|
|
case DEFORMV_NORMAL:
|
|
//normal does not actually move the verts, but it does change the normals array
|
|
//we don't currently support that.
|
|
if (src != (const avec4_t*)dst)
|
|
memcpy(dst, src, sizeof(*src)*cnt);
|
|
/*
|
|
args[0] = deformv->args[1] * shaderstate.curtime;
|
|
|
|
for ( j = 0; j < cnt; j++ )
|
|
{
|
|
args[1] = normalsArray[j][2] * args[0];
|
|
|
|
deflect = deformv->args[0] * R_FastSin(args[1]);
|
|
normalsArray[j][0] *= deflect;
|
|
deflect = deformv->args[0] * R_FastSin(args[1] + 0.25);
|
|
normalsArray[j][1] *= deflect;
|
|
VectorNormalizeFast(normalsArray[j]);
|
|
}
|
|
*/ break;
|
|
|
|
case DEFORMV_MOVE:
|
|
table = FTableForFunc(deformv->func.type);
|
|
deflect = deformv->func.args[2] + shaderstate.curtime * deformv->func.args[3];
|
|
deflect = FTABLE_EVALUATE(table, deflect) * deformv->func.args[1] + deformv->func.args[0];
|
|
|
|
for ( j = 0; j < cnt; j++ )
|
|
VectorMA(src[j], deflect, deformv->args, dst[j]);
|
|
break;
|
|
|
|
case DEFORMV_BULGE:
|
|
args[0] = deformv->args[0]/(2*M_PI);
|
|
args[1] = deformv->args[1];
|
|
args[2] = shaderstate.curtime * deformv->args[2]/(2*M_PI);
|
|
|
|
for (j = 0; j < cnt; j++)
|
|
{
|
|
deflect = R_FastSin(mesh->st_array[j][0]*args[0] + args[2])*args[1];
|
|
dst[j][0] = src[j][0]+deflect*mesh->normals_array[j][0];
|
|
dst[j][1] = src[j][1]+deflect*mesh->normals_array[j][1];
|
|
dst[j][2] = src[j][2]+deflect*mesh->normals_array[j][2];
|
|
}
|
|
break;
|
|
|
|
case DEFORMV_AUTOSPRITE:
|
|
if (mesh->numindexes < 6)
|
|
break;
|
|
|
|
for (j = 0; j < cnt-3; j+=4, src+=4, dst+=4)
|
|
{
|
|
vec3_t mid, d;
|
|
float radius;
|
|
mid[0] = 0.25*(src[0][0] + src[1][0] + src[2][0] + src[3][0]);
|
|
mid[1] = 0.25*(src[0][1] + src[1][1] + src[2][1] + src[3][1]);
|
|
mid[2] = 0.25*(src[0][2] + src[1][2] + src[2][2] + src[3][2]);
|
|
VectorSubtract(src[0], mid, d);
|
|
radius = 2*VectorLength(d);
|
|
|
|
for (k = 0; k < 4; k++)
|
|
{
|
|
dst[k][0] = mid[0] + radius*((mesh->st_array[k][0]-0.5)*r_refdef.m_view[0+0]-(mesh->st_array[k][1]-0.5)*r_refdef.m_view[0+1]);
|
|
dst[k][1] = mid[1] + radius*((mesh->st_array[k][0]-0.5)*r_refdef.m_view[4+0]-(mesh->st_array[k][1]-0.5)*r_refdef.m_view[4+1]);
|
|
dst[k][2] = mid[2] + radius*((mesh->st_array[k][0]-0.5)*r_refdef.m_view[8+0]-(mesh->st_array[k][1]-0.5)*r_refdef.m_view[8+1]);
|
|
}
|
|
}
|
|
break;
|
|
|
|
case DEFORMV_AUTOSPRITE2:
|
|
if (mesh->numindexes < 6)
|
|
break;
|
|
|
|
for (k = 0; k < mesh->numindexes; k += 6)
|
|
{
|
|
int long_axis, short_axis;
|
|
vec3_t axis;
|
|
float len[3];
|
|
mat3_t m0, m1, m2, result;
|
|
float *quad[4];
|
|
vec3_t rot_centre, tv;
|
|
|
|
quad[0] = (float *)(dst + mesh->indexes[k+0]);
|
|
quad[1] = (float *)(dst + mesh->indexes[k+1]);
|
|
quad[2] = (float *)(dst + mesh->indexes[k+2]);
|
|
|
|
for (j = 2; j >= 0; j--)
|
|
{
|
|
quad[3] = (float *)(dst + mesh->indexes[k+3+j]);
|
|
if (!VectorEquals (quad[3], quad[0]) &&
|
|
!VectorEquals (quad[3], quad[1]) &&
|
|
!VectorEquals (quad[3], quad[2]))
|
|
{
|
|
break;
|
|
}
|
|
}
|
|
|
|
// build a matrix were the longest axis of the billboard is the Y-Axis
|
|
VectorSubtract(quad[1], quad[0], m0[0]);
|
|
VectorSubtract(quad[2], quad[0], m0[1]);
|
|
VectorSubtract(quad[2], quad[1], m0[2]);
|
|
len[0] = DotProduct(m0[0], m0[0]);
|
|
len[1] = DotProduct(m0[1], m0[1]);
|
|
len[2] = DotProduct(m0[2], m0[2]);
|
|
|
|
if ((len[2] > len[1]) && (len[2] > len[0]))
|
|
{
|
|
if (len[1] > len[0])
|
|
{
|
|
long_axis = 1;
|
|
short_axis = 0;
|
|
}
|
|
else
|
|
{
|
|
long_axis = 0;
|
|
short_axis = 1;
|
|
}
|
|
}
|
|
else if ((len[1] > len[2]) && (len[1] > len[0]))
|
|
{
|
|
if (len[2] > len[0])
|
|
{
|
|
long_axis = 2;
|
|
short_axis = 0;
|
|
}
|
|
else
|
|
{
|
|
long_axis = 0;
|
|
short_axis = 2;
|
|
}
|
|
}
|
|
else //if ( (len[0] > len[1]) && (len[0] > len[2]) )
|
|
{
|
|
if (len[2] > len[1])
|
|
{
|
|
long_axis = 2;
|
|
short_axis = 1;
|
|
}
|
|
else
|
|
{
|
|
long_axis = 1;
|
|
short_axis = 2;
|
|
}
|
|
}
|
|
|
|
if (DotProduct(m0[long_axis], m0[short_axis]))
|
|
{
|
|
VectorNormalize2(m0[long_axis], axis);
|
|
VectorCopy(axis, m0[1]);
|
|
|
|
if (axis[0] || axis[1])
|
|
{
|
|
VectorVectors(m0[1], m0[2], m0[0]);
|
|
}
|
|
else
|
|
{
|
|
VectorVectors(m0[1], m0[0], m0[2]);
|
|
}
|
|
}
|
|
else
|
|
{
|
|
VectorNormalize2(m0[long_axis], axis);
|
|
VectorNormalize2(m0[short_axis], m0[0]);
|
|
VectorCopy(axis, m0[1]);
|
|
CrossProduct(m0[0], m0[1], m0[2]);
|
|
}
|
|
|
|
for (j = 0; j < 3; j++)
|
|
rot_centre[j] = (quad[0][j] + quad[1][j] + quad[2][j] + quad[3][j]) * 0.25;
|
|
|
|
if (shaderstate.curentity)
|
|
{
|
|
VectorAdd(shaderstate.curentity->origin, rot_centre, tv);
|
|
}
|
|
else
|
|
{
|
|
VectorCopy(rot_centre, tv);
|
|
}
|
|
VectorSubtract(r_origin, tv, tv);
|
|
|
|
// filter any longest-axis-parts off the camera-direction
|
|
deflect = -DotProduct(tv, axis);
|
|
|
|
VectorMA(tv, deflect, axis, m1[2]);
|
|
VectorNormalizeFast(m1[2]);
|
|
VectorCopy(axis, m1[1]);
|
|
CrossProduct(m1[1], m1[2], m1[0]);
|
|
|
|
Matrix3_Transpose(m1, m2);
|
|
Matrix3_Multiply(m2, m0, result);
|
|
|
|
for (j = 0; j < 4; j++)
|
|
{
|
|
VectorSubtract(quad[j], rot_centre, tv);
|
|
Matrix3_Multiply_Vec3(result, tv, quad[j]);
|
|
VectorAdd(rot_centre, quad[j], quad[j]);
|
|
}
|
|
}
|
|
break;
|
|
|
|
// case DEFORMV_PROJECTION_SHADOW:
|
|
// break;
|
|
}
|
|
}
|
|
|
|
|
|
|
|
/*does not do the draw call, does not consider indicies (except for billboard generation) */
|
|
static qboolean BE_DrawMeshChain_SetupPass(shaderpass_t *pass, unsigned int vertcount)
|
|
{
|
|
int vdec;
|
|
void *map;
|
|
int i;
|
|
unsigned int passno = 0, tmu;
|
|
|
|
int lastpass = pass->numMergedPasses;
|
|
|
|
for (i = 0; i < lastpass; i++)
|
|
{
|
|
if (pass[i].texgen == T_GEN_UPPEROVERLAY && !TEXVALID(shaderstate.curtexnums->upperoverlay))
|
|
continue;
|
|
if (pass[i].texgen == T_GEN_LOWEROVERLAY && !TEXVALID(shaderstate.curtexnums->loweroverlay))
|
|
continue;
|
|
if (pass[i].texgen == T_GEN_FULLBRIGHT && !TEXVALID(shaderstate.curtexnums->fullbright))
|
|
continue;
|
|
break;
|
|
}
|
|
if (i == lastpass)
|
|
return false;
|
|
|
|
/*all meshes in a chain must have the same features*/
|
|
|
|
vdec = 0;
|
|
|
|
/*we only use one colour, generated from the first pass*/
|
|
vdec |= BE_GenerateColourMods(vertcount, pass);
|
|
|
|
tmu = 0;
|
|
/*activate tmus*/
|
|
for (passno = 0; passno < lastpass; passno++)
|
|
{
|
|
if (pass[passno].texgen == T_GEN_UPPEROVERLAY && !TEXVALID(shaderstate.curtexnums->upperoverlay))
|
|
continue;
|
|
if (pass[passno].texgen == T_GEN_LOWEROVERLAY && !TEXVALID(shaderstate.curtexnums->loweroverlay))
|
|
continue;
|
|
if (pass[passno].texgen == T_GEN_FULLBRIGHT && !TEXVALID(shaderstate.curtexnums->fullbright))
|
|
continue;
|
|
|
|
SelectPassTexture(tmu, pass+passno);
|
|
|
|
vdec |= D3D_VDEC_ST0<<tmu;
|
|
allocvertexbuffer(shaderstate.dynst_buff[tmu], shaderstate.dynst_size, &shaderstate.dynst_offs[tmu], &map, vertcount*sizeof(vec2_t));
|
|
GenerateTCMods(pass+passno, map);
|
|
d3dcheck(IDirect3DVertexBuffer9_Unlock(shaderstate.dynst_buff[tmu]));
|
|
d3dcheck(IDirect3DDevice9_SetStreamSource(pD3DDev9, 5+tmu, shaderstate.dynst_buff[tmu], shaderstate.dynst_offs[tmu] - vertcount*sizeof(vec2_t), sizeof(vec2_t)));
|
|
tmu++;
|
|
}
|
|
/*deactivate any extras*/
|
|
for (; tmu < shaderstate.lastpasscount; )
|
|
{
|
|
d3dcheck(IDirect3DDevice9_SetStreamSource(pD3DDev9, 5+tmu, NULL, 0, 0));
|
|
BindTexture(tmu, NULL);
|
|
d3dcheck(IDirect3DDevice9_SetTextureStageState(pD3DDev9, tmu, D3DTSS_COLOROP, D3DTOP_DISABLE));
|
|
tmu++;
|
|
}
|
|
shaderstate.lastpasscount = tmu;
|
|
|
|
// if (meshchain->normals_array &&
|
|
// meshchain->2 &&
|
|
// meshchain->tnormals_array)
|
|
// vdec |= D3D_VDEC_NORMS;
|
|
|
|
if (vdec != shaderstate.curvertdecl)
|
|
{
|
|
shaderstate.curvertdecl = vdec;
|
|
d3dcheck(IDirect3DDevice9_SetVertexDeclaration(pD3DDev9, vertexdecls[shaderstate.curvertdecl]));
|
|
}
|
|
|
|
D3DBE_ApplyShaderBits(pass->shaderbits);
|
|
return true;
|
|
}
|
|
|
|
static void BE_Cull(unsigned int cullflags)
|
|
{
|
|
cullflags |= r_refdef.flipcull;
|
|
if (shaderstate.curcull != cullflags)
|
|
{
|
|
shaderstate.curcull = cullflags;
|
|
if (shaderstate.curcull & 1)
|
|
{
|
|
if (shaderstate.curcull & SHADER_CULL_FRONT)
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_CULLMODE, D3DCULL_CW);
|
|
else if (shaderstate.curcull & SHADER_CULL_BACK)
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_CULLMODE, D3DCULL_CCW);
|
|
else
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_CULLMODE, D3DCULL_NONE);
|
|
}
|
|
else
|
|
{
|
|
if (shaderstate.curcull & SHADER_CULL_FRONT)
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_CULLMODE, D3DCULL_CCW);
|
|
else if (shaderstate.curcull & SHADER_CULL_BACK)
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_CULLMODE, D3DCULL_CW);
|
|
else
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_CULLMODE, D3DCULL_NONE);
|
|
}
|
|
}
|
|
}
|
|
|
|
static void BE_DrawMeshChain_Internal(void)
|
|
{
|
|
unsigned int vertcount, idxcount, idxfirst;
|
|
mesh_t *m;
|
|
void *map;
|
|
int i;
|
|
unsigned int mno;
|
|
unsigned int passno = 0;
|
|
shaderpass_t *pass = shaderstate.curshader->passes;
|
|
|
|
BE_Cull(shaderstate.curshader->flags & (SHADER_CULL_FRONT | SHADER_CULL_BACK));
|
|
|
|
for (mno = 0, vertcount = 0, idxcount = 0; mno < shaderstate.nummeshes; mno++)
|
|
{
|
|
m = shaderstate.meshlist[mno];
|
|
vertcount += m->numvertexes;
|
|
idxcount += m->numindexes;
|
|
}
|
|
|
|
/*vertex buffers are common to all passes*/
|
|
allocvertexbuffer(shaderstate.dynxyz_buff, shaderstate.dynxyz_size, &shaderstate.dynxyz_offs, &map, vertcount*sizeof(vecV_t));
|
|
for (mno = 0, vertcount = 0; mno < shaderstate.nummeshes; mno++)
|
|
{
|
|
vecV_t *dest = (vecV_t*)((char*)map+vertcount*sizeof(vecV_t));
|
|
m = shaderstate.meshlist[mno];
|
|
deformgen(&shaderstate.curshader->deforms[0], m->numvertexes, m->xyz_array, dest, m);
|
|
for (i = 1; i < shaderstate.curshader->numdeforms; i++)
|
|
{
|
|
deformgen(&shaderstate.curshader->deforms[i], m->numvertexes, dest, dest, m);
|
|
}
|
|
vertcount += m->numvertexes;
|
|
}
|
|
d3dcheck(IDirect3DVertexBuffer9_Unlock(shaderstate.dynxyz_buff));
|
|
d3dcheck(IDirect3DDevice9_SetStreamSource(pD3DDev9, 0, shaderstate.dynxyz_buff, shaderstate.dynxyz_offs - vertcount*sizeof(vecV_t), sizeof(vecV_t)));
|
|
|
|
/*so are index buffers*/
|
|
idxfirst = allocindexbuffer(&map, idxcount);
|
|
for (mno = 0, vertcount = 0; mno < shaderstate.nummeshes; mno++)
|
|
{
|
|
m = shaderstate.meshlist[mno];
|
|
for (i = 0; i < m->numindexes; i++)
|
|
((index_t*)map)[i] = m->indexes[i]+vertcount;
|
|
map = (char*)map + m->numindexes*sizeof(index_t);
|
|
vertcount += m->numvertexes;
|
|
}
|
|
d3dcheck(IDirect3DIndexBuffer9_Unlock(shaderstate.dynidx_buff));
|
|
d3dcheck(IDirect3DDevice9_SetIndices(pD3DDev9, shaderstate.dynidx_buff));
|
|
|
|
if (shaderstate.mode == BEM_DEPTHONLY)
|
|
{
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_COLORWRITEENABLE, 0);
|
|
/*deactivate any extras*/
|
|
for (passno = 0; passno < shaderstate.lastpasscount; )
|
|
{
|
|
d3dcheck(IDirect3DDevice9_SetStreamSource(pD3DDev9, 5+passno, NULL, 0, 0));
|
|
BindTexture(passno, NULL);
|
|
d3dcheck(IDirect3DDevice9_SetTextureStageState(pD3DDev9, passno, D3DTSS_COLOROP, D3DTOP_DISABLE));
|
|
passno++;
|
|
}
|
|
shaderstate.lastpasscount = 0;
|
|
d3dcheck(IDirect3DDevice9_DrawIndexedPrimitive(pD3DDev9, D3DPT_TRIANGLELIST, 0, 0, vertcount, idxfirst, idxcount/3));
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_COLORWRITEENABLE, D3DCOLORWRITEENABLE_RED|D3DCOLORWRITEENABLE_GREEN|D3DCOLORWRITEENABLE_BLUE|D3DCOLORWRITEENABLE_ALPHA);
|
|
}
|
|
else
|
|
{
|
|
/*now go through and flush each pass*/
|
|
for (passno = 0; passno < shaderstate.curshader->numpasses; passno += pass->numMergedPasses)
|
|
{
|
|
if (!BE_DrawMeshChain_SetupPass(pass+passno, vertcount))
|
|
continue;
|
|
#ifdef BENCH
|
|
shaderstate.bench.draws++;
|
|
if (shaderstate.bench.clamp && shaderstate.bench.clamp < shaderstate.bench.draws)
|
|
continue;
|
|
#endif
|
|
d3dcheck(IDirect3DDevice9_DrawIndexedPrimitive(pD3DDev9, D3DPT_TRIANGLELIST, 0, 0, vertcount, idxfirst, idxcount/3));
|
|
}
|
|
}
|
|
}
|
|
|
|
void BE_SelectMode(backendmode_t mode, unsigned int flags)
|
|
{
|
|
shaderstate.mode = mode;
|
|
shaderstate.flags = flags;
|
|
}
|
|
|
|
void _CrtCheckMemory(void);
|
|
|
|
/*Generates an optimised vbo for each of the given model's textures*/
|
|
void BE_GenBrushModelVBO(model_t *mod)
|
|
{
|
|
unsigned int maxvboverts;
|
|
unsigned int maxvboelements;
|
|
|
|
unsigned int t;
|
|
unsigned int i;
|
|
unsigned int v;
|
|
unsigned int vcount, ecount;
|
|
unsigned int pervertsize; //erm, that name wasn't intentional
|
|
unsigned int meshes;
|
|
|
|
vbo_t *vbo;
|
|
char *vboedata;
|
|
mesh_t *m;
|
|
char *vbovdata;
|
|
|
|
if (!mod->numsurfaces)
|
|
return;
|
|
|
|
for (t = 0; t < mod->numtextures; t++)
|
|
{
|
|
if (!mod->textures[t])
|
|
continue;
|
|
vbo = &mod->textures[t]->vbo;
|
|
BE_ClearVBO(vbo);
|
|
|
|
maxvboverts = 0;
|
|
maxvboelements = 0;
|
|
meshes = 0;
|
|
for (i=0 ; i<mod->numsurfaces ; i++)
|
|
{
|
|
if (mod->surfaces[i].texinfo->texture != mod->textures[t])
|
|
continue;
|
|
m = mod->surfaces[i].mesh;
|
|
if (!m)
|
|
continue;
|
|
|
|
meshes++;
|
|
maxvboelements += m->numindexes;
|
|
maxvboverts += m->numvertexes;
|
|
}
|
|
#if sizeof_index_t == 2
|
|
if (maxvboverts > (1<<(sizeof(index_t)*8))-1)
|
|
continue;
|
|
#endif
|
|
if (!maxvboverts)
|
|
continue;
|
|
|
|
//fixme: stop this from leaking!
|
|
vcount = 0;
|
|
ecount = 0;
|
|
|
|
pervertsize = sizeof(vecV_t)+ //coord
|
|
sizeof(vec2_t)+ //tex
|
|
sizeof(vec2_t)+ //lm
|
|
sizeof(vec3_t)+ //normal
|
|
sizeof(vec3_t)+ //sdir
|
|
sizeof(vec3_t)+ //tdir
|
|
sizeof(vec4_t); //colours
|
|
|
|
vbovdata = BZ_Malloc(maxvboverts*pervertsize);
|
|
vboedata = BZ_Malloc(maxvboelements*sizeof(index_t));
|
|
|
|
vbo->coord = (vecV_t*)(vbovdata);
|
|
vbo->texcoord = (vec2_t*)((char*)vbo->coord+maxvboverts*sizeof(*vbo->coord));
|
|
vbo->lmcoord = (vec2_t*)((char*)vbo->texcoord+maxvboverts*sizeof(*vbo->texcoord));
|
|
vbo->normals = (vec3_t*)((char*)vbo->lmcoord+maxvboverts*sizeof(*vbo->lmcoord));
|
|
vbo->svector = (vec3_t*)((char*)vbo->normals+maxvboverts*sizeof(*vbo->normals));
|
|
vbo->tvector = (vec3_t*)((char*)vbo->svector+maxvboverts*sizeof(*vbo->svector));
|
|
vbo->colours4f = (vec4_t*)((char*)vbo->tvector+maxvboverts*sizeof(*vbo->tvector));
|
|
vbo->indicies = (index_t*)vboedata;
|
|
|
|
vbo->meshcount = meshes;
|
|
vbo->meshlist = BZ_Malloc(meshes*sizeof(*vbo->meshlist));
|
|
|
|
meshes = 0;
|
|
|
|
for (i=0 ; i<mod->numsurfaces ; i++)
|
|
{
|
|
if (mod->surfaces[i].texinfo->texture != mod->textures[t])
|
|
continue;
|
|
m = mod->surfaces[i].mesh;
|
|
if (!m)
|
|
continue;
|
|
|
|
mod->surfaces[i].mark = &vbo->meshlist[meshes++];
|
|
*mod->surfaces[i].mark = NULL;
|
|
|
|
m->vbofirstvert = vcount;
|
|
m->vbofirstelement = ecount;
|
|
for (v = 0; v < m->numindexes; v++)
|
|
vbo->indicies[ecount++] = vcount + m->indexes[v];
|
|
for (v = 0; v < m->numvertexes; v++)
|
|
{
|
|
vbo->coord[vcount+v][0] = m->xyz_array[v][0];
|
|
vbo->coord[vcount+v][1] = m->xyz_array[v][1];
|
|
vbo->coord[vcount+v][2] = m->xyz_array[v][2];
|
|
if (m->st_array)
|
|
{
|
|
vbo->texcoord[vcount+v][0] = m->st_array[v][0];
|
|
vbo->texcoord[vcount+v][1] = m->st_array[v][1];
|
|
}
|
|
if (m->lmst_array)
|
|
{
|
|
vbo->lmcoord[vcount+v][0] = m->lmst_array[v][0];
|
|
vbo->lmcoord[vcount+v][1] = m->lmst_array[v][1];
|
|
}
|
|
if (m->normals_array)
|
|
{
|
|
vbo->normals[vcount+v][0] = m->normals_array[v][0];
|
|
vbo->normals[vcount+v][1] = m->normals_array[v][1];
|
|
vbo->normals[vcount+v][2] = m->normals_array[v][2];
|
|
}
|
|
if (m->snormals_array)
|
|
{
|
|
vbo->svector[vcount+v][0] = m->snormals_array[v][0];
|
|
vbo->svector[vcount+v][1] = m->snormals_array[v][1];
|
|
vbo->svector[vcount+v][2] = m->snormals_array[v][2];
|
|
}
|
|
if (m->tnormals_array)
|
|
{
|
|
vbo->tvector[vcount+v][0] = m->tnormals_array[v][0];
|
|
vbo->tvector[vcount+v][1] = m->tnormals_array[v][1];
|
|
vbo->tvector[vcount+v][2] = m->tnormals_array[v][2];
|
|
}
|
|
if (m->colors4f_array)
|
|
{
|
|
vbo->colours4f[vcount+v][0] = m->colors4f_array[v][0];
|
|
vbo->colours4f[vcount+v][1] = m->colors4f_array[v][1];
|
|
vbo->colours4f[vcount+v][2] = m->colors4f_array[v][2];
|
|
vbo->colours4f[vcount+v][3] = m->colors4f_array[v][3];
|
|
}
|
|
}
|
|
vcount += v;
|
|
}
|
|
|
|
// if (GL_BuildVBO(vbo, vbovdata, vcount*pervertsize, vboedata, ecount*sizeof(index_t)))
|
|
{
|
|
BZ_Free(vbovdata);
|
|
BZ_Free(vboedata);
|
|
}
|
|
}
|
|
//for (i=0 ; i<mod->numsurfaces ; i++)
|
|
//{
|
|
// if (!mod->surfaces[i].mark)
|
|
// Host_EndGame("Surfaces with bad textures detected\n");
|
|
//}
|
|
}
|
|
/*Wipes a vbo*/
|
|
void BE_ClearVBO(vbo_t *vbo)
|
|
{
|
|
}
|
|
|
|
/*upload all lightmaps at the start to reduce lags*/
|
|
void BE_UploadLightmaps(qboolean force)
|
|
{
|
|
int i;
|
|
|
|
for (i = 0; i < numlightmaps; i++)
|
|
{
|
|
if (!lightmap[i])
|
|
continue;
|
|
|
|
if (force)
|
|
{
|
|
lightmap[i]->rectchange.l = 0;
|
|
lightmap[i]->rectchange.t = 0;
|
|
lightmap[i]->rectchange.w = LMBLOCK_WIDTH;
|
|
lightmap[i]->rectchange.h = LMBLOCK_HEIGHT;
|
|
}
|
|
|
|
if (lightmap[i]->modified)
|
|
{
|
|
IDirect3DTexture9 *tex = lightmap_textures[i].ptr;
|
|
D3DLOCKED_RECT lock;
|
|
RECT rect;
|
|
glRect_t *theRect = &lightmap[i]->rectchange;
|
|
int r;
|
|
if (tex)
|
|
{
|
|
lightmap[i]->modified = 0;
|
|
rect.left = theRect->l;
|
|
rect.right = theRect->l + theRect->w;
|
|
rect.top = theRect->t;
|
|
rect.bottom = theRect->t + theRect->h;
|
|
|
|
IDirect3DTexture9_LockRect(tex, 0, &lock, &rect, 0);
|
|
for (r = 0; r < lightmap[i]->rectchange.h; r++)
|
|
{
|
|
memcpy((char*)lock.pBits + r*lock.Pitch, lightmap[i]->lightmaps+(theRect->l+((r+theRect->t)*LMBLOCK_WIDTH))*lightmap_bytes, lightmap[i]->rectchange.w*lightmap_bytes);
|
|
}
|
|
IDirect3DTexture9_UnlockRect(tex, 0);
|
|
theRect->l = LMBLOCK_WIDTH;
|
|
theRect->t = LMBLOCK_HEIGHT;
|
|
theRect->h = 0;
|
|
theRect->w = 0;
|
|
}
|
|
else
|
|
lightmap_textures[i] = R_AllocNewTexture(LMBLOCK_WIDTH, LMBLOCK_HEIGHT);
|
|
}
|
|
}
|
|
}
|
|
|
|
void BE_UploadAllLightmaps(void)
|
|
{
|
|
BE_UploadLightmaps(true);
|
|
}
|
|
|
|
qboolean BE_LightCullModel(vec3_t org, model_t *model)
|
|
{
|
|
#ifdef RTLIGHTS
|
|
if ((shaderstate.mode == BEM_LIGHT || shaderstate.mode == BEM_STENCIL))
|
|
{
|
|
/*true if hidden from current light*/
|
|
/*we have no rtlight support, so mneh*/
|
|
}
|
|
#endif
|
|
return false;
|
|
}
|
|
|
|
batch_t *BE_GetTempBatch(void)
|
|
{
|
|
if (shaderstate.wbatch >= shaderstate.maxwbatches)
|
|
{
|
|
shaderstate.wbatch++;
|
|
return NULL;
|
|
}
|
|
return &shaderstate.wbatches[shaderstate.wbatch++];
|
|
}
|
|
|
|
static void BE_RotateForEntity (const entity_t *e, const model_t *mod)
|
|
{
|
|
float mv[16];
|
|
float m[16];
|
|
|
|
shaderstate.curentity = e;
|
|
|
|
m[0] = e->axis[0][0];
|
|
m[1] = e->axis[0][1];
|
|
m[2] = e->axis[0][2];
|
|
m[3] = 0;
|
|
|
|
m[4] = e->axis[1][0];
|
|
m[5] = e->axis[1][1];
|
|
m[6] = e->axis[1][2];
|
|
m[7] = 0;
|
|
|
|
m[8] = e->axis[2][0];
|
|
m[9] = e->axis[2][1];
|
|
m[10] = e->axis[2][2];
|
|
m[11] = 0;
|
|
|
|
m[12] = e->origin[0];
|
|
m[13] = e->origin[1];
|
|
m[14] = e->origin[2];
|
|
m[15] = 1;
|
|
|
|
if (e->scale != 1 && e->scale != 0) //hexen 2 stuff
|
|
{
|
|
float z;
|
|
float escale;
|
|
escale = e->scale;
|
|
switch(e->drawflags&SCALE_TYPE_MASKIN)
|
|
{
|
|
default:
|
|
case SCALE_TYPE_UNIFORM:
|
|
VectorScale((m+0), escale, (m+0));
|
|
VectorScale((m+4), escale, (m+4));
|
|
VectorScale((m+8), escale, (m+8));
|
|
break;
|
|
case SCALE_TYPE_XYONLY:
|
|
VectorScale((m+0), escale, (m+0));
|
|
VectorScale((m+4), escale, (m+4));
|
|
break;
|
|
case SCALE_TYPE_ZONLY:
|
|
VectorScale((m+8), escale, (m+8));
|
|
break;
|
|
}
|
|
if (mod && (e->drawflags&SCALE_TYPE_MASKIN) != SCALE_TYPE_XYONLY)
|
|
{
|
|
switch(e->drawflags&SCALE_ORIGIN_MASKIN)
|
|
{
|
|
case SCALE_ORIGIN_CENTER:
|
|
z = ((mod->maxs[2] + mod->mins[2]) * (1-escale))/2;
|
|
VectorMA((m+12), z, e->axis[2], (m+12));
|
|
break;
|
|
case SCALE_ORIGIN_BOTTOM:
|
|
VectorMA((m+12), mod->mins[2]*(1-escale), e->axis[2], (m+12));
|
|
break;
|
|
case SCALE_ORIGIN_TOP:
|
|
VectorMA((m+12), -mod->maxs[2], e->axis[2], (m+12));
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
else if (mod && !strcmp(mod->name, "progs/eyes.mdl"))
|
|
{
|
|
/*resize eyes, to make them easier to see*/
|
|
m[14] -= (22 + 8);
|
|
VectorScale((m+0), 2, (m+0));
|
|
VectorScale((m+4), 2, (m+4));
|
|
VectorScale((m+8), 2, (m+8));
|
|
}
|
|
if (mod && !ruleset_allow_larger_models.ival && mod->clampscale != 1)
|
|
{ //possibly this should be on a per-frame basis, but that's a real pain to do
|
|
Con_DPrintf("Rescaling %s by %f\n", mod->name, mod->clampscale);
|
|
VectorScale((m+0), mod->clampscale, (m+0));
|
|
VectorScale((m+4), mod->clampscale, (m+4));
|
|
VectorScale((m+8), mod->clampscale, (m+8));
|
|
}
|
|
|
|
if (e->flags & Q2RF_WEAPONMODEL && r_refdef.currentplayernum>=0)
|
|
{
|
|
float *Matrix4_NewRotation(float a, float x, float y, float z);
|
|
/*FIXME: no bob*/
|
|
float iv[16];
|
|
Matrix4_Invert(r_refdef.m_view, iv);
|
|
Matrix4_NewRotation(90, 1, 0, 0);
|
|
Matrix4_Multiply(iv, m, mv);
|
|
Matrix4_Multiply(mv, Matrix4_NewRotation(-90, 1, 0, 0), iv);
|
|
Matrix4_Multiply(iv, Matrix4_NewRotation(90, 0, 0, 1), mv);
|
|
|
|
m[2] *= 0.1;
|
|
m[6] *= 0.1;
|
|
m[10] *= 0.1;
|
|
IDirect3DDevice9_SetTransform(pD3DDev9, D3DTS_WORLD, (D3DMATRIX*)mv);
|
|
}
|
|
else
|
|
{
|
|
IDirect3DDevice9_SetTransform(pD3DDev9, D3DTS_WORLD, (D3DMATRIX*)m);
|
|
}
|
|
}
|
|
|
|
void BE_SubmitBatch(batch_t *batch)
|
|
{
|
|
shaderstate.nummeshes = batch->meshes - batch->firstmesh;
|
|
if (!shaderstate.nummeshes)
|
|
return;
|
|
//FIXME: Why does this seem to work in GL?
|
|
if (batch->shader->flags & SHADER_FLARE)
|
|
return;
|
|
if (shaderstate.curentity != batch->ent)
|
|
{
|
|
BE_RotateForEntity(batch->ent, batch->ent->model);
|
|
shaderstate.curtime = r_refdef.time - shaderstate.curentity->shaderTime;
|
|
}
|
|
shaderstate.meshlist = batch->mesh + batch->firstmesh;
|
|
shaderstate.curshader = batch->shader;
|
|
shaderstate.curtexnums = batch->skin;
|
|
if (batch->lightmap < 0)
|
|
shaderstate.curlightmap = r_nulltex;
|
|
else
|
|
shaderstate.curlightmap = lightmap_textures[batch->lightmap];
|
|
|
|
BE_DrawMeshChain_Internal();
|
|
}
|
|
|
|
void BE_DrawMesh_List(shader_t *shader, int nummeshes, mesh_t **meshlist, vbo_t *vbo, texnums_t *texnums)
|
|
{
|
|
shaderstate.curshader = shader;
|
|
shaderstate.curtexnums = texnums;
|
|
shaderstate.curlightmap = r_nulltex;
|
|
shaderstate.meshlist = meshlist;
|
|
shaderstate.nummeshes = nummeshes;
|
|
|
|
BE_DrawMeshChain_Internal();
|
|
}
|
|
|
|
void BE_DrawMesh_Single(shader_t *shader, mesh_t *meshchain, vbo_t *vbo, texnums_t *texnums)
|
|
{
|
|
shaderstate.curtime = realtime;
|
|
shaderstate.curshader = shader;
|
|
shaderstate.curtexnums = texnums?texnums:&shader->defaulttextures;
|
|
shaderstate.curlightmap = r_nulltex;
|
|
shaderstate.meshlist = &meshchain;
|
|
shaderstate.nummeshes = 1;
|
|
|
|
BE_DrawMeshChain_Internal();
|
|
}
|
|
|
|
qboolean BE_ShouldDraw(entity_t *e)
|
|
{
|
|
if (!r_refdef.externalview && (e->externalmodelview & (1<<r_refdef.currentplayernum)))
|
|
return false;
|
|
if (!Cam_DrawPlayer(r_refdef.currentplayernum, e->keynum-1))
|
|
return false;
|
|
return true;
|
|
}
|
|
|
|
|
|
|
|
#ifdef Q3CLIENT
|
|
|
|
//q3 lightning gun
|
|
static void R_DrawLightning(entity_t *e)
|
|
{
|
|
vec3_t v;
|
|
vec3_t dir, cr;
|
|
float scale = e->scale;
|
|
float length;
|
|
|
|
vecV_t points[4];
|
|
vec2_t texcoords[4] = {{0, 0}, {0, 1}, {1, 1}, {1, 0}};
|
|
static index_t indexarray[6] = {0, 1, 2, 0, 2, 3};
|
|
|
|
mesh_t mesh;
|
|
|
|
if (!e->forcedshader)
|
|
return;
|
|
|
|
if (!scale)
|
|
scale = 10;
|
|
|
|
|
|
VectorSubtract(e->origin, e->oldorigin, dir);
|
|
length = Length(dir);
|
|
|
|
//this seems to be about right.
|
|
texcoords[2][0] = length/128;
|
|
texcoords[3][0] = length/128;
|
|
|
|
VectorSubtract(r_refdef.vieworg, e->origin, v);
|
|
CrossProduct(v, dir, cr);
|
|
VectorNormalize(cr);
|
|
|
|
VectorMA(e->origin, -scale/2, cr, points[0]);
|
|
VectorMA(e->origin, scale/2, cr, points[1]);
|
|
|
|
VectorSubtract(r_refdef.vieworg, e->oldorigin, v);
|
|
CrossProduct(v, dir, cr);
|
|
VectorNormalize(cr);
|
|
|
|
VectorMA(e->oldorigin, scale/2, cr, points[2]);
|
|
VectorMA(e->oldorigin, -scale/2, cr, points[3]);
|
|
|
|
memset(&mesh, 0, sizeof(mesh));
|
|
mesh.vbofirstelement = 0;
|
|
mesh.vbofirstvert = 0;
|
|
mesh.xyz_array = points;
|
|
mesh.indexes = indexarray;
|
|
mesh.numindexes = sizeof(indexarray)/sizeof(indexarray[0]);
|
|
mesh.colors4f_array = NULL;
|
|
mesh.lmst_array = NULL;
|
|
mesh.normals_array = NULL;
|
|
mesh.numvertexes = 4;
|
|
mesh.st_array = texcoords;
|
|
BE_DrawMesh_Single(e->forcedshader, &mesh, NULL, NULL);
|
|
}
|
|
//q3 railgun beam
|
|
static void R_DrawRailCore(entity_t *e)
|
|
{
|
|
vec3_t v;
|
|
vec3_t dir, cr;
|
|
float scale = e->scale;
|
|
float length;
|
|
|
|
mesh_t mesh;
|
|
vecV_t points[4];
|
|
vec2_t texcoords[4] = {{0, 0}, {0, 1}, {1, 1}, {1, 0}};
|
|
static index_t indexarray[6] = {0, 1, 2, 0, 2, 3};
|
|
vec4_t colors[4];
|
|
|
|
if (!e->forcedshader)
|
|
return;
|
|
|
|
if (!scale)
|
|
scale = 10;
|
|
|
|
|
|
VectorSubtract(e->origin, e->oldorigin, dir);
|
|
length = Length(dir);
|
|
|
|
//this seems to be about right.
|
|
texcoords[2][0] = length/128;
|
|
texcoords[3][0] = length/128;
|
|
|
|
VectorSubtract(r_refdef.vieworg, e->origin, v);
|
|
CrossProduct(v, dir, cr);
|
|
VectorNormalize(cr);
|
|
|
|
VectorMA(e->origin, -scale/2, cr, points[0]);
|
|
VectorMA(e->origin, scale/2, cr, points[1]);
|
|
|
|
VectorSubtract(r_refdef.vieworg, e->oldorigin, v);
|
|
CrossProduct(v, dir, cr);
|
|
VectorNormalize(cr);
|
|
|
|
VectorMA(e->oldorigin, scale/2, cr, points[2]);
|
|
VectorMA(e->oldorigin, -scale/2, cr, points[3]);
|
|
|
|
Vector4Copy(e->shaderRGBAf, colors[0]);
|
|
Vector4Copy(e->shaderRGBAf, colors[1]);
|
|
Vector4Copy(e->shaderRGBAf, colors[2]);
|
|
Vector4Copy(e->shaderRGBAf, colors[3]);
|
|
|
|
memset(&mesh, 0, sizeof(mesh));
|
|
mesh.vbofirstelement = 0;
|
|
mesh.vbofirstvert = 0;
|
|
mesh.xyz_array = points;
|
|
mesh.indexes = indexarray;
|
|
mesh.numindexes = sizeof(indexarray)/sizeof(indexarray[0]);
|
|
mesh.colors4f_array = (vec4_t*)colors;
|
|
mesh.lmst_array = NULL;
|
|
mesh.normals_array = NULL;
|
|
mesh.numvertexes = 4;
|
|
mesh.st_array = texcoords;
|
|
|
|
BE_DrawMesh_Single(e->forcedshader, &mesh, NULL, NULL);
|
|
}
|
|
#endif
|
|
static void BE_GenModelBatches(batch_t **batches)
|
|
{
|
|
int i;
|
|
entity_t *ent;
|
|
|
|
/*clear the batch list*/
|
|
for (i = 0; i < SHADER_SORT_COUNT; i++)
|
|
batches[i] = NULL;
|
|
|
|
if (!r_drawentities.ival)
|
|
return;
|
|
|
|
// draw sprites seperately, because of alpha blending
|
|
for (i=0 ; i<cl_numvisedicts ; i++)
|
|
{
|
|
ent = &cl_visedicts[i];
|
|
|
|
if (!BE_ShouldDraw(ent))
|
|
continue;
|
|
|
|
switch(ent->rtype)
|
|
{
|
|
case RT_MODEL:
|
|
default:
|
|
if (!ent->model)
|
|
continue;
|
|
if (ent->model->needload)
|
|
continue;
|
|
switch(ent->model->type)
|
|
{
|
|
case mod_brush:
|
|
if (r_drawentities.ival == 2)
|
|
continue;
|
|
Surf_GenBrushBatches(batches, ent);
|
|
break;
|
|
case mod_alias:
|
|
if (r_drawentities.ival == 3)
|
|
continue;
|
|
R_GAlias_GenerateBatches(ent, batches);
|
|
break;
|
|
case mod_sprite:
|
|
break;
|
|
}
|
|
break;
|
|
case RT_SPRITE:
|
|
//RQ_AddDistReorder(GLR_DrawSprite, currententity, NULL, currententity->origin);
|
|
break;
|
|
|
|
#ifdef Q3CLIENT
|
|
case RT_BEAM:
|
|
case RT_RAIL_RINGS:
|
|
case RT_LIGHTNING:
|
|
R_DrawLightning(ent);
|
|
continue;
|
|
case RT_RAIL_CORE:
|
|
R_DrawRailCore(ent);
|
|
continue;
|
|
#endif
|
|
|
|
case RT_POLY:
|
|
/*not implemented*/
|
|
break;
|
|
case RT_PORTALSURFACE:
|
|
/*nothing*/
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
static void BE_SubmitMeshesSortList(batch_t *sortlist)
|
|
{
|
|
batch_t *batch;
|
|
for (batch = sortlist; batch; batch = batch->next)
|
|
{
|
|
if (batch->meshes == batch->firstmesh)
|
|
continue;
|
|
|
|
if (batch->buildmeshes)
|
|
batch->buildmeshes(batch);
|
|
else
|
|
{
|
|
batch->shader = R_TextureAnimation(batch->ent->framestate.g[FS_REG].frame[0], batch->texture)->shader;
|
|
batch->skin = &batch->shader->defaulttextures;
|
|
}
|
|
|
|
if (batch->shader->flags & SHADER_NODLIGHT)
|
|
if (shaderstate.mode == BEM_LIGHT || shaderstate.mode == BEM_SMAPLIGHT)
|
|
continue;
|
|
|
|
if (batch->shader->flags & SHADER_SKY)
|
|
{
|
|
if (shaderstate.mode == BEM_STANDARD)
|
|
R_DrawSkyChain (batch);
|
|
continue;
|
|
}
|
|
|
|
BE_SubmitBatch(batch);
|
|
}
|
|
}
|
|
|
|
|
|
/*generates a new modelview matrix, as well as vpn vectors*/
|
|
static void R_MirrorMatrix(plane_t *plane)
|
|
{
|
|
float mirror[16];
|
|
float view[16];
|
|
float result[16];
|
|
|
|
vec3_t pnorm;
|
|
VectorNegate(plane->normal, pnorm);
|
|
|
|
mirror[0] = 1-2*pnorm[0]*pnorm[0];
|
|
mirror[1] = -2*pnorm[0]*pnorm[1];
|
|
mirror[2] = -2*pnorm[0]*pnorm[2];
|
|
mirror[3] = 0;
|
|
|
|
mirror[4] = -2*pnorm[1]*pnorm[0];
|
|
mirror[5] = 1-2*pnorm[1]*pnorm[1];
|
|
mirror[6] = -2*pnorm[1]*pnorm[2] ;
|
|
mirror[7] = 0;
|
|
|
|
mirror[8] = -2*pnorm[2]*pnorm[0];
|
|
mirror[9] = -2*pnorm[2]*pnorm[1];
|
|
mirror[10] = 1-2*pnorm[2]*pnorm[2];
|
|
mirror[11] = 0;
|
|
|
|
mirror[12] = -2*pnorm[0]*plane->dist;
|
|
mirror[13] = -2*pnorm[1]*plane->dist;
|
|
mirror[14] = -2*pnorm[2]*plane->dist;
|
|
mirror[15] = 1;
|
|
|
|
view[0] = vpn[0];
|
|
view[1] = vpn[1];
|
|
view[2] = vpn[2];
|
|
view[3] = 0;
|
|
|
|
view[4] = -vright[0];
|
|
view[5] = -vright[1];
|
|
view[6] = -vright[2];
|
|
view[7] = 0;
|
|
|
|
view[8] = vup[0];
|
|
view[9] = vup[1];
|
|
view[10] = vup[2];
|
|
view[11] = 0;
|
|
|
|
view[12] = r_refdef.vieworg[0];
|
|
view[13] = r_refdef.vieworg[1];
|
|
view[14] = r_refdef.vieworg[2];
|
|
view[15] = 1;
|
|
|
|
VectorMA(r_refdef.vieworg, 0.25, plane->normal, r_refdef.pvsorigin);
|
|
|
|
Matrix4_Multiply(mirror, view, result);
|
|
|
|
vpn[0] = result[0];
|
|
vpn[1] = result[1];
|
|
vpn[2] = result[2];
|
|
|
|
vright[0] = -result[4];
|
|
vright[1] = -result[5];
|
|
vright[2] = -result[6];
|
|
|
|
vup[0] = result[8];
|
|
vup[1] = result[9];
|
|
vup[2] = result[10];
|
|
|
|
r_refdef.vieworg[0] = result[12];
|
|
r_refdef.vieworg[1] = result[13];
|
|
r_refdef.vieworg[2] = result[14];
|
|
}
|
|
static entity_t *R_NearestPortal(plane_t *plane)
|
|
{
|
|
int i;
|
|
entity_t *best = NULL;
|
|
float dist, bestd = 0;
|
|
for (i = 0; i < cl_numvisedicts; i++)
|
|
{
|
|
if (cl_visedicts[i].rtype == RT_PORTALSURFACE)
|
|
{
|
|
dist = DotProduct(cl_visedicts[i].origin, plane->normal)-plane->dist;
|
|
dist = fabs(dist);
|
|
if (dist < 64 && (!best || dist < bestd))
|
|
best = &cl_visedicts[i];
|
|
}
|
|
}
|
|
return best;
|
|
}
|
|
|
|
static void TransformCoord(vec3_t in, vec3_t planea[3], vec3_t planeo, vec3_t viewa[3], vec3_t viewo, vec3_t result)
|
|
{
|
|
int i;
|
|
vec3_t local;
|
|
vec3_t transformed;
|
|
float d;
|
|
|
|
local[0] = in[0] - planeo[0];
|
|
local[1] = in[1] - planeo[1];
|
|
local[2] = in[2] - planeo[2];
|
|
|
|
VectorClear(transformed);
|
|
for ( i = 0 ; i < 3 ; i++ )
|
|
{
|
|
d = DotProduct(local, planea[i]);
|
|
VectorMA(transformed, d, viewa[i], transformed);
|
|
}
|
|
|
|
result[0] = transformed[0] + viewo[0];
|
|
result[1] = transformed[1] + viewo[1];
|
|
result[2] = transformed[2] + viewo[2];
|
|
}
|
|
static void TransformDir(vec3_t in, vec3_t planea[3], vec3_t viewa[3], vec3_t result)
|
|
{
|
|
int i;
|
|
float d;
|
|
vec3_t tmp;
|
|
|
|
VectorCopy(in, tmp);
|
|
|
|
VectorClear(result);
|
|
for ( i = 0 ; i < 3 ; i++ )
|
|
{
|
|
d = DotProduct(tmp, planea[i]);
|
|
VectorMA(result, d, viewa[i], result);
|
|
}
|
|
}
|
|
static R_RenderScene(void)
|
|
{
|
|
IDirect3DDevice9_SetTransform(pD3DDev9, D3DTS_PROJECTION, (D3DMATRIX*)r_refdef.m_projection);
|
|
IDirect3DDevice9_SetTransform(pD3DDev9, D3DTS_VIEW, (D3DMATRIX*)r_refdef.m_view);
|
|
R_SetFrustum (r_refdef.m_projection, r_refdef.m_view);
|
|
Surf_DrawWorld();
|
|
}
|
|
|
|
static void R_DrawPortal(batch_t *batch, batch_t **blist)
|
|
{
|
|
entity_t *view;
|
|
float glplane[4];
|
|
plane_t plane;
|
|
refdef_t oldrefdef;
|
|
mesh_t *mesh = batch->mesh[batch->firstmesh];
|
|
int sort;
|
|
|
|
if (r_refdef.recurse)
|
|
return;
|
|
|
|
VectorCopy(mesh->normals_array[0], plane.normal);
|
|
plane.dist = DotProduct(mesh->xyz_array[0], plane.normal);
|
|
|
|
//if we're too far away from the surface, don't draw anything
|
|
if (batch->shader->flags & SHADER_AGEN_PORTAL)
|
|
{
|
|
/*there's a portal alpha blend on that surface, that fades out after this distance*/
|
|
if (DotProduct(r_refdef.vieworg, plane.normal)-plane.dist > batch->shader->portaldist)
|
|
return;
|
|
}
|
|
//if we're behind it, then also don't draw anything.
|
|
if (DotProduct(r_refdef.vieworg, plane.normal)-plane.dist < 0)
|
|
return;
|
|
|
|
view = R_NearestPortal(&plane);
|
|
//if (!view)
|
|
// return;
|
|
|
|
oldrefdef = r_refdef;
|
|
r_refdef.recurse = true;
|
|
|
|
r_refdef.externalview = true;
|
|
|
|
if (!view || VectorCompare(view->origin, view->oldorigin))
|
|
{
|
|
r_refdef.flipcull ^= true;
|
|
R_MirrorMatrix(&plane);
|
|
}
|
|
else
|
|
{
|
|
float d;
|
|
vec3_t paxis[3], porigin, vaxis[3], vorg;
|
|
void PerpendicularVector( vec3_t dst, const vec3_t src );
|
|
|
|
/*calculate where the surface is meant to be*/
|
|
VectorCopy(mesh->normals_array[0], paxis[0]);
|
|
PerpendicularVector(paxis[1], paxis[0]);
|
|
CrossProduct(paxis[0], paxis[1], paxis[2]);
|
|
d = DotProduct(view->origin, plane.normal) - plane.dist;
|
|
VectorMA(view->origin, -d, paxis[0], porigin);
|
|
|
|
/*grab the camera origin*/
|
|
VectorNegate(view->axis[0], vaxis[0]);
|
|
VectorNegate(view->axis[1], vaxis[1]);
|
|
VectorCopy(view->axis[2], vaxis[2]);
|
|
VectorCopy(view->oldorigin, vorg);
|
|
|
|
VectorCopy(vorg, r_refdef.pvsorigin);
|
|
|
|
/*rotate it a bit*/
|
|
RotatePointAroundVector(vaxis[1], vaxis[0], view->axis[1], sin(realtime)*4);
|
|
CrossProduct(vaxis[0], vaxis[1], vaxis[2]);
|
|
|
|
TransformCoord(oldrefdef.vieworg, paxis, porigin, vaxis, vorg, r_refdef.vieworg);
|
|
TransformDir(vpn, paxis, vaxis, vpn);
|
|
TransformDir(vright, paxis, vaxis, vright);
|
|
TransformDir(vup, paxis, vaxis, vup);
|
|
}
|
|
Matrix4_ModelViewMatrixFromAxis(r_refdef.m_view, vpn, vright, vup, r_refdef.vieworg);
|
|
VectorAngles(vpn, vup, r_refdef.viewangles);
|
|
VectorCopy(r_refdef.vieworg, r_origin);
|
|
|
|
/*FIXME: the batch stuff should be done in renderscene*/
|
|
|
|
/*fixup the first mesh index*/
|
|
for (sort = 0; sort < SHADER_SORT_COUNT; sort++)
|
|
for (batch = blist[sort]; batch; batch = batch->next)
|
|
{
|
|
batch->firstmesh = batch->meshes;
|
|
}
|
|
|
|
/*FIXME: can we get away with stenciling the screen?*/
|
|
/*Add to frustum culling instead of clip planes?*/
|
|
glplane[0] = plane.normal[0];
|
|
glplane[1] = plane.normal[1];
|
|
glplane[2] = plane.normal[2];
|
|
glplane[3] = -plane.dist;
|
|
IDirect3DDevice9_SetClipPlane(pD3DDev9, 0, glplane);
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_CLIPPLANEENABLE, D3DCLIPPLANE0);
|
|
R_RenderScene();
|
|
IDirect3DDevice9_SetRenderState(pD3DDev9, D3DRS_CLIPPLANEENABLE, 0);
|
|
|
|
for (sort = 0; sort < SHADER_SORT_COUNT; sort++)
|
|
for (batch = blist[sort]; batch; batch = batch->next)
|
|
{
|
|
batch->firstmesh = 0;
|
|
}
|
|
r_refdef = oldrefdef;
|
|
|
|
/*broken stuff*/
|
|
AngleVectors (r_refdef.viewangles, vpn, vright, vup);
|
|
VectorCopy (r_refdef.vieworg, r_origin);
|
|
|
|
IDirect3DDevice9_SetTransform(pD3DDev9, D3DTS_PROJECTION, (D3DMATRIX*)r_refdef.m_projection);
|
|
IDirect3DDevice9_SetTransform(pD3DDev9, D3DTS_VIEW, (D3DMATRIX*)r_refdef.m_view);
|
|
R_SetFrustum (r_refdef.m_projection, r_refdef.m_view);
|
|
}
|
|
|
|
static void BE_SubmitMeshesPortals(batch_t **worldlist, batch_t *dynamiclist)
|
|
{
|
|
batch_t *batch, *old;
|
|
int i;
|
|
/*attempt to draw portal shaders*/
|
|
if (shaderstate.mode == BEM_STANDARD)
|
|
{
|
|
for (i = 0; i < 2; i++)
|
|
{
|
|
for (batch = i?dynamiclist:worldlist[SHADER_SORT_PORTAL]; batch; batch = batch->next)
|
|
{
|
|
if (batch->meshes == batch->firstmesh)
|
|
continue;
|
|
|
|
if (batch->buildmeshes)
|
|
batch->buildmeshes(batch);
|
|
else
|
|
batch->shader = R_TextureAnimation(batch->ent->framestate.g[FS_REG].frame[0], batch->texture)->shader;
|
|
|
|
|
|
/*draw already-drawn portals as depth-only, to ensure that their contents are not harmed*/
|
|
BE_SelectMode(BEM_DEPTHONLY, 0);
|
|
for (old = worldlist[SHADER_SORT_PORTAL]; old && old != batch; old = old->next)
|
|
{
|
|
if (old->meshes == old->firstmesh)
|
|
continue;
|
|
BE_SubmitBatch(old);
|
|
}
|
|
if (!old)
|
|
{
|
|
for (old = dynamiclist; old != batch; old = old->next)
|
|
{
|
|
if (old->meshes == old->firstmesh)
|
|
continue;
|
|
BE_SubmitBatch(old);
|
|
}
|
|
}
|
|
BE_SelectMode(BEM_STANDARD, 0);
|
|
|
|
R_DrawPortal(batch, worldlist);
|
|
|
|
/*clear depth again*/
|
|
IDirect3DDevice9_Clear(pD3DDev9, 0, NULL, D3DCLEAR_ZBUFFER, D3DCOLOR_XRGB(0,0,0), 1, 0);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
void BE_SubmitMeshes (qboolean drawworld, batch_t **blist)
|
|
{
|
|
model_t *model = cl.worldmodel;
|
|
int i;
|
|
|
|
for (i = SHADER_SORT_PORTAL; i < SHADER_SORT_COUNT; i++)
|
|
{
|
|
if (drawworld)
|
|
{
|
|
if (i == SHADER_SORT_PORTAL /*&& !r_noportals.ival*/ && !r_refdef.recurse)
|
|
BE_SubmitMeshesPortals(model->batches, blist[i]);
|
|
|
|
BE_SubmitMeshesSortList(model->batches[i]);
|
|
}
|
|
BE_SubmitMeshesSortList(blist[i]);
|
|
}
|
|
}
|
|
|
|
void BE_DrawWorld (qbyte *vis)
|
|
{
|
|
batch_t *batches[SHADER_SORT_COUNT];
|
|
RSpeedLocals();
|
|
|
|
shaderstate.curentity = NULL;
|
|
|
|
if (!r_refdef.recurse)
|
|
{
|
|
if (shaderstate.wbatch > shaderstate.maxwbatches)
|
|
{
|
|
int newm = shaderstate.wbatch;
|
|
shaderstate.wbatches = BZ_Realloc(shaderstate.wbatches, newm * sizeof(*shaderstate.wbatches));
|
|
memset(shaderstate.wbatches + shaderstate.maxwbatches, 0, (newm - shaderstate.maxwbatches) * sizeof(*shaderstate.wbatches));
|
|
shaderstate.maxwbatches = newm;
|
|
}
|
|
shaderstate.wbatch = 0;
|
|
}
|
|
|
|
BE_GenModelBatches(batches);
|
|
|
|
if (vis)
|
|
{
|
|
BE_UploadLightmaps(false);
|
|
|
|
//make sure the world draws correctly
|
|
r_worldentity.shaderRGBAf[0] = 1;
|
|
r_worldentity.shaderRGBAf[1] = 1;
|
|
r_worldentity.shaderRGBAf[2] = 1;
|
|
r_worldentity.shaderRGBAf[3] = 1;
|
|
r_worldentity.axis[0][0] = 1;
|
|
r_worldentity.axis[1][1] = 1;
|
|
r_worldentity.axis[2][2] = 1;
|
|
|
|
BE_SelectMode(BEM_STANDARD, 0);
|
|
|
|
RSpeedRemark();
|
|
BE_SubmitMeshes(true, batches);
|
|
RSpeedEnd(RSPEED_WORLD);
|
|
}
|
|
else
|
|
{
|
|
RSpeedRemark();
|
|
BE_SubmitMeshes(false, batches);
|
|
RSpeedEnd(RSPEED_DRAWENTITIES);
|
|
}
|
|
|
|
BE_RotateForEntity(&r_worldentity, NULL);
|
|
}
|
|
|
|
|
|
#endif
|