Mercurial > libavcodec.hg
annotate s3tc.c @ 11352:6e0af2cfdcfe libavcodec
Do MC and IDCT in coding (hilbert) order
This increases the slice size to 64 pixels, due to having to decode an
entire chroma superblock row per slice.
This can be up to 6% slower depending on clip and CPU, but is necessary
for future optimizations that gain significantly more than was lost.
author | conrad |
---|---|
date | Wed, 03 Mar 2010 23:27:40 +0000 |
parents | 2acf0ae7b041 |
children |
rev | line source |
---|---|
4933 | 1 /* |
2 * S3 Texture Compression (S3TC) decoding functions | |
3 * Copyright (c) 2007 by Ivo van Poorten | |
4 * | |
5214 | 5 * see also: http://wiki.multimedia.cx/index.php?title=S3TC |
6 * | |
4933 | 7 * This file is part of FFmpeg. |
8 * | |
9 * FFmpeg is free software; you can redistribute it and/or | |
10 * modify it under the terms of the GNU Lesser General Public | |
11 * License as published by the Free Software Foundation; either | |
12 * version 2.1 of the License, or (at your option) any later version. | |
13 * | |
14 * FFmpeg is distributed in the hope that it will be useful, | |
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
17 * Lesser General Public License for more details. | |
18 * | |
19 * You should have received a copy of the GNU Lesser General Public | |
20 * License along with FFmpeg; if not, write to the Free Software | |
21 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | |
22 */ | |
23 | |
8573
2acf0ae7b041
Fix build: Add intreadwrite.h and bswap.h #includes where necessary.
diego
parents:
6306
diff
changeset
|
24 #include "libavutil/intreadwrite.h" |
4933 | 25 #include "avcodec.h" |
26 #include "s3tc.h" | |
27 | |
28 static inline void dxt1_decode_pixels(const uint8_t *s, uint32_t *d, | |
29 unsigned int qstride, unsigned int flag, | |
30 uint64_t alpha) { | |
31 unsigned int x, y, c0, c1, a = (!flag * 255) << 24; | |
32 unsigned int rb0, rb1, rb2, rb3, g0, g1, g2, g3; | |
33 uint32_t colors[4], pixels; | |
34 | |
6306 | 35 c0 = AV_RL16(s); |
36 c1 = AV_RL16(s+2); | |
4933 | 37 |
38 rb0 = (c0<<3 | c0<<8) & 0xf800f8; | |
39 rb1 = (c1<<3 | c1<<8) & 0xf800f8; | |
40 rb0 += (rb0>>5) & 0x070007; | |
41 rb1 += (rb1>>5) & 0x070007; | |
42 g0 = (c0 <<5) & 0x00fc00; | |
43 g1 = (c1 <<5) & 0x00fc00; | |
44 g0 += (g0 >>6) & 0x000300; | |
45 g1 += (g1 >>6) & 0x000300; | |
46 | |
47 colors[0] = rb0 + g0 + a; | |
48 colors[1] = rb1 + g1 + a; | |
49 | |
50 if (c0 > c1 || flag) { | |
51 rb2 = (((2*rb0+rb1) * 21) >> 6) & 0xff00ff; | |
52 rb3 = (((2*rb1+rb0) * 21) >> 6) & 0xff00ff; | |
53 g2 = (((2*g0 +g1 ) * 21) >> 6) & 0x00ff00; | |
54 g3 = (((2*g1 +g0 ) * 21) >> 6) & 0x00ff00; | |
55 colors[3] = rb3 + g3 + a; | |
56 } else { | |
57 rb2 = ((rb0+rb1) >> 1) & 0xff00ff; | |
58 g2 = ((g0 +g1 ) >> 1) & 0x00ff00; | |
59 colors[3] = 0; | |
60 } | |
61 | |
62 colors[2] = rb2 + g2 + a; | |
63 | |
6306 | 64 pixels = AV_RL32(s+4); |
4933 | 65 for (y=0; y<4; y++) { |
66 for (x=0; x<4; x++) { | |
67 a = (alpha & 0x0f) << 28; | |
68 a += a >> 4; | |
69 d[x] = a + colors[pixels&3]; | |
70 pixels >>= 2; | |
71 alpha >>= 4; | |
72 } | |
73 d += qstride; | |
74 } | |
75 } | |
76 | |
77 void ff_decode_dxt1(const uint8_t *s, uint8_t *dst, | |
78 const unsigned int w, const unsigned int h, | |
79 const unsigned int stride) { | |
80 unsigned int bx, by, qstride = stride/4; | |
81 uint32_t *d = (uint32_t *) dst; | |
82 | |
83 for (by=0; by < h/4; by++, d += stride-w) | |
84 for (bx=0; bx < w/4; bx++, s+=8, d+=4) | |
85 dxt1_decode_pixels(s, d, qstride, 0, 0LL); | |
86 } | |
87 | |
88 void ff_decode_dxt3(const uint8_t *s, uint8_t *dst, | |
89 const unsigned int w, const unsigned int h, | |
90 const unsigned int stride) { | |
91 unsigned int bx, by, qstride = stride/4; | |
92 uint32_t *d = (uint32_t *) dst; | |
93 | |
94 for (by=0; by < h/4; by++, d += stride-w) | |
95 for (bx=0; bx < w/4; bx++, s+=16, d+=4) | |
96 dxt1_decode_pixels(s+8, d, qstride, 1, AV_RL64(s)); | |
97 } |