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.

744 lines
34KB

  1. /*
  2. * H.26L/H.264/AVC/JVT/14496-10/... loop filter
  3. * Copyright (c) 2003 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. * @file libavcodec/h264_loopfilter.c
  23. * H.264 / AVC / MPEG4 part10 loop filter.
  24. * @author Michael Niedermayer <michaelni@gmx.at>
  25. */
  26. #include "internal.h"
  27. #include "dsputil.h"
  28. #include "avcodec.h"
  29. #include "mpegvideo.h"
  30. #include "h264.h"
  31. #include "mathops.h"
  32. #include "rectangle.h"
  33. //#undef NDEBUG
  34. #include <assert.h>
  35. /* Deblocking filter (p153) */
  36. static const uint8_t alpha_table[52*3] = {
  37. 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  38. 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  39. 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  40. 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  41. 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  42. 0, 0, 0, 0, 0, 0, 4, 4, 5, 6,
  43. 7, 8, 9, 10, 12, 13, 15, 17, 20, 22,
  44. 25, 28, 32, 36, 40, 45, 50, 56, 63, 71,
  45. 80, 90,101,113,127,144,162,182,203,226,
  46. 255,255,
  47. 255,255,255,255,255,255,255,255,255,255,255,255,255,
  48. 255,255,255,255,255,255,255,255,255,255,255,255,255,
  49. 255,255,255,255,255,255,255,255,255,255,255,255,255,
  50. 255,255,255,255,255,255,255,255,255,255,255,255,255,
  51. };
  52. static const uint8_t beta_table[52*3] = {
  53. 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  54. 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  55. 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  56. 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  57. 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  58. 0, 0, 0, 0, 0, 0, 2, 2, 2, 3,
  59. 3, 3, 3, 4, 4, 4, 6, 6, 7, 7,
  60. 8, 8, 9, 9, 10, 10, 11, 11, 12, 12,
  61. 13, 13, 14, 14, 15, 15, 16, 16, 17, 17,
  62. 18, 18,
  63. 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
  64. 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
  65. 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
  66. 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
  67. };
  68. static const uint8_t tc0_table[52*3][4] = {
  69. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
  70. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
  71. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
  72. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
  73. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
  74. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
  75. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
  76. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
  77. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
  78. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
  79. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
  80. {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 1 },
  81. {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 1, 1 }, {-1, 0, 1, 1 }, {-1, 1, 1, 1 },
  82. {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 },
  83. {-1, 1, 1, 2 }, {-1, 1, 2, 3 }, {-1, 1, 2, 3 }, {-1, 2, 2, 3 }, {-1, 2, 2, 4 }, {-1, 2, 3, 4 },
  84. {-1, 2, 3, 4 }, {-1, 3, 3, 5 }, {-1, 3, 4, 6 }, {-1, 3, 4, 6 }, {-1, 4, 5, 7 }, {-1, 4, 5, 8 },
  85. {-1, 4, 6, 9 }, {-1, 5, 7,10 }, {-1, 6, 8,11 }, {-1, 6, 8,13 }, {-1, 7,10,14 }, {-1, 8,11,16 },
  86. {-1, 9,12,18 }, {-1,10,13,20 }, {-1,11,15,23 }, {-1,13,17,25 },
  87. {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
  88. {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
  89. {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
  90. {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
  91. {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
  92. {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
  93. {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
  94. {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
  95. {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
  96. };
  97. static void av_always_inline filter_mb_edgev( uint8_t *pix, int stride, int16_t bS[4], unsigned int qp, H264Context *h) {
  98. const unsigned int index_a = qp + h->slice_alpha_c0_offset;
  99. const int alpha = alpha_table[index_a];
  100. const int beta = beta_table[qp + h->slice_beta_offset];
  101. if (alpha ==0 || beta == 0) return;
  102. if( bS[0] < 4 ) {
  103. int8_t tc[4];
  104. tc[0] = tc0_table[index_a][bS[0]];
  105. tc[1] = tc0_table[index_a][bS[1]];
  106. tc[2] = tc0_table[index_a][bS[2]];
  107. tc[3] = tc0_table[index_a][bS[3]];
  108. h->s.dsp.h264_h_loop_filter_luma(pix, stride, alpha, beta, tc);
  109. } else {
  110. h->s.dsp.h264_h_loop_filter_luma_intra(pix, stride, alpha, beta);
  111. }
  112. }
  113. static void av_always_inline filter_mb_edgecv( uint8_t *pix, int stride, int16_t bS[4], unsigned int qp, H264Context *h ) {
  114. const unsigned int index_a = qp + h->slice_alpha_c0_offset;
  115. const int alpha = alpha_table[index_a];
  116. const int beta = beta_table[qp + h->slice_beta_offset];
  117. if (alpha ==0 || beta == 0) return;
  118. if( bS[0] < 4 ) {
  119. int8_t tc[4];
  120. tc[0] = tc0_table[index_a][bS[0]]+1;
  121. tc[1] = tc0_table[index_a][bS[1]]+1;
  122. tc[2] = tc0_table[index_a][bS[2]]+1;
  123. tc[3] = tc0_table[index_a][bS[3]]+1;
  124. h->s.dsp.h264_h_loop_filter_chroma(pix, stride, alpha, beta, tc);
  125. } else {
  126. h->s.dsp.h264_h_loop_filter_chroma_intra(pix, stride, alpha, beta);
  127. }
  128. }
  129. static void filter_mb_mbaff_edgev( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int bsi, int qp ) {
  130. int i;
  131. int index_a = qp + h->slice_alpha_c0_offset;
  132. int alpha = alpha_table[index_a];
  133. int beta = beta_table[qp + h->slice_beta_offset];
  134. for( i = 0; i < 8; i++, pix += stride) {
  135. const int bS_index = (i >> 1) * bsi;
  136. if( bS[bS_index] == 0 ) {
  137. continue;
  138. }
  139. if( bS[bS_index] < 4 ) {
  140. const int tc0 = tc0_table[index_a][bS[bS_index]];
  141. const int p0 = pix[-1];
  142. const int p1 = pix[-2];
  143. const int p2 = pix[-3];
  144. const int q0 = pix[0];
  145. const int q1 = pix[1];
  146. const int q2 = pix[2];
  147. if( FFABS( p0 - q0 ) < alpha &&
  148. FFABS( p1 - p0 ) < beta &&
  149. FFABS( q1 - q0 ) < beta ) {
  150. int tc = tc0;
  151. int i_delta;
  152. if( FFABS( p2 - p0 ) < beta ) {
  153. if(tc0)
  154. pix[-2] = p1 + av_clip( ( p2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( p1 << 1 ) ) >> 1, -tc0, tc0 );
  155. tc++;
  156. }
  157. if( FFABS( q2 - q0 ) < beta ) {
  158. if(tc0)
  159. pix[1] = q1 + av_clip( ( q2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( q1 << 1 ) ) >> 1, -tc0, tc0 );
  160. tc++;
  161. }
  162. i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
  163. pix[-1] = av_clip_uint8( p0 + i_delta ); /* p0' */
  164. pix[0] = av_clip_uint8( q0 - i_delta ); /* q0' */
  165. tprintf(h->s.avctx, "filter_mb_mbaff_edgev i:%d, qp:%d, indexA:%d, alpha:%d, beta:%d, tc:%d\n# bS:%d -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, qp[qp_index], index_a, alpha, beta, tc, bS[bS_index], pix[-3], p1, p0, q0, q1, pix[2], p1, pix[-1], pix[0], q1);
  166. }
  167. }else{
  168. const int p0 = pix[-1];
  169. const int p1 = pix[-2];
  170. const int p2 = pix[-3];
  171. const int q0 = pix[0];
  172. const int q1 = pix[1];
  173. const int q2 = pix[2];
  174. if( FFABS( p0 - q0 ) < alpha &&
  175. FFABS( p1 - p0 ) < beta &&
  176. FFABS( q1 - q0 ) < beta ) {
  177. if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
  178. if( FFABS( p2 - p0 ) < beta)
  179. {
  180. const int p3 = pix[-4];
  181. /* p0', p1', p2' */
  182. pix[-1] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
  183. pix[-2] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
  184. pix[-3] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
  185. } else {
  186. /* p0' */
  187. pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
  188. }
  189. if( FFABS( q2 - q0 ) < beta)
  190. {
  191. const int q3 = pix[3];
  192. /* q0', q1', q2' */
  193. pix[0] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
  194. pix[1] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
  195. pix[2] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
  196. } else {
  197. /* q0' */
  198. pix[0] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
  199. }
  200. }else{
  201. /* p0', q0' */
  202. pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
  203. pix[ 0] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
  204. }
  205. tprintf(h->s.avctx, "filter_mb_mbaff_edgev i:%d, qp:%d, indexA:%d, alpha:%d, beta:%d\n# bS:4 -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x, %02x, %02x]\n", i, qp[qp_index], index_a, alpha, beta, p2, p1, p0, q0, q1, q2, pix[-3], pix[-2], pix[-1], pix[0], pix[1], pix[2]);
  206. }
  207. }
  208. }
  209. }
  210. static void filter_mb_mbaff_edgecv( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int bsi, int qp ) {
  211. int i;
  212. int index_a = qp + h->slice_alpha_c0_offset;
  213. int alpha = alpha_table[index_a];
  214. int beta = beta_table[qp + h->slice_beta_offset];
  215. for( i = 0; i < 4; i++, pix += stride) {
  216. const int bS_index = i*bsi;
  217. if( bS[bS_index] == 0 ) {
  218. continue;
  219. }
  220. if( bS[bS_index] < 4 ) {
  221. const int tc = tc0_table[index_a][bS[bS_index]] + 1;
  222. const int p0 = pix[-1];
  223. const int p1 = pix[-2];
  224. const int q0 = pix[0];
  225. const int q1 = pix[1];
  226. if( FFABS( p0 - q0 ) < alpha &&
  227. FFABS( p1 - p0 ) < beta &&
  228. FFABS( q1 - q0 ) < beta ) {
  229. const int i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
  230. pix[-1] = av_clip_uint8( p0 + i_delta ); /* p0' */
  231. pix[0] = av_clip_uint8( q0 - i_delta ); /* q0' */
  232. tprintf(h->s.avctx, "filter_mb_mbaff_edgecv i:%d, qp:%d, indexA:%d, alpha:%d, beta:%d, tc:%d\n# bS:%d -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, qp[qp_index], index_a, alpha, beta, tc, bS[bS_index], pix[-3], p1, p0, q0, q1, pix[2], p1, pix[-1], pix[0], q1);
  233. }
  234. }else{
  235. const int p0 = pix[-1];
  236. const int p1 = pix[-2];
  237. const int q0 = pix[0];
  238. const int q1 = pix[1];
  239. if( FFABS( p0 - q0 ) < alpha &&
  240. FFABS( p1 - p0 ) < beta &&
  241. FFABS( q1 - q0 ) < beta ) {
  242. pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2; /* p0' */
  243. pix[0] = ( 2*q1 + q0 + p1 + 2 ) >> 2; /* q0' */
  244. tprintf(h->s.avctx, "filter_mb_mbaff_edgecv i:%d\n# bS:4 -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x, %02x, %02x]\n", i, pix[-3], p1, p0, q0, q1, pix[2], pix[-3], pix[-2], pix[-1], pix[0], pix[1], pix[2]);
  245. }
  246. }
  247. }
  248. }
  249. static void av_always_inline filter_mb_edgeh( uint8_t *pix, int stride, int16_t bS[4], unsigned int qp, H264Context *h ) {
  250. const unsigned int index_a = qp + h->slice_alpha_c0_offset;
  251. const int alpha = alpha_table[index_a];
  252. const int beta = beta_table[qp + h->slice_beta_offset];
  253. if (alpha ==0 || beta == 0) return;
  254. if( bS[0] < 4 ) {
  255. int8_t tc[4];
  256. tc[0] = tc0_table[index_a][bS[0]];
  257. tc[1] = tc0_table[index_a][bS[1]];
  258. tc[2] = tc0_table[index_a][bS[2]];
  259. tc[3] = tc0_table[index_a][bS[3]];
  260. h->s.dsp.h264_v_loop_filter_luma(pix, stride, alpha, beta, tc);
  261. } else {
  262. h->s.dsp.h264_v_loop_filter_luma_intra(pix, stride, alpha, beta);
  263. }
  264. }
  265. static void av_always_inline filter_mb_edgech( uint8_t *pix, int stride, int16_t bS[4], unsigned int qp, H264Context *h ) {
  266. const unsigned int index_a = qp + h->slice_alpha_c0_offset;
  267. const int alpha = alpha_table[index_a];
  268. const int beta = beta_table[qp + h->slice_beta_offset];
  269. if (alpha ==0 || beta == 0) return;
  270. if( bS[0] < 4 ) {
  271. int8_t tc[4];
  272. tc[0] = tc0_table[index_a][bS[0]]+1;
  273. tc[1] = tc0_table[index_a][bS[1]]+1;
  274. tc[2] = tc0_table[index_a][bS[2]]+1;
  275. tc[3] = tc0_table[index_a][bS[3]]+1;
  276. h->s.dsp.h264_v_loop_filter_chroma(pix, stride, alpha, beta, tc);
  277. } else {
  278. h->s.dsp.h264_v_loop_filter_chroma_intra(pix, stride, alpha, beta);
  279. }
  280. }
  281. void ff_h264_filter_mb_fast( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize) {
  282. MpegEncContext * const s = &h->s;
  283. int mb_xy;
  284. int mb_type, left_type;
  285. int qp, qp0, qp1, qpc, qpc0, qpc1, qp_thresh;
  286. mb_xy = h->mb_xy;
  287. if(!h->top_type || !s->dsp.h264_loop_filter_strength || h->pps.chroma_qp_diff) {
  288. ff_h264_filter_mb(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize);
  289. return;
  290. }
  291. assert(!FRAME_MBAFF);
  292. left_type= h->left_type[0];
  293. mb_type = s->current_picture.mb_type[mb_xy];
  294. qp = s->current_picture.qscale_table[mb_xy];
  295. qp0 = s->current_picture.qscale_table[mb_xy-1];
  296. qp1 = s->current_picture.qscale_table[h->top_mb_xy];
  297. qpc = get_chroma_qp( h, 0, qp );
  298. qpc0 = get_chroma_qp( h, 0, qp0 );
  299. qpc1 = get_chroma_qp( h, 0, qp1 );
  300. qp0 = (qp + qp0 + 1) >> 1;
  301. qp1 = (qp + qp1 + 1) >> 1;
  302. qpc0 = (qpc + qpc0 + 1) >> 1;
  303. qpc1 = (qpc + qpc1 + 1) >> 1;
  304. qp_thresh = 15+52 - h->slice_alpha_c0_offset;
  305. if(qp <= qp_thresh && qp0 <= qp_thresh && qp1 <= qp_thresh &&
  306. qpc <= qp_thresh && qpc0 <= qp_thresh && qpc1 <= qp_thresh)
  307. return;
  308. if( IS_INTRA(mb_type) ) {
  309. int16_t bS4[4] = {4,4,4,4};
  310. int16_t bS3[4] = {3,3,3,3};
  311. int16_t *bSH = FIELD_PICTURE ? bS3 : bS4;
  312. if(left_type)
  313. filter_mb_edgev( &img_y[4*0], linesize, bS4, qp0, h);
  314. if( IS_8x8DCT(mb_type) ) {
  315. filter_mb_edgev( &img_y[4*2], linesize, bS3, qp, h);
  316. filter_mb_edgeh( &img_y[4*0*linesize], linesize, bSH, qp1, h);
  317. filter_mb_edgeh( &img_y[4*2*linesize], linesize, bS3, qp, h);
  318. } else {
  319. filter_mb_edgev( &img_y[4*1], linesize, bS3, qp, h);
  320. filter_mb_edgev( &img_y[4*2], linesize, bS3, qp, h);
  321. filter_mb_edgev( &img_y[4*3], linesize, bS3, qp, h);
  322. filter_mb_edgeh( &img_y[4*0*linesize], linesize, bSH, qp1, h);
  323. filter_mb_edgeh( &img_y[4*1*linesize], linesize, bS3, qp, h);
  324. filter_mb_edgeh( &img_y[4*2*linesize], linesize, bS3, qp, h);
  325. filter_mb_edgeh( &img_y[4*3*linesize], linesize, bS3, qp, h);
  326. }
  327. if(left_type){
  328. filter_mb_edgecv( &img_cb[2*0], uvlinesize, bS4, qpc0, h);
  329. filter_mb_edgecv( &img_cr[2*0], uvlinesize, bS4, qpc0, h);
  330. }
  331. filter_mb_edgecv( &img_cb[2*2], uvlinesize, bS3, qpc, h);
  332. filter_mb_edgecv( &img_cr[2*2], uvlinesize, bS3, qpc, h);
  333. filter_mb_edgech( &img_cb[2*0*uvlinesize], uvlinesize, bSH, qpc1, h);
  334. filter_mb_edgech( &img_cb[2*2*uvlinesize], uvlinesize, bS3, qpc, h);
  335. filter_mb_edgech( &img_cr[2*0*uvlinesize], uvlinesize, bSH, qpc1, h);
  336. filter_mb_edgech( &img_cr[2*2*uvlinesize], uvlinesize, bS3, qpc, h);
  337. return;
  338. } else {
  339. DECLARE_ALIGNED_8(int16_t, bS)[2][4][4];
  340. uint64_t (*bSv)[4] = (uint64_t(*)[4])bS;
  341. int edges;
  342. if( IS_8x8DCT(mb_type) && (h->cbp&7) == 7 ) {
  343. edges = 4;
  344. bSv[0][0] = bSv[0][2] = bSv[1][0] = bSv[1][2] = 0x0002000200020002ULL;
  345. } else {
  346. int mask_edge1 = (3*(((5*mb_type)>>5)&1)) | (mb_type>>4); //(mb_type & (MB_TYPE_16x16 | MB_TYPE_8x16)) ? 3 : (mb_type & MB_TYPE_16x8) ? 1 : 0;
  347. int mask_edge0 = 3*((mask_edge1>>1) & ((5*left_type)>>5)&1); // (mb_type & (MB_TYPE_16x16 | MB_TYPE_8x16)) && (h->left_type[0] & (MB_TYPE_16x16 | MB_TYPE_8x16)) ? 3 : 0;
  348. int step = 1+(mb_type>>24); //IS_8x8DCT(mb_type) ? 2 : 1;
  349. edges = 4 - 3*((mb_type>>3) & !(h->cbp & 15)); //(mb_type & MB_TYPE_16x16) && !(h->cbp & 15) ? 1 : 4;
  350. s->dsp.h264_loop_filter_strength( bS, h->non_zero_count_cache, h->ref_cache, h->mv_cache,
  351. h->list_count==2, edges, step, mask_edge0, mask_edge1, FIELD_PICTURE);
  352. }
  353. if( IS_INTRA(left_type) )
  354. bSv[0][0] = 0x0004000400040004ULL;
  355. if( IS_INTRA(h->top_type) )
  356. bSv[1][0] = FIELD_PICTURE ? 0x0003000300030003ULL : 0x0004000400040004ULL;
  357. #define FILTER(hv,dir,edge)\
  358. if(bSv[dir][edge]) {\
  359. filter_mb_edge##hv( &img_y[4*edge*(dir?linesize:1)], linesize, bS[dir][edge], edge ? qp : qp##dir, h );\
  360. if(!(edge&1)) {\
  361. filter_mb_edgec##hv( &img_cb[2*edge*(dir?uvlinesize:1)], uvlinesize, bS[dir][edge], edge ? qpc : qpc##dir, h );\
  362. filter_mb_edgec##hv( &img_cr[2*edge*(dir?uvlinesize:1)], uvlinesize, bS[dir][edge], edge ? qpc : qpc##dir, h );\
  363. }\
  364. }
  365. if(left_type)
  366. FILTER(v,0,0);
  367. if( edges == 1 ) {
  368. FILTER(h,1,0);
  369. } else if( IS_8x8DCT(mb_type) ) {
  370. FILTER(v,0,2);
  371. FILTER(h,1,0);
  372. FILTER(h,1,2);
  373. } else {
  374. FILTER(v,0,1);
  375. FILTER(v,0,2);
  376. FILTER(v,0,3);
  377. FILTER(h,1,0);
  378. FILTER(h,1,1);
  379. FILTER(h,1,2);
  380. FILTER(h,1,3);
  381. }
  382. #undef FILTER
  383. }
  384. }
  385. static int check_mv(H264Context *h, long b_idx, long bn_idx, int mvy_limit){
  386. int v;
  387. v= h->ref_cache[0][b_idx] != h->ref_cache[0][bn_idx];
  388. if(!v && h->ref_cache[0][b_idx]!=-1)
  389. v= h->mv_cache[0][b_idx][0] - h->mv_cache[0][bn_idx][0] + 3 >= 7U |
  390. FFABS( h->mv_cache[0][b_idx][1] - h->mv_cache[0][bn_idx][1] ) >= mvy_limit;
  391. if(h->list_count==2){
  392. if(!v)
  393. v = h->ref_cache[1][b_idx] != h->ref_cache[1][bn_idx] |
  394. h->mv_cache[1][b_idx][0] - h->mv_cache[1][bn_idx][0] + 3 >= 7U |
  395. FFABS( h->mv_cache[1][b_idx][1] - h->mv_cache[1][bn_idx][1] ) >= mvy_limit;
  396. if(v){
  397. if(h->ref_cache[0][b_idx] != h->ref_cache[1][bn_idx] |
  398. h->ref_cache[1][b_idx] != h->ref_cache[0][bn_idx])
  399. return 1;
  400. return
  401. h->mv_cache[0][b_idx][0] - h->mv_cache[1][bn_idx][0] + 3 >= 7U |
  402. FFABS( h->mv_cache[0][b_idx][1] - h->mv_cache[1][bn_idx][1] ) >= mvy_limit |
  403. h->mv_cache[1][b_idx][0] - h->mv_cache[0][bn_idx][0] + 3 >= 7U |
  404. FFABS( h->mv_cache[1][b_idx][1] - h->mv_cache[0][bn_idx][1] ) >= mvy_limit;
  405. }
  406. }
  407. return v;
  408. }
  409. static av_always_inline void filter_mb_dir(H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize, int mb_xy, int mb_type, int mvy_limit, int first_vertical_edge_done, int dir) {
  410. MpegEncContext * const s = &h->s;
  411. int edge;
  412. const int mbm_xy = dir == 0 ? mb_xy -1 : h->top_mb_xy;
  413. const int mbm_type = dir == 0 ? h->left_type[0] : h->top_type;
  414. // how often to recheck mv-based bS when iterating between edges
  415. static const uint8_t mask_edge_tab[2][8]={{0,3,3,3,1,1,1,1},
  416. {0,3,1,1,3,3,3,3}};
  417. const int mask_edge = mask_edge_tab[dir][(mb_type>>3)&7];
  418. const int edges = mask_edge== 3 && !(h->cbp&15) ? 1 : 4;
  419. // how often to recheck mv-based bS when iterating along each edge
  420. const int mask_par0 = mb_type & (MB_TYPE_16x16 | (MB_TYPE_8x16 >> dir));
  421. if(mbm_type && !first_vertical_edge_done){
  422. if (FRAME_MBAFF && (dir == 1) && ((mb_y&1) == 0)
  423. && IS_INTERLACED(mbm_type&~mb_type)
  424. ) {
  425. // This is a special case in the norm where the filtering must
  426. // be done twice (one each of the field) even if we are in a
  427. // frame macroblock.
  428. //
  429. unsigned int tmp_linesize = 2 * linesize;
  430. unsigned int tmp_uvlinesize = 2 * uvlinesize;
  431. int mbn_xy = mb_xy - 2 * s->mb_stride;
  432. int j;
  433. for(j=0; j<2; j++, mbn_xy += s->mb_stride){
  434. DECLARE_ALIGNED_8(int16_t, bS)[4];
  435. int qp;
  436. if( IS_INTRA(mb_type|s->current_picture.mb_type[mbn_xy]) ) {
  437. *(uint64_t*)bS= 0x0003000300030003ULL;
  438. } else {
  439. const uint8_t *mbn_nnz = h->non_zero_count[mbn_xy] + 4+3*8; //FIXME 8x8dct?
  440. int i;
  441. for( i = 0; i < 4; i++ ) {
  442. bS[i] = 1 + !!(h->non_zero_count_cache[scan8[0]+i] | mbn_nnz[i]);
  443. }
  444. }
  445. // Do not use s->qscale as luma quantizer because it has not the same
  446. // value in IPCM macroblocks.
  447. qp = ( s->current_picture.qscale_table[mb_xy] + s->current_picture.qscale_table[mbn_xy] + 1 ) >> 1;
  448. tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d ls:%d uvls:%d", mb_x, mb_y, dir, edge, qp, tmp_linesize, tmp_uvlinesize);
  449. { int i; for (i = 0; i < 4; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); }
  450. filter_mb_edgeh( &img_y[j*linesize], tmp_linesize, bS, qp, h );
  451. filter_mb_edgech( &img_cb[j*uvlinesize], tmp_uvlinesize, bS,
  452. ( h->chroma_qp[0] + get_chroma_qp( h, 0, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1, h);
  453. filter_mb_edgech( &img_cr[j*uvlinesize], tmp_uvlinesize, bS,
  454. ( h->chroma_qp[1] + get_chroma_qp( h, 1, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1, h);
  455. }
  456. }else{
  457. DECLARE_ALIGNED_8(int16_t, bS)[4];
  458. int qp;
  459. if( IS_INTRA(mb_type|mbm_type)) {
  460. *(uint64_t*)bS= 0x0003000300030003ULL;
  461. if ( (!IS_INTERLACED(mb_type|mbm_type))
  462. || ((FRAME_MBAFF || (s->picture_structure != PICT_FRAME)) && (dir == 0))
  463. )
  464. *(uint64_t*)bS= 0x0004000400040004ULL;
  465. } else {
  466. int i, l;
  467. int mv_done;
  468. if( FRAME_MBAFF && IS_INTERLACED(mb_type ^ mbm_type)) { //FIXME not posible left
  469. *(uint64_t*)bS= 0x0001000100010001ULL;
  470. mv_done = 1;
  471. }
  472. else if( mask_par0 && ((mbm_type & (MB_TYPE_16x16 | (MB_TYPE_8x16 >> dir)))) ) {
  473. int b_idx= 8 + 4;
  474. int bn_idx= b_idx - (dir ? 8:1);
  475. bS[0] = bS[1] = bS[2] = bS[3] = check_mv(h, 8 + 4, bn_idx, mvy_limit);
  476. mv_done = 1;
  477. }
  478. else
  479. mv_done = 0;
  480. for( i = 0; i < 4; i++ ) {
  481. int x = dir == 0 ? 0 : i;
  482. int y = dir == 0 ? i : 0;
  483. int b_idx= 8 + 4 + x + 8*y;
  484. int bn_idx= b_idx - (dir ? 8:1);
  485. if( h->non_zero_count_cache[b_idx] |
  486. h->non_zero_count_cache[bn_idx] ) {
  487. bS[i] = 2;
  488. }
  489. else if(!mv_done)
  490. {
  491. bS[i] = check_mv(h, b_idx, bn_idx, mvy_limit);
  492. }
  493. }
  494. }
  495. /* Filter edge */
  496. // Do not use s->qscale as luma quantizer because it has not the same
  497. // value in IPCM macroblocks.
  498. if(bS[0]+bS[1]+bS[2]+bS[3]){
  499. qp = ( s->current_picture.qscale_table[mb_xy] + s->current_picture.qscale_table[mbm_xy] + 1 ) >> 1;
  500. //tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d, QPc:%d, QPcn:%d\n", mb_x, mb_y, dir, edge, qp, h->chroma_qp[0], s->current_picture.qscale_table[mbn_xy]);
  501. tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d ls:%d uvls:%d", mb_x, mb_y, dir, edge, qp, linesize, uvlinesize);
  502. //{ int i; for (i = 0; i < 4; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); }
  503. if( dir == 0 ) {
  504. filter_mb_edgev( &img_y[0], linesize, bS, qp, h );
  505. {
  506. int qp= ( h->chroma_qp[0] + get_chroma_qp( h, 0, s->current_picture.qscale_table[mbm_xy] ) + 1 ) >> 1;
  507. filter_mb_edgecv( &img_cb[0], uvlinesize, bS, qp, h);
  508. if(h->pps.chroma_qp_diff)
  509. qp= ( h->chroma_qp[1] + get_chroma_qp( h, 1, s->current_picture.qscale_table[mbm_xy] ) + 1 ) >> 1;
  510. filter_mb_edgecv( &img_cr[0], uvlinesize, bS, qp, h);
  511. }
  512. } else {
  513. filter_mb_edgeh( &img_y[0], linesize, bS, qp, h );
  514. {
  515. int qp= ( h->chroma_qp[0] + get_chroma_qp( h, 0, s->current_picture.qscale_table[mbm_xy] ) + 1 ) >> 1;
  516. filter_mb_edgech( &img_cb[0], uvlinesize, bS, qp, h);
  517. if(h->pps.chroma_qp_diff)
  518. qp= ( h->chroma_qp[1] + get_chroma_qp( h, 1, s->current_picture.qscale_table[mbm_xy] ) + 1 ) >> 1;
  519. filter_mb_edgech( &img_cr[0], uvlinesize, bS, qp, h);
  520. }
  521. }
  522. }
  523. }
  524. }
  525. /* Calculate bS */
  526. for( edge = 1; edge < edges; edge++ ) {
  527. DECLARE_ALIGNED_8(int16_t, bS)[4];
  528. int qp;
  529. if( IS_8x8DCT(mb_type & (edge<<24)) ) // (edge&1) && IS_8x8DCT(mb_type)
  530. continue;
  531. if( IS_INTRA(mb_type)) {
  532. *(uint64_t*)bS= 0x0003000300030003ULL;
  533. } else {
  534. int i, l;
  535. int mv_done;
  536. if( edge & mask_edge ) {
  537. *(uint64_t*)bS= 0;
  538. mv_done = 1;
  539. }
  540. else if( mask_par0 ) {
  541. int b_idx= 8 + 4 + edge * (dir ? 8:1);
  542. int bn_idx= b_idx - (dir ? 8:1);
  543. bS[0] = bS[1] = bS[2] = bS[3] = check_mv(h, b_idx, bn_idx, mvy_limit);
  544. mv_done = 1;
  545. }
  546. else
  547. mv_done = 0;
  548. for( i = 0; i < 4; i++ ) {
  549. int x = dir == 0 ? edge : i;
  550. int y = dir == 0 ? i : edge;
  551. int b_idx= 8 + 4 + x + 8*y;
  552. int bn_idx= b_idx - (dir ? 8:1);
  553. if( h->non_zero_count_cache[b_idx] |
  554. h->non_zero_count_cache[bn_idx] ) {
  555. bS[i] = 2;
  556. }
  557. else if(!mv_done)
  558. {
  559. bS[i] = check_mv(h, b_idx, bn_idx, mvy_limit);
  560. }
  561. }
  562. if(bS[0]+bS[1]+bS[2]+bS[3] == 0)
  563. continue;
  564. }
  565. /* Filter edge */
  566. // Do not use s->qscale as luma quantizer because it has not the same
  567. // value in IPCM macroblocks.
  568. qp = s->current_picture.qscale_table[mb_xy];
  569. //tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d, QPc:%d, QPcn:%d\n", mb_x, mb_y, dir, edge, qp, h->chroma_qp[0], s->current_picture.qscale_table[mbn_xy]);
  570. tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d ls:%d uvls:%d", mb_x, mb_y, dir, edge, qp, linesize, uvlinesize);
  571. //{ int i; for (i = 0; i < 4; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); }
  572. if( dir == 0 ) {
  573. filter_mb_edgev( &img_y[4*edge], linesize, bS, qp, h );
  574. if( (edge&1) == 0 ) {
  575. filter_mb_edgecv( &img_cb[2*edge], uvlinesize, bS, h->chroma_qp[0], h);
  576. filter_mb_edgecv( &img_cr[2*edge], uvlinesize, bS, h->chroma_qp[1], h);
  577. }
  578. } else {
  579. filter_mb_edgeh( &img_y[4*edge*linesize], linesize, bS, qp, h );
  580. if( (edge&1) == 0 ) {
  581. filter_mb_edgech( &img_cb[2*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[0], h);
  582. filter_mb_edgech( &img_cr[2*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[1], h);
  583. }
  584. }
  585. }
  586. }
  587. void ff_h264_filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize) {
  588. MpegEncContext * const s = &h->s;
  589. const int mb_xy= mb_x + mb_y*s->mb_stride;
  590. const int mb_type = s->current_picture.mb_type[mb_xy];
  591. const int mvy_limit = IS_INTERLACED(mb_type) ? 2 : 4;
  592. int first_vertical_edge_done = 0;
  593. av_unused int dir;
  594. int list;
  595. if (FRAME_MBAFF
  596. // and current and left pair do not have the same interlaced type
  597. && IS_INTERLACED(mb_type^h->left_type[0])
  598. // and left mb is in available to us
  599. && h->left_type[0]) {
  600. /* First vertical edge is different in MBAFF frames
  601. * There are 8 different bS to compute and 2 different Qp
  602. */
  603. DECLARE_ALIGNED_8(int16_t, bS)[8];
  604. int qp[2];
  605. int bqp[2];
  606. int rqp[2];
  607. int mb_qp, mbn0_qp, mbn1_qp;
  608. int i;
  609. first_vertical_edge_done = 1;
  610. if( IS_INTRA(mb_type) )
  611. *(uint64_t*)&bS[0]=
  612. *(uint64_t*)&bS[4]= 0x0004000400040004ULL;
  613. else {
  614. static const uint8_t offset[2][2][8]={
  615. {
  616. {7+8*0, 7+8*0, 7+8*0, 7+8*0, 7+8*1, 7+8*1, 7+8*1, 7+8*1},
  617. {7+8*2, 7+8*2, 7+8*2, 7+8*2, 7+8*3, 7+8*3, 7+8*3, 7+8*3},
  618. },{
  619. {7+8*0, 7+8*1, 7+8*2, 7+8*3, 7+8*0, 7+8*1, 7+8*2, 7+8*3},
  620. {7+8*0, 7+8*1, 7+8*2, 7+8*3, 7+8*0, 7+8*1, 7+8*2, 7+8*3},
  621. }
  622. };
  623. const uint8_t *off= offset[MB_FIELD][mb_y&1];
  624. for( i = 0; i < 8; i++ ) {
  625. int j= MB_FIELD ? i>>2 : i&1;
  626. int mbn_xy = h->left_mb_xy[j];
  627. int mbn_type= h->left_type[j];
  628. if( IS_INTRA( mbn_type ) )
  629. bS[i] = 4;
  630. else{
  631. bS[i] = 1 + !!(h->non_zero_count_cache[12+8*(i>>1)] |
  632. ((!h->pps.cabac && IS_8x8DCT(mbn_type)) ?
  633. (h->cbp_table[mbn_xy] & ((MB_FIELD ? (i&2) : (mb_y&1)) ? 8 : 2))
  634. :
  635. h->non_zero_count[mbn_xy][ off[i] ]));
  636. }
  637. }
  638. }
  639. mb_qp = s->current_picture.qscale_table[mb_xy];
  640. mbn0_qp = s->current_picture.qscale_table[h->left_mb_xy[0]];
  641. mbn1_qp = s->current_picture.qscale_table[h->left_mb_xy[1]];
  642. qp[0] = ( mb_qp + mbn0_qp + 1 ) >> 1;
  643. bqp[0] = ( get_chroma_qp( h, 0, mb_qp ) +
  644. get_chroma_qp( h, 0, mbn0_qp ) + 1 ) >> 1;
  645. rqp[0] = ( get_chroma_qp( h, 1, mb_qp ) +
  646. get_chroma_qp( h, 1, mbn0_qp ) + 1 ) >> 1;
  647. qp[1] = ( mb_qp + mbn1_qp + 1 ) >> 1;
  648. bqp[1] = ( get_chroma_qp( h, 0, mb_qp ) +
  649. get_chroma_qp( h, 0, mbn1_qp ) + 1 ) >> 1;
  650. rqp[1] = ( get_chroma_qp( h, 1, mb_qp ) +
  651. get_chroma_qp( h, 1, mbn1_qp ) + 1 ) >> 1;
  652. /* Filter edge */
  653. tprintf(s->avctx, "filter mb:%d/%d MBAFF, QPy:%d/%d, QPb:%d/%d QPr:%d/%d ls:%d uvls:%d", mb_x, mb_y, qp[0], qp[1], bqp[0], bqp[1], rqp[0], rqp[1], linesize, uvlinesize);
  654. { int i; for (i = 0; i < 8; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); }
  655. if(MB_FIELD){
  656. filter_mb_mbaff_edgev ( h, img_y , linesize, bS , 1, qp [0] );
  657. filter_mb_mbaff_edgev ( h, img_y + 8* linesize, linesize, bS+4, 1, qp [1] );
  658. filter_mb_mbaff_edgecv( h, img_cb, uvlinesize, bS , 1, bqp[0] );
  659. filter_mb_mbaff_edgecv( h, img_cb + 4*uvlinesize, uvlinesize, bS+4, 1, bqp[1] );
  660. filter_mb_mbaff_edgecv( h, img_cr, uvlinesize, bS , 1, rqp[0] );
  661. filter_mb_mbaff_edgecv( h, img_cr + 4*uvlinesize, uvlinesize, bS+4, 1, rqp[1] );
  662. }else{
  663. filter_mb_mbaff_edgev ( h, img_y , 2* linesize, bS , 2, qp [0] );
  664. filter_mb_mbaff_edgev ( h, img_y + linesize, 2* linesize, bS+1, 2, qp [1] );
  665. filter_mb_mbaff_edgecv( h, img_cb, 2*uvlinesize, bS , 2, bqp[0] );
  666. filter_mb_mbaff_edgecv( h, img_cb + uvlinesize, 2*uvlinesize, bS+1, 2, bqp[1] );
  667. filter_mb_mbaff_edgecv( h, img_cr, 2*uvlinesize, bS , 2, rqp[0] );
  668. filter_mb_mbaff_edgecv( h, img_cr + uvlinesize, 2*uvlinesize, bS+1, 2, rqp[1] );
  669. }
  670. }
  671. #if CONFIG_SMALL
  672. for( dir = 0; dir < 2; dir++ )
  673. filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, dir ? 0 : first_vertical_edge_done, dir);
  674. #else
  675. filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, first_vertical_edge_done, 0);
  676. filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, 0, 1);
  677. #endif
  678. }