godot/modules/pvr/texture_loader_pvr.cpp
Rémi Verschelde 3f3f5a5359 Merge remote-tracking branch 'origin/gles3' into gles3-on-master
Various merge conflicts have been fixed manually and some mistakes
might have been made - time will tell :)
2017-01-02 21:52:26 +01:00

712 lines
17 KiB
C++

/*************************************************************************/
/* texture_loader_pvr.cpp */
/*************************************************************************/
/* This file is part of: */
/* GODOT ENGINE */
/* http://www.godotengine.org */
/*************************************************************************/
/* Copyright (c) 2007-2017 Juan Linietsky, Ariel Manzur. */
/* */
/* Permission is hereby granted, free of charge, to any person obtaining */
/* a copy of this software and associated documentation files (the */
/* "Software"), to deal in the Software without restriction, including */
/* without limitation the rights to use, copy, modify, merge, publish, */
/* distribute, sublicense, and/or sell copies of the Software, and to */
/* permit persons to whom the Software is furnished to do so, subject to */
/* the following conditions: */
/* */
/* The above copyright notice and this permission notice shall be */
/* included in all copies or substantial portions of the Software. */
/* */
/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */
/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */
/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */
/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */
/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */
/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */
/*************************************************************************/
#include "texture_loader_pvr.h"
#include "os/file_access.h"
#include <string.h>
#include "PvrTcEncoder.h"
#include "RgbaBitmap.h"
static void _pvrtc_decompress(Image* p_img);
enum PVRFLags {
PVR_HAS_MIPMAPS=0x00000100,
PVR_TWIDDLED=0x00000200,
PVR_NORMAL_MAP=0x00000400,
PVR_BORDER=0x00000800,
PVR_CUBE_MAP=0x00001000,
PVR_FALSE_MIPMAPS=0x00002000,
PVR_VOLUME_TEXTURES=0x00004000,
PVR_HAS_ALPHA=0x00008000,
PVR_VFLIP=0x00010000
};
RES ResourceFormatPVR::load(const String &p_path,const String& p_original_path,Error *r_error) {
if (r_error)
*r_error=ERR_CANT_OPEN;
Error err;
FileAccess *f = FileAccess::open(p_path,FileAccess::READ,&err);
if (!f)
return RES();
FileAccessRef faref(f);
ERR_FAIL_COND_V(err,RES());
if (r_error)
*r_error=ERR_FILE_CORRUPT;
uint32_t hsize = f->get_32();
ERR_FAIL_COND_V(hsize!=52,RES());
uint32_t height = f->get_32();
uint32_t width = f->get_32();
uint32_t mipmaps = f->get_32();
uint32_t flags = f->get_32();
uint32_t surfsize = f->get_32();
uint32_t bpp = f->get_32();
uint32_t rmask = f->get_32();
uint32_t gmask = f->get_32();
uint32_t bmask = f->get_32();
uint32_t amask = f->get_32();
uint8_t pvrid[5]={0,0,0,0,0};
f->get_buffer(pvrid,4);
ERR_FAIL_COND_V(String((char*)pvrid)!="PVR!",RES());
uint32_t surfcount = f->get_32();
/*
print_line("height: "+itos(height));
print_line("width: "+itos(width));
print_line("mipmaps: "+itos(mipmaps));
print_line("flags: "+itos(flags));
print_line("surfsize: "+itos(surfsize));
print_line("bpp: "+itos(bpp));
print_line("rmask: "+itos(rmask));
print_line("gmask: "+itos(gmask));
print_line("bmask: "+itos(bmask));
print_line("amask: "+itos(amask));
print_line("surfcount: "+itos(surfcount));
*/
DVector<uint8_t> data;
data.resize(surfsize);
ERR_FAIL_COND_V(data.size()==0,RES());
DVector<uint8_t>::Write w = data.write();
f->get_buffer(&w[0],surfsize);
err = f->get_error();
ERR_FAIL_COND_V(err!=OK,RES());
Image::Format format=Image::FORMAT_MAX;
switch(flags&0xFF) {
case 0x18:
case 0xC: format=(flags&PVR_HAS_ALPHA)?Image::FORMAT_PVRTC2A:Image::FORMAT_PVRTC2; break;
case 0x19:
case 0xD: format=(flags&PVR_HAS_ALPHA)?Image::FORMAT_PVRTC4A:Image::FORMAT_PVRTC4; break;
case 0x16:
format=Image::FORMAT_L8; break;
case 0x17:
format=Image::FORMAT_LA8; break;
case 0x20:
case 0x80:
case 0x81:
format=Image::FORMAT_DXT1; break;
case 0x21:
case 0x22:
case 0x82:
case 0x83:
format=Image::FORMAT_DXT3; break;
case 0x23:
case 0x24:
case 0x84:
case 0x85:
format=Image::FORMAT_DXT5; break;
case 0x4:
case 0x15:
format=Image::FORMAT_RGB8; break;
case 0x5:
case 0x12:
format=Image::FORMAT_RGBA8; break;
case 0x36:
format=Image::FORMAT_ETC; break;
default:
ERR_EXPLAIN("Unsupported format in PVR texture: "+itos(flags&0xFF));
ERR_FAIL_V(RES());
}
w = DVector<uint8_t>::Write();
int tex_flags=Texture::FLAG_FILTER|Texture::FLAG_REPEAT;
if (mipmaps)
tex_flags|=Texture::FLAG_MIPMAPS;
print_line("flip: "+itos(flags&PVR_VFLIP));
Image image(width,height,mipmaps,format,data);
ERR_FAIL_COND_V(image.empty(),RES());
Ref<ImageTexture> texture = memnew( ImageTexture );
texture->create_from_image(image,tex_flags);
if (r_error)
*r_error=OK;
return texture;
}
void ResourceFormatPVR::get_recognized_extensions(List<String> *p_extensions) const {
p_extensions->push_back("pvr");
}
bool ResourceFormatPVR::handles_type(const String& p_type) const {
return ObjectTypeDB::is_type(p_type,"Texture");
}
String ResourceFormatPVR::get_resource_type(const String &p_path) const {
if (p_path.extension().to_lower()=="pvr")
return "Texture";
return "";
}
static void _compress_pvrtc4(Image * p_img) {
Image img = *p_img;
bool make_mipmaps=false;
if (img.get_width()%8 || img.get_height()%8) {
make_mipmaps=img.has_mipmaps();
img.resize(img.get_width()+(8-(img.get_width()%8)),img.get_height()+(8-(img.get_height()%8)));
}
img.convert(Image::FORMAT_RGBA8);
if (!img.has_mipmaps() && make_mipmaps)
img.generate_mipmaps();
bool use_alpha=img.detect_alpha();
Image new_img;
new_img.create(img.get_width(),img.get_height(),true,use_alpha?Image::FORMAT_PVRTC4A:Image::FORMAT_PVRTC4);
DVector<uint8_t> data=new_img.get_data();
{
DVector<uint8_t>::Write wr=data.write();
DVector<uint8_t>::Read r=img.get_data().read();
for(int i=0;i<=new_img.get_mipmap_count();i++) {
int ofs,size,w,h;
img.get_mipmap_offset_size_and_dimensions(i,ofs,size,w,h);
Javelin::RgbaBitmap bm(w,h);
copymem(bm.GetData(),&r[ofs],size);
{
Javelin::ColorRgba<unsigned char> *dp = bm.GetData();
for(int j=0;j<size/4;j++) {
SWAP(dp[j].r,dp[j].b);
}
}
new_img.get_mipmap_offset_size_and_dimensions(i,ofs,size,w,h);
Javelin::PvrTcEncoder::EncodeRgba4Bpp(&wr[ofs],bm);
}
}
*p_img = Image(new_img.get_width(),new_img.get_height(),new_img.has_mipmaps(),new_img.get_format(),data);
}
ResourceFormatPVR::ResourceFormatPVR() {
Image::_image_decompress_pvrtc=_pvrtc_decompress;
Image::_image_compress_pvrtc4_func=_compress_pvrtc4;
Image::_image_compress_pvrtc2_func=_compress_pvrtc4;
}
/////////////////////////////////////////////////////////
//PVRTC decompressor, Based on PVRTC decompressor by IMGTEC.
/////////////////////////////////////////////////////////
#define PT_INDEX 2
#define BLK_Y_SIZE 4
#define BLK_X_MAX 8
#define BLK_X_2BPP 8
#define BLK_X_4BPP 4
#define WRAP_COORD(Val, Size) ((Val) & ((Size)-1))
/*
Define an expression to either wrap or clamp large or small vals to the
legal coordinate range
*/
#define LIMIT_COORD(Val, Size, p_tiled) \
((p_tiled)? WRAP_COORD((Val), (Size)): CLAMP((Val), 0, (Size)-1))
struct PVRTCBlock {
//blocks are 64 bits
uint32_t data[2];
};
_FORCE_INLINE_ bool is_po2( uint32_t p_input ) {
if( p_input==0 )
return 0;
uint32_t minus1=p_input- 1;
return ((p_input|minus1)==(p_input^minus1))?1:0;
}
static void unpack_5554(const PVRTCBlock *p_block, int p_ab_colors[2][4]) {
uint32_t raw_bits[2];
raw_bits[0] = p_block->data[1] & (0xFFFE);
raw_bits[1] = p_block->data[1] >> 16;
for(int i=0;i<2;i++) {
if(raw_bits[i] & (1<<15)) {
p_ab_colors[i][0]= (raw_bits[i] >> 10) & 0x1F;
p_ab_colors[i][1]= (raw_bits[i] >> 5) & 0x1F;
p_ab_colors[i][2]= raw_bits[i] & 0x1F;
if(i==0)
p_ab_colors[0][2]|= p_ab_colors[0][2] >> 4;
p_ab_colors[i][3] = 0xF;
} else {
p_ab_colors[i][0] = (raw_bits[i] >> (8-1)) & 0x1E;
p_ab_colors[i][1] = (raw_bits[i] >> (4-1)) & 0x1E;
p_ab_colors[i][0] |= p_ab_colors[i][0] >> 4;
p_ab_colors[i][1] |= p_ab_colors[i][1] >> 4;
p_ab_colors[i][2] = (raw_bits[i] & 0xF) << 1;
if(i==0)
p_ab_colors[0][2] |= p_ab_colors[0][2] >> 3;
else
p_ab_colors[0][2] |= p_ab_colors[0][2] >> 4;
p_ab_colors[i][3] = (raw_bits[i] >> 11) & 0xE;
}
}
}
static void unpack_modulations(const PVRTCBlock *p_block, const int p_2bit, int p_modulation[8][16], int p_modulation_modes[8][16], int p_x, int p_y) {
int block_mod_mode = p_block->data[1] & 1;
uint32_t modulation_bits = p_block->data[0];
if(p_2bit && block_mod_mode) {
for(int y = 0; y < BLK_Y_SIZE; y++) {
for(int x = 0; x < BLK_X_2BPP; x++) {
p_modulation_modes[y+p_y][x+p_x] = block_mod_mode;
if(((x^y)&1) == 0) {
p_modulation[y+p_y][x+p_x] = modulation_bits & 3;
modulation_bits >>= 2;
}
}
}
} else if(p_2bit) {
for(int y = 0; y < BLK_Y_SIZE; y++) {
for(int x = 0; x < BLK_X_2BPP; x++) {
p_modulation_modes[y+p_y][x+p_x] = block_mod_mode;
if(modulation_bits & 1)
p_modulation[y+p_y][x+p_x] = 0x3;
else
p_modulation[y+p_y][x+p_x] = 0x0;
modulation_bits >>= 1;
}
}
} else {
for(int y = 0; y < BLK_Y_SIZE; y++) {
for(int x = 0; x < BLK_X_4BPP; x++) {
p_modulation_modes[y+p_y][x+p_x] = block_mod_mode;
p_modulation[y+p_y][x+p_x] = modulation_bits & 3;
modulation_bits >>= 2;
}
}
}
ERR_FAIL_COND(modulation_bits!=0);
}
static void interpolate_colors(const int p_colorp[4], const int p_colorq[4], const int p_colorr[4], const int p_colors[4], bool p_2bit, const int x, const int y, int r_result[4]) {
int u, v, uscale;
int k;
int tmp1, tmp2;
int P[4], Q[4], R[4], S[4];
for(k = 0; k < 4; k++) {
P[k] = p_colorp[k];
Q[k] = p_colorq[k];
R[k] = p_colorr[k];
S[k] = p_colors[k];
}
v = (y & 0x3) | ((~y & 0x2) << 1);
if(p_2bit)
u = (x & 0x7) | ((~x & 0x4) << 1);
else
u = (x & 0x3) | ((~x & 0x2) << 1);
v = v - BLK_Y_SIZE/2;
if(p_2bit) {
u = u - BLK_X_2BPP/2;
uscale = 8;
} else {
u = u - BLK_X_4BPP/2;
uscale = 4;
}
for(k = 0; k < 4; k++) {
tmp1 = P[k] * uscale + u * (Q[k] - P[k]);
tmp2 = R[k] * uscale + u * (S[k] - R[k]);
tmp1 = tmp1 * 4 + v * (tmp2 - tmp1);
r_result[k] = tmp1;
}
if(p_2bit) {
for(k = 0; k < 3; k++) {
r_result[k] >>= 2;
}
r_result[3] >>= 1;
} else {
for(k = 0; k < 3; k++) {
r_result[k] >>= 1;
}
}
for(k = 0; k < 4; k++) {
ERR_FAIL_COND(r_result[k] >= 256);
}
for(k = 0; k < 3; k++) {
r_result[k] += r_result[k] >> 5;
}
r_result[3] += r_result[3] >> 4;
for(k = 0; k < 4; k++) {
ERR_FAIL_COND(r_result[k] >= 256);
}
}
static void get_modulation_value(int x, int y, const int p_2bit, const int p_modulation[8][16], const int p_modulation_modes[8][16], int *r_mod, int *p_dopt)
{
static const int rep_vals0[4] = {0, 3, 5, 8};
static const int rep_vals1[4] = {0, 4, 4, 8};
int mod_val;
y = (y & 0x3) | ((~y & 0x2) << 1);
if(p_2bit)
x = (x & 0x7) | ((~x & 0x4) << 1);
else
x = (x & 0x3) | ((~x & 0x2) << 1);
*p_dopt = 0;
if(p_modulation_modes[y][x]==0) {
mod_val = rep_vals0[p_modulation[y][x]];
} else if(p_2bit) {
if(((x^y)&1)==0)
mod_val = rep_vals0[p_modulation[y][x]];
else if(p_modulation_modes[y][x] == 1) {
mod_val = (rep_vals0[p_modulation[y-1][x]] +
rep_vals0[p_modulation[y+1][x]] +
rep_vals0[p_modulation[y][x-1]] +
rep_vals0[p_modulation[y][x+1]] + 2) / 4;
} else if(p_modulation_modes[y][x] == 2) {
mod_val = (rep_vals0[p_modulation[y][x-1]] +
rep_vals0[p_modulation[y][x+1]] + 1) / 2;
} else {
mod_val = (rep_vals0[p_modulation[y-1][x]] +
rep_vals0[p_modulation[y+1][x]] + 1) / 2;
}
} else {
mod_val = rep_vals1[p_modulation[y][x]];
*p_dopt = p_modulation[y][x] == PT_INDEX;
}
*r_mod =mod_val;
}
static int disable_twiddling = 0;
static uint32_t twiddle_uv(uint32_t p_height, uint32_t p_width, uint32_t p_y, uint32_t p_x) {
uint32_t twiddled;
uint32_t min_dimension;
uint32_t max_value;
uint32_t scr_bit_pos;
uint32_t dst_bit_pos;
int shift_count;
ERR_FAIL_COND_V(p_y >= p_height,0);
ERR_FAIL_COND_V(p_x >= p_width,0);
ERR_FAIL_COND_V(!is_po2(p_height),0);
ERR_FAIL_COND_V(!is_po2(p_width),0);
if(p_height < p_width) {
min_dimension = p_height;
max_value = p_x;
} else {
min_dimension = p_width;
max_value = p_y;
}
if(disable_twiddling)
return (p_y* p_width + p_x);
scr_bit_pos = 1;
dst_bit_pos = 1;
twiddled = 0;
shift_count = 0;
while(scr_bit_pos < min_dimension) {
if(p_y & scr_bit_pos) {
twiddled |= dst_bit_pos;
}
if(p_x & scr_bit_pos) {
twiddled |= (dst_bit_pos << 1);
}
scr_bit_pos <<= 1;
dst_bit_pos <<= 2;
shift_count += 1;
}
max_value >>= shift_count;
twiddled |= (max_value << (2*shift_count));
return twiddled;
}
static void decompress_pvrtc(PVRTCBlock *p_comp_img, const int p_2bit, const int p_width, const int p_height, const int p_tiled, unsigned char* p_dst) {
int x, y;
int i, j;
int block_x, blk_y;
int block_xp1, blk_yp1;
int x_block_size;
int block_width, block_height;
int p_x, p_y;
int p_modulation[8][16];
int p_modulation_modes[8][16];
int Mod, DoPT;
unsigned int u_pos;
// local neighbourhood of blocks
PVRTCBlock *p_blocks[2][2];
PVRTCBlock *prev[2][2] = {{NULL, NULL}, {NULL, NULL}};
struct
{
int Reps[2][4];
}colors5554[2][2];
int ASig[4], BSig[4];
int r_result[4];
if(p_2bit)
x_block_size = BLK_X_2BPP;
else
x_block_size = BLK_X_4BPP;
block_width = MAX(2, p_width / x_block_size);
block_height = MAX(2, p_height / BLK_Y_SIZE);
for(y = 0; y < p_height; y++)
{
for(x = 0; x < p_width; x++)
{
block_x = (x - x_block_size/2);
blk_y = (y - BLK_Y_SIZE/2);
block_x = LIMIT_COORD(block_x, p_width, p_tiled);
blk_y = LIMIT_COORD(blk_y, p_height, p_tiled);
block_x /= x_block_size;
blk_y /= BLK_Y_SIZE;
block_xp1 = LIMIT_COORD(block_x+1, block_width, p_tiled);
blk_yp1 = LIMIT_COORD(blk_y+1, block_height, p_tiled);
p_blocks[0][0] = p_comp_img +twiddle_uv(block_height, block_width, blk_y, block_x);
p_blocks[0][1] = p_comp_img +twiddle_uv(block_height, block_width, blk_y, block_xp1);
p_blocks[1][0] = p_comp_img +twiddle_uv(block_height, block_width, blk_yp1, block_x);
p_blocks[1][1] = p_comp_img +twiddle_uv(block_height, block_width, blk_yp1, block_xp1);
if(memcmp(prev, p_blocks, 4*sizeof(void*)) != 0) {
p_y = 0;
for(i = 0; i < 2; i++) {
p_x = 0;
for(j = 0; j < 2; j++) {
unpack_5554(p_blocks[i][j], colors5554[i][j].Reps);
unpack_modulations(p_blocks[i][j],
p_2bit,
p_modulation,
p_modulation_modes,
p_x, p_y);
p_x += x_block_size;
}
p_y += BLK_Y_SIZE;
}
memcpy(prev, p_blocks, 4*sizeof(void*));
}
interpolate_colors(colors5554[0][0].Reps[0],
colors5554[0][1].Reps[0],
colors5554[1][0].Reps[0],
colors5554[1][1].Reps[0],
p_2bit, x, y,
ASig);
interpolate_colors(colors5554[0][0].Reps[1],
colors5554[0][1].Reps[1],
colors5554[1][0].Reps[1],
colors5554[1][1].Reps[1],
p_2bit, x, y,
BSig);
get_modulation_value(x,y, p_2bit, (const int (*)[16])p_modulation, (const int (*)[16])p_modulation_modes,
&Mod, &DoPT);
for(i = 0; i < 4; i++) {
r_result[i] = ASig[i] * 8 + Mod * (BSig[i] - ASig[i]);
r_result[i] >>= 3;
}
if(DoPT)
r_result[3] = 0;
u_pos = (x+y*p_width)<<2;
p_dst[u_pos+0] = (uint8_t)r_result[0];
p_dst[u_pos+1] = (uint8_t)r_result[1];
p_dst[u_pos+2] = (uint8_t)r_result[2];
p_dst[u_pos+3] = (uint8_t)r_result[3];
}
}
}
static void _pvrtc_decompress(Image* p_img) {
// static void decompress_pvrtc(const void *p_comp_img, const int p_2bit, const int p_width, const int p_height, unsigned char* p_dst) {
// decompress_pvrtc((PVRTCBlock*)p_comp_img,p_2bit,p_width,p_height,1,p_dst);
// }
ERR_FAIL_COND( p_img->get_format()!=Image::FORMAT_PVRTC2 && p_img->get_format()!=Image::FORMAT_PVRTC2A && p_img->get_format()!=Image::FORMAT_PVRTC4 && p_img->get_format()!=Image::FORMAT_PVRTC4A);
bool _2bit = (p_img->get_format()==Image::FORMAT_PVRTC2 || p_img->get_format()==Image::FORMAT_PVRTC2A );
DVector<uint8_t> data = p_img->get_data();
DVector<uint8_t>::Read r = data.read();
DVector<uint8_t> newdata;
newdata.resize( p_img->get_width() * p_img->get_height() * 4);
DVector<uint8_t>::Write w=newdata.write();
decompress_pvrtc((PVRTCBlock*)r.ptr(),_2bit,p_img->get_width(),p_img->get_height(),0,(unsigned char*)w.ptr());
//for(int i=0;i<newdata.size();i++) {
// print_line(itos(w[i]));
//}
w=DVector<uint8_t>::Write();
r=DVector<uint8_t>::Read();
bool make_mipmaps=p_img->has_mipmaps();
Image newimg(p_img->get_width(),p_img->get_height(),false,Image::FORMAT_RGBA8,newdata);
if (make_mipmaps)
newimg.generate_mipmaps();
*p_img=newimg;
}