quakeforge/libs/video/renderer/gl/gl_mod_alias.c
Bill Currie 7ead5a91f8 don't let gcc automaticly inline functions but make gcc inline bigger
functions when told to. also make gcc warn if it can't inline a function.

Explicitly inline several functions (including moving VectorNormalize to
mathlib.h so it /can/ be) resulting in a 5.5% speedup for spam2 (88 to 92
fps)
2003-08-11 06:05:07 +00:00

782 lines
18 KiB
C

/*
gl_mod_alias.c
(description)
Copyright (C) 1996-1997 Id Software, Inc.
This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to:
Free Software Foundation, Inc.
59 Temple Place - Suite 330
Boston, MA 02111-1307, USA
*/
#ifdef HAVE_CONFIG_H
# include "config.h"
#endif
static __attribute__ ((unused)) const char rcsid[] =
"$Id$";
#ifdef HAVE_STRING_H
# include <string.h>
#endif
#ifdef HAVE_STRINGS_H
# include <strings.h>
#endif
#include <math.h>
#include <stdio.h>
#include <stdlib.h>
#include "QF/console.h"
#include "QF/cvar.h"
#include "QF/locs.h"
#include "QF/mathlib.h"
#include "QF/qargs.h"
#include "QF/render.h"
#include "QF/skin.h"
#include "QF/sound.h"
#include "QF/sys.h"
#include "QF/vid.h"
#include "QF/GL/defines.h"
#include "QF/GL/funcs.h"
#include "QF/GL/qf_rlight.h"
#include "QF/GL/qf_rmain.h"
#include "QF/GL/qf_rsurf.h"
#include "QF/GL/qf_screen.h"
#include "QF/GL/qf_vid.h"
#include "compat.h"
#include "r_cvar.h"
#include "r_dynamic.h"
#include "r_local.h"
#include "view.h"
typedef struct {
vec3_t vert;
float lightdot;
} blended_vert_t;
typedef struct {
blended_vert_t *verts;
int *order;
tex_coord_t *tex_coord;
int count;
} vert_order_t;
float r_avertexnormals[NUMVERTEXNORMALS][3] = {
#include "anorms.h"
};
// precalculated dot products for quantized angles
#define SHADEDOT_QUANT 16
float r_avertexnormal_dots[SHADEDOT_QUANT][256] =
#include "anorm_dots.h"
;
vec3_t shadevector;
static inline void
GL_DrawAliasFrameTri (vert_order_t *vo)
{
float color[4];
int count;
blended_vert_t *verts;
tex_coord_t *tex_coord;
verts = vo->verts;
tex_coord = vo->tex_coord;
color[3] = modelalpha;
count = vo->count;
qfglBegin (GL_TRIANGLES);
do {
qfglTexCoord2fv (tex_coord->st);
VectorMultAdd (ambientcolor, verts->lightdot, shadecolor, color);
qfglColor4fv (color);
qfglVertex3fv (verts->vert);
tex_coord++;
verts++;
} while (count--);
qfglEnd();
}
static inline void
GL_DrawAliasFrameTri_fb (vert_order_t *vo)
{
int count;
float color[4] = { 1.0, 1.0, 1.0, 0.0};
blended_vert_t *verts;
tex_coord_t *tex_coord;
verts = vo->verts;
color[3] = modelalpha * 1.0;
count = vo->count;
tex_coord = vo->tex_coord;
qfglBegin (GL_TRIANGLES);
do {
qfglTexCoord2fv (tex_coord->st);
qfglColor4fv (color);
qfglVertex3fv (verts->vert);
tex_coord++;
verts++;
} while (--count);
qfglEnd();
}
static inline void
GL_DrawAliasFrameTriMulti (vert_order_t *vo)
{
float color[4];
int count;
blended_vert_t *verts;
tex_coord_t *tex_coord;
verts = vo->verts;
tex_coord = vo->tex_coord;
color[3] = modelalpha;
count = vo->count;
qfglBegin (GL_TRIANGLES);
do {
// texture coordinates come from the draw list
qglMultiTexCoord2fv (gl_mtex_enum + 0, tex_coord->st);
qglMultiTexCoord2fv (gl_mtex_enum + 1, tex_coord->st);
tex_coord++;
// normals and vertexes come from the frame list
VectorMultAdd (ambientcolor, verts->lightdot, shadecolor, color);
qfglColor4fv (color);
qfglVertex3fv (verts->vert);
verts++;
} while (--count);
qfglEnd ();
}
static inline void
GL_DrawAliasFrame (vert_order_t *vo)
{
float color[4];
int count;
int *order;
blended_vert_t *verts;
verts = vo->verts;
order = vo->order;
color[3] = modelalpha;
while ((count = *order++)) {
// get the vertex count and primitive type
if (count < 0) {
count = -count;
qfglBegin (GL_TRIANGLE_FAN);
} else {
qfglBegin (GL_TRIANGLE_STRIP);
}
do {
// texture coordinates come from the draw list
qfglTexCoord2fv ((float *) order);
order += 2;
// normals and vertexes come from the frame list
VectorMultAdd (ambientcolor, verts->lightdot, shadecolor, color);
qfglColor4fv (color);
qfglVertex3fv (verts->vert);
verts++;
} while (--count);
qfglEnd ();
}
}
static inline void
GL_DrawAliasFrame_fb (vert_order_t *vo)
{
int count;
int *order;
blended_vert_t *verts;
verts = vo->verts;
order = vo->order;
color_white[3] = modelalpha * 255;
qfglColor4ubv (color_white);
while ((count = *order++)) {
// get the vertex count and primitive type
if (count < 0) {
count = -count;
qfglBegin (GL_TRIANGLE_FAN);
} else {
qfglBegin (GL_TRIANGLE_STRIP);
}
do {
// texture coordinates come from the draw list
qfglTexCoord2fv ((float *) order);
order += 2;
qfglVertex3fv (verts->vert);
verts++;
} while (--count);
qfglEnd ();
}
}
static inline void
GL_DrawAliasFrameMulti (vert_order_t *vo)
{
float color[4];
int count;
int *order;
blended_vert_t *verts;
verts = vo->verts;
order = vo->order;
color[3] = modelalpha;
while ((count = *order++)) {
// get the vertex count and primitive type
if (count < 0) {
count = -count;
qfglBegin (GL_TRIANGLE_FAN);
} else {
qfglBegin (GL_TRIANGLE_STRIP);
}
do {
// texture coordinates come from the draw list
qglMultiTexCoord2fv (gl_mtex_enum + 0, (float *) order);
qglMultiTexCoord2fv (gl_mtex_enum + 1, (float *) order);
order += 2;
// normals and vertexes come from the frame list
VectorMultAdd (ambientcolor, verts->lightdot, shadecolor, color);
qfglColor4fv (color);
qfglVertex3fv (verts->vert);
verts++;
} while (--count);
qfglEnd ();
}
}
/*
GL_DrawAliasShadow
Standard shadow drawing
*/
static void
GL_DrawAliasShadow (aliashdr_t *paliashdr, vert_order_t *vo)
{
float height, lheight;
int count;
int *order;
vec3_t point;
blended_vert_t *verts;
verts = vo->verts;
order = vo->order;
lheight = currententity->origin[2] - lightspot[2];
height = -lheight + 1.0;
while ((count = *order++)) {
// get the vertex count and primitive type
if (count < 0) {
count = -count;
qfglBegin (GL_TRIANGLE_FAN);
} else
qfglBegin (GL_TRIANGLE_STRIP);
do {
order += 2; // skip texture coords
// normals and vertexes come from the frame list
point[0] =
verts->vert[0] * paliashdr->mdl.scale[0] +
paliashdr->mdl.scale_origin[0];
point[1] =
verts->vert[1] * paliashdr->mdl.scale[1] +
paliashdr->mdl.scale_origin[1];
point[2] =
verts->vert[2] * paliashdr->mdl.scale[2] +
paliashdr->mdl.scale_origin[2] + lheight;
point[0] -= shadevector[0] * point[2];
point[1] -= shadevector[1] * point[2];
point[2] = height;
qfglVertex3fv (point);
verts++;
} while (--count);
qfglEnd ();
}
}
static inline vert_order_t *
GL_GetAliasFrameVerts16 (int frame, aliashdr_t *paliashdr, entity_t *e)
{
float interval;
int count, numposes, pose, i;
trivertx16_t *verts;
vert_order_t *vo;
blended_vert_t *vo_v;
if ((frame >= paliashdr->mdl.numframes) || (frame < 0)) {
if (developer->int_val)
Con_Printf ("R_AliasSetupFrame: no such frame %d %s\n", frame,
currententity->model->name);
frame = 0;
}
pose = paliashdr->frames[frame].firstpose;
numposes = paliashdr->frames[frame].numposes;
verts = (trivertx16_t *) ((byte *) paliashdr + paliashdr->posedata);
count = paliashdr->poseverts;
vo = Hunk_TempAlloc (sizeof (*vo) + count * sizeof (blended_vert_t));
vo->order = (int *) ((byte *) paliashdr + paliashdr->commands);
vo->verts = (blended_vert_t *) &vo[1];
if (paliashdr->tex_coord) {
vo->tex_coord = (tex_coord_t *) ((byte *) paliashdr + paliashdr->tex_coord);
} else {
vo->tex_coord = NULL;
}
vo->count = count;
if (numposes > 1) {
interval = paliashdr->frames[frame].interval;
pose += (int) (r_realtime / interval) % numposes;
} else {
/*
One tenth of a second is good for most Quake animations. If
the nextthink is longer then the animation is usually meant
to pause (e.g. check out the shambler magic animation in
shambler.qc). If its shorter then things will still be
smoothed partly, and the jumps will be less noticable
because of the shorter time. So, this is probably a good
assumption.
*/
interval = 0.1;
}
if (gl_lerp_anim->int_val) {
trivertx16_t *verts1, *verts2;
float blend;
vec3_t v1, v2;
e->frame_interval = interval;
if (e->pose2 != pose) {
e->frame_start_time = r_realtime;
if (e->pose2 == -1) {
e->pose1 = pose;
} else {
e->pose1 = e->pose2;
}
e->pose2 = pose;
blend = 0.0;
} else {
blend = (r_realtime - e->frame_start_time) / e->frame_interval;
}
// wierd things start happening if blend passes 1
if (r_paused || blend > 1.0)
blend = 1.0;
if (blend == 0.0) {
verts = verts + e->pose1 * count;
} else if (blend == 1.0) {
verts = verts + e->pose2 * count;
} else {
verts1 = verts + e->pose1 * count;
verts2 = verts + e->pose2 * count;
for (i = 0, vo_v = vo->verts; i < count;
i++, vo_v++, verts1++, verts2++) {
float *n1, *n2;
float d1, d2;
VectorBlend (v1, v2, blend, vo_v->vert);
n1 = r_avertexnormals[verts1->lightnormalindex];
n2 = r_avertexnormals[verts2->lightnormalindex];
d1 = DotProduct (shadevector, n1);
d2 = DotProduct (shadevector, n2);
vo_v->lightdot = max (0, d1 * (1.0 - blend) + d2 * blend);
}
return vo;
}
} else {
verts += pose * count;
}
for (i = 0, vo_v = vo->verts; i < count; i++, vo_v++, verts++) {
float *n;
float d;
n = r_avertexnormals[verts->lightnormalindex];
d = DotProduct (shadevector, n);
vo_v->lightdot = max (0.0, d);
}
return vo;
}
static inline vert_order_t *
GL_GetAliasFrameVerts (int frame, aliashdr_t *paliashdr, entity_t *e)
{
float interval;
int count, numposes, pose, i;
trivertx_t *verts;
vert_order_t *vo;
blended_vert_t *vo_v;
if ((frame >= paliashdr->mdl.numframes) || (frame < 0)) {
if (developer->int_val)
Con_Printf ("R_AliasSetupFrame: no such frame %d %s\n", frame,
currententity->model->name);
frame = 0;
}
pose = paliashdr->frames[frame].firstpose;
numposes = paliashdr->frames[frame].numposes;
verts = (trivertx_t *) ((byte *) paliashdr + paliashdr->posedata);
count = paliashdr->poseverts;
vo = Hunk_TempAlloc (sizeof (*vo) + count * sizeof (blended_vert_t));
vo->order = (int *) ((byte *) paliashdr + paliashdr->commands);
vo->verts = (blended_vert_t *) &vo[1];
if (paliashdr->tex_coord) {
vo->tex_coord = (tex_coord_t *) ((byte *) paliashdr + paliashdr->tex_coord);
} else {
vo->tex_coord = NULL;
}
vo->count = count;
if (numposes > 1) {
interval = paliashdr->frames[frame].interval;
pose += (int) (r_realtime / interval) % numposes;
} else {
/*
One tenth of a second is good for most Quake animations. If
the nextthink is longer then the animation is usually meant
to pause (e.g. check out the shambler magic animation in
shambler.qc). If its shorter then things will still be
smoothed partly, and the jumps will be less noticable
because of the shorter time. So, this is probably a good
assumption.
*/
interval = 0.1;
}
if (gl_lerp_anim->int_val) {
trivertx_t *verts1, *verts2;
float blend;
e->frame_interval = interval;
if (e->pose2 != pose) {
e->frame_start_time = r_realtime;
if (e->pose2 == -1) {
e->pose1 = pose;
} else {
e->pose1 = e->pose2;
}
e->pose2 = pose;
blend = 0;
} else {
blend = (r_realtime - e->frame_start_time) / e->frame_interval;
}
// wierd things start happening if blend passes 1
if (r_paused || blend > 1.0)
blend = 1.0;
if (blend == 0.0) {
verts = verts + e->pose1 * count;
} else if (blend == 1.0) {
verts = verts + e->pose2 * count;
} else {
verts1 = verts + e->pose1 * count;
verts2 = verts + e->pose2 * count;
for (i = 0, vo_v = vo->verts; i < count;
i++, vo_v++, verts1++, verts2++) {
float *n1, *n2;
float d1, d2;
VectorBlend (verts1->v, verts2->v, blend, vo_v->vert);
n1 = r_avertexnormals[verts1->lightnormalindex];
n2 = r_avertexnormals[verts2->lightnormalindex];
d1 = DotProduct (shadevector, n1);
d2 = DotProduct (shadevector, n2);
vo_v->lightdot = max (0, d1 * (1.0 - blend) + d2 * blend);
}
return vo;
}
} else {
verts += pose * count;
}
for (i = 0, vo_v = vo->verts; i < count; i++, vo_v++, verts++) {
float *n;
float d;
VectorCopy (verts->v, vo_v->vert);
n = r_avertexnormals[verts->lightnormalindex];
d = DotProduct (shadevector, n);
vo_v->lightdot = max (0.0, d);
}
return vo;
}
static maliasskindesc_t *
R_AliasGetSkindesc (int skinnum, aliashdr_t *ahdr)
{
maliasskindesc_t *pskindesc;
maliasskingroup_t *paliasskingroup;
if ((skinnum >= ahdr->mdl.numskins) || (skinnum < 0)) {
Con_DPrintf ("R_AliasSetupSkin: no such skin # %d\n", skinnum);
skinnum = 0;
}
pskindesc = ((maliasskindesc_t *)
((byte *) ahdr + ahdr->skindesc)) + skinnum;
if (pskindesc->type == ALIAS_SKIN_GROUP) {
float fullskininterval;
int i;
int numskins;
float skintargettime, skintime;
float *pskinintervals;
paliasskingroup = (maliasskingroup_t *) ((byte *) ahdr +
pskindesc->skin);
pskinintervals = (float *)
((byte *) ahdr + paliasskingroup->intervals);
numskins = paliasskingroup->numskins;
fullskininterval = pskinintervals[numskins - 1];
skintime = r_realtime + currententity->syncbase;
skintargettime = skintime -
((int) (skintime / fullskininterval)) * fullskininterval;
for (i = 0; i < (numskins - 1); i++) {
if (pskinintervals[i] > skintargettime)
break;
}
pskindesc = &paliasskingroup->skindescs[i];
}
return pskindesc;
}
void
R_DrawAliasModel (entity_t *e)
{
float add, an, minshade, radius, shade;
int texture, i;
int fb_texture = 0;
unsigned int lnum;
aliashdr_t *paliashdr;
model_t *model;
vec3_t dist, scale;
vert_order_t *vo;
model = e->model;
radius = model->radius;
if (e->scale != 1.0)
radius *= e->scale;
if (R_CullSphere (e->origin, radius))
return;
VectorSubtract (r_origin, e->origin, modelorg);
if (!model->fullbright) {
// get lighting information
R_LightPoint (e->origin);
ambientcolor[0] *= e->colormod[0];
ambientcolor[1] *= e->colormod[1];
ambientcolor[2] *= e->colormod[2];
VectorScale (ambientcolor, 0.005, ambientcolor);
VectorCopy (ambientcolor, shadecolor);
for (lnum = 0; lnum < r_maxdlights; lnum++) {
if (r_dlights[lnum].die >= r_realtime) {
float d;
VectorSubtract (e->origin, r_dlights[lnum].origin, dist);
d = DotProduct (dist, dist);
d = max (d, 64.0) * 200.0;
add = r_dlights[lnum].radius * r_dlights[lnum].radius * 8.0 /
d;
if (add > 0.0)
VectorMultAdd (ambientcolor, add, r_dlights[lnum].color,
ambientcolor);
}
}
// clamp lighting so it doesn't overbright as much
for (i = 0; i < 3; i++) {
ambientcolor[i] = min (ambientcolor[i], 128.0 / 200.0);
if (ambientcolor[i] + shadecolor[i] > 1)
shadecolor[i] = 1 - ambientcolor[i];
}
// always give the gun some light
shade = max (ambientcolor[0], max (ambientcolor[1], ambientcolor[2]));
minshade = model->min_light;
if (shade < minshade) {
ambientcolor[0] += minshade - shade;
ambientcolor[1] += minshade - shade;
ambientcolor[2] += minshade - shade;
}
an = e->angles[1] * (M_PI / 180.0);
shadevector[0] = cos (-an);
shadevector[1] = sin (-an);
shadevector[2] = 1.0;
VectorNormalize (shadevector);
}
modelalpha = e->colormod[3];
// locate the proper data
paliashdr = Cache_Get (&e->model->cache);
c_alias_polys += paliashdr->mdl.numtris;
// if the model has a colorised/external skin, use it, otherwise use
// the skin embedded in the model data
if (e->skin && !gl_nocolors->int_val) {
skin_t *skin = e->skin;
texture = skin->texture;
if (gl_fb_models->int_val) {
fb_texture = skin->fb_texture;
}
} else {
maliasskindesc_t *skindesc;
skindesc = R_AliasGetSkindesc (e->skinnum, paliashdr);
texture = skindesc->texnum;
if (gl_fb_models->int_val && !model->fullbright)
fb_texture = skindesc->fb_texnum;
}
if (paliashdr->mdl.ident == POLYHEADER16) {
VectorScale (paliashdr->mdl.scale, e->scale / 256.0, scale);
vo = GL_GetAliasFrameVerts16 (e->frame, paliashdr, e);
} else {
VectorScale (paliashdr->mdl.scale, e->scale, scale);
vo = GL_GetAliasFrameVerts (e->frame, paliashdr, e);
}
// setup the transform
qfglPushMatrix ();
R_RotateForEntity (e);
qfglTranslatef (paliashdr->mdl.scale_origin[0],
paliashdr->mdl.scale_origin[1],
paliashdr->mdl.scale_origin[2]);
qfglScalef (scale[0], scale[1], scale[2]);
if (modelalpha < 1.0)
qfglDepthMask (GL_FALSE);
// draw all the triangles
if (model->fullbright) {
qfglBindTexture (GL_TEXTURE_2D, texture);
if (vo->tex_coord)
GL_DrawAliasFrameTri_fb (vo);
else
GL_DrawAliasFrame_fb (vo);
} else if (!fb_texture) {
// Model has no fullbrights, don't bother with multi
qfglBindTexture (GL_TEXTURE_2D, texture);
if (vo->tex_coord)
GL_DrawAliasFrameTri (vo);
else
GL_DrawAliasFrame (vo);
} else { // try multitexture
if (gl_mtex_active) { // set up the textures
qglActiveTexture (gl_mtex_enum + 0);
qfglBindTexture (GL_TEXTURE_2D, texture);
qglActiveTexture (gl_mtex_enum + 1);
qfglBindTexture (GL_TEXTURE_2D, fb_texture);
qfglTexEnvf (GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_DECAL);
qfglEnable (GL_TEXTURE_2D);
// do the heavy lifting
if (vo->tex_coord)
GL_DrawAliasFrameTriMulti (vo);
else
GL_DrawAliasFrameMulti (vo);
// restore the settings
qfglDisable (GL_TEXTURE_2D);
qglActiveTexture (gl_mtex_enum + 0);
} else {
if (vo->tex_coord) {
qfglBindTexture (GL_TEXTURE_2D, texture);
GL_DrawAliasFrameTri (vo);
qfglBindTexture (GL_TEXTURE_2D, fb_texture);
GL_DrawAliasFrameTri_fb (vo);
} else {
qfglBindTexture (GL_TEXTURE_2D, texture);
GL_DrawAliasFrame (vo);
qfglBindTexture (GL_TEXTURE_2D, fb_texture);
GL_DrawAliasFrame_fb (vo);
}
}
}
qfglPopMatrix ();
// FIXME: Translucent objects should cast colored shadows
// torches, grenades, and lightning bolts do not have shadows
if (r_shadows->int_val && model->shadow_alpha) {
qfglPushMatrix ();
R_RotateForEntity (e);
qfglDisable (GL_TEXTURE_2D);
qfglDepthMask (GL_FALSE);
color_black[3] = modelalpha * model->shadow_alpha;
qfglColor4ubv (color_black);
GL_DrawAliasShadow (paliashdr, vo);
qfglDepthMask (GL_TRUE);
qfglEnable (GL_TEXTURE_2D);
qfglPopMatrix ();
} else if (modelalpha < 1.0)
qfglDepthMask (GL_TRUE);
Cache_Release (&e->model->cache);
}