mirror of
https://github.com/reactos/reactos.git
synced 2024-06-30 09:50:07 +00:00
12b78f182d
This is no official system DLL and WineD3D is its only user. But latest WineD3D prefers the libtxc_dxtn codebase instead of the alternative library we used. This is also what Mesa uses and Fedora ships now that the patents have expired, so we should stick to the same. I'm importing the libtxc_dxtn-1.0.1 codebase from https://people.freedesktop.org/~cbrill/libtxc_dxtn/ It compiles warning-free in our tree without modifying a single line! I'm merging this Pull Request based on the positive response in CORE-12759. I lack a proper test case myself, but my version at least shouldn't make things worse. In my opinion, the previous one was broken anyway due to using stdcall imports when WineD3D called cdecl ones. This also removes the NSWPAT option from our buildsystem as dxtn was its latest user.
244 lines
8.5 KiB
C
244 lines
8.5 KiB
C
/*
|
|
* libtxc_dxtn
|
|
* Version: 1.0
|
|
*
|
|
* Copyright (C) 2004 Roland Scheidegger All Rights Reserved.
|
|
*
|
|
* 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
|
|
* BRIAN PAUL 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 <stdio.h>
|
|
#include "txc_dxtn.h"
|
|
|
|
#define EXP5TO8R(packedcol) \
|
|
((((packedcol) >> 8) & 0xf8) | (((packedcol) >> 13) & 0x7))
|
|
|
|
#define EXP6TO8G(packedcol) \
|
|
((((packedcol) >> 3) & 0xfc) | (((packedcol) >> 9) & 0x3))
|
|
|
|
#define EXP5TO8B(packedcol) \
|
|
((((packedcol) << 3) & 0xf8) | (((packedcol) >> 2) & 0x7))
|
|
|
|
#define EXP4TO8(col) \
|
|
((col) | ((col) << 4))
|
|
|
|
/* inefficient. To be efficient, it would be necessary to decode 16 pixels at once */
|
|
|
|
static void dxt135_decode_imageblock ( const GLubyte *img_block_src,
|
|
GLint i, GLint j, GLuint dxt_type, GLvoid *texel ) {
|
|
GLchan *rgba = (GLchan *) texel;
|
|
const GLushort color0 = img_block_src[0] | (img_block_src[1] << 8);
|
|
const GLushort color1 = img_block_src[2] | (img_block_src[3] << 8);
|
|
const GLuint bits = img_block_src[4] | (img_block_src[5] << 8) |
|
|
(img_block_src[6] << 16) | (img_block_src[7] << 24);
|
|
/* What about big/little endian? */
|
|
GLubyte bit_pos = 2 * (j * 4 + i) ;
|
|
GLubyte code = (GLubyte) ((bits >> bit_pos) & 3);
|
|
|
|
rgba[ACOMP] = CHAN_MAX;
|
|
switch (code) {
|
|
case 0:
|
|
rgba[RCOMP] = UBYTE_TO_CHAN( EXP5TO8R(color0) );
|
|
rgba[GCOMP] = UBYTE_TO_CHAN( EXP6TO8G(color0) );
|
|
rgba[BCOMP] = UBYTE_TO_CHAN( EXP5TO8B(color0) );
|
|
break;
|
|
case 1:
|
|
rgba[RCOMP] = UBYTE_TO_CHAN( EXP5TO8R(color1) );
|
|
rgba[GCOMP] = UBYTE_TO_CHAN( EXP6TO8G(color1) );
|
|
rgba[BCOMP] = UBYTE_TO_CHAN( EXP5TO8B(color1) );
|
|
break;
|
|
case 2:
|
|
if ((dxt_type > 1) || (color0 > color1)) {
|
|
rgba[RCOMP] = UBYTE_TO_CHAN( ((EXP5TO8R(color0) * 2 + EXP5TO8R(color1)) / 3) );
|
|
rgba[GCOMP] = UBYTE_TO_CHAN( ((EXP6TO8G(color0) * 2 + EXP6TO8G(color1)) / 3) );
|
|
rgba[BCOMP] = UBYTE_TO_CHAN( ((EXP5TO8B(color0) * 2 + EXP5TO8B(color1)) / 3) );
|
|
}
|
|
else {
|
|
rgba[RCOMP] = UBYTE_TO_CHAN( ((EXP5TO8R(color0) + EXP5TO8R(color1)) / 2) );
|
|
rgba[GCOMP] = UBYTE_TO_CHAN( ((EXP6TO8G(color0) + EXP6TO8G(color1)) / 2) );
|
|
rgba[BCOMP] = UBYTE_TO_CHAN( ((EXP5TO8B(color0) + EXP5TO8B(color1)) / 2) );
|
|
}
|
|
break;
|
|
case 3:
|
|
if ((dxt_type > 1) || (color0 > color1)) {
|
|
rgba[RCOMP] = UBYTE_TO_CHAN( ((EXP5TO8R(color0) + EXP5TO8R(color1) * 2) / 3) );
|
|
rgba[GCOMP] = UBYTE_TO_CHAN( ((EXP6TO8G(color0) + EXP6TO8G(color1) * 2) / 3) );
|
|
rgba[BCOMP] = UBYTE_TO_CHAN( ((EXP5TO8B(color0) + EXP5TO8B(color1) * 2) / 3) );
|
|
}
|
|
else {
|
|
rgba[RCOMP] = 0;
|
|
rgba[GCOMP] = 0;
|
|
rgba[BCOMP] = 0;
|
|
if (dxt_type == 1) rgba[ACOMP] = UBYTE_TO_CHAN(0);
|
|
}
|
|
break;
|
|
default:
|
|
/* CANNOT happen (I hope) */
|
|
break;
|
|
}
|
|
}
|
|
|
|
|
|
void fetch_2d_texel_rgb_dxt1(GLint srcRowStride, const GLubyte *pixdata,
|
|
GLint i, GLint j, GLvoid *texel)
|
|
{
|
|
/* Extract the (i,j) pixel from pixdata and return it
|
|
* in texel[RCOMP], texel[GCOMP], texel[BCOMP], texel[ACOMP].
|
|
*/
|
|
|
|
const GLubyte *blksrc = (pixdata + ((srcRowStride + 3) / 4 * (j / 4) + (i / 4)) * 8);
|
|
dxt135_decode_imageblock(blksrc, (i&3), (j&3), 0, texel);
|
|
}
|
|
|
|
|
|
void fetch_2d_texel_rgba_dxt1(GLint srcRowStride, const GLubyte *pixdata,
|
|
GLint i, GLint j, GLvoid *texel)
|
|
{
|
|
/* Extract the (i,j) pixel from pixdata and return it
|
|
* in texel[RCOMP], texel[GCOMP], texel[BCOMP], texel[ACOMP].
|
|
*/
|
|
|
|
const GLubyte *blksrc = (pixdata + ((srcRowStride + 3) / 4 * (j / 4) + (i / 4)) * 8);
|
|
dxt135_decode_imageblock(blksrc, (i&3), (j&3), 1, texel);
|
|
}
|
|
|
|
void fetch_2d_texel_rgba_dxt3(GLint srcRowStride, const GLubyte *pixdata,
|
|
GLint i, GLint j, GLvoid *texel) {
|
|
|
|
/* Extract the (i,j) pixel from pixdata and return it
|
|
* in texel[RCOMP], texel[GCOMP], texel[BCOMP], texel[ACOMP].
|
|
*/
|
|
|
|
GLchan *rgba = (GLchan *) texel;
|
|
const GLubyte *blksrc = (pixdata + ((srcRowStride + 3) / 4 * (j / 4) + (i / 4)) * 16);
|
|
#if 0
|
|
/* Simple 32bit version. */
|
|
/* that's pretty brain-dead for a single pixel, isn't it? */
|
|
const GLubyte bit_pos = 4 * ((j&3) * 4 + (i&3));
|
|
const GLuint alpha_low = blksrc[0] | (blksrc[1] << 8) | (blksrc[2] << 16) | (blksrc[3] << 24);
|
|
const GLuint alpha_high = blksrc[4] | (blksrc[5] << 8) | (blksrc[6] << 16) | (blksrc[7] << 24);
|
|
|
|
dxt135_decode_imageblock(blksrc + 8, (i&3), (j&3), 2, texel);
|
|
if (bit_pos < 32)
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( (GLubyte)(EXP4TO8((alpha_low >> bit_pos) & 15)) );
|
|
else
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( (GLubyte)(EXP4TO8((alpha_high >> (bit_pos - 32)) & 15)) );
|
|
#endif
|
|
#if 1
|
|
/* TODO test this! */
|
|
const GLubyte anibble = (blksrc[((j&3) * 4 + (i&3)) / 2] >> (4 * (i&1))) & 0xf;
|
|
dxt135_decode_imageblock(blksrc + 8, (i&3), (j&3), 2, texel);
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( (GLubyte)(EXP4TO8(anibble)) );
|
|
#endif
|
|
|
|
}
|
|
|
|
void fetch_2d_texel_rgba_dxt5(GLint srcRowStride, const GLubyte *pixdata,
|
|
GLint i, GLint j, GLvoid *texel) {
|
|
|
|
/* Extract the (i,j) pixel from pixdata and return it
|
|
* in texel[RCOMP], texel[GCOMP], texel[BCOMP], texel[ACOMP].
|
|
*/
|
|
|
|
GLchan *rgba = (GLchan *) texel;
|
|
const GLubyte *blksrc = (pixdata + ((srcRowStride + 3) / 4 * (j / 4) + (i / 4)) * 16);
|
|
const GLubyte alpha0 = blksrc[0];
|
|
const GLubyte alpha1 = blksrc[1];
|
|
#if 0
|
|
const GLubyte bit_pos = 3 * ((j&3) * 4 + (i&3));
|
|
/* simple 32bit version */
|
|
const GLuint bits_low = blksrc[2] | (blksrc[3] << 8) | (blksrc[4] << 16) | (blksrc[5] << 24);
|
|
const GLuint bits_high = blksrc[6] | (blksrc[7] << 8);
|
|
GLubyte code;
|
|
|
|
if (bit_pos < 30)
|
|
code = (GLubyte) ((bits_low >> bit_pos) & 7);
|
|
else if (bit_pos == 30)
|
|
code = (GLubyte) ((bits_low >> 30) & 3) | ((bits_high << 2) & 4);
|
|
else
|
|
code = (GLubyte) ((bits_high >> (bit_pos - 32)) & 7);
|
|
#endif
|
|
#if 1
|
|
/* TODO test this! */
|
|
const GLubyte bit_pos = ((j&3) * 4 + (i&3)) * 3;
|
|
const GLubyte acodelow = blksrc[2 + bit_pos / 8];
|
|
const GLubyte acodehigh = blksrc[3 + bit_pos / 8];
|
|
const GLubyte code = (acodelow >> (bit_pos & 0x7) |
|
|
(acodehigh << (8 - (bit_pos & 0x7)))) & 0x7;
|
|
#endif
|
|
dxt135_decode_imageblock(blksrc + 8, (i&3), (j&3), 2, texel);
|
|
#if 0
|
|
if (alpha0 > alpha1) {
|
|
switch (code) {
|
|
case 0:
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( alpha0 );
|
|
break;
|
|
case 1:
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( alpha1 );
|
|
break;
|
|
case 2:
|
|
case 3:
|
|
case 4:
|
|
case 5:
|
|
case 6:
|
|
case 7:
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( ((alpha0 * (8 - code) + (alpha1 * (code - 1))) / 7) );
|
|
break;
|
|
}
|
|
}
|
|
else {
|
|
switch (code) {
|
|
case 0:
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( alpha0 );
|
|
break;
|
|
case 1:
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( alpha1 );
|
|
break;
|
|
case 2:
|
|
case 3:
|
|
case 4:
|
|
case 5:
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( ((alpha0 * (6 - code) + (alpha1 * (code - 1))) / 5) );
|
|
break;
|
|
case 6:
|
|
rgba[ACOMP] = 0;
|
|
break;
|
|
case 7:
|
|
rgba[ACOMP] = CHAN_MAX;
|
|
break;
|
|
}
|
|
}
|
|
#endif
|
|
/* not sure. Which version is faster? */
|
|
#if 1
|
|
/* TODO test this */
|
|
if (code == 0)
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( alpha0 );
|
|
else if (code == 1)
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( alpha1 );
|
|
else if (alpha0 > alpha1)
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( ((alpha0 * (8 - code) + (alpha1 * (code - 1))) / 7) );
|
|
else if (code < 6)
|
|
rgba[ACOMP] = UBYTE_TO_CHAN( ((alpha0 * (6 - code) + (alpha1 * (code - 1))) / 5) );
|
|
else if (code == 6)
|
|
rgba[ACOMP] = 0;
|
|
else
|
|
rgba[ACOMP] = CHAN_MAX;
|
|
#endif
|
|
}
|