falkTX
/
FFmpeg
mirror of https://github.com/falkTX/FFmpeg.git
1
0
Fork 0
Code Issues 0 Releases 338 Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
28457 Commits
32 Branches
1.1GB
Tree: 44ca80df34
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '44ca80df34'
${ noResults }
FFmpeg/libavcodec/h264dsp_template.c

317 lines
11KB
Raw Blame History 

  1. /*

  2.  * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder

  3.  * Copyright (c) 2003-2010 Michael Niedermayer <michaelni@gmx.at>

  4.  *

  5.  * This file is part of Libav.

  6.  *

  7.  * Libav is free software; you can redistribute it and/or

  8.  * modify it under the terms of the GNU Lesser General Public

  9.  * License as published by the Free Software Foundation; either

  10.  * version 2.1 of the License, or (at your option) any later version.

  11.  *

  12.  * Libav is distributed in the hope that it will be useful,

  13.  * but WITHOUT ANY WARRANTY; without even the implied warranty of

  14.  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU

  15.  * Lesser General Public License for more details.

  16.  *

  17.  * You should have received a copy of the GNU Lesser General Public

  18.  * License along with Libav; if not, write to the Free Software

  19.  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA

  20.  */

  21. 

  22. /**

  23.  * @file

  24.  * H.264 / AVC / MPEG4 part10 DSP functions.

  25.  * @author Michael Niedermayer <michaelni@gmx.at>

  26.  */

  27. 

  28. #define BIT_DEPTH 8

  29. #define pixel uint8_t

  30. #define av_clip_pixel av_clip_uint8

  31. #define FUNCC(a) a ## _c

  32. 

  33. #define op_scale1(x)  block[x] = av_clip_pixel( (block[x]*weight + offset) >> log2_denom )

  34. #define op_scale2(x)  dst[x] = av_clip_pixel( (src[x]*weights + dst[x]*weightd + offset) >> (log2_denom+1))

  35. #define H264_WEIGHT(W,H) \

  36. static void FUNCC(weight_h264_pixels ## W ## x ## H)(uint8_t *_block, int stride, int log2_denom, int weight, int offset){ \

  37.     int y; \

  38.     pixel *block = (pixel*)_block; \

  39.     stride /= sizeof(pixel); \

  40.     offset <<= (log2_denom + (BIT_DEPTH-8)); \

  41.     if(log2_denom) offset += 1<<(log2_denom-1); \

  42.     for(y=0; y<H; y++, block += stride){ \

  43.         op_scale1(0); \

  44.         op_scale1(1); \

  45.         if(W==2) continue; \

  46.         op_scale1(2); \

  47.         op_scale1(3); \

  48.         if(W==4) continue; \

  49.         op_scale1(4); \

  50.         op_scale1(5); \

  51.         op_scale1(6); \

  52.         op_scale1(7); \

  53.         if(W==8) continue; \

  54.         op_scale1(8); \

  55.         op_scale1(9); \

  56.         op_scale1(10); \

  57.         op_scale1(11); \

  58.         op_scale1(12); \

  59.         op_scale1(13); \

  60.         op_scale1(14); \

  61.         op_scale1(15); \

  62.     } \

  63. } \

  64. static void FUNCC(biweight_h264_pixels ## W ## x ## H)(uint8_t *_dst, uint8_t *_src, int stride, int log2_denom, int weightd, int weights, int offset){ \

  65.     int y; \

  66.     pixel *dst = (pixel*)_dst; \

  67.     pixel *src = (pixel*)_src; \

  68.     stride /= sizeof(pixel); \

  69.     offset = ((offset + 1) | 1) << log2_denom; \

  70.     for(y=0; y<H; y++, dst += stride, src += stride){ \

  71.         op_scale2(0); \

  72.         op_scale2(1); \

  73.         if(W==2) continue; \

  74.         op_scale2(2); \

  75.         op_scale2(3); \

  76.         if(W==4) continue; \

  77.         op_scale2(4); \

  78.         op_scale2(5); \

  79.         op_scale2(6); \

  80.         op_scale2(7); \

  81.         if(W==8) continue; \

  82.         op_scale2(8); \

  83.         op_scale2(9); \

  84.         op_scale2(10); \

  85.         op_scale2(11); \

  86.         op_scale2(12); \

  87.         op_scale2(13); \

  88.         op_scale2(14); \

  89.         op_scale2(15); \

  90.     } \

  91. }

  92. 

  93. H264_WEIGHT(16,16)

  94. H264_WEIGHT(16,8)

  95. H264_WEIGHT(8,16)

  96. H264_WEIGHT(8,8)

  97. H264_WEIGHT(8,4)

  98. H264_WEIGHT(4,8)

  99. H264_WEIGHT(4,4)

  100. H264_WEIGHT(4,2)

  101. H264_WEIGHT(2,4)

  102. H264_WEIGHT(2,2)

  103. 

  104. #undef op_scale1

  105. #undef op_scale2

  106. #undef H264_WEIGHT

  107. 

  108. static av_always_inline av_flatten void FUNCC(h264_loop_filter_luma)(uint8_t *_pix, int xstride, int ystride, int inner_iters, int alpha, int beta, int8_t *tc0)

  109. {

  110.     pixel *pix = (pixel*)_pix;

  111.     int i, d;

  112.     xstride /= sizeof(pixel);

  113.     ystride /= sizeof(pixel);

  114.     alpha <<= BIT_DEPTH - 8;

  115.     beta  <<= BIT_DEPTH - 8;

  116.     for( i = 0; i < 4; i++ ) {

  117.         const int tc_orig = tc0[i] << (BIT_DEPTH - 8);

  118.         if( tc_orig < 0 ) {

  119.             pix += inner_iters*ystride;

  120.             continue;

  121.         }

  122.         for( d = 0; d < inner_iters; d++ ) {

  123.             const int p0 = pix[-1*xstride];

  124.             const int p1 = pix[-2*xstride];

  125.             const int p2 = pix[-3*xstride];

  126.             const int q0 = pix[0];

  127.             const int q1 = pix[1*xstride];

  128.             const int q2 = pix[2*xstride];

  129. 

  130.             if( FFABS( p0 - q0 ) < alpha &&

  131.                 FFABS( p1 - p0 ) < beta &&

  132.                 FFABS( q1 - q0 ) < beta ) {

  133. 

  134.                 int tc = tc_orig;

  135.                 int i_delta;

  136. 

  137.                 if( FFABS( p2 - p0 ) < beta ) {

  138.                     if(tc_orig)

  139.                     pix[-2*xstride] = p1 + av_clip( (( p2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - p1, -tc_orig, tc_orig );

  140.                     tc++;

  141.                 }

  142.                 if( FFABS( q2 - q0 ) < beta ) {

  143.                     if(tc_orig)

  144.                     pix[   xstride] = q1 + av_clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc_orig, tc_orig );

  145.                     tc++;

  146.                 }

  147. 

  148.                 i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );

  149.                 pix[-xstride] = av_clip_pixel( p0 + i_delta );    /* p0' */

  150.                 pix[0]        = av_clip_pixel( q0 - i_delta );    /* q0' */

  151.             }

  152.             pix += ystride;

  153.         }

  154.     }

  155. }

  156. static void FUNCC(h264_v_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)

  157. {

  158.     FUNCC(h264_loop_filter_luma)(pix, stride, sizeof(pixel), 4, alpha, beta, tc0);

  159. }

  160. static void FUNCC(h264_h_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)

  161. {

  162.     FUNCC(h264_loop_filter_luma)(pix, sizeof(pixel), stride, 4, alpha, beta, tc0);

  163. }

  164. static void FUNCC(h264_h_loop_filter_luma_mbaff)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)

  165. {

  166.     FUNCC(h264_loop_filter_luma)(pix, sizeof(pixel), stride, 2, alpha, beta, tc0);

  167. }

  168. 

  169. static av_always_inline av_flatten void FUNCC(h264_loop_filter_luma_intra)(uint8_t *_pix, int xstride, int ystride, int inner_iters, int alpha, int beta)

  170. {

  171.     pixel *pix = (pixel*)_pix;

  172.     int d;

  173.     xstride /= sizeof(pixel);

  174.     ystride /= sizeof(pixel);

  175.     alpha <<= BIT_DEPTH - 8;

  176.     beta  <<= BIT_DEPTH - 8;

  177.     for( d = 0; d < 4 * inner_iters; d++ ) {

  178.         const int p2 = pix[-3*xstride];

  179.         const int p1 = pix[-2*xstride];

  180.         const int p0 = pix[-1*xstride];

  181. 

  182.         const int q0 = pix[ 0*xstride];

  183.         const int q1 = pix[ 1*xstride];

  184.         const int q2 = pix[ 2*xstride];

  185. 

  186.         if( FFABS( p0 - q0 ) < alpha &&

  187.             FFABS( p1 - p0 ) < beta &&

  188.             FFABS( q1 - q0 ) < beta ) {

  189. 

  190.             if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){

  191.                 if( FFABS( p2 - p0 ) < beta)

  192.                 {

  193.                     const int p3 = pix[-4*xstride];

  194.                     /* p0', p1', p2' */

  195.                     pix[-1*xstride] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;

  196.                     pix[-2*xstride] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;

  197.                     pix[-3*xstride] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;

  198.                 } else {

  199.                     /* p0' */

  200.                     pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;

  201.                 }

  202.                 if( FFABS( q2 - q0 ) < beta)

  203.                 {

  204.                     const int q3 = pix[3*xstride];

  205.                     /* q0', q1', q2' */

  206.                     pix[0*xstride] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;

  207.                     pix[1*xstride] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;

  208.                     pix[2*xstride] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;

  209.                 } else {

  210.                     /* q0' */

  211.                     pix[0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;

  212.                 }

  213.             }else{

  214.                 /* p0', q0' */

  215.                 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;

  216.                 pix[ 0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;

  217.             }

  218.         }

  219.         pix += ystride;

  220.     }

  221. }

  222. static void FUNCC(h264_v_loop_filter_luma_intra)(uint8_t *pix, int stride, int alpha, int beta)

  223. {

  224.     FUNCC(h264_loop_filter_luma_intra)(pix, stride, sizeof(pixel), 4, alpha, beta);

  225. }

  226. static void FUNCC(h264_h_loop_filter_luma_intra)(uint8_t *pix, int stride, int alpha, int beta)

  227. {

  228.     FUNCC(h264_loop_filter_luma_intra)(pix, sizeof(pixel), stride, 4, alpha, beta);

  229. }

  230. static void FUNCC(h264_h_loop_filter_luma_mbaff_intra)(uint8_t *pix, int stride, int alpha, int beta)

  231. {

  232.     FUNCC(h264_loop_filter_luma_intra)(pix, sizeof(pixel), stride, 2, alpha, beta);

  233. }

  234. 

  235. static av_always_inline av_flatten void FUNCC(h264_loop_filter_chroma)(uint8_t *_pix, int xstride, int ystride, int inner_iters, int alpha, int beta, int8_t *tc0)

  236. {

  237.     pixel *pix = (pixel*)_pix;

  238.     int i, d;

  239.     xstride /= sizeof(pixel);

  240.     ystride /= sizeof(pixel);

  241.     alpha <<= BIT_DEPTH - 8;

  242.     beta  <<= BIT_DEPTH - 8;

  243.     for( i = 0; i < 4; i++ ) {

  244.         const int tc = ((tc0[i] - 1) << (BIT_DEPTH - 8)) + 1;

  245.         if( tc <= 0 ) {

  246.             pix += inner_iters*ystride;

  247.             continue;

  248.         }

  249.         for( d = 0; d < inner_iters; d++ ) {

  250.             const int p0 = pix[-1*xstride];

  251.             const int p1 = pix[-2*xstride];

  252.             const int q0 = pix[0];

  253.             const int q1 = pix[1*xstride];

  254. 

  255.             if( FFABS( p0 - q0 ) < alpha &&

  256.                 FFABS( p1 - p0 ) < beta &&

  257.                 FFABS( q1 - q0 ) < beta ) {

  258. 

  259.                 int delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );

  260. 

  261.                 pix[-xstride] = av_clip_pixel( p0 + delta );    /* p0' */

  262.                 pix[0]        = av_clip_pixel( q0 - delta );    /* q0' */

  263.             }

  264.             pix += ystride;

  265.         }

  266.     }

  267. }

  268. static void FUNCC(h264_v_loop_filter_chroma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)

  269. {

  270.     FUNCC(h264_loop_filter_chroma)(pix, stride, sizeof(pixel), 2, alpha, beta, tc0);

  271. }

  272. static void FUNCC(h264_h_loop_filter_chroma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)

  273. {

  274.     FUNCC(h264_loop_filter_chroma)(pix, sizeof(pixel), stride, 2, alpha, beta, tc0);

  275. }

  276. static void FUNCC(h264_h_loop_filter_chroma_mbaff)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)

  277. {

  278.     FUNCC(h264_loop_filter_chroma)(pix, sizeof(pixel), stride, 1, alpha, beta, tc0);

  279. }

  280. 

  281. static av_always_inline av_flatten void FUNCC(h264_loop_filter_chroma_intra)(uint8_t *_pix, int xstride, int ystride, int inner_iters, int alpha, int beta)

  282. {

  283.     pixel *pix = (pixel*)_pix;

  284.     int d;

  285.     xstride /= sizeof(pixel);

  286.     ystride /= sizeof(pixel);

  287.     alpha <<= BIT_DEPTH - 8;

  288.     beta  <<= BIT_DEPTH - 8;

  289.     for( d = 0; d < 4 * inner_iters; d++ ) {

  290.         const int p0 = pix[-1*xstride];

  291.         const int p1 = pix[-2*xstride];

  292.         const int q0 = pix[0];

  293.         const int q1 = pix[1*xstride];

  294. 

  295.         if( FFABS( p0 - q0 ) < alpha &&

  296.             FFABS( p1 - p0 ) < beta &&

  297.             FFABS( q1 - q0 ) < beta ) {

  298. 

  299.             pix[-xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;   /* p0' */

  300.             pix[0]        = ( 2*q1 + q0 + p1 + 2 ) >> 2;   /* q0' */

  301.         }

  302.         pix += ystride;

  303.     }

  304. }

  305. static void FUNCC(h264_v_loop_filter_chroma_intra)(uint8_t *pix, int stride, int alpha, int beta)

  306. {

  307.     FUNCC(h264_loop_filter_chroma_intra)(pix, stride, sizeof(pixel), 2, alpha, beta);

  308. }

  309. static void FUNCC(h264_h_loop_filter_chroma_intra)(uint8_t *pix, int stride, int alpha, int beta)

  310. {

  311.     FUNCC(h264_loop_filter_chroma_intra)(pix, sizeof(pixel), stride, 2, alpha, beta);

  312. }

  313. static void FUNCC(h264_h_loop_filter_chroma_mbaff_intra)(uint8_t *pix, int stride, int alpha, int beta)

  314. {

  315.     FUNCC(h264_loop_filter_chroma_intra)(pix, sizeof(pixel), stride, 1, alpha, beta);

  316. }