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.
33139 Commits
32 Branches
745MB
Tree: a75b254c74
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'a75b254c74'
${ noResults }
FFmpeg/libavcodec/h264dsp_template.c

331 lines
12KB
Raw Blame History 

  1. /*

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

  3.  * Copyright (c) 2003-2011 Michael Niedermayer <michaelni@gmx.at>

  4.  *

  5.  * This file is part of FFmpeg.

  6.  *

  7.  * FFmpeg 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.  * FFmpeg 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 FFmpeg; 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. #include "bit_depth_template.c"

  29. 

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

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

  32. #define H264_WEIGHT(W,H) \

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

  34.     int y; \

  35.     pixel *block = (pixel*)p_block; \

  36.     stride >>= sizeof(pixel)-1; \

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

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

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

  40.         op_scale1(0); \

  41.         op_scale1(1); \

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

  43.         op_scale1(2); \

  44.         op_scale1(3); \

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

  46.         op_scale1(4); \

  47.         op_scale1(5); \

  48.         op_scale1(6); \

  49.         op_scale1(7); \

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

  51.         op_scale1(8); \

  52.         op_scale1(9); \

  53.         op_scale1(10); \

  54.         op_scale1(11); \

  55.         op_scale1(12); \

  56.         op_scale1(13); \

  57.         op_scale1(14); \

  58.         op_scale1(15); \

  59.     } \

  60. } \

  61. 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){ \

  62.     int y; \

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

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

  65.     stride >>= sizeof(pixel)-1; \

  66.     offset <<= (BIT_DEPTH-8); \

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

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

  69.         op_scale2(0); \

  70.         op_scale2(1); \

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

  72.         op_scale2(2); \

  73.         op_scale2(3); \

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

  75.         op_scale2(4); \

  76.         op_scale2(5); \

  77.         op_scale2(6); \

  78.         op_scale2(7); \

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

  80.         op_scale2(8); \

  81.         op_scale2(9); \

  82.         op_scale2(10); \

  83.         op_scale2(11); \

  84.         op_scale2(12); \

  85.         op_scale2(13); \

  86.         op_scale2(14); \

  87.         op_scale2(15); \

  88.     } \

  89. }

  90. 

  91. H264_WEIGHT(16,16)

  92. H264_WEIGHT(16,8)

  93. H264_WEIGHT(8,16)

  94. H264_WEIGHT(8,8)

  95. H264_WEIGHT(8,4)

  96. H264_WEIGHT(4,8)

  97. H264_WEIGHT(4,4)

  98. H264_WEIGHT(4,2)

  99. H264_WEIGHT(2,4)

  100. H264_WEIGHT(2,2)

  101. 

  102. #undef op_scale1

  103. #undef op_scale2

  104. #undef H264_WEIGHT

  105. 

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

  107. {

  108.     pixel *pix = (pixel*)p_pix;

  109.     int i, d;

  110.     xstride >>= sizeof(pixel)-1;

  111.     ystride >>= sizeof(pixel)-1;

  112.     alpha <<= BIT_DEPTH - 8;

  113.     beta  <<= BIT_DEPTH - 8;

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

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

  116.         if( tc_orig < 0 ) {

  117.             pix += inner_iters*ystride;

  118.             continue;

  119.         }

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

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

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

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

  124.             const int q0 = pix[0];

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

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

  127. 

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

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

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

  131. 

  132.                 int tc = tc_orig;

  133.                 int i_delta;

  134. 

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

  136.                     if(tc_orig)

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

  138.                     tc++;

  139.                 }

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

  141.                     if(tc_orig)

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

  143.                     tc++;

  144.                 }

  145. 

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

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

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

  149.             }

  150.             pix += ystride;

  151.         }

  152.     }

  153. }

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

  155. {

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

  157. }

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

  159. {

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

  161. }

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

  163. {

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

  165. }

  166. 

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

  168. {

  169.     pixel *pix = (pixel*)p_pix;

  170.     int d;

  171.     xstride >>= sizeof(pixel)-1;

  172.     ystride >>= sizeof(pixel)-1;

  173.     alpha <<= BIT_DEPTH - 8;

  174.     beta  <<= BIT_DEPTH - 8;

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

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

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

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

  179. 

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

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

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

  183. 

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

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

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

  187. 

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

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

  190.                 {

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

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

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

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

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

  196.                 } else {

  197.                     /* p0' */

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

  199.                 }

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

  201.                 {

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

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

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

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

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

  207.                 } else {

  208.                     /* q0' */

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

  210.                 }

  211.             }else{

  212.                 /* p0', q0' */

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

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

  215.             }

  216.         }

  217.         pix += ystride;

  218.     }

  219. }

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

  221. {

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

  223. }

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

  225. {

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

  227. }

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

  229. {

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

  231. }

  232. 

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

  234. {

  235.     pixel *pix = (pixel*)p_pix;

  236.     int i, d;

  237.     alpha <<= BIT_DEPTH - 8;

  238.     beta  <<= BIT_DEPTH - 8;

  239.     xstride >>= sizeof(pixel)-1;

  240.     ystride >>= sizeof(pixel)-1;

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

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

  243.         if( tc <= 0 ) {

  244.             pix += inner_iters*ystride;

  245.             continue;

  246.         }

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

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

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

  250.             const int q0 = pix[0];

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

  252. 

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

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

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

  256. 

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

  258. 

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

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

  261.             }

  262.             pix += ystride;

  263.         }

  264.     }

  265. }

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

  267. {

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

  269. }

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

  271. {

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

  273. }

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

  275. {

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

  277. }

  278. static void FUNCC(h264_h_loop_filter_chroma422)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)

  279. {

  280.     FUNCC(h264_loop_filter_chroma)(pix, sizeof(pixel), stride, 4, alpha, beta, tc0);

  281. }

  282. static void FUNCC(h264_h_loop_filter_chroma422_mbaff)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)

  283. {

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

  285. }

  286. 

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

  288. {

  289.     pixel *pix = (pixel*)p_pix;

  290.     int d;

  291.     xstride >>= sizeof(pixel)-1;

  292.     ystride >>= sizeof(pixel)-1;

  293.     alpha <<= BIT_DEPTH - 8;

  294.     beta  <<= BIT_DEPTH - 8;

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

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

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

  298.         const int q0 = pix[0];

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

  300. 

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

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

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

  304. 

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

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

  307.         }

  308.         pix += ystride;

  309.     }

  310. }

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

  312. {

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

  314. }

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

  316. {

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

  318. }

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

  320. {

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

  322. }

  323. static void FUNCC(h264_h_loop_filter_chroma422_intra)(uint8_t *pix, int stride, int alpha, int beta)

  324. {

  325.     FUNCC(h264_loop_filter_chroma_intra)(pix, sizeof(pixel), stride, 4, alpha, beta);

  326. }

  327. static void FUNCC(h264_h_loop_filter_chroma422_mbaff_intra)(uint8_t *pix, int stride, int alpha, int beta)

  328. {

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

  330. }