h264_loopfilter.c
Go to the documentation of this file.
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 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 
28 #include "libavutil/intreadwrite.h"
29 #include "internal.h"
30 #include "dsputil.h"
31 #include "avcodec.h"
32 #include "mpegvideo.h"
33 #include "h264.h"
34 #include "mathops.h"
35 #include "rectangle.h"
36 
37 //#undef NDEBUG
38 #include <assert.h>
39 
40 /* Deblocking filter (p153) */
41 static const uint8_t alpha_table[52*3] = {
42  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
43  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
44  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
45  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
46  0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
47  0, 0, 0, 0, 0, 0, 4, 4, 5, 6,
48  7, 8, 9, 10, 12, 13, 15, 17, 20, 22,
49  25, 28, 32, 36, 40, 45, 50, 56, 63, 71,
50  80, 90,101,113,127,144,162,182,203,226,
51  255,255,
52  255,255,255,255,255,255,255,255,255,255,255,255,255,
53  255,255,255,255,255,255,255,255,255,255,255,255,255,
54  255,255,255,255,255,255,255,255,255,255,255,255,255,
55  255,255,255,255,255,255,255,255,255,255,255,255,255,
56 };
57 static const uint8_t beta_table[52*3] = {
58  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
59  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
60  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
61  0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
62  0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
63  0, 0, 0, 0, 0, 0, 2, 2, 2, 3,
64  3, 3, 3, 4, 4, 4, 6, 6, 7, 7,
65  8, 8, 9, 9, 10, 10, 11, 11, 12, 12,
66  13, 13, 14, 14, 15, 15, 16, 16, 17, 17,
67  18, 18,
68  18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
69  18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
70  18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
71  18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
72 };
73 static const uint8_t tc0_table[52*3][4] = {
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 }, {-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, 0 },
81  {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
82  {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
83  {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
84  {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
85  {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 1 },
86  {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 1, 1 }, {-1, 0, 1, 1 }, {-1, 1, 1, 1 },
87  {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 },
88  {-1, 1, 1, 2 }, {-1, 1, 2, 3 }, {-1, 1, 2, 3 }, {-1, 2, 2, 3 }, {-1, 2, 2, 4 }, {-1, 2, 3, 4 },
89  {-1, 2, 3, 4 }, {-1, 3, 3, 5 }, {-1, 3, 4, 6 }, {-1, 3, 4, 6 }, {-1, 4, 5, 7 }, {-1, 4, 5, 8 },
90  {-1, 4, 6, 9 }, {-1, 5, 7,10 }, {-1, 6, 8,11 }, {-1, 6, 8,13 }, {-1, 7,10,14 }, {-1, 8,11,16 },
91  {-1, 9,12,18 }, {-1,10,13,20 }, {-1,11,15,23 }, {-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 }, {-1,13,17,25 }, {-1,13,17,25 },
96  {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
97  {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
98  {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
99  {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
100  {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
101 };
102 
103 /* intra: 0 if this loopfilter call is guaranteed to be inter (bS < 4), 1 if it might be intra (bS == 4) */
104 static av_always_inline void filter_mb_edgev(uint8_t *pix, int stride,
105  const int16_t bS[4],
106  unsigned int qp, int a, int b,
107  H264Context *h, int intra)
108 {
109  const unsigned int index_a = qp + a;
110  const int alpha = alpha_table[index_a];
111  const int beta = beta_table[qp + b];
112  if (alpha ==0 || beta == 0) return;
113 
114  if( bS[0] < 4 || !intra ) {
115  int8_t tc[4];
116  tc[0] = tc0_table[index_a][bS[0]];
117  tc[1] = tc0_table[index_a][bS[1]];
118  tc[2] = tc0_table[index_a][bS[2]];
119  tc[3] = tc0_table[index_a][bS[3]];
120  h->h264dsp.h264_h_loop_filter_luma(pix, stride, alpha, beta, tc);
121  } else {
122  h->h264dsp.h264_h_loop_filter_luma_intra(pix, stride, alpha, beta);
123  }
124 }
125 
126 static av_always_inline void filter_mb_edgecv(uint8_t *pix, int stride,
127  const int16_t bS[4],
128  unsigned int qp, int a, int b,
129  H264Context *h, int intra)
130 {
131  const unsigned int index_a = qp + a;
132  const int alpha = alpha_table[index_a];
133  const int beta = beta_table[qp + b];
134  if (alpha ==0 || beta == 0) return;
135 
136  if( bS[0] < 4 || !intra ) {
137  int8_t tc[4];
138  tc[0] = tc0_table[index_a][bS[0]]+1;
139  tc[1] = tc0_table[index_a][bS[1]]+1;
140  tc[2] = tc0_table[index_a][bS[2]]+1;
141  tc[3] = tc0_table[index_a][bS[3]]+1;
142  h->h264dsp.h264_h_loop_filter_chroma(pix, stride, alpha, beta, tc);
143  } else {
144  h->h264dsp.h264_h_loop_filter_chroma_intra(pix, stride, alpha, beta);
145  }
146 }
147 
149  int stride,
150  const int16_t bS[7], int bsi,
151  int qp, int a, int b,
152  int intra)
153 {
154  const unsigned int index_a = qp + a;
155  const int alpha = alpha_table[index_a];
156  const int beta = beta_table[qp + b];
157  if (alpha ==0 || beta == 0) return;
158 
159  if( bS[0] < 4 || !intra ) {
160  int8_t tc[4];
161  tc[0] = tc0_table[index_a][bS[0*bsi]];
162  tc[1] = tc0_table[index_a][bS[1*bsi]];
163  tc[2] = tc0_table[index_a][bS[2*bsi]];
164  tc[3] = tc0_table[index_a][bS[3*bsi]];
165  h->h264dsp.h264_h_loop_filter_luma_mbaff(pix, stride, alpha, beta, tc);
166  } else {
167  h->h264dsp.h264_h_loop_filter_luma_mbaff_intra(pix, stride, alpha, beta);
168  }
169 }
170 
172  uint8_t *pix, int stride,
173  const int16_t bS[7],
174  int bsi, int qp, int a,
175  int b, int intra)
176 {
177  const unsigned int index_a = qp + a;
178  const int alpha = alpha_table[index_a];
179  const int beta = beta_table[qp + b];
180  if (alpha ==0 || beta == 0) return;
181 
182  if( bS[0] < 4 || !intra ) {
183  int8_t tc[4];
184  tc[0] = tc0_table[index_a][bS[0*bsi]] + 1;
185  tc[1] = tc0_table[index_a][bS[1*bsi]] + 1;
186  tc[2] = tc0_table[index_a][bS[2*bsi]] + 1;
187  tc[3] = tc0_table[index_a][bS[3*bsi]] + 1;
188  h->h264dsp.h264_h_loop_filter_chroma_mbaff(pix, stride, alpha, beta, tc);
189  } else {
190  h->h264dsp.h264_h_loop_filter_chroma_mbaff_intra(pix, stride, alpha, beta);
191  }
192 }
193 
194 static av_always_inline void filter_mb_edgeh(uint8_t *pix, int stride,
195  const int16_t bS[4],
196  unsigned int qp, int a, int b,
197  H264Context *h, int intra)
198 {
199  const unsigned int index_a = qp + a;
200  const int alpha = alpha_table[index_a];
201  const int beta = beta_table[qp + b];
202  if (alpha ==0 || beta == 0) return;
203 
204  if( bS[0] < 4 || !intra ) {
205  int8_t tc[4];
206  tc[0] = tc0_table[index_a][bS[0]];
207  tc[1] = tc0_table[index_a][bS[1]];
208  tc[2] = tc0_table[index_a][bS[2]];
209  tc[3] = tc0_table[index_a][bS[3]];
210  h->h264dsp.h264_v_loop_filter_luma(pix, stride, alpha, beta, tc);
211  } else {
212  h->h264dsp.h264_v_loop_filter_luma_intra(pix, stride, alpha, beta);
213  }
214 }
215 
216 static av_always_inline void filter_mb_edgech(uint8_t *pix, int stride,
217  const int16_t bS[4],
218  unsigned int qp, int a, int b,
219  H264Context *h, int intra)
220 {
221  const unsigned int index_a = qp + a;
222  const int alpha = alpha_table[index_a];
223  const int beta = beta_table[qp + b];
224  if (alpha ==0 || beta == 0) return;
225 
226  if( bS[0] < 4 || !intra ) {
227  int8_t tc[4];
228  tc[0] = tc0_table[index_a][bS[0]]+1;
229  tc[1] = tc0_table[index_a][bS[1]]+1;
230  tc[2] = tc0_table[index_a][bS[2]]+1;
231  tc[3] = tc0_table[index_a][bS[3]]+1;
232  h->h264dsp.h264_v_loop_filter_chroma(pix, stride, alpha, beta, tc);
233  } else {
234  h->h264dsp.h264_v_loop_filter_chroma_intra(pix, stride, alpha, beta);
235  }
236 }
237 
239  int mb_x, int mb_y,
240  uint8_t *img_y,
241  uint8_t *img_cb,
242  uint8_t *img_cr,
243  unsigned int linesize,
244  unsigned int uvlinesize,
245  int pixel_shift)
246 {
247  MpegEncContext * const s = &h->s;
248  int chroma = !(CONFIG_GRAY && (s->flags&CODEC_FLAG_GRAY));
249  int chroma444 = CHROMA444;
250  int chroma422 = CHROMA422;
251 
252  int mb_xy = h->mb_xy;
253  int left_type= h->left_type[LTOP];
254  int top_type= h->top_type;
255 
256  int qp_bd_offset = 6 * (h->sps.bit_depth_luma - 8);
257  int a = 52 + h->slice_alpha_c0_offset - qp_bd_offset;
258  int b = 52 + h->slice_beta_offset - qp_bd_offset;
259 
260  int mb_type = s->current_picture.f.mb_type[mb_xy];
261  int qp = s->current_picture.f.qscale_table[mb_xy];
262  int qp0 = s->current_picture.f.qscale_table[mb_xy - 1];
263  int qp1 = s->current_picture.f.qscale_table[h->top_mb_xy];
264  int qpc = get_chroma_qp( h, 0, qp );
265  int qpc0 = get_chroma_qp( h, 0, qp0 );
266  int qpc1 = get_chroma_qp( h, 0, qp1 );
267  qp0 = (qp + qp0 + 1) >> 1;
268  qp1 = (qp + qp1 + 1) >> 1;
269  qpc0 = (qpc + qpc0 + 1) >> 1;
270  qpc1 = (qpc + qpc1 + 1) >> 1;
271 
272  if( IS_INTRA(mb_type) ) {
273  static const int16_t bS4[4] = {4,4,4,4};
274  static const int16_t bS3[4] = {3,3,3,3};
275  const int16_t *bSH = FIELD_PICTURE ? bS3 : bS4;
276  if(left_type)
277  filter_mb_edgev( &img_y[4*0<<pixel_shift], linesize, bS4, qp0, a, b, h, 1);
278  if( IS_8x8DCT(mb_type) ) {
279  filter_mb_edgev( &img_y[4*2<<pixel_shift], linesize, bS3, qp, a, b, h, 0);
280  if(top_type){
281  filter_mb_edgeh( &img_y[4*0*linesize], linesize, bSH, qp1, a, b, h, 1);
282  }
283  filter_mb_edgeh( &img_y[4*2*linesize], linesize, bS3, qp, a, b, h, 0);
284  } else {
285  filter_mb_edgev( &img_y[4*1<<pixel_shift], linesize, bS3, qp, a, b, h, 0);
286  filter_mb_edgev( &img_y[4*2<<pixel_shift], linesize, bS3, qp, a, b, h, 0);
287  filter_mb_edgev( &img_y[4*3<<pixel_shift], linesize, bS3, qp, a, b, h, 0);
288  if(top_type){
289  filter_mb_edgeh( &img_y[4*0*linesize], linesize, bSH, qp1, a, b, h, 1);
290  }
291  filter_mb_edgeh( &img_y[4*1*linesize], linesize, bS3, qp, a, b, h, 0);
292  filter_mb_edgeh( &img_y[4*2*linesize], linesize, bS3, qp, a, b, h, 0);
293  filter_mb_edgeh( &img_y[4*3*linesize], linesize, bS3, qp, a, b, h, 0);
294  }
295  if(chroma){
296  if(chroma444){
297  if(left_type){
298  filter_mb_edgev( &img_cb[4*0<<pixel_shift], linesize, bS4, qpc0, a, b, h, 1);
299  filter_mb_edgev( &img_cr[4*0<<pixel_shift], linesize, bS4, qpc0, a, b, h, 1);
300  }
301  if( IS_8x8DCT(mb_type) ) {
302  filter_mb_edgev( &img_cb[4*2<<pixel_shift], linesize, bS3, qpc, a, b, h, 0);
303  filter_mb_edgev( &img_cr[4*2<<pixel_shift], linesize, bS3, qpc, a, b, h, 0);
304  if(top_type){
305  filter_mb_edgeh( &img_cb[4*0*linesize], linesize, bSH, qpc1, a, b, h, 1 );
306  filter_mb_edgeh( &img_cr[4*0*linesize], linesize, bSH, qpc1, a, b, h, 1 );
307  }
308  filter_mb_edgeh( &img_cb[4*2*linesize], linesize, bS3, qpc, a, b, h, 0);
309  filter_mb_edgeh( &img_cr[4*2*linesize], linesize, bS3, qpc, a, b, h, 0);
310  } else {
311  filter_mb_edgev( &img_cb[4*1<<pixel_shift], linesize, bS3, qpc, a, b, h, 0);
312  filter_mb_edgev( &img_cr[4*1<<pixel_shift], linesize, bS3, qpc, a, b, h, 0);
313  filter_mb_edgev( &img_cb[4*2<<pixel_shift], linesize, bS3, qpc, a, b, h, 0);
314  filter_mb_edgev( &img_cr[4*2<<pixel_shift], linesize, bS3, qpc, a, b, h, 0);
315  filter_mb_edgev( &img_cb[4*3<<pixel_shift], linesize, bS3, qpc, a, b, h, 0);
316  filter_mb_edgev( &img_cr[4*3<<pixel_shift], linesize, bS3, qpc, a, b, h, 0);
317  if(top_type){
318  filter_mb_edgeh( &img_cb[4*0*linesize], linesize, bSH, qpc1, a, b, h, 1);
319  filter_mb_edgeh( &img_cr[4*0*linesize], linesize, bSH, qpc1, a, b, h, 1);
320  }
321  filter_mb_edgeh( &img_cb[4*1*linesize], linesize, bS3, qpc, a, b, h, 0);
322  filter_mb_edgeh( &img_cr[4*1*linesize], linesize, bS3, qpc, a, b, h, 0);
323  filter_mb_edgeh( &img_cb[4*2*linesize], linesize, bS3, qpc, a, b, h, 0);
324  filter_mb_edgeh( &img_cr[4*2*linesize], linesize, bS3, qpc, a, b, h, 0);
325  filter_mb_edgeh( &img_cb[4*3*linesize], linesize, bS3, qpc, a, b, h, 0);
326  filter_mb_edgeh( &img_cr[4*3*linesize], linesize, bS3, qpc, a, b, h, 0);
327  }
328  }else if(chroma422){
329  if(left_type){
330  filter_mb_edgecv(&img_cb[2*0<<pixel_shift], uvlinesize, bS4, qpc0, a, b, h, 1);
331  filter_mb_edgecv(&img_cr[2*0<<pixel_shift], uvlinesize, bS4, qpc0, a, b, h, 1);
332  }
333  filter_mb_edgecv(&img_cb[2*2<<pixel_shift], uvlinesize, bS3, qpc, a, b, h, 0);
334  filter_mb_edgecv(&img_cr[2*2<<pixel_shift], uvlinesize, bS3, qpc, a, b, h, 0);
335  if(top_type){
336  filter_mb_edgech(&img_cb[4*0*uvlinesize], uvlinesize, bSH, qpc1, a, b, h, 1);
337  filter_mb_edgech(&img_cr[4*0*uvlinesize], uvlinesize, bSH, qpc1, a, b, h, 1);
338  }
339  filter_mb_edgech(&img_cb[4*1*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);
340  filter_mb_edgech(&img_cr[4*1*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);
341  filter_mb_edgech(&img_cb[4*2*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);
342  filter_mb_edgech(&img_cr[4*2*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);
343  filter_mb_edgech(&img_cb[4*3*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);
344  filter_mb_edgech(&img_cr[4*3*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);
345  }else{
346  if(left_type){
347  filter_mb_edgecv( &img_cb[2*0<<pixel_shift], uvlinesize, bS4, qpc0, a, b, h, 1);
348  filter_mb_edgecv( &img_cr[2*0<<pixel_shift], uvlinesize, bS4, qpc0, a, b, h, 1);
349  }
350  filter_mb_edgecv( &img_cb[2*2<<pixel_shift], uvlinesize, bS3, qpc, a, b, h, 0);
351  filter_mb_edgecv( &img_cr[2*2<<pixel_shift], uvlinesize, bS3, qpc, a, b, h, 0);
352  if(top_type){
353  filter_mb_edgech( &img_cb[2*0*uvlinesize], uvlinesize, bSH, qpc1, a, b, h, 1);
354  filter_mb_edgech( &img_cr[2*0*uvlinesize], uvlinesize, bSH, qpc1, a, b, h, 1);
355  }
356  filter_mb_edgech( &img_cb[2*2*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);
357  filter_mb_edgech( &img_cr[2*2*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);
358  }
359  }
360  return;
361  } else {
362  LOCAL_ALIGNED_8(int16_t, bS, [2], [4][4]);
363  int edges;
364  if( IS_8x8DCT(mb_type) && (h->cbp&7) == 7 && !chroma444 ) {
365  edges = 4;
366  AV_WN64A(bS[0][0], 0x0002000200020002ULL);
367  AV_WN64A(bS[0][2], 0x0002000200020002ULL);
368  AV_WN64A(bS[1][0], 0x0002000200020002ULL);
369  AV_WN64A(bS[1][2], 0x0002000200020002ULL);
370  } else {
371  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;
372  int mask_edge0 = 3*((mask_edge1>>1) & ((5*left_type)>>5)&1); // (mb_type & (MB_TYPE_16x16 | MB_TYPE_8x16)) && (h->left_type[LTOP] & (MB_TYPE_16x16 | MB_TYPE_8x16)) ? 3 : 0;
373  int step = 1+(mb_type>>24); //IS_8x8DCT(mb_type) ? 2 : 1;
374  edges = 4 - 3*((mb_type>>3) & !(h->cbp & 15)); //(mb_type & MB_TYPE_16x16) && !(h->cbp & 15) ? 1 : 4;
376  h->list_count==2, edges, step, mask_edge0, mask_edge1, FIELD_PICTURE);
377  }
378  if( IS_INTRA(left_type) )
379  AV_WN64A(bS[0][0], 0x0004000400040004ULL);
380  if( IS_INTRA(top_type) )
381  AV_WN64A(bS[1][0], FIELD_PICTURE ? 0x0003000300030003ULL : 0x0004000400040004ULL);
382 
383 #define FILTER(hv,dir,edge,intra)\
384  if(AV_RN64A(bS[dir][edge])) { \
385  filter_mb_edge##hv( &img_y[4*edge*(dir?linesize:1<<pixel_shift)], linesize, bS[dir][edge], edge ? qp : qp##dir, a, b, h, intra );\
386  if(chroma){\
387  if(chroma444){\
388  filter_mb_edge##hv( &img_cb[4*edge*(dir?linesize:1<<pixel_shift)], linesize, bS[dir][edge], edge ? qpc : qpc##dir, a, b, h, intra );\
389  filter_mb_edge##hv( &img_cr[4*edge*(dir?linesize:1<<pixel_shift)], linesize, bS[dir][edge], edge ? qpc : qpc##dir, a, b, h, intra );\
390  } else if(!(edge&1)) {\
391  filter_mb_edgec##hv( &img_cb[2*edge*(dir?uvlinesize:1<<pixel_shift)], uvlinesize, bS[dir][edge], edge ? qpc : qpc##dir, a, b, h, intra );\
392  filter_mb_edgec##hv( &img_cr[2*edge*(dir?uvlinesize:1<<pixel_shift)], uvlinesize, bS[dir][edge], edge ? qpc : qpc##dir, a, b, h, intra );\
393  }\
394  }\
395  }
396  if(left_type)
397  FILTER(v,0,0,1);
398  if( edges == 1 ) {
399  if(top_type)
400  FILTER(h,1,0,1);
401  } else if( IS_8x8DCT(mb_type) ) {
402  FILTER(v,0,2,0);
403  if(top_type)
404  FILTER(h,1,0,1);
405  FILTER(h,1,2,0);
406  } else {
407  FILTER(v,0,1,0);
408  FILTER(v,0,2,0);
409  FILTER(v,0,3,0);
410  if(top_type)
411  FILTER(h,1,0,1);
412  FILTER(h,1,1,0);
413  FILTER(h,1,2,0);
414  FILTER(h,1,3,0);
415  }
416 #undef FILTER
417  }
418 }
419 
420 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) {
421  assert(!FRAME_MBAFF);
423  ff_h264_filter_mb(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize);
424  return;
425  }
426 
427 #if CONFIG_SMALL
428  h264_filter_mb_fast_internal(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, h->pixel_shift);
429 #else
430  if(h->pixel_shift){
431  h264_filter_mb_fast_internal(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, 1);
432  }else{
433  h264_filter_mb_fast_internal(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, 0);
434  }
435 #endif
436 }
437 
438 static int check_mv(H264Context *h, long b_idx, long bn_idx, int mvy_limit){
439  int v;
440 
441  v= h->ref_cache[0][b_idx] != h->ref_cache[0][bn_idx];
442  if(!v && h->ref_cache[0][b_idx]!=-1)
443  v= h->mv_cache[0][b_idx][0] - h->mv_cache[0][bn_idx][0] + 3 >= 7U |
444  FFABS( h->mv_cache[0][b_idx][1] - h->mv_cache[0][bn_idx][1] ) >= mvy_limit;
445 
446  if(h->list_count==2){
447  if(!v)
448  v = h->ref_cache[1][b_idx] != h->ref_cache[1][bn_idx] |
449  h->mv_cache[1][b_idx][0] - h->mv_cache[1][bn_idx][0] + 3 >= 7U |
450  FFABS( h->mv_cache[1][b_idx][1] - h->mv_cache[1][bn_idx][1] ) >= mvy_limit;
451 
452  if(v){
453  if(h->ref_cache[0][b_idx] != h->ref_cache[1][bn_idx] |
454  h->ref_cache[1][b_idx] != h->ref_cache[0][bn_idx])
455  return 1;
456  return
457  h->mv_cache[0][b_idx][0] - h->mv_cache[1][bn_idx][0] + 3 >= 7U |
458  FFABS( h->mv_cache[0][b_idx][1] - h->mv_cache[1][bn_idx][1] ) >= mvy_limit |
459  h->mv_cache[1][b_idx][0] - h->mv_cache[0][bn_idx][0] + 3 >= 7U |
460  FFABS( h->mv_cache[1][b_idx][1] - h->mv_cache[0][bn_idx][1] ) >= mvy_limit;
461  }
462  }
463 
464  return v;
465 }
466 
467 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 a, int b, int chroma, int dir) {
468  MpegEncContext * const s = &h->s;
469  int edge;
470  int chroma_qp_avg[2];
471  int chroma444 = CHROMA444;
472  int chroma422 = CHROMA422;
473  const int mbm_xy = dir == 0 ? mb_xy -1 : h->top_mb_xy;
474  const int mbm_type = dir == 0 ? h->left_type[LTOP] : h->top_type;
475 
476  // how often to recheck mv-based bS when iterating between edges
477  static const uint8_t mask_edge_tab[2][8]={{0,3,3,3,1,1,1,1},
478  {0,3,1,1,3,3,3,3}};
479  const int mask_edge = mask_edge_tab[dir][(mb_type>>3)&7];
480  const int edges = mask_edge== 3 && !(h->cbp&15) ? 1 : 4;
481 
482  // how often to recheck mv-based bS when iterating along each edge
483  const int mask_par0 = mb_type & (MB_TYPE_16x16 | (MB_TYPE_8x16 >> dir));
484 
485  if(mbm_type && !first_vertical_edge_done){
486 
487  if (FRAME_MBAFF && (dir == 1) && ((mb_y&1) == 0)
488  && IS_INTERLACED(mbm_type&~mb_type)
489  ) {
490  // This is a special case in the norm where the filtering must
491  // be done twice (one each of the field) even if we are in a
492  // frame macroblock.
493  //
494  unsigned int tmp_linesize = 2 * linesize;
495  unsigned int tmp_uvlinesize = 2 * uvlinesize;
496  int mbn_xy = mb_xy - 2 * s->mb_stride;
497  int j;
498 
499  for(j=0; j<2; j++, mbn_xy += s->mb_stride){
500  DECLARE_ALIGNED(8, int16_t, bS)[4];
501  int qp;
502  if (IS_INTRA(mb_type | s->current_picture.f.mb_type[mbn_xy])) {
503  AV_WN64A(bS, 0x0003000300030003ULL);
504  } else {
505  if (!CABAC && IS_8x8DCT(s->current_picture.f.mb_type[mbn_xy])) {
506  bS[0]= 1+((h->cbp_table[mbn_xy] & 0x4000)||h->non_zero_count_cache[scan8[0]+0]);
507  bS[1]= 1+((h->cbp_table[mbn_xy] & 0x4000)||h->non_zero_count_cache[scan8[0]+1]);
508  bS[2]= 1+((h->cbp_table[mbn_xy] & 0x8000)||h->non_zero_count_cache[scan8[0]+2]);
509  bS[3]= 1+((h->cbp_table[mbn_xy] & 0x8000)||h->non_zero_count_cache[scan8[0]+3]);
510  }else{
511  const uint8_t *mbn_nnz = h->non_zero_count[mbn_xy] + 3*4;
512  int i;
513  for( i = 0; i < 4; i++ ) {
514  bS[i] = 1 + !!(h->non_zero_count_cache[scan8[0]+i] | mbn_nnz[i]);
515  }
516  }
517  }
518  // Do not use s->qscale as luma quantizer because it has not the same
519  // value in IPCM macroblocks.
520  qp = (s->current_picture.f.qscale_table[mb_xy] + s->current_picture.f.qscale_table[mbn_xy] + 1) >> 1;
521  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);
522  { int i; for (i = 0; i < 4; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); }
523  filter_mb_edgeh( &img_y[j*linesize], tmp_linesize, bS, qp, a, b, h, 0 );
524  chroma_qp_avg[0] = (h->chroma_qp[0] + get_chroma_qp(h, 0, s->current_picture.f.qscale_table[mbn_xy]) + 1) >> 1;
525  chroma_qp_avg[1] = (h->chroma_qp[1] + get_chroma_qp(h, 1, s->current_picture.f.qscale_table[mbn_xy]) + 1) >> 1;
526  if (chroma) {
527  if (chroma444) {
528  filter_mb_edgeh (&img_cb[j*uvlinesize], tmp_uvlinesize, bS, chroma_qp_avg[0], a, b, h, 0);
529  filter_mb_edgeh (&img_cr[j*uvlinesize], tmp_uvlinesize, bS, chroma_qp_avg[1], a, b, h, 0);
530  } else {
531  filter_mb_edgech(&img_cb[j*uvlinesize], tmp_uvlinesize, bS, chroma_qp_avg[0], a, b, h, 0);
532  filter_mb_edgech(&img_cr[j*uvlinesize], tmp_uvlinesize, bS, chroma_qp_avg[1], a, b, h, 0);
533  }
534  }
535  }
536  }else{
537  DECLARE_ALIGNED(8, int16_t, bS)[4];
538  int qp;
539 
540  if( IS_INTRA(mb_type|mbm_type)) {
541  AV_WN64A(bS, 0x0003000300030003ULL);
542  if ( (!IS_INTERLACED(mb_type|mbm_type))
543  || ((FRAME_MBAFF || (s->picture_structure != PICT_FRAME)) && (dir == 0))
544  )
545  AV_WN64A(bS, 0x0004000400040004ULL);
546  } else {
547  int i;
548  int mv_done;
549 
550  if( dir && FRAME_MBAFF && IS_INTERLACED(mb_type ^ mbm_type)) {
551  AV_WN64A(bS, 0x0001000100010001ULL);
552  mv_done = 1;
553  }
554  else if( mask_par0 && ((mbm_type & (MB_TYPE_16x16 | (MB_TYPE_8x16 >> dir)))) ) {
555  int b_idx= 8 + 4;
556  int bn_idx= b_idx - (dir ? 8:1);
557 
558  bS[0] = bS[1] = bS[2] = bS[3] = check_mv(h, 8 + 4, bn_idx, mvy_limit);
559  mv_done = 1;
560  }
561  else
562  mv_done = 0;
563 
564  for( i = 0; i < 4; i++ ) {
565  int x = dir == 0 ? 0 : i;
566  int y = dir == 0 ? i : 0;
567  int b_idx= 8 + 4 + x + 8*y;
568  int bn_idx= b_idx - (dir ? 8:1);
569 
570  if( h->non_zero_count_cache[b_idx] |
571  h->non_zero_count_cache[bn_idx] ) {
572  bS[i] = 2;
573  }
574  else if(!mv_done)
575  {
576  bS[i] = check_mv(h, b_idx, bn_idx, mvy_limit);
577  }
578  }
579  }
580 
581  /* Filter edge */
582  // Do not use s->qscale as luma quantizer because it has not the same
583  // value in IPCM macroblocks.
584  if(bS[0]+bS[1]+bS[2]+bS[3]){
585  qp = (s->current_picture.f.qscale_table[mb_xy] + s->current_picture.f.qscale_table[mbm_xy] + 1) >> 1;
586  //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]);
587  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);
588  //{ int i; for (i = 0; i < 4; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); }
589  chroma_qp_avg[0] = (h->chroma_qp[0] + get_chroma_qp(h, 0, s->current_picture.f.qscale_table[mbm_xy]) + 1) >> 1;
590  chroma_qp_avg[1] = (h->chroma_qp[1] + get_chroma_qp(h, 1, s->current_picture.f.qscale_table[mbm_xy]) + 1) >> 1;
591  if( dir == 0 ) {
592  filter_mb_edgev( &img_y[0], linesize, bS, qp, a, b, h, 1 );
593  if (chroma) {
594  if (chroma444) {
595  filter_mb_edgev ( &img_cb[0], uvlinesize, bS, chroma_qp_avg[0], a, b, h, 1);
596  filter_mb_edgev ( &img_cr[0], uvlinesize, bS, chroma_qp_avg[1], a, b, h, 1);
597  } else {
598  filter_mb_edgecv( &img_cb[0], uvlinesize, bS, chroma_qp_avg[0], a, b, h, 1);
599  filter_mb_edgecv( &img_cr[0], uvlinesize, bS, chroma_qp_avg[1], a, b, h, 1);
600  }
601  }
602  } else {
603  filter_mb_edgeh( &img_y[0], linesize, bS, qp, a, b, h, 1 );
604  if (chroma) {
605  if (chroma444) {
606  filter_mb_edgeh ( &img_cb[0], uvlinesize, bS, chroma_qp_avg[0], a, b, h, 1);
607  filter_mb_edgeh ( &img_cr[0], uvlinesize, bS, chroma_qp_avg[1], a, b, h, 1);
608  } else {
609  filter_mb_edgech( &img_cb[0], uvlinesize, bS, chroma_qp_avg[0], a, b, h, 1);
610  filter_mb_edgech( &img_cr[0], uvlinesize, bS, chroma_qp_avg[1], a, b, h, 1);
611  }
612  }
613  }
614  }
615  }
616  }
617 
618  /* Calculate bS */
619  for( edge = 1; edge < edges; edge++ ) {
620  DECLARE_ALIGNED(8, int16_t, bS)[4];
621  int qp;
622  const int deblock_edge = !IS_8x8DCT(mb_type & (edge<<24)); // (edge&1) && IS_8x8DCT(mb_type)
623 
624  if (!deblock_edge && (!chroma422 || dir == 0))
625  continue;
626 
627  if( IS_INTRA(mb_type)) {
628  AV_WN64A(bS, 0x0003000300030003ULL);
629  } else {
630  int i;
631  int mv_done;
632 
633  if( edge & mask_edge ) {
634  AV_ZERO64(bS);
635  mv_done = 1;
636  }
637  else if( mask_par0 ) {
638  int b_idx= 8 + 4 + edge * (dir ? 8:1);
639  int bn_idx= b_idx - (dir ? 8:1);
640 
641  bS[0] = bS[1] = bS[2] = bS[3] = check_mv(h, b_idx, bn_idx, mvy_limit);
642  mv_done = 1;
643  }
644  else
645  mv_done = 0;
646 
647  for( i = 0; i < 4; i++ ) {
648  int x = dir == 0 ? edge : i;
649  int y = dir == 0 ? i : edge;
650  int b_idx= 8 + 4 + x + 8*y;
651  int bn_idx= b_idx - (dir ? 8:1);
652 
653  if( h->non_zero_count_cache[b_idx] |
654  h->non_zero_count_cache[bn_idx] ) {
655  bS[i] = 2;
656  }
657  else if(!mv_done)
658  {
659  bS[i] = check_mv(h, b_idx, bn_idx, mvy_limit);
660  }
661  }
662 
663  if(bS[0]+bS[1]+bS[2]+bS[3] == 0)
664  continue;
665  }
666 
667  /* Filter edge */
668  // Do not use s->qscale as luma quantizer because it has not the same
669  // value in IPCM macroblocks.
671  //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]);
672  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);
673  //{ int i; for (i = 0; i < 4; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); }
674  if( dir == 0 ) {
675  filter_mb_edgev( &img_y[4*edge << h->pixel_shift], linesize, bS, qp, a, b, h, 0 );
676  if (chroma) {
677  if (chroma444) {
678  filter_mb_edgev ( &img_cb[4*edge << h->pixel_shift], uvlinesize, bS, h->chroma_qp[0], a, b, h, 0);
679  filter_mb_edgev ( &img_cr[4*edge << h->pixel_shift], uvlinesize, bS, h->chroma_qp[1], a, b, h, 0);
680  } else if( (edge&1) == 0 ) {
681  filter_mb_edgecv( &img_cb[2*edge << h->pixel_shift], uvlinesize, bS, h->chroma_qp[0], a, b, h, 0);
682  filter_mb_edgecv( &img_cr[2*edge << h->pixel_shift], uvlinesize, bS, h->chroma_qp[1], a, b, h, 0);
683  }
684  }
685  } else {
686  if (chroma422) {
687  if (deblock_edge)
688  filter_mb_edgeh(&img_y[4*edge*linesize], linesize, bS, qp, a, b, h, 0);
689  if (chroma) {
690  filter_mb_edgech(&img_cb[4*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[0], a, b, h, 0);
691  filter_mb_edgech(&img_cr[4*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[1], a, b, h, 0);
692  }
693  } else {
694  filter_mb_edgeh(&img_y[4*edge*linesize], linesize, bS, qp, a, b, h, 0);
695  if (chroma) {
696  if (chroma444) {
697  filter_mb_edgeh (&img_cb[4*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[0], a, b, h, 0);
698  filter_mb_edgeh (&img_cr[4*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[1], a, b, h, 0);
699  } else if ((edge&1) == 0) {
700  filter_mb_edgech(&img_cb[2*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[0], a, b, h, 0);
701  filter_mb_edgech(&img_cr[2*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[1], a, b, h, 0);
702  }
703  }
704  }
705  }
706  }
707 }
708 
709 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) {
710  MpegEncContext * const s = &h->s;
711  const int mb_xy= mb_x + mb_y*s->mb_stride;
712  const int mb_type = s->current_picture.f.mb_type[mb_xy];
713  const int mvy_limit = IS_INTERLACED(mb_type) ? 2 : 4;
714  int first_vertical_edge_done = 0;
715  av_unused int dir;
716  int chroma = !(CONFIG_GRAY && (s->flags&CODEC_FLAG_GRAY));
717  int qp_bd_offset = 6 * (h->sps.bit_depth_luma - 8);
718  int a = 52 + h->slice_alpha_c0_offset - qp_bd_offset;
719  int b = 52 + h->slice_beta_offset - qp_bd_offset;
720 
721  if (FRAME_MBAFF
722  // and current and left pair do not have the same interlaced type
723  && IS_INTERLACED(mb_type^h->left_type[LTOP])
724  // and left mb is in available to us
725  && h->left_type[LTOP]) {
726  /* First vertical edge is different in MBAFF frames
727  * There are 8 different bS to compute and 2 different Qp
728  */
729  DECLARE_ALIGNED(8, int16_t, bS)[8];
730  int qp[2];
731  int bqp[2];
732  int rqp[2];
733  int mb_qp, mbn0_qp, mbn1_qp;
734  int i;
735  first_vertical_edge_done = 1;
736 
737  if( IS_INTRA(mb_type) ) {
738  AV_WN64A(&bS[0], 0x0004000400040004ULL);
739  AV_WN64A(&bS[4], 0x0004000400040004ULL);
740  } else {
741  static const uint8_t offset[2][2][8]={
742  {
743  {3+4*0, 3+4*0, 3+4*0, 3+4*0, 3+4*1, 3+4*1, 3+4*1, 3+4*1},
744  {3+4*2, 3+4*2, 3+4*2, 3+4*2, 3+4*3, 3+4*3, 3+4*3, 3+4*3},
745  },{
746  {3+4*0, 3+4*1, 3+4*2, 3+4*3, 3+4*0, 3+4*1, 3+4*2, 3+4*3},
747  {3+4*0, 3+4*1, 3+4*2, 3+4*3, 3+4*0, 3+4*1, 3+4*2, 3+4*3},
748  }
749  };
750  const uint8_t *off= offset[MB_FIELD][mb_y&1];
751  for( i = 0; i < 8; i++ ) {
752  int j= MB_FIELD ? i>>2 : i&1;
753  int mbn_xy = h->left_mb_xy[LEFT(j)];
754  int mbn_type= h->left_type[LEFT(j)];
755 
756  if( IS_INTRA( mbn_type ) )
757  bS[i] = 4;
758  else{
759  bS[i] = 1 + !!(h->non_zero_count_cache[12+8*(i>>1)] |
760  ((!h->pps.cabac && IS_8x8DCT(mbn_type)) ?
761  (h->cbp_table[mbn_xy] & (((MB_FIELD ? (i&2) : (mb_y&1)) ? 8 : 2) << 12))
762  :
763  h->non_zero_count[mbn_xy][ off[i] ]));
764  }
765  }
766  }
767 
768  mb_qp = s->current_picture.f.qscale_table[mb_xy];
769  mbn0_qp = s->current_picture.f.qscale_table[h->left_mb_xy[0]];
770  mbn1_qp = s->current_picture.f.qscale_table[h->left_mb_xy[1]];
771  qp[0] = ( mb_qp + mbn0_qp + 1 ) >> 1;
772  bqp[0] = ( get_chroma_qp( h, 0, mb_qp ) +
773  get_chroma_qp( h, 0, mbn0_qp ) + 1 ) >> 1;
774  rqp[0] = ( get_chroma_qp( h, 1, mb_qp ) +
775  get_chroma_qp( h, 1, mbn0_qp ) + 1 ) >> 1;
776  qp[1] = ( mb_qp + mbn1_qp + 1 ) >> 1;
777  bqp[1] = ( get_chroma_qp( h, 0, mb_qp ) +
778  get_chroma_qp( h, 0, mbn1_qp ) + 1 ) >> 1;
779  rqp[1] = ( get_chroma_qp( h, 1, mb_qp ) +
780  get_chroma_qp( h, 1, mbn1_qp ) + 1 ) >> 1;
781 
782  /* Filter edge */
783  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);
784  { int i; for (i = 0; i < 8; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); }
785  if(MB_FIELD){
786  filter_mb_mbaff_edgev ( h, img_y , linesize, bS , 1, qp [0], a, b, 1 );
787  filter_mb_mbaff_edgev ( h, img_y + 8* linesize, linesize, bS+4, 1, qp [1], a, b, 1 );
788  if (chroma){
789  if (CHROMA444) {
790  filter_mb_mbaff_edgev ( h, img_cb, uvlinesize, bS , 1, bqp[0], a, b, 1 );
791  filter_mb_mbaff_edgev ( h, img_cb + 8*uvlinesize, uvlinesize, bS+4, 1, bqp[1], a, b, 1 );
792  filter_mb_mbaff_edgev ( h, img_cr, uvlinesize, bS , 1, rqp[0], a, b, 1 );
793  filter_mb_mbaff_edgev ( h, img_cr + 8*uvlinesize, uvlinesize, bS+4, 1, rqp[1], a, b, 1 );
794  } else if (CHROMA422) {
795  filter_mb_mbaff_edgecv(h, img_cb, uvlinesize, bS , 1, bqp[0], a, b, 1);
796  filter_mb_mbaff_edgecv(h, img_cb + 8*uvlinesize, uvlinesize, bS+4, 1, bqp[1], a, b, 1);
797  filter_mb_mbaff_edgecv(h, img_cr, uvlinesize, bS , 1, rqp[0], a, b, 1);
798  filter_mb_mbaff_edgecv(h, img_cr + 8*uvlinesize, uvlinesize, bS+4, 1, rqp[1], a, b, 1);
799  }else{
800  filter_mb_mbaff_edgecv( h, img_cb, uvlinesize, bS , 1, bqp[0], a, b, 1 );
801  filter_mb_mbaff_edgecv( h, img_cb + 4*uvlinesize, uvlinesize, bS+4, 1, bqp[1], a, b, 1 );
802  filter_mb_mbaff_edgecv( h, img_cr, uvlinesize, bS , 1, rqp[0], a, b, 1 );
803  filter_mb_mbaff_edgecv( h, img_cr + 4*uvlinesize, uvlinesize, bS+4, 1, rqp[1], a, b, 1 );
804  }
805  }
806  }else{
807  filter_mb_mbaff_edgev ( h, img_y , 2* linesize, bS , 2, qp [0], a, b, 1 );
808  filter_mb_mbaff_edgev ( h, img_y + linesize, 2* linesize, bS+1, 2, qp [1], a, b, 1 );
809  if (chroma){
810  if (CHROMA444) {
811  filter_mb_mbaff_edgev ( h, img_cb, 2*uvlinesize, bS , 2, bqp[0], a, b, 1 );
812  filter_mb_mbaff_edgev ( h, img_cb + uvlinesize, 2*uvlinesize, bS+1, 2, bqp[1], a, b, 1 );
813  filter_mb_mbaff_edgev ( h, img_cr, 2*uvlinesize, bS , 2, rqp[0], a, b, 1 );
814  filter_mb_mbaff_edgev ( h, img_cr + uvlinesize, 2*uvlinesize, bS+1, 2, rqp[1], a, b, 1 );
815  }else{
816  filter_mb_mbaff_edgecv( h, img_cb, 2*uvlinesize, bS , 2, bqp[0], a, b, 1 );
817  filter_mb_mbaff_edgecv( h, img_cb + uvlinesize, 2*uvlinesize, bS+1, 2, bqp[1], a, b, 1 );
818  filter_mb_mbaff_edgecv( h, img_cr, 2*uvlinesize, bS , 2, rqp[0], a, b, 1 );
819  filter_mb_mbaff_edgecv( h, img_cr + uvlinesize, 2*uvlinesize, bS+1, 2, rqp[1], a, b, 1 );
820  }
821  }
822  }
823  }
824 
825 #if CONFIG_SMALL
826  for( dir = 0; dir < 2; dir++ )
827  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, a, b, chroma, dir);
828 #else
829  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, a, b, chroma, 0);
830  filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, 0, a, b, chroma, 1);
831 #endif
832 }