nzp-team/dquakeplus
0
0
Fork 0
mirror of https://github.com/nzp-team/dquakeplus.git synced 2025-01-31 11:40:36 +00:00
Code Issues Projects Releases Packages Wiki Activity

Refactor clut4 conversion to its own file, use 8bpp resampling to better reuse code

Browse source
This commit is contained in:
Shpuld Shpuldson 2023-09-22 08:26:11 +03:00
parent b2c43cbe5a
commit eab5ddd082
6 changed files with 292 additions and 264 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

3
MakePHAT
View file

@ -103,7 +103,8 @@ HARDWARE_VIDEO_ONLY_OBJS = \
source/psp/video_hardware_surface.o \
source/psp/video_hardware_warp.o \
source/psp/video_hardware_fog.o \
source/psp/video_hardware_dxtn.o
source/psp/video_hardware_dxtn.o \
source/psp/video_hardware_colorquant.o
HARDWARE_VIDEO_ONLY_FLAGS = -DPSP_HARDWARE_VIDEO
OBJS = $(COMMON_OBJS) $(HARDWARE_VIDEO_ONLY_OBJS)

3
MakeSLIM
View file

@ -103,7 +103,8 @@ HARDWARE_VIDEO_ONLY_OBJS = \
source/psp/video_hardware_surface.o \
source/psp/video_hardware_warp.o \
source/psp/video_hardware_fog.o \
source/psp/video_hardware_dxtn.o
source/psp/video_hardware_dxtn.o \
source/psp/video_hardware_colorquant.o
HARDWARE_VIDEO_ONLY_FLAGS = -DPSP_HARDWARE_VIDEO
OBJS = $(COMMON_OBJS) $(HARDWARE_VIDEO_ONLY_OBJS)

4
source/psp/video_hardware.h
View file

@ -30,6 +30,7 @@ void GL_Upload16(int texture_index, const byte *data, int width, int height);
int GL_LoadTexture(const char *identifier, int width, int height, const byte *data, qboolean stretch_to_power_of_two, int filter, int mipmap_level);
// CLUT4
int GL_LoadTexture4(const char *identifier, unsigned int width, unsigned int height, const byte *data, int filter, qboolean swizzled);
int GL_LoadTexture8to4(const char *identifier, unsigned int width, unsigned int height, const byte *data, const byte *pal, int filter);
int GL_LoadTextureLM (const char *identifier, int width, int height, const byte *data, int bpp, int filter, qboolean update, int forcopy);
int GL_LoadImages (const char *identifier, int width, int height, const byte *data, qboolean stretch_to_power_of_two, int filter, int mipmap_level, int bpp);
@ -421,5 +422,6 @@ int D_DrawParticleBuffered (psp_particle* vertices, particle2_t *pparticl
extern int zombie_skins[2][2];
extern int faces_rejected, faces_checked, faces_clipped;
extern int faces_rejected, faces_checked, faces_clipped;
void convert_8bpp_to_4bpp(const byte* indata, const byte* inpal, int width, int height, byte* outdata, byte* outpal);

240
source/psp/video_hardware_colorquant.cpp Normal file
View file

@ -0,0 +1,240 @@
/*
Copyright (C) 2023 Shpuld and NZP Team.
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 the Free Software
Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*/
extern "C"
{
#include "../quakedef.h"
}
#include <pspgu.h>
#include <malloc.h>
struct colentry_t {
byte r;
byte g;
byte b;
byte a;
unsigned short pixcount;
};
colentry_t colentries[256];
typedef byte bucket_t[256];
bucket_t buckets[16];
int bucket_counts[16];
int num_buckets = 0;
byte byte_to_halfbyte_map[256];
void add_to_bucket(int bucket, int color) {
buckets[bucket][bucket_counts[bucket]] = color;
bucket_counts[bucket]++;
}
void remove_last_from_bucket(int bucket) {
bucket_counts[bucket]--;
buckets[bucket][bucket_counts[bucket]] = 0;
}
void init_buckets() {
memset(buckets, 0, sizeof(buckets));
memset(bucket_counts, 0, sizeof(bucket_counts));
memset(colentries, 0, sizeof(colentries));
memset(byte_to_halfbyte_map, 0, sizeof(byte_to_halfbyte_map));
num_buckets = 0;
}
int get_new_bucket() {
num_buckets += 1;
return num_buckets - 1;
}
colentry_t * color_of_bucket(int bucket, int index) {
return &(colentries[buckets[bucket][index]]);
}
int get_color_index(int bucket, int index) {
return buckets[bucket][index];
}
void sort_and_cut_bucket(int current_bucket, int depth, int target) {
if (num_buckets >= target) {
return;
}
int count = bucket_counts[current_bucket];
// find the highest range and count pixels in bucket
byte minR = 255;
byte maxR = 0;
byte minG = 255;
byte maxG = 0;
byte minB = 255;
byte maxB = 0;
int pixelsInBucket = 0;
for (int i = 0; i < count; i++) {
colentry_t * col = color_of_bucket(current_bucket, i);
pixelsInBucket += col->pixcount;
if (col->r < minR) minR = col->r;
if (col->r > maxR) maxR = col->r;
if (col->g < minG) minG = col->g;
if (col->g > maxG) maxG = col->g;
if (col->b < minB) minB = col->b;
if (col->b > maxB) maxB = col->b;
}
int rangeR = maxR - minR;
int rangeG = maxG - minG;
int rangeB = maxB - minB;
// 0 = r, 1 = g, 2 = b
int sortBy = 0;
if (rangeG > rangeR && rangeG > rangeB) sortBy = 1;
if (rangeB > rangeR && rangeB > rangeG) sortBy = 2;
// bubble sort, replace with comb sort or heap sort
bool sorted = false;
while (!sorted) {
sorted = true; // this will change if a swap is done
for (int i = 0; i < (count - 1); i++) {
colentry_t * col1 = color_of_bucket(current_bucket, i);
colentry_t * col2 = color_of_bucket(current_bucket, i + 1);
bool swap = false;
switch (sortBy) {
case 0: swap = col1->r > col2->r; break;
case 1: swap = col1->g > col2->g; break;
case 2: swap = col1->b > col2->b; break;
}
if (swap) {
byte prevval = buckets[current_bucket][i];
byte nextval = buckets[current_bucket][i + 1];
buckets[current_bucket][i] = nextval;
buckets[current_bucket][i + 1] = prevval;
sorted = false;
}
}
}
// median cut
int pixelsCounted = 0;
int split = 0;
for (int i = 0; i < count; i++) {
if (pixelsCounted > pixelsInBucket * 0.5) {
split = i;
break;
}
byte_to_halfbyte_map[get_color_index(current_bucket, i)] = current_bucket;
colentry_t * col = color_of_bucket(current_bucket, i);
pixelsCounted += col->pixcount;
}
int next_bucket = get_new_bucket();
for (int i = count-1; i >= split; i--) {
byte colorindex = get_color_index(current_bucket, i);
byte_to_halfbyte_map[colorindex] = next_bucket;
add_to_bucket(next_bucket, colorindex);
remove_last_from_bucket(current_bucket);
}
if (depth >= 3) return;
sort_and_cut_bucket(current_bucket, depth + 1, target);
sort_and_cut_bucket(next_bucket, depth + 1, target);
}
void convert_8bpp_to_4bpp(const byte* indata, const byte* inpal, int width, int height, byte* outdata, byte* outpal)
{
init_buckets();
int current_bucket = get_new_bucket();
// Set bucket pixelcounts
for (int pixi = 0; pixi < width * height; pixi++) {
byte color = indata[pixi];
colentries[color].pixcount += 1;
}
int MAX_COLORS = 256;
bool has_transparency = false;
byte transparent_index = 0;
int colors_used = 0;
// Set colors and fill first bucket
for (int i = 0; i < MAX_COLORS; i++) {
colentry_t * ce = &(colentries[i]);
if (ce->pixcount == 0) {
continue; // speed up by not processing unused colors
}
colors_used++;
ce->r = inpal[i * 3 + 0];
ce->g = inpal[i * 3 + 1];
ce->b = inpal[i * 3 + 2];
ce->a = 255;
if (ce->r == 0 && ce->g == 0 && ce->b == 255) {
ce->r = ce->g = ce->b = 128;
ce->a = 255;
has_transparency = true;
transparent_index = i;
continue; // don't add transparencies to buckets
}
add_to_bucket(current_bucket, i);
}
// could check if bucket has 16 colors or less and early out
sort_and_cut_bucket(current_bucket, 0, has_transparency ? 15 : 16);
if (has_transparency) {
int transparent_bucket = get_new_bucket();
byte_to_halfbyte_map[transparent_index] = transparent_bucket; // last index
add_to_bucket(transparent_bucket, transparent_index);
}
int image_size = width * height * 0.5;
for (int i = 0; i < image_size; i++) {
byte color_index = indata[i*2];
outdata[i] = byte_to_halfbyte_map[color_index];
color_index = indata[i*2 + 1];
outdata[i] += byte_to_halfbyte_map[color_index] << 4;
}
for (int i = 0; i < num_buckets; i++) {
int r = 0;
int g = 0;
int b = 0;
int alpha = 255;
colentry_t * col;
for (int j = 0; j < bucket_counts[i]; j++) {
byte colindex = get_color_index(i, j);
if (has_transparency && transparent_index == colindex) {
r = 128;
g = 128;
b = 128;
alpha = 0;
break;
}
col = color_of_bucket(i, j);
r += col->r;
g += col->g;
b += col->b;
}
r /= bucket_counts[i];
g /= bucket_counts[i];
b /= bucket_counts[i];
((unsigned int*)outpal)[i] = (alpha << 24) + (b << 16) + (g << 8) + r;
}
}

44
source/psp/video_hardware_draw.cpp
View file

@ -3347,8 +3347,6 @@ void GL_Upload4(int texture_index, const byte *data, int width, int height)
int GL_LoadTexture4(const char *identifier, unsigned int width, unsigned int height, const byte *data, int filter, qboolean swizzled)
{
int texture_index = -1;
tex_scale_down = r_tex_scale_down.value == qtrue;
if (identifier[0])
{
@ -3424,3 +3422,45 @@ int GL_LoadTexture4(const char *identifier, unsigned int width, unsigned int hei
// Done.
return texture_index;
}
int GL_LoadTexture8to4(const char *identifier, unsigned int width, unsigned int height, const byte *data, const byte *pal, int filter)
{
tex_scale_down = r_tex_scale_down.value == qtrue;
int new_width = width;
int new_height = height;
if (tex_scale_down == true)
{
new_width = std::max(round_down(width), 32U);
new_height = std::max(round_down(height),16U);
}
else
{
new_width = std::max(round_up(width), 32U);
new_height = std::max(round_up(height),16U);
}
std::size_t resamp_size = new_width * new_height;
byte * resamp_data = static_cast<byte*>(memalign(16, resamp_size));
// Resamp required?
if ((new_width != width) || (new_height != height)) {
GL_ResampleTexture(data, width, height, resamp_data, new_width, new_height);
} else {
memcpy(resamp_data, data, resamp_size);
}
std::size_t buffer_size = resamp_size * 0.5;
byte * clut4data = static_cast<byte*>(memalign(16, buffer_size + 16 * 4));
byte * clut4pal = &(clut4data[buffer_size]);
byte * unswizzled_data = static_cast<byte*>(memalign(16, buffer_size));
convert_8bpp_to_4bpp(resamp_data, pal, new_width, new_height, unswizzled_data, clut4pal);
swizzle_fast(clut4data, unswizzled_data, new_width * 0.5, new_height);
free(unswizzled_data);
int id = GL_LoadTexture4(identifier, new_width, new_height, clut4data, filter, qtrue);
free(clut4data);
return id;
}

262
source/psp/wad3.cpp
View file

@ -189,140 +189,6 @@ int ConvertWad3ToRGBA(miptex_t *tex)
return index;
}
struct colentry_t {
byte r;
byte g;
byte b;
byte a;
unsigned short pixcount;
};
colentry_t colentries[256];
typedef byte bucket_t[256];
bucket_t buckets[16];
int bucket_counts[16];
int num_buckets = 0;
byte byte_to_halfbyte_map[256];
void add_to_bucket(int bucket, int color) {
buckets[bucket][bucket_counts[bucket]] = color;
bucket_counts[bucket]++;
}
void remove_last_from_bucket(int bucket) {
bucket_counts[bucket]--;
buckets[bucket][bucket_counts[bucket]] = 0;
}
void init_buckets() {
memset(buckets, 0, sizeof(buckets));
memset(bucket_counts, 0, sizeof(bucket_counts));
memset(colentries, 0, sizeof(colentries));
memset(byte_to_halfbyte_map, 0, sizeof(byte_to_halfbyte_map));
num_buckets = 0;
}
int get_new_bucket() {
// Con_Printf("Create new bucket %d\n", num_buckets);
num_buckets += 1;
return num_buckets - 1;
}
colentry_t * color_of_bucket(int bucket, int index) {
return &(colentries[buckets[bucket][index]]);
}
int get_color_index(int bucket, int index) {
return buckets[bucket][index];
}
void sort_and_cut_bucket(int current_bucket, int depth, int target) {
if (num_buckets >= target) {
// Con_Printf("quit due to too many buckets %d / %d\n", num_buckets, target);
return;
}
int count = bucket_counts[current_bucket];
// find the highest range and count pixels in bucket
byte minR = 255;
byte maxR = 0;
byte minG = 255;
byte maxG = 0;
byte minB = 255;
byte maxB = 0;
int pixelsInBucket = 0;
for (int i = 0; i < count; i++) {
colentry_t * col = color_of_bucket(current_bucket, i);
pixelsInBucket += col->pixcount;
if (col->r < minR) minR = col->r;
if (col->r > maxR) maxR = col->r;
if (col->g < minG) minG = col->g;
if (col->g > maxG) maxG = col->g;
if (col->b < minB) minB = col->b;
if (col->b > maxB) maxB = col->b;
}
int rangeR = maxR - minR;
int rangeG = maxG - minG;
int rangeB = maxB - minB;
// 0 = r, 1 = g, 2 = b
int sortBy = 0;
if (rangeG > rangeR && rangeG > rangeB) sortBy = 1;
if (rangeB > rangeR && rangeB > rangeG) sortBy = 2;
// Con_Printf("%d: depth: %d, count: %d, pixels: %d, sortBy: %d\n", current_bucket, depth, count, pixelsInBucket, sortBy);
// bubble sort, replace with comb sort or heap sort
bool sorted = false;
while (!sorted) {
sorted = true; // this will change if a swap is done
for (int i = 0; i < (count - 1); i++) {
colentry_t * col1 = color_of_bucket(current_bucket, i);
colentry_t * col2 = color_of_bucket(current_bucket, i + 1);
bool swap = false;
switch (sortBy) {
case 0: swap = col1->r > col2->r; break;
case 1: swap = col1->g > col2->g; break;
case 2: swap = col1->b > col2->b; break;
}
if (swap) {
byte prevval = buckets[current_bucket][i];
byte nextval = buckets[current_bucket][i + 1];
buckets[current_bucket][i] = nextval;
buckets[current_bucket][i + 1] = prevval;
sorted = false;
}
}
}
// median cut
int pixelsCounted = 0;
int split = 0;
for (int i = 0; i < count; i++) {
if (pixelsCounted > pixelsInBucket * 0.5) {
split = i;
break;
}
byte_to_halfbyte_map[get_color_index(current_bucket, i)] = current_bucket;
colentry_t * col = color_of_bucket(current_bucket, i);
pixelsCounted += col->pixcount;
}
int next_bucket = get_new_bucket();
for (int i = count-1; i >= split; i--) {
byte colorindex = get_color_index(current_bucket, i);
byte_to_halfbyte_map[colorindex] = next_bucket;
add_to_bucket(next_bucket, colorindex);
remove_last_from_bucket(current_bucket);
}
if (depth >= 3) return;
sort_and_cut_bucket(current_bucket, depth + 1, target);
sort_and_cut_bucket(next_bucket, depth + 1, target);
}
int ConvertWad3ToClut4(miptex_t *tex)
{
// Check that texture has data
@ -331,134 +197,12 @@ int ConvertWad3ToClut4(miptex_t *tex)
}
// Get pointers to WAD3 data and palette
byte* wadData = ((byte*)tex) + tex->offsets[0];
byte* palette = ((byte*)tex) + tex->offsets[3] + (tex->width>>3)*(tex->height>>3) + 2;
//byte* palette = wadData + tex->offsets[MIPLEVELS]; // Palette starts 2 bytes after the last mipmap
const byte* wadData = ((byte*)tex) + tex->offsets[0];
const byte* palette = ((byte*)tex) + tex->offsets[3] + (tex->width>>3)*(tex->height>>3) + 2;
init_buckets();
int current_bucket = get_new_bucket();
// Set bucket pixelcounts
for (int pixi = 0; pixi < tex->width * tex->height; pixi++) {
byte color = wadData[pixi];
colentries[color].pixcount += 1;
}
int MAX_COLORS = 256;
bool has_transparency = false;
byte transparent_index = 0;
int colors_used = 0;
// Set colors and fill first bucket
for (int i = 0; i < MAX_COLORS; i++) {
colentry_t * ce = &(colentries[i]);
if (ce->pixcount == 0) {
continue; // speed up by not processing unused colors
}
colors_used++;
ce->r = palette[i * 3 + 0];
ce->g = palette[i * 3 + 1];
ce->b = palette[i * 3 + 2];
ce->a = 255;
if (ce->r == 0 && ce->g == 0 && ce->b == 255) {
ce->r = ce->g = ce->b = 128;
ce->a = 255;
has_transparency = true;
transparent_index = i;
continue; // don't add transparencies to buckets
}
add_to_bucket(current_bucket, i);
}
// Con_Printf("Initial bucket %d: %d colors\n", current_bucket, bucket_counts[current_bucket]);
// could check if bucket has 16 colors or less and early out
sort_and_cut_bucket(current_bucket, 0, has_transparency ? 15 : 16);
if (has_transparency) {
int transparent_bucket = get_new_bucket();
byte_to_halfbyte_map[transparent_index] = transparent_bucket; // last index
add_to_bucket(transparent_bucket, transparent_index);
}
// Con_Printf("num buckets: %d, has transparency: %d, colors actually used: %d\n", num_buckets, has_transparency, colors_used);
/* hack to resample width/height if it's less than 32 */
bool stretch_sideways = false;
if (tex->width <= 16) {
tex->width *= 2;
stretch_sideways = true;
}
int imageSize = tex->width * tex->height * 0.5;
byte* texData = (byte*)memalign(16, 16 * 4 + imageSize);
byte* unswizzled = (byte*)memalign(16, imageSize);
unsigned int * newPal = (unsigned int*)&(texData[imageSize]);
/*
for (int y = 0; y < tex->height; y++) {
for (int x = 0; x < tex->width; x++) {
byte colorIndex = wadData[y * (tex->width / width_mult) / height_mult + x / width_mult * 2];
unswizzled[y * tex->width + x] = byte_to_halfbyte_map[colorIndex];
colorIndex = wadData[y * (tex->width / width_mult) / height_mult + x / width_mult * 2 + 1];
unswizzled[y * tex->width + x] = byte_to_halfbyte_map[colorIndex] << 4;
}
}
*/
if (stretch_sideways) {
for (int i = 0; i < imageSize; i++) {
byte colorIndex = wadData[i];
unswizzled[i] = byte_to_halfbyte_map[colorIndex] + (byte_to_halfbyte_map[colorIndex] << 4);
}
} else {
for (int i = 0; i < imageSize; i++) {
byte colorIndex = wadData[i*2];
unswizzled[i] = byte_to_halfbyte_map[colorIndex];
colorIndex = wadData[i*2 + 1];
unswizzled[i] += byte_to_halfbyte_map[colorIndex] << 4;
}
}
for (int i = 0; i < num_buckets; i++) {
int r = 0;
int g = 0;
int b = 0;
int alpha = 255;
colentry_t * col;
for (int j = 0; j < bucket_counts[i]; j++) {
byte colindex = get_color_index(i, j);
if (has_transparency && transparent_index == colindex) {
r = 128;
g = 128;
b = 128;
alpha = 0;
break;
}
col = color_of_bucket(i, j);
r += col->r;
g += col->g;
b += col->b;
}
r /= bucket_counts[i];
g /= bucket_counts[i];
b /= bucket_counts[i];
newPal[i] = (alpha << 24) + (b << 16) + (g << 8) + r;
//Con_Printf("Bucket %d count: %d\n", i, bucket_counts[i]);
//Con_Printf("Color %d: %d %d %d %x\n", i, r, g, b, newPal[i]);
}
//Con_Printf("_ _ _ \n");
swizzle_fast(texData, unswizzled, tex->width * 0.5, tex->height);
// memcpy(texData, unswizzled, imageSize);
int index = GL_LoadTexture4(tex->name, tex->width, tex->height, texData, GU_LINEAR, qtrue);
free(texData);
free(unswizzled);
return index;
return GL_LoadTexture8to4(tex->name, tex->width, tex->height, wadData, palette, GU_LINEAR);
}
int WAD3_LoadTexture(miptex_t *mt)
{
char texname[MAX_QPATH];