FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
motion_est.c
Go to the documentation of this file.
1 /*
2  * Motion estimation
3  * Copyright (c) 2000,2001 Fabrice Bellard
4  * Copyright (c) 2002-2004 Michael Niedermayer
5  *
6  * new motion estimation (X1/EPZS) by Michael Niedermayer <michaelni@gmx.at>
7  *
8  * This file is part of FFmpeg.
9  *
10  * FFmpeg is free software; you can redistribute it and/or
11  * modify it under the terms of the GNU Lesser General Public
12  * License as published by the Free Software Foundation; either
13  * version 2.1 of the License, or (at your option) any later version.
14  *
15  * FFmpeg is distributed in the hope that it will be useful,
16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18  * Lesser General Public License for more details.
19  *
20  * You should have received a copy of the GNU Lesser General Public
21  * License along with FFmpeg; if not, write to the Free Software
22  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23  */
24 
25 /**
26  * @file
27  * Motion estimation.
28  */
29 
30 #include <stdlib.h>
31 #include <stdio.h>
32 #include <limits.h>
33 
34 #include "avcodec.h"
35 #include "mathops.h"
36 #include "mpegvideo.h"
37 
38 #undef NDEBUG
39 #include <assert.h>
40 
41 #define P_LEFT P[1]
42 #define P_TOP P[2]
43 #define P_TOPRIGHT P[3]
44 #define P_MEDIAN P[4]
45 #define P_MV1 P[9]
46 
47 static int sad_hpel_motion_search(MpegEncContext * s,
48  int *mx_ptr, int *my_ptr, int dmin,
49  int src_index, int ref_index,
50  int size, int h);
51 
52 static inline unsigned update_map_generation(MotionEstContext *c)
53 {
54  c->map_generation+= 1<<(ME_MAP_MV_BITS*2);
55  if(c->map_generation==0){
56  c->map_generation= 1<<(ME_MAP_MV_BITS*2);
57  memset(c->map, 0, sizeof(uint32_t)*ME_MAP_SIZE);
58  }
59  return c->map_generation;
60 }
61 
62 /* shape adaptive search stuff */
63 typedef struct Minima{
64  int height;
65  int x, y;
66  int checked;
67 }Minima;
68 
69 static int minima_cmp(const void *a, const void *b){
70  const Minima *da = (const Minima *) a;
71  const Minima *db = (const Minima *) b;
72 
73  return da->height - db->height;
74 }
75 
76 #define FLAG_QPEL 1 //must be 1
77 #define FLAG_CHROMA 2
78 #define FLAG_DIRECT 4
79 
80 static inline void init_ref(MotionEstContext *c, uint8_t *src[3], uint8_t *ref[3], uint8_t *ref2[3], int x, int y, int ref_index){
81  const int offset[3]= {
82  y*c-> stride + x,
83  ((y*c->uvstride + x)>>1),
84  ((y*c->uvstride + x)>>1),
85  };
86  int i;
87  for(i=0; i<3; i++){
88  c->src[0][i]= src [i] + offset[i];
89  c->ref[0][i]= ref [i] + offset[i];
90  }
91  if(ref_index){
92  for(i=0; i<3; i++){
93  c->ref[ref_index][i]= ref2[i] + offset[i];
94  }
95  }
96 }
97 
98 static int get_flags(MotionEstContext *c, int direct, int chroma){
99  return ((c->avctx->flags&CODEC_FLAG_QPEL) ? FLAG_QPEL : 0)
100  + (direct ? FLAG_DIRECT : 0)
101  + (chroma ? FLAG_CHROMA : 0);
102 }
103 
104 static av_always_inline int cmp_direct_inline(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
105  const int size, const int h, int ref_index, int src_index,
106  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, int qpel){
107  MotionEstContext * const c= &s->me;
108  const int stride= c->stride;
109  const int hx= subx + (x<<(1+qpel));
110  const int hy= suby + (y<<(1+qpel));
111  uint8_t * const * const ref= c->ref[ref_index];
112  uint8_t * const * const src= c->src[src_index];
113  int d;
114  //FIXME check chroma 4mv, (no crashes ...)
115  av_assert2(x >= c->xmin && hx <= c->xmax<<(qpel+1) && y >= c->ymin && hy <= c->ymax<<(qpel+1));
116  if(x >= c->xmin && hx <= c->xmax<<(qpel+1) && y >= c->ymin && hy <= c->ymax<<(qpel+1)){
117  const int time_pp= s->pp_time;
118  const int time_pb= s->pb_time;
119  const int mask= 2*qpel+1;
120  if(s->mv_type==MV_TYPE_8X8){
121  int i;
122  for(i=0; i<4; i++){
123  int fx = c->direct_basis_mv[i][0] + hx;
124  int fy = c->direct_basis_mv[i][1] + hy;
125  int bx = hx ? fx - c->co_located_mv[i][0] : c->co_located_mv[i][0]*(time_pb - time_pp)/time_pp + ((i &1)<<(qpel+4));
126  int by = hy ? fy - c->co_located_mv[i][1] : c->co_located_mv[i][1]*(time_pb - time_pp)/time_pp + ((i>>1)<<(qpel+4));
127  int fxy= (fx&mask) + ((fy&mask)<<(qpel+1));
128  int bxy= (bx&mask) + ((by&mask)<<(qpel+1));
129 
130  uint8_t *dst= c->temp + 8*(i&1) + 8*stride*(i>>1);
131  if(qpel){
132  c->qpel_put[1][fxy](dst, ref[0] + (fx>>2) + (fy>>2)*stride, stride);
133  c->qpel_avg[1][bxy](dst, ref[8] + (bx>>2) + (by>>2)*stride, stride);
134  }else{
135  c->hpel_put[1][fxy](dst, ref[0] + (fx>>1) + (fy>>1)*stride, stride, 8);
136  c->hpel_avg[1][bxy](dst, ref[8] + (bx>>1) + (by>>1)*stride, stride, 8);
137  }
138  }
139  }else{
140  int fx = c->direct_basis_mv[0][0] + hx;
141  int fy = c->direct_basis_mv[0][1] + hy;
142  int bx = hx ? fx - c->co_located_mv[0][0] : (c->co_located_mv[0][0]*(time_pb - time_pp)/time_pp);
143  int by = hy ? fy - c->co_located_mv[0][1] : (c->co_located_mv[0][1]*(time_pb - time_pp)/time_pp);
144  int fxy= (fx&mask) + ((fy&mask)<<(qpel+1));
145  int bxy= (bx&mask) + ((by&mask)<<(qpel+1));
146 
147  if(qpel){
148  c->qpel_put[1][fxy](c->temp , ref[0] + (fx>>2) + (fy>>2)*stride , stride);
149  c->qpel_put[1][fxy](c->temp + 8 , ref[0] + (fx>>2) + (fy>>2)*stride + 8 , stride);
150  c->qpel_put[1][fxy](c->temp + 8*stride, ref[0] + (fx>>2) + (fy>>2)*stride + 8*stride, stride);
151  c->qpel_put[1][fxy](c->temp + 8 + 8*stride, ref[0] + (fx>>2) + (fy>>2)*stride + 8 + 8*stride, stride);
152  c->qpel_avg[1][bxy](c->temp , ref[8] + (bx>>2) + (by>>2)*stride , stride);
153  c->qpel_avg[1][bxy](c->temp + 8 , ref[8] + (bx>>2) + (by>>2)*stride + 8 , stride);
154  c->qpel_avg[1][bxy](c->temp + 8*stride, ref[8] + (bx>>2) + (by>>2)*stride + 8*stride, stride);
155  c->qpel_avg[1][bxy](c->temp + 8 + 8*stride, ref[8] + (bx>>2) + (by>>2)*stride + 8 + 8*stride, stride);
156  }else{
157  av_assert2((fx>>1) + 16*s->mb_x >= -16);
158  av_assert2((fy>>1) + 16*s->mb_y >= -16);
159  av_assert2((fx>>1) + 16*s->mb_x <= s->width);
160  av_assert2((fy>>1) + 16*s->mb_y <= s->height);
161  av_assert2((bx>>1) + 16*s->mb_x >= -16);
162  av_assert2((by>>1) + 16*s->mb_y >= -16);
163  av_assert2((bx>>1) + 16*s->mb_x <= s->width);
164  av_assert2((by>>1) + 16*s->mb_y <= s->height);
165 
166  c->hpel_put[0][fxy](c->temp, ref[0] + (fx>>1) + (fy>>1)*stride, stride, 16);
167  c->hpel_avg[0][bxy](c->temp, ref[8] + (bx>>1) + (by>>1)*stride, stride, 16);
168  }
169  }
170  d = cmp_func(s, c->temp, src[0], stride, 16);
171  }else
172  d= 256*256*256*32;
173  return d;
174 }
175 
176 static av_always_inline int cmp_inline(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
177  const int size, const int h, int ref_index, int src_index,
178  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, int qpel, int chroma){
179  MotionEstContext * const c= &s->me;
180  const int stride= c->stride;
181  const int uvstride= c->uvstride;
182  const int dxy= subx + (suby<<(1+qpel)); //FIXME log2_subpel?
183  const int hx= subx + (x<<(1+qpel));
184  const int hy= suby + (y<<(1+qpel));
185  uint8_t * const * const ref= c->ref[ref_index];
186  uint8_t * const * const src= c->src[src_index];
187  int d;
188  //FIXME check chroma 4mv, (no crashes ...)
189  int uvdxy; /* no, it might not be used uninitialized */
190  if(dxy){
191  if(qpel){
192  c->qpel_put[size][dxy](c->temp, ref[0] + x + y*stride, stride); //FIXME prototype (add h)
193  if(chroma){
194  int cx= hx/2;
195  int cy= hy/2;
196  cx= (cx>>1)|(cx&1);
197  cy= (cy>>1)|(cy&1);
198  uvdxy= (cx&1) + 2*(cy&1);
199  //FIXME x/y wrong, but mpeg4 qpel is sick anyway, we should drop as much of it as possible in favor for h264
200  }
201  }else{
202  c->hpel_put[size][dxy](c->temp, ref[0] + x + y*stride, stride, h);
203  if(chroma)
204  uvdxy= dxy | (x&1) | (2*(y&1));
205  }
206  d = cmp_func(s, c->temp, src[0], stride, h);
207  }else{
208  d = cmp_func(s, src[0], ref[0] + x + y*stride, stride, h);
209  if(chroma)
210  uvdxy= (x&1) + 2*(y&1);
211  }
212  if(chroma){
213  uint8_t * const uvtemp= c->temp + 16*stride;
214  c->hpel_put[size+1][uvdxy](uvtemp , ref[1] + (x>>1) + (y>>1)*uvstride, uvstride, h>>1);
215  c->hpel_put[size+1][uvdxy](uvtemp+8, ref[2] + (x>>1) + (y>>1)*uvstride, uvstride, h>>1);
216  d += chroma_cmp_func(s, uvtemp , src[1], uvstride, h>>1);
217  d += chroma_cmp_func(s, uvtemp+8, src[2], uvstride, h>>1);
218  }
219  return d;
220 }
221 
222 static int cmp_simple(MpegEncContext *s, const int x, const int y,
223  int ref_index, int src_index,
224  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func){
225  return cmp_inline(s,x,y,0,0,0,16,ref_index,src_index, cmp_func, chroma_cmp_func, 0, 0);
226 }
227 
228 static int cmp_fpel_internal(MpegEncContext *s, const int x, const int y,
229  const int size, const int h, int ref_index, int src_index,
230  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
231  if(flags&FLAG_DIRECT){
232  return cmp_direct_inline(s,x,y,0,0,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, flags&FLAG_QPEL);
233  }else{
234  return cmp_inline(s,x,y,0,0,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 0, flags&FLAG_CHROMA);
235  }
236 }
237 
238 static int cmp_internal(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
239  const int size, const int h, int ref_index, int src_index,
240  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
241  if(flags&FLAG_DIRECT){
242  return cmp_direct_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, flags&FLAG_QPEL);
243  }else{
244  return cmp_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, flags&FLAG_QPEL, flags&FLAG_CHROMA);
245  }
246 }
247 
248 /** @brief compares a block (either a full macroblock or a partition thereof)
249  against a proposed motion-compensated prediction of that block
250  */
251 static av_always_inline int cmp(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
252  const int size, const int h, int ref_index, int src_index,
253  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
254  if(av_builtin_constant_p(flags) && av_builtin_constant_p(h) && av_builtin_constant_p(size)
255  && av_builtin_constant_p(subx) && av_builtin_constant_p(suby)
256  && flags==0 && h==16 && size==0 && subx==0 && suby==0){
257  return cmp_simple(s,x,y,ref_index,src_index, cmp_func, chroma_cmp_func);
258  }else if(av_builtin_constant_p(subx) && av_builtin_constant_p(suby)
259  && subx==0 && suby==0){
260  return cmp_fpel_internal(s,x,y,size,h,ref_index,src_index, cmp_func, chroma_cmp_func,flags);
261  }else{
262  return cmp_internal(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, flags);
263  }
264 }
265 
266 static int cmp_hpel(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
267  const int size, const int h, int ref_index, int src_index,
268  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
269  if(flags&FLAG_DIRECT){
270  return cmp_direct_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 0);
271  }else{
272  return cmp_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 0, flags&FLAG_CHROMA);
273  }
274 }
275 
276 static int cmp_qpel(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
277  const int size, const int h, int ref_index, int src_index,
278  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
279  if(flags&FLAG_DIRECT){
280  return cmp_direct_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 1);
281  }else{
282  return cmp_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 1, flags&FLAG_CHROMA);
283  }
284 }
285 
286 #include "motion_est_template.c"
287 
288 static int zero_cmp(void *s, uint8_t *a, uint8_t *b, int stride, int h){
289  return 0;
290 }
291 
292 static void zero_hpel(uint8_t *a, const uint8_t *b, ptrdiff_t stride, int h){
293 }
294 
295 int ff_init_me(MpegEncContext *s){
296  MotionEstContext * const c= &s->me;
297  int cache_size= FFMIN(ME_MAP_SIZE>>ME_MAP_SHIFT, 1<<ME_MAP_SHIFT);
298  int dia_size= FFMAX(FFABS(s->avctx->dia_size)&255, FFABS(s->avctx->pre_dia_size)&255);
299 
300  if(FFMIN(s->avctx->dia_size, s->avctx->pre_dia_size) < -FFMIN(ME_MAP_SIZE, MAX_SAB_SIZE)){
301  av_log(s->avctx, AV_LOG_ERROR, "ME_MAP size is too small for SAB diamond\n");
302  return -1;
303  }
304  //special case of snow is needed because snow uses its own iterative ME code
305  if(s->me_method!=ME_ZERO && s->me_method!=ME_EPZS && s->me_method!=ME_X1 && s->avctx->codec_id != AV_CODEC_ID_SNOW){
306  av_log(s->avctx, AV_LOG_ERROR, "me_method is only allowed to be set to zero and epzs; for hex,umh,full and others see dia_size\n");
307  return -1;
308  }
309 
310  c->avctx= s->avctx;
311 
312  if(cache_size < 2*dia_size && !c->stride){
313  av_log(s->avctx, AV_LOG_INFO, "ME_MAP size may be a little small for the selected diamond size\n");
314  }
315 
316  ff_set_cmp(&s->dsp, s->dsp.me_pre_cmp, c->avctx->me_pre_cmp);
317  ff_set_cmp(&s->dsp, s->dsp.me_cmp, c->avctx->me_cmp);
318  ff_set_cmp(&s->dsp, s->dsp.me_sub_cmp, c->avctx->me_sub_cmp);
319  ff_set_cmp(&s->dsp, s->dsp.mb_cmp, c->avctx->mb_cmp);
320 
321  c->flags = get_flags(c, 0, c->avctx->me_cmp &FF_CMP_CHROMA);
322  c->sub_flags= get_flags(c, 0, c->avctx->me_sub_cmp&FF_CMP_CHROMA);
323  c->mb_flags = get_flags(c, 0, c->avctx->mb_cmp &FF_CMP_CHROMA);
324 
325 /*FIXME s->no_rounding b_type*/
326  if(s->flags&CODEC_FLAG_QPEL){
327  c->sub_motion_search= qpel_motion_search;
328  c->qpel_avg= s->dsp.avg_qpel_pixels_tab;
329  if(s->no_rounding) c->qpel_put= s->dsp.put_no_rnd_qpel_pixels_tab;
330  else c->qpel_put= s->dsp.put_qpel_pixels_tab;
331  }else{
332  if(c->avctx->me_sub_cmp&FF_CMP_CHROMA)
333  c->sub_motion_search= hpel_motion_search;
334  else if( c->avctx->me_sub_cmp == FF_CMP_SAD
335  && c->avctx-> me_cmp == FF_CMP_SAD
336  && c->avctx-> mb_cmp == FF_CMP_SAD)
337  c->sub_motion_search= sad_hpel_motion_search; // 2050 vs. 2450 cycles
338  else
339  c->sub_motion_search= hpel_motion_search;
340  }
341  c->hpel_avg = s->hdsp.avg_pixels_tab;
342  if (s->no_rounding)
343  c->hpel_put = s->hdsp.put_no_rnd_pixels_tab;
344  else
345  c->hpel_put = s->hdsp.put_pixels_tab;
346 
347  if(s->linesize){
348  c->stride = s->linesize;
349  c->uvstride= s->uvlinesize;
350  }else{
351  c->stride = 16*s->mb_width + 32;
352  c->uvstride= 8*s->mb_width + 16;
353  }
354 
355  /* 8x8 fullpel search would need a 4x4 chroma compare, which we do
356  * not have yet, and even if we had, the motion estimation code
357  * does not expect it. */
358  if(s->codec_id != AV_CODEC_ID_SNOW){
359  if((c->avctx->me_cmp&FF_CMP_CHROMA)/* && !s->dsp.me_cmp[2]*/){
360  s->dsp.me_cmp[2]= zero_cmp;
361  }
362  if((c->avctx->me_sub_cmp&FF_CMP_CHROMA) && !s->dsp.me_sub_cmp[2]){
363  s->dsp.me_sub_cmp[2]= zero_cmp;
364  }
365  c->hpel_put[2][0]= c->hpel_put[2][1]=
366  c->hpel_put[2][2]= c->hpel_put[2][3]= zero_hpel;
367  }
368 
369  if(s->codec_id == AV_CODEC_ID_H261){
370  c->sub_motion_search= no_sub_motion_search;
371  }
372 
373  return 0;
374 }
375 
376 #define CHECK_SAD_HALF_MV(suffix, x, y) \
377 {\
378  d= s->dsp.pix_abs[size][(x?1:0)+(y?2:0)](NULL, pix, ptr+((x)>>1), stride, h);\
379  d += (mv_penalty[pen_x + x] + mv_penalty[pen_y + y])*penalty_factor;\
380  COPY3_IF_LT(dminh, d, dx, x, dy, y)\
381 }
382 
383 static int sad_hpel_motion_search(MpegEncContext * s,
384  int *mx_ptr, int *my_ptr, int dmin,
385  int src_index, int ref_index,
386  int size, int h)
387 {
388  MotionEstContext * const c= &s->me;
389  const int penalty_factor= c->sub_penalty_factor;
390  int mx, my, dminh;
391  uint8_t *pix, *ptr;
392  int stride= c->stride;
393  LOAD_COMMON
394 
395  av_assert2(c->sub_flags == 0);
396 
397  if(c->skip){
398  *mx_ptr = 0;
399  *my_ptr = 0;
400  return dmin;
401  }
402 
403  pix = c->src[src_index][0];
404 
405  mx = *mx_ptr;
406  my = *my_ptr;
407  ptr = c->ref[ref_index][0] + (my * stride) + mx;
408 
409  dminh = dmin;
410 
411  if (mx > xmin && mx < xmax &&
412  my > ymin && my < ymax) {
413  int dx=0, dy=0;
414  int d, pen_x, pen_y;
415  const int index= (my<<ME_MAP_SHIFT) + mx;
416  const int t= score_map[(index-(1<<ME_MAP_SHIFT))&(ME_MAP_SIZE-1)];
417  const int l= score_map[(index- 1 )&(ME_MAP_SIZE-1)];
418  const int r= score_map[(index+ 1 )&(ME_MAP_SIZE-1)];
419  const int b= score_map[(index+(1<<ME_MAP_SHIFT))&(ME_MAP_SIZE-1)];
420  mx<<=1;
421  my<<=1;
422 
423 
424  pen_x= pred_x + mx;
425  pen_y= pred_y + my;
426 
427  ptr-= stride;
428  if(t<=b){
429  CHECK_SAD_HALF_MV(y2 , 0, -1)
430  if(l<=r){
431  CHECK_SAD_HALF_MV(xy2, -1, -1)
432  if(t+r<=b+l){
433  CHECK_SAD_HALF_MV(xy2, +1, -1)
434  ptr+= stride;
435  }else{
436  ptr+= stride;
437  CHECK_SAD_HALF_MV(xy2, -1, +1)
438  }
439  CHECK_SAD_HALF_MV(x2 , -1, 0)
440  }else{
441  CHECK_SAD_HALF_MV(xy2, +1, -1)
442  if(t+l<=b+r){
443  CHECK_SAD_HALF_MV(xy2, -1, -1)
444  ptr+= stride;
445  }else{
446  ptr+= stride;
447  CHECK_SAD_HALF_MV(xy2, +1, +1)
448  }
449  CHECK_SAD_HALF_MV(x2 , +1, 0)
450  }
451  }else{
452  if(l<=r){
453  if(t+l<=b+r){
454  CHECK_SAD_HALF_MV(xy2, -1, -1)
455  ptr+= stride;
456  }else{
457  ptr+= stride;
458  CHECK_SAD_HALF_MV(xy2, +1, +1)
459  }
460  CHECK_SAD_HALF_MV(x2 , -1, 0)
461  CHECK_SAD_HALF_MV(xy2, -1, +1)
462  }else{
463  if(t+r<=b+l){
464  CHECK_SAD_HALF_MV(xy2, +1, -1)
465  ptr+= stride;
466  }else{
467  ptr+= stride;
468  CHECK_SAD_HALF_MV(xy2, -1, +1)
469  }
470  CHECK_SAD_HALF_MV(x2 , +1, 0)
471  CHECK_SAD_HALF_MV(xy2, +1, +1)
472  }
473  CHECK_SAD_HALF_MV(y2 , 0, +1)
474  }
475  mx+=dx;
476  my+=dy;
477 
478  }else{
479  mx<<=1;
480  my<<=1;
481  }
482 
483  *mx_ptr = mx;
484  *my_ptr = my;
485  return dminh;
486 }
487 
488 static inline void set_p_mv_tables(MpegEncContext * s, int mx, int my, int mv4)
489 {
490  const int xy= s->mb_x + s->mb_y*s->mb_stride;
491 
492  s->p_mv_table[xy][0] = mx;
493  s->p_mv_table[xy][1] = my;
494 
495  /* has already been set to the 4 MV if 4MV is done */
496  if(mv4){
497  int mot_xy= s->block_index[0];
498 
499  s->current_picture.motion_val[0][mot_xy ][0] = mx;
500  s->current_picture.motion_val[0][mot_xy ][1] = my;
501  s->current_picture.motion_val[0][mot_xy + 1][0] = mx;
502  s->current_picture.motion_val[0][mot_xy + 1][1] = my;
503 
504  mot_xy += s->b8_stride;
505  s->current_picture.motion_val[0][mot_xy ][0] = mx;
506  s->current_picture.motion_val[0][mot_xy ][1] = my;
507  s->current_picture.motion_val[0][mot_xy + 1][0] = mx;
508  s->current_picture.motion_val[0][mot_xy + 1][1] = my;
509  }
510 }
511 
512 /**
513  * get fullpel ME search limits.
514  */
515 static inline void get_limits(MpegEncContext *s, int x, int y)
516 {
517  MotionEstContext * const c= &s->me;
518  int range= c->avctx->me_range >> (1 + !!(c->flags&FLAG_QPEL));
519  int max_range = MAX_MV >> (1 + !!(c->flags&FLAG_QPEL));
520 /*
521  if(c->avctx->me_range) c->range= c->avctx->me_range >> 1;
522  else c->range= 16;
523 */
524  if (s->unrestricted_mv) {
525  c->xmin = - x - 16;
526  c->ymin = - y - 16;
527  c->xmax = - x + s->width;
528  c->ymax = - y + s->height;
529  } else if (s->out_format == FMT_H261){
530  // Search range of H261 is different from other codec standards
531  c->xmin = (x > 15) ? - 15 : 0;
532  c->ymin = (y > 15) ? - 15 : 0;
533  c->xmax = (x < s->mb_width * 16 - 16) ? 15 : 0;
534  c->ymax = (y < s->mb_height * 16 - 16) ? 15 : 0;
535  } else {
536  c->xmin = - x;
537  c->ymin = - y;
538  c->xmax = - x + s->mb_width *16 - 16;
539  c->ymax = - y + s->mb_height*16 - 16;
540  }
541  if(!range || range > max_range)
542  range = max_range;
543  if(range){
544  c->xmin = FFMAX(c->xmin,-range);
545  c->xmax = FFMIN(c->xmax, range);
546  c->ymin = FFMAX(c->ymin,-range);
547  c->ymax = FFMIN(c->ymax, range);
548  }
549 }
550 
551 static inline void init_mv4_ref(MotionEstContext *c){
552  const int stride= c->stride;
553 
554  c->ref[1][0] = c->ref[0][0] + 8;
555  c->ref[2][0] = c->ref[0][0] + 8*stride;
556  c->ref[3][0] = c->ref[2][0] + 8;
557  c->src[1][0] = c->src[0][0] + 8;
558  c->src[2][0] = c->src[0][0] + 8*stride;
559  c->src[3][0] = c->src[2][0] + 8;
560 }
561 
562 static inline int h263_mv4_search(MpegEncContext *s, int mx, int my, int shift)
563 {
564  MotionEstContext * const c= &s->me;
565  const int size= 1;
566  const int h=8;
567  int block;
568  int P[10][2];
569  int dmin_sum=0, mx4_sum=0, my4_sum=0, i;
570  int same=1;
571  const int stride= c->stride;
572  uint8_t *mv_penalty= c->current_mv_penalty;
573  int saftey_cliping= s->unrestricted_mv && (s->width&15) && (s->height&15);
574 
575  init_mv4_ref(c);
576 
577  for(block=0; block<4; block++){
578  int mx4, my4;
579  int pred_x4, pred_y4;
580  int dmin4;
581  static const int off[4]= {2, 1, 1, -1};
582  const int mot_stride = s->b8_stride;
583  const int mot_xy = s->block_index[block];
584 
585  if(saftey_cliping){
586  c->xmax = - 16*s->mb_x + s->width - 8*(block &1);
587  c->ymax = - 16*s->mb_y + s->height - 8*(block>>1);
588  }
589 
590  P_LEFT[0] = s->current_picture.motion_val[0][mot_xy - 1][0];
591  P_LEFT[1] = s->current_picture.motion_val[0][mot_xy - 1][1];
592 
593  if(P_LEFT[0] > (c->xmax<<shift)) P_LEFT[0] = (c->xmax<<shift);
594 
595  /* special case for first line */
596  if (s->first_slice_line && block<2) {
597  c->pred_x= pred_x4= P_LEFT[0];
598  c->pred_y= pred_y4= P_LEFT[1];
599  } else {
600  P_TOP[0] = s->current_picture.motion_val[0][mot_xy - mot_stride ][0];
601  P_TOP[1] = s->current_picture.motion_val[0][mot_xy - mot_stride ][1];
602  P_TOPRIGHT[0] = s->current_picture.motion_val[0][mot_xy - mot_stride + off[block]][0];
603  P_TOPRIGHT[1] = s->current_picture.motion_val[0][mot_xy - mot_stride + off[block]][1];
604  if(P_TOP[1] > (c->ymax<<shift)) P_TOP[1] = (c->ymax<<shift);
605  if(P_TOPRIGHT[0] < (c->xmin<<shift)) P_TOPRIGHT[0]= (c->xmin<<shift);
606  if(P_TOPRIGHT[0] > (c->xmax<<shift)) P_TOPRIGHT[0]= (c->xmax<<shift);
607  if(P_TOPRIGHT[1] > (c->ymax<<shift)) P_TOPRIGHT[1]= (c->ymax<<shift);
608 
609  P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
610  P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
611 
612  c->pred_x= pred_x4 = P_MEDIAN[0];
613  c->pred_y= pred_y4 = P_MEDIAN[1];
614  }
615  P_MV1[0]= mx;
616  P_MV1[1]= my;
617  if(saftey_cliping)
618  for(i=0; i<10; i++){
619  if(P[i][0] > (c->xmax<<shift)) P[i][0]= (c->xmax<<shift);
620  if(P[i][1] > (c->ymax<<shift)) P[i][1]= (c->ymax<<shift);
621  }
622 
623  dmin4 = epzs_motion_search4(s, &mx4, &my4, P, block, block, s->p_mv_table, (1<<16)>>shift);
624 
625  dmin4= c->sub_motion_search(s, &mx4, &my4, dmin4, block, block, size, h);
626 
627  if(s->dsp.me_sub_cmp[0] != s->dsp.mb_cmp[0]){
628  int dxy;
629  const int offset= ((block&1) + (block>>1)*stride)*8;
630  uint8_t *dest_y = c->scratchpad + offset;
631  if(s->quarter_sample){
632  uint8_t *ref= c->ref[block][0] + (mx4>>2) + (my4>>2)*stride;
633  dxy = ((my4 & 3) << 2) | (mx4 & 3);
634 
635  if(s->no_rounding)
636  s->dsp.put_no_rnd_qpel_pixels_tab[1][dxy](dest_y , ref , stride);
637  else
638  s->dsp.put_qpel_pixels_tab [1][dxy](dest_y , ref , stride);
639  }else{
640  uint8_t *ref= c->ref[block][0] + (mx4>>1) + (my4>>1)*stride;
641  dxy = ((my4 & 1) << 1) | (mx4 & 1);
642 
643  if(s->no_rounding)
644  s->hdsp.put_no_rnd_pixels_tab[1][dxy](dest_y , ref , stride, h);
645  else
646  s->hdsp.put_pixels_tab [1][dxy](dest_y , ref , stride, h);
647  }
648  dmin_sum+= (mv_penalty[mx4-pred_x4] + mv_penalty[my4-pred_y4])*c->mb_penalty_factor;
649  }else
650  dmin_sum+= dmin4;
651 
652  if(s->quarter_sample){
653  mx4_sum+= mx4/2;
654  my4_sum+= my4/2;
655  }else{
656  mx4_sum+= mx4;
657  my4_sum+= my4;
658  }
659 
660  s->current_picture.motion_val[0][s->block_index[block]][0] = mx4;
661  s->current_picture.motion_val[0][s->block_index[block]][1] = my4;
662 
663  if(mx4 != mx || my4 != my) same=0;
664  }
665 
666  if(same)
667  return INT_MAX;
668 
669  if(s->dsp.me_sub_cmp[0] != s->dsp.mb_cmp[0]){
670  dmin_sum += s->dsp.mb_cmp[0](s, s->new_picture.f.data[0] + s->mb_x*16 + s->mb_y*16*stride, c->scratchpad, stride, 16);
671  }
672 
673  if(c->avctx->mb_cmp&FF_CMP_CHROMA){
674  int dxy;
675  int mx, my;
676  int offset;
677 
678  mx= ff_h263_round_chroma(mx4_sum);
679  my= ff_h263_round_chroma(my4_sum);
680  dxy = ((my & 1) << 1) | (mx & 1);
681 
682  offset= (s->mb_x*8 + (mx>>1)) + (s->mb_y*8 + (my>>1))*s->uvlinesize;
683 
684  if(s->no_rounding){
685  s->hdsp.put_no_rnd_pixels_tab[1][dxy](c->scratchpad , s->last_picture.f.data[1] + offset, s->uvlinesize, 8);
686  s->hdsp.put_no_rnd_pixels_tab[1][dxy](c->scratchpad + 8, s->last_picture.f.data[2] + offset, s->uvlinesize, 8);
687  }else{
688  s->hdsp.put_pixels_tab [1][dxy](c->scratchpad , s->last_picture.f.data[1] + offset, s->uvlinesize, 8);
689  s->hdsp.put_pixels_tab [1][dxy](c->scratchpad + 8, s->last_picture.f.data[2] + offset, s->uvlinesize, 8);
690  }
691 
692  dmin_sum += s->dsp.mb_cmp[1](s, s->new_picture.f.data[1] + s->mb_x*8 + s->mb_y*8*s->uvlinesize, c->scratchpad , s->uvlinesize, 8);
693  dmin_sum += s->dsp.mb_cmp[1](s, s->new_picture.f.data[2] + s->mb_x*8 + s->mb_y*8*s->uvlinesize, c->scratchpad+8, s->uvlinesize, 8);
694  }
695 
696  c->pred_x= mx;
697  c->pred_y= my;
698 
699  switch(c->avctx->mb_cmp&0xFF){
700  /*case FF_CMP_SSE:
701  return dmin_sum+ 32*s->qscale*s->qscale;*/
702  case FF_CMP_RD:
703  return dmin_sum;
704  default:
705  return dmin_sum+ 11*c->mb_penalty_factor;
706  }
707 }
708 
709 static inline void init_interlaced_ref(MpegEncContext *s, int ref_index){
710  MotionEstContext * const c= &s->me;
711 
712  c->ref[1+ref_index][0] = c->ref[0+ref_index][0] + s->linesize;
713  c->src[1][0] = c->src[0][0] + s->linesize;
714  if(c->flags & FLAG_CHROMA){
715  c->ref[1+ref_index][1] = c->ref[0+ref_index][1] + s->uvlinesize;
716  c->ref[1+ref_index][2] = c->ref[0+ref_index][2] + s->uvlinesize;
717  c->src[1][1] = c->src[0][1] + s->uvlinesize;
718  c->src[1][2] = c->src[0][2] + s->uvlinesize;
719  }
720 }
721 
722 static int interlaced_search(MpegEncContext *s, int ref_index,
723  int16_t (*mv_tables[2][2])[2], uint8_t *field_select_tables[2], int mx, int my, int user_field_select)
724 {
725  MotionEstContext * const c= &s->me;
726  const int size=0;
727  const int h=8;
728  int block;
729  int P[10][2];
730  uint8_t * const mv_penalty= c->current_mv_penalty;
731  int same=1;
732  const int stride= 2*s->linesize;
733  int dmin_sum= 0;
734  const int mot_stride= s->mb_stride;
735  const int xy= s->mb_x + s->mb_y*mot_stride;
736 
737  c->ymin>>=1;
738  c->ymax>>=1;
739  c->stride<<=1;
740  c->uvstride<<=1;
741  init_interlaced_ref(s, ref_index);
742 
743  for(block=0; block<2; block++){
744  int field_select;
745  int best_dmin= INT_MAX;
746  int best_field= -1;
747 
748  for(field_select=0; field_select<2; field_select++){
749  int dmin, mx_i, my_i;
750  int16_t (*mv_table)[2]= mv_tables[block][field_select];
751 
752  if(user_field_select){
753  av_assert1(field_select==0 || field_select==1);
754  av_assert1(field_select_tables[block][xy]==0 || field_select_tables[block][xy]==1);
755  if(field_select_tables[block][xy] != field_select)
756  continue;
757  }
758 
759  P_LEFT[0] = mv_table[xy - 1][0];
760  P_LEFT[1] = mv_table[xy - 1][1];
761  if(P_LEFT[0] > (c->xmax<<1)) P_LEFT[0] = (c->xmax<<1);
762 
763  c->pred_x= P_LEFT[0];
764  c->pred_y= P_LEFT[1];
765 
766  if(!s->first_slice_line){
767  P_TOP[0] = mv_table[xy - mot_stride][0];
768  P_TOP[1] = mv_table[xy - mot_stride][1];
769  P_TOPRIGHT[0] = mv_table[xy - mot_stride + 1][0];
770  P_TOPRIGHT[1] = mv_table[xy - mot_stride + 1][1];
771  if(P_TOP[1] > (c->ymax<<1)) P_TOP[1] = (c->ymax<<1);
772  if(P_TOPRIGHT[0] < (c->xmin<<1)) P_TOPRIGHT[0]= (c->xmin<<1);
773  if(P_TOPRIGHT[0] > (c->xmax<<1)) P_TOPRIGHT[0]= (c->xmax<<1);
774  if(P_TOPRIGHT[1] > (c->ymax<<1)) P_TOPRIGHT[1]= (c->ymax<<1);
775 
776  P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
777  P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
778  }
779  P_MV1[0]= mx; //FIXME not correct if block != field_select
780  P_MV1[1]= my / 2;
781 
782  dmin = epzs_motion_search2(s, &mx_i, &my_i, P, block, field_select+ref_index, mv_table, (1<<16)>>1);
783 
784  dmin= c->sub_motion_search(s, &mx_i, &my_i, dmin, block, field_select+ref_index, size, h);
785 
786  mv_table[xy][0]= mx_i;
787  mv_table[xy][1]= my_i;
788 
789  if(s->dsp.me_sub_cmp[0] != s->dsp.mb_cmp[0]){
790  int dxy;
791 
792  //FIXME chroma ME
793  uint8_t *ref= c->ref[field_select+ref_index][0] + (mx_i>>1) + (my_i>>1)*stride;
794  dxy = ((my_i & 1) << 1) | (mx_i & 1);
795 
796  if(s->no_rounding){
797  s->hdsp.put_no_rnd_pixels_tab[size][dxy](c->scratchpad, ref , stride, h);
798  }else{
799  s->hdsp.put_pixels_tab [size][dxy](c->scratchpad, ref , stride, h);
800  }
801  dmin= s->dsp.mb_cmp[size](s, c->src[block][0], c->scratchpad, stride, h);
802  dmin+= (mv_penalty[mx_i-c->pred_x] + mv_penalty[my_i-c->pred_y] + 1)*c->mb_penalty_factor;
803  }else
804  dmin+= c->mb_penalty_factor; //field_select bits
805 
806  dmin += field_select != block; //slightly prefer same field
807 
808  if(dmin < best_dmin){
809  best_dmin= dmin;
810  best_field= field_select;
811  }
812  }
813  {
814  int16_t (*mv_table)[2]= mv_tables[block][best_field];
815 
816  if(mv_table[xy][0] != mx) same=0; //FIXME check if these checks work and are any good at all
817  if(mv_table[xy][1]&1) same=0;
818  if(mv_table[xy][1]*2 != my) same=0;
819  if(best_field != block) same=0;
820  }
821 
822  field_select_tables[block][xy]= best_field;
823  dmin_sum += best_dmin;
824  }
825 
826  c->ymin<<=1;
827  c->ymax<<=1;
828  c->stride>>=1;
829  c->uvstride>>=1;
830 
831  if(same)
832  return INT_MAX;
833 
834  switch(c->avctx->mb_cmp&0xFF){
835  /*case FF_CMP_SSE:
836  return dmin_sum+ 32*s->qscale*s->qscale;*/
837  case FF_CMP_RD:
838  return dmin_sum;
839  default:
840  return dmin_sum+ 11*c->mb_penalty_factor;
841  }
842 }
843 
844 static inline int get_penalty_factor(int lambda, int lambda2, int type){
845  switch(type&0xFF){
846  default:
847  case FF_CMP_SAD:
848  return lambda>>FF_LAMBDA_SHIFT;
849  case FF_CMP_DCT:
850  return (3*lambda)>>(FF_LAMBDA_SHIFT+1);
851  case FF_CMP_W53:
852  return (4*lambda)>>(FF_LAMBDA_SHIFT);
853  case FF_CMP_W97:
854  return (2*lambda)>>(FF_LAMBDA_SHIFT);
855  case FF_CMP_SATD:
856  case FF_CMP_DCT264:
857  return (2*lambda)>>FF_LAMBDA_SHIFT;
858  case FF_CMP_RD:
859  case FF_CMP_PSNR:
860  case FF_CMP_SSE:
861  case FF_CMP_NSSE:
862  return lambda2>>FF_LAMBDA_SHIFT;
863  case FF_CMP_BIT:
864  return 1;
865  }
866 }
867 
868 void ff_estimate_p_frame_motion(MpegEncContext * s,
869  int mb_x, int mb_y)
870 {
871  MotionEstContext * const c= &s->me;
872  uint8_t *pix, *ppix;
873  int sum, mx, my, dmin;
874  int varc; ///< the variance of the block (sum of squared (p[y][x]-average))
875  int vard; ///< sum of squared differences with the estimated motion vector
876  int P[10][2];
877  const int shift= 1+s->quarter_sample;
878  int mb_type=0;
879  Picture * const pic= &s->current_picture;
880 
881  init_ref(c, s->new_picture.f.data, s->last_picture.f.data, NULL, 16*mb_x, 16*mb_y, 0);
882 
883  av_assert0(s->quarter_sample==0 || s->quarter_sample==1);
884  av_assert0(s->linesize == c->stride);
885  av_assert0(s->uvlinesize == c->uvstride);
886 
887  c->penalty_factor = get_penalty_factor(s->lambda, s->lambda2, c->avctx->me_cmp);
888  c->sub_penalty_factor= get_penalty_factor(s->lambda, s->lambda2, c->avctx->me_sub_cmp);
889  c->mb_penalty_factor = get_penalty_factor(s->lambda, s->lambda2, c->avctx->mb_cmp);
890  c->current_mv_penalty= c->mv_penalty[s->f_code] + MAX_MV;
891 
892  get_limits(s, 16*mb_x, 16*mb_y);
893  c->skip=0;
894 
895  /* intra / predictive decision */
896  pix = c->src[0][0];
897  sum = s->dsp.pix_sum(pix, s->linesize);
898  varc = s->dsp.pix_norm1(pix, s->linesize) - (((unsigned)sum*sum)>>8) + 500;
899 
900  pic->mb_mean[s->mb_stride * mb_y + mb_x] = (sum+128)>>8;
901  pic->mb_var [s->mb_stride * mb_y + mb_x] = (varc+128)>>8;
902  c->mb_var_sum_temp += (varc+128)>>8;
903 
904  switch(s->me_method) {
905  case ME_ZERO:
906  default:
907  mx = 0;
908  my = 0;
909  dmin = 0;
910  break;
911  case ME_X1:
912  case ME_EPZS:
913  {
914  const int mot_stride = s->b8_stride;
915  const int mot_xy = s->block_index[0];
916 
917  P_LEFT[0] = s->current_picture.motion_val[0][mot_xy - 1][0];
918  P_LEFT[1] = s->current_picture.motion_val[0][mot_xy - 1][1];
919 
920  if(P_LEFT[0] > (c->xmax<<shift)) P_LEFT[0] = (c->xmax<<shift);
921 
922  if(!s->first_slice_line) {
923  P_TOP[0] = s->current_picture.motion_val[0][mot_xy - mot_stride ][0];
924  P_TOP[1] = s->current_picture.motion_val[0][mot_xy - mot_stride ][1];
925  P_TOPRIGHT[0] = s->current_picture.motion_val[0][mot_xy - mot_stride + 2][0];
926  P_TOPRIGHT[1] = s->current_picture.motion_val[0][mot_xy - mot_stride + 2][1];
927  if(P_TOP[1] > (c->ymax<<shift)) P_TOP[1] = (c->ymax<<shift);
928  if(P_TOPRIGHT[0] < (c->xmin<<shift)) P_TOPRIGHT[0]= (c->xmin<<shift);
929  if(P_TOPRIGHT[1] > (c->ymax<<shift)) P_TOPRIGHT[1]= (c->ymax<<shift);
930 
931  P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
932  P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
933 
934  if(s->out_format == FMT_H263){
935  c->pred_x = P_MEDIAN[0];
936  c->pred_y = P_MEDIAN[1];
937  }else { /* mpeg1 at least */
938  c->pred_x= P_LEFT[0];
939  c->pred_y= P_LEFT[1];
940  }
941  }else{
942  c->pred_x= P_LEFT[0];
943  c->pred_y= P_LEFT[1];
944  }
945 
946  }
947  dmin = ff_epzs_motion_search(s, &mx, &my, P, 0, 0, s->p_mv_table, (1<<16)>>shift, 0, 16);
948 
949  break;
950  }
951 
952  /* At this point (mx,my) are full-pell and the relative displacement */
953  ppix = c->ref[0][0] + (my * s->linesize) + mx;
954 
955  vard = s->dsp.sse[0](NULL, pix, ppix, s->linesize, 16);
956 
957  pic->mc_mb_var[s->mb_stride * mb_y + mb_x] = (vard+128)>>8;
958  c->mc_mb_var_sum_temp += (vard+128)>>8;
959 
960  if(mb_type){
961  int p_score= FFMIN(vard, varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*100);
962  int i_score= varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*20;
963  c->scene_change_score+= ff_sqrt(p_score) - ff_sqrt(i_score);
964 
965  if(mb_type == CANDIDATE_MB_TYPE_INTER){
966  c->sub_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
967  set_p_mv_tables(s, mx, my, 1);
968  }else{
969  mx <<=shift;
970  my <<=shift;
971  }
972  if(mb_type == CANDIDATE_MB_TYPE_INTER4V){
973  h263_mv4_search(s, mx, my, shift);
974 
975  set_p_mv_tables(s, mx, my, 0);
976  }
977  if(mb_type == CANDIDATE_MB_TYPE_INTER_I){
978  interlaced_search(s, 0, s->p_field_mv_table, s->p_field_select_table, mx, my, 1);
979  }
980  }else if(c->avctx->mb_decision > FF_MB_DECISION_SIMPLE){
981  int p_score= FFMIN(vard, varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*100);
982  int i_score= varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*20;
983  c->scene_change_score+= ff_sqrt(p_score) - ff_sqrt(i_score);
984 
985  if (vard*2 + 200*256 > varc)
986  mb_type|= CANDIDATE_MB_TYPE_INTRA;
987  if (varc*2 + 200*256 > vard || s->qscale > 24){
988 // if (varc*2 + 200*256 + 50*(s->lambda2>>FF_LAMBDA_SHIFT) > vard){
989  mb_type|= CANDIDATE_MB_TYPE_INTER;
990  c->sub_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
991  if(s->flags&CODEC_FLAG_MV0)
992  if(mx || my)
993  mb_type |= CANDIDATE_MB_TYPE_SKIPPED; //FIXME check difference
994  }else{
995  mx <<=shift;
996  my <<=shift;
997  }
998  if((s->flags&CODEC_FLAG_4MV)
999  && !c->skip && varc>50<<8 && vard>10<<8){
1000  if(h263_mv4_search(s, mx, my, shift) < INT_MAX)
1001  mb_type|=CANDIDATE_MB_TYPE_INTER4V;
1002 
1003  set_p_mv_tables(s, mx, my, 0);
1004  }else
1005  set_p_mv_tables(s, mx, my, 1);
1006  if((s->flags&CODEC_FLAG_INTERLACED_ME)
1007  && !c->skip){ //FIXME varc/d checks
1008  if(interlaced_search(s, 0, s->p_field_mv_table, s->p_field_select_table, mx, my, 0) < INT_MAX)
1009  mb_type |= CANDIDATE_MB_TYPE_INTER_I;
1010  }
1011  }else{
1012  int intra_score, i;
1013  mb_type= CANDIDATE_MB_TYPE_INTER;
1014 
1015  dmin= c->sub_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
1016  if(c->avctx->me_sub_cmp != c->avctx->mb_cmp && !c->skip)
1017  dmin= get_mb_score(s, mx, my, 0, 0, 0, 16, 1);
1018 
1019  if((s->flags&CODEC_FLAG_4MV)
1020  && !c->skip && varc>50<<8 && vard>10<<8){
1021  int dmin4= h263_mv4_search(s, mx, my, shift);
1022  if(dmin4 < dmin){
1023  mb_type= CANDIDATE_MB_TYPE_INTER4V;
1024  dmin=dmin4;
1025  }
1026  }
1027  if((s->flags&CODEC_FLAG_INTERLACED_ME)
1028  && !c->skip){ //FIXME varc/d checks
1029  int dmin_i= interlaced_search(s, 0, s->p_field_mv_table, s->p_field_select_table, mx, my, 0);
1030  if(dmin_i < dmin){
1031  mb_type = CANDIDATE_MB_TYPE_INTER_I;
1032  dmin= dmin_i;
1033  }
1034  }
1035 
1036  set_p_mv_tables(s, mx, my, mb_type!=CANDIDATE_MB_TYPE_INTER4V);
1037 
1038  /* get intra luma score */
1039  if((c->avctx->mb_cmp&0xFF)==FF_CMP_SSE){
1040  intra_score= varc - 500;
1041  }else{
1042  unsigned mean = (sum+128)>>8;
1043  mean*= 0x01010101;
1044 
1045  for(i=0; i<16; i++){
1046  *(uint32_t*)(&c->scratchpad[i*s->linesize+ 0]) = mean;
1047  *(uint32_t*)(&c->scratchpad[i*s->linesize+ 4]) = mean;
1048  *(uint32_t*)(&c->scratchpad[i*s->linesize+ 8]) = mean;
1049  *(uint32_t*)(&c->scratchpad[i*s->linesize+12]) = mean;
1050  }
1051 
1052  intra_score= s->dsp.mb_cmp[0](s, c->scratchpad, pix, s->linesize, 16);
1053  }
1054  intra_score += c->mb_penalty_factor*16;
1055 
1056  if(intra_score < dmin){
1057  mb_type= CANDIDATE_MB_TYPE_INTRA;
1058  s->current_picture.mb_type[mb_y*s->mb_stride + mb_x] = CANDIDATE_MB_TYPE_INTRA; //FIXME cleanup
1059  }else
1060  s->current_picture.mb_type[mb_y*s->mb_stride + mb_x] = 0;
1061 
1062  {
1063  int p_score= FFMIN(vard, varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*100);
1064  int i_score= varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*20;
1065  c->scene_change_score+= ff_sqrt(p_score) - ff_sqrt(i_score);
1066  }
1067  }
1068 
1069  s->mb_type[mb_y*s->mb_stride + mb_x]= mb_type;
1070 }
1071 
1072 int ff_pre_estimate_p_frame_motion(MpegEncContext * s,
1073  int mb_x, int mb_y)
1074 {
1075  MotionEstContext * const c= &s->me;
1076  int mx, my, dmin;
1077  int P[10][2];
1078  const int shift= 1+s->quarter_sample;
1079  const int xy= mb_x + mb_y*s->mb_stride;
1080  init_ref(c, s->new_picture.f.data, s->last_picture.f.data, NULL, 16*mb_x, 16*mb_y, 0);
1081 
1082  av_assert0(s->quarter_sample==0 || s->quarter_sample==1);
1083 
1084  c->pre_penalty_factor = get_penalty_factor(s->lambda, s->lambda2, c->avctx->me_pre_cmp);
1085  c->current_mv_penalty= c->mv_penalty[s->f_code] + MAX_MV;
1086 
1087  get_limits(s, 16*mb_x, 16*mb_y);
1088  c->skip=0;
1089 
1090  P_LEFT[0] = s->p_mv_table[xy + 1][0];
1091  P_LEFT[1] = s->p_mv_table[xy + 1][1];
1092 
1093  if(P_LEFT[0] < (c->xmin<<shift)) P_LEFT[0] = (c->xmin<<shift);
1094 
1095  /* special case for first line */
1096  if (s->first_slice_line) {
1097  c->pred_x= P_LEFT[0];
1098  c->pred_y= P_LEFT[1];
1099  P_TOP[0]= P_TOPRIGHT[0]= P_MEDIAN[0]=
1100  P_TOP[1]= P_TOPRIGHT[1]= P_MEDIAN[1]= 0; //FIXME
1101  } else {
1102  P_TOP[0] = s->p_mv_table[xy + s->mb_stride ][0];
1103  P_TOP[1] = s->p_mv_table[xy + s->mb_stride ][1];
1104  P_TOPRIGHT[0] = s->p_mv_table[xy + s->mb_stride - 1][0];
1105  P_TOPRIGHT[1] = s->p_mv_table[xy + s->mb_stride - 1][1];
1106  if(P_TOP[1] < (c->ymin<<shift)) P_TOP[1] = (c->ymin<<shift);
1107  if(P_TOPRIGHT[0] > (c->xmax<<shift)) P_TOPRIGHT[0]= (c->xmax<<shift);
1108  if(P_TOPRIGHT[1] < (c->ymin<<shift)) P_TOPRIGHT[1]= (c->ymin<<shift);
1109 
1110  P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
1111  P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
1112 
1113  c->pred_x = P_MEDIAN[0];
1114  c->pred_y = P_MEDIAN[1];
1115  }
1116 
1117  dmin = ff_epzs_motion_search(s, &mx, &my, P, 0, 0, s->p_mv_table, (1<<16)>>shift, 0, 16);
1118 
1119  s->p_mv_table[xy][0] = mx<<shift;
1120  s->p_mv_table[xy][1] = my<<shift;
1121 
1122  return dmin;
1123 }
1124 
1125 static int estimate_motion_b(MpegEncContext *s, int mb_x, int mb_y,
1126  int16_t (*mv_table)[2], int ref_index, int f_code)
1127 {
1128  MotionEstContext * const c= &s->me;
1129  int mx, my, dmin;
1130  int P[10][2];
1131  const int shift= 1+s->quarter_sample;
1132  const int mot_stride = s->mb_stride;
1133  const int mot_xy = mb_y*mot_stride + mb_x;
1134  uint8_t * const mv_penalty= c->mv_penalty[f_code] + MAX_MV;
1135  int mv_scale;
1136 
1137  c->penalty_factor = get_penalty_factor(s->lambda, s->lambda2, c->avctx->me_cmp);
1138  c->sub_penalty_factor= get_penalty_factor(s->lambda, s->lambda2, c->avctx->me_sub_cmp);
1139  c->mb_penalty_factor = get_penalty_factor(s->lambda, s->lambda2, c->avctx->mb_cmp);
1140  c->current_mv_penalty= mv_penalty;
1141 
1142  get_limits(s, 16*mb_x, 16*mb_y);
1143 
1144  switch(s->me_method) {
1145  case ME_ZERO:
1146  default:
1147  mx = 0;
1148  my = 0;
1149  dmin = 0;
1150  break;
1151  case ME_X1:
1152  case ME_EPZS:
1153  P_LEFT[0] = mv_table[mot_xy - 1][0];
1154  P_LEFT[1] = mv_table[mot_xy - 1][1];
1155 
1156  if (P_LEFT[0] > (c->xmax << shift)) P_LEFT[0] = (c->xmax << shift);
1157 
1158  /* special case for first line */
1159  if (!s->first_slice_line) {
1160  P_TOP[0] = mv_table[mot_xy - mot_stride ][0];
1161  P_TOP[1] = mv_table[mot_xy - mot_stride ][1];
1162  P_TOPRIGHT[0] = mv_table[mot_xy - mot_stride + 1][0];
1163  P_TOPRIGHT[1] = mv_table[mot_xy - mot_stride + 1][1];
1164  if (P_TOP[1] > (c->ymax << shift)) P_TOP[1] = (c->ymax << shift);
1165  if (P_TOPRIGHT[0] < (c->xmin << shift)) P_TOPRIGHT[0] = (c->xmin << shift);
1166  if (P_TOPRIGHT[1] > (c->ymax << shift)) P_TOPRIGHT[1] = (c->ymax << shift);
1167 
1168  P_MEDIAN[0] = mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
1169  P_MEDIAN[1] = mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
1170  }
1171  c->pred_x = P_LEFT[0];
1172  c->pred_y = P_LEFT[1];
1173 
1174  if(mv_table == s->b_forw_mv_table){
1175  mv_scale= (s->pb_time<<16) / (s->pp_time<<shift);
1176  }else{
1177  mv_scale= ((s->pb_time - s->pp_time)<<16) / (s->pp_time<<shift);
1178  }
1179 
1180  dmin = ff_epzs_motion_search(s, &mx, &my, P, 0, ref_index, s->p_mv_table, mv_scale, 0, 16);
1181 
1182  break;
1183  }
1184 
1185  dmin= c->sub_motion_search(s, &mx, &my, dmin, 0, ref_index, 0, 16);
1186 
1187  if(c->avctx->me_sub_cmp != c->avctx->mb_cmp && !c->skip)
1188  dmin= get_mb_score(s, mx, my, 0, ref_index, 0, 16, 1);
1189 
1190 // s->mb_type[mb_y*s->mb_width + mb_x]= mb_type;
1191  mv_table[mot_xy][0]= mx;
1192  mv_table[mot_xy][1]= my;
1193 
1194  return dmin;
1195 }
1196 
1197 static inline int check_bidir_mv(MpegEncContext * s,
1198  int motion_fx, int motion_fy,
1199  int motion_bx, int motion_by,
1200  int pred_fx, int pred_fy,
1201  int pred_bx, int pred_by,
1202  int size, int h)
1203 {
1204  //FIXME optimize?
1205  //FIXME better f_code prediction (max mv & distance)
1206  //FIXME pointers
1207  MotionEstContext * const c= &s->me;
1208  uint8_t * const mv_penalty_f= c->mv_penalty[s->f_code] + MAX_MV; // f_code of the prev frame
1209  uint8_t * const mv_penalty_b= c->mv_penalty[s->b_code] + MAX_MV; // f_code of the prev frame
1210  int stride= c->stride;
1211  uint8_t *dest_y = c->scratchpad;
1212  uint8_t *ptr;
1213  int dxy;
1214  int src_x, src_y;
1215  int fbmin;
1216  uint8_t **src_data= c->src[0];
1217  uint8_t **ref_data= c->ref[0];
1218  uint8_t **ref2_data= c->ref[2];
1219 
1220  if(s->quarter_sample){
1221  dxy = ((motion_fy & 3) << 2) | (motion_fx & 3);
1222  src_x = motion_fx >> 2;
1223  src_y = motion_fy >> 2;
1224 
1225  ptr = ref_data[0] + (src_y * stride) + src_x;
1226  s->dsp.put_qpel_pixels_tab[0][dxy](dest_y , ptr , stride);
1227 
1228  dxy = ((motion_by & 3) << 2) | (motion_bx & 3);
1229  src_x = motion_bx >> 2;
1230  src_y = motion_by >> 2;
1231 
1232  ptr = ref2_data[0] + (src_y * stride) + src_x;
1233  s->dsp.avg_qpel_pixels_tab[size][dxy](dest_y , ptr , stride);
1234  }else{
1235  dxy = ((motion_fy & 1) << 1) | (motion_fx & 1);
1236  src_x = motion_fx >> 1;
1237  src_y = motion_fy >> 1;
1238 
1239  ptr = ref_data[0] + (src_y * stride) + src_x;
1240  s->hdsp.put_pixels_tab[size][dxy](dest_y , ptr , stride, h);
1241 
1242  dxy = ((motion_by & 1) << 1) | (motion_bx & 1);
1243  src_x = motion_bx >> 1;
1244  src_y = motion_by >> 1;
1245 
1246  ptr = ref2_data[0] + (src_y * stride) + src_x;
1247  s->hdsp.avg_pixels_tab[size][dxy](dest_y , ptr , stride, h);
1248  }
1249 
1250  fbmin = (mv_penalty_f[motion_fx-pred_fx] + mv_penalty_f[motion_fy-pred_fy])*c->mb_penalty_factor
1251  +(mv_penalty_b[motion_bx-pred_bx] + mv_penalty_b[motion_by-pred_by])*c->mb_penalty_factor
1252  + s->dsp.mb_cmp[size](s, src_data[0], dest_y, stride, h); //FIXME new_pic
1253 
1254  if(c->avctx->mb_cmp&FF_CMP_CHROMA){
1255  }
1256  //FIXME CHROMA !!!
1257 
1258  return fbmin;
1259 }
1260 
1261 /* refine the bidir vectors in hq mode and return the score in both lq & hq mode*/
1262 static inline int bidir_refine(MpegEncContext * s, int mb_x, int mb_y)
1263 {
1264  MotionEstContext * const c= &s->me;
1265  const int mot_stride = s->mb_stride;
1266  const int xy = mb_y *mot_stride + mb_x;
1267  int fbmin;
1268  int pred_fx= s->b_bidir_forw_mv_table[xy-1][0];
1269  int pred_fy= s->b_bidir_forw_mv_table[xy-1][1];
1270  int pred_bx= s->b_bidir_back_mv_table[xy-1][0];
1271  int pred_by= s->b_bidir_back_mv_table[xy-1][1];
1272  int motion_fx= s->b_bidir_forw_mv_table[xy][0]= s->b_forw_mv_table[xy][0];
1273  int motion_fy= s->b_bidir_forw_mv_table[xy][1]= s->b_forw_mv_table[xy][1];
1274  int motion_bx= s->b_bidir_back_mv_table[xy][0]= s->b_back_mv_table[xy][0];
1275  int motion_by= s->b_bidir_back_mv_table[xy][1]= s->b_back_mv_table[xy][1];
1276  const int flags= c->sub_flags;
1277  const int qpel= flags&FLAG_QPEL;
1278  const int shift= 1+qpel;
1279  const int xmin= c->xmin<<shift;
1280  const int ymin= c->ymin<<shift;
1281  const int xmax= c->xmax<<shift;
1282  const int ymax= c->ymax<<shift;
1283 #define HASH(fx,fy,bx,by) ((fx)+17*(fy)+63*(bx)+117*(by))
1284 #define HASH8(fx,fy,bx,by) ((uint8_t)HASH(fx,fy,bx,by))
1285  int hashidx= HASH(motion_fx,motion_fy, motion_bx, motion_by);
1286  uint8_t map[256] = { 0 };
1287 
1288  map[hashidx&255] = 1;
1289 
1290  fbmin= check_bidir_mv(s, motion_fx, motion_fy,
1291  motion_bx, motion_by,
1292  pred_fx, pred_fy,
1293  pred_bx, pred_by,
1294  0, 16);
1295 
1296  if(s->avctx->bidir_refine){
1297  int end;
1298  static const uint8_t limittab[5]={0,8,32,64,80};
1299  const int limit= limittab[s->avctx->bidir_refine];
1300  static const int8_t vect[][4]={
1301 { 0, 0, 0, 1}, { 0, 0, 0,-1}, { 0, 0, 1, 0}, { 0, 0,-1, 0}, { 0, 1, 0, 0}, { 0,-1, 0, 0}, { 1, 0, 0, 0}, {-1, 0, 0, 0},
1302 
1303 { 0, 0, 1, 1}, { 0, 0,-1,-1}, { 0, 1, 1, 0}, { 0,-1,-1, 0}, { 1, 1, 0, 0}, {-1,-1, 0, 0}, { 1, 0, 0, 1}, {-1, 0, 0,-1},
1304 { 0, 1, 0, 1}, { 0,-1, 0,-1}, { 1, 0, 1, 0}, {-1, 0,-1, 0},
1305 { 0, 0,-1, 1}, { 0, 0, 1,-1}, { 0,-1, 1, 0}, { 0, 1,-1, 0}, {-1, 1, 0, 0}, { 1,-1, 0, 0}, { 1, 0, 0,-1}, {-1, 0, 0, 1},
1306 { 0,-1, 0, 1}, { 0, 1, 0,-1}, {-1, 0, 1, 0}, { 1, 0,-1, 0},
1307 
1308 { 0, 1, 1, 1}, { 0,-1,-1,-1}, { 1, 1, 1, 0}, {-1,-1,-1, 0}, { 1, 1, 0, 1}, {-1,-1, 0,-1}, { 1, 0, 1, 1}, {-1, 0,-1,-1},
1309 { 0,-1, 1, 1}, { 0, 1,-1,-1}, {-1, 1, 1, 0}, { 1,-1,-1, 0}, { 1, 1, 0,-1}, {-1,-1, 0, 1}, { 1, 0,-1, 1}, {-1, 0, 1,-1},
1310 { 0, 1,-1, 1}, { 0,-1, 1,-1}, { 1,-1, 1, 0}, {-1, 1,-1, 0}, {-1, 1, 0, 1}, { 1,-1, 0,-1}, { 1, 0, 1,-1}, {-1, 0,-1, 1},
1311 { 0, 1, 1,-1}, { 0,-1,-1, 1}, { 1, 1,-1, 0}, {-1,-1, 1, 0}, { 1,-1, 0, 1}, {-1, 1, 0,-1}, {-1, 0, 1, 1}, { 1, 0,-1,-1},
1312 
1313 { 1, 1, 1, 1}, {-1,-1,-1,-1},
1314 { 1, 1, 1,-1}, {-1,-1,-1, 1}, { 1, 1,-1, 1}, {-1,-1, 1,-1}, { 1,-1, 1, 1}, {-1, 1,-1,-1}, {-1, 1, 1, 1}, { 1,-1,-1,-1},
1315 { 1, 1,-1,-1}, {-1,-1, 1, 1}, { 1,-1,-1, 1}, {-1, 1, 1,-1}, { 1,-1, 1,-1}, {-1, 1,-1, 1},
1316  };
1317  static const uint8_t hash[]={
1318 HASH8( 0, 0, 0, 1), HASH8( 0, 0, 0,-1), HASH8( 0, 0, 1, 0), HASH8( 0, 0,-1, 0), HASH8( 0, 1, 0, 0), HASH8( 0,-1, 0, 0), HASH8( 1, 0, 0, 0), HASH8(-1, 0, 0, 0),
1319 
1320 HASH8( 0, 0, 1, 1), HASH8( 0, 0,-1,-1), HASH8( 0, 1, 1, 0), HASH8( 0,-1,-1, 0), HASH8( 1, 1, 0, 0), HASH8(-1,-1, 0, 0), HASH8( 1, 0, 0, 1), HASH8(-1, 0, 0,-1),
1321 HASH8( 0, 1, 0, 1), HASH8( 0,-1, 0,-1), HASH8( 1, 0, 1, 0), HASH8(-1, 0,-1, 0),
1322 HASH8( 0, 0,-1, 1), HASH8( 0, 0, 1,-1), HASH8( 0,-1, 1, 0), HASH8( 0, 1,-1, 0), HASH8(-1, 1, 0, 0), HASH8( 1,-1, 0, 0), HASH8( 1, 0, 0,-1), HASH8(-1, 0, 0, 1),
1323 HASH8( 0,-1, 0, 1), HASH8( 0, 1, 0,-1), HASH8(-1, 0, 1, 0), HASH8( 1, 0,-1, 0),
1324 
1325 HASH8( 0, 1, 1, 1), HASH8( 0,-1,-1,-1), HASH8( 1, 1, 1, 0), HASH8(-1,-1,-1, 0), HASH8( 1, 1, 0, 1), HASH8(-1,-1, 0,-1), HASH8( 1, 0, 1, 1), HASH8(-1, 0,-1,-1),
1326 HASH8( 0,-1, 1, 1), HASH8( 0, 1,-1,-1), HASH8(-1, 1, 1, 0), HASH8( 1,-1,-1, 0), HASH8( 1, 1, 0,-1), HASH8(-1,-1, 0, 1), HASH8( 1, 0,-1, 1), HASH8(-1, 0, 1,-1),
1327 HASH8( 0, 1,-1, 1), HASH8( 0,-1, 1,-1), HASH8( 1,-1, 1, 0), HASH8(-1, 1,-1, 0), HASH8(-1, 1, 0, 1), HASH8( 1,-1, 0,-1), HASH8( 1, 0, 1,-1), HASH8(-1, 0,-1, 1),
1328 HASH8( 0, 1, 1,-1), HASH8( 0,-1,-1, 1), HASH8( 1, 1,-1, 0), HASH8(-1,-1, 1, 0), HASH8( 1,-1, 0, 1), HASH8(-1, 1, 0,-1), HASH8(-1, 0, 1, 1), HASH8( 1, 0,-1,-1),
1329 
1330 HASH8( 1, 1, 1, 1), HASH8(-1,-1,-1,-1),
1331 HASH8( 1, 1, 1,-1), HASH8(-1,-1,-1, 1), HASH8( 1, 1,-1, 1), HASH8(-1,-1, 1,-1), HASH8( 1,-1, 1, 1), HASH8(-1, 1,-1,-1), HASH8(-1, 1, 1, 1), HASH8( 1,-1,-1,-1),
1332 HASH8( 1, 1,-1,-1), HASH8(-1,-1, 1, 1), HASH8( 1,-1,-1, 1), HASH8(-1, 1, 1,-1), HASH8( 1,-1, 1,-1), HASH8(-1, 1,-1, 1),
1333 };
1334 
1335 #define CHECK_BIDIR(fx,fy,bx,by)\
1336  if( !map[(hashidx+HASH(fx,fy,bx,by))&255]\
1337  &&(fx<=0 || motion_fx+fx<=xmax) && (fy<=0 || motion_fy+fy<=ymax) && (bx<=0 || motion_bx+bx<=xmax) && (by<=0 || motion_by+by<=ymax)\
1338  &&(fx>=0 || motion_fx+fx>=xmin) && (fy>=0 || motion_fy+fy>=ymin) && (bx>=0 || motion_bx+bx>=xmin) && (by>=0 || motion_by+by>=ymin)){\
1339  int score;\
1340  map[(hashidx+HASH(fx,fy,bx,by))&255] = 1;\
1341  score= check_bidir_mv(s, motion_fx+fx, motion_fy+fy, motion_bx+bx, motion_by+by, pred_fx, pred_fy, pred_bx, pred_by, 0, 16);\
1342  if(score < fbmin){\
1343  hashidx += HASH(fx,fy,bx,by);\
1344  fbmin= score;\
1345  motion_fx+=fx;\
1346  motion_fy+=fy;\
1347  motion_bx+=bx;\
1348  motion_by+=by;\
1349  end=0;\
1350  }\
1351  }
1352 #define CHECK_BIDIR2(a,b,c,d)\
1353 CHECK_BIDIR(a,b,c,d)\
1354 CHECK_BIDIR(-(a),-(b),-(c),-(d))
1355 
1356  do{
1357  int i;
1358  int borderdist=0;
1359  end=1;
1360 
1361  CHECK_BIDIR2(0,0,0,1)
1362  CHECK_BIDIR2(0,0,1,0)
1363  CHECK_BIDIR2(0,1,0,0)
1364  CHECK_BIDIR2(1,0,0,0)
1365 
1366  for(i=8; i<limit; i++){
1367  int fx= motion_fx+vect[i][0];
1368  int fy= motion_fy+vect[i][1];
1369  int bx= motion_bx+vect[i][2];
1370  int by= motion_by+vect[i][3];
1371  if(borderdist<=0){
1372  int a= (xmax - FFMAX(fx,bx))|(FFMIN(fx,bx) - xmin);
1373  int b= (ymax - FFMAX(fy,by))|(FFMIN(fy,by) - ymin);
1374  if((a|b) < 0)
1375  map[(hashidx+hash[i])&255] = 1;
1376  }
1377  if(!map[(hashidx+hash[i])&255]){
1378  int score;
1379  map[(hashidx+hash[i])&255] = 1;
1380  score= check_bidir_mv(s, fx, fy, bx, by, pred_fx, pred_fy, pred_bx, pred_by, 0, 16);
1381  if(score < fbmin){
1382  hashidx += hash[i];
1383  fbmin= score;
1384  motion_fx=fx;
1385  motion_fy=fy;
1386  motion_bx=bx;
1387  motion_by=by;
1388  end=0;
1389  borderdist--;
1390  if(borderdist<=0){
1391  int a= FFMIN(xmax - FFMAX(fx,bx), FFMIN(fx,bx) - xmin);
1392  int b= FFMIN(ymax - FFMAX(fy,by), FFMIN(fy,by) - ymin);
1393  borderdist= FFMIN(a,b);
1394  }
1395  }
1396  }
1397  }
1398  }while(!end);
1399  }
1400 
1401  s->b_bidir_forw_mv_table[xy][0]= motion_fx;
1402  s->b_bidir_forw_mv_table[xy][1]= motion_fy;
1403  s->b_bidir_back_mv_table[xy][0]= motion_bx;
1404  s->b_bidir_back_mv_table[xy][1]= motion_by;
1405 
1406  return fbmin;
1407 }
1408 
1409 static inline int direct_search(MpegEncContext * s, int mb_x, int mb_y)
1410 {
1411  MotionEstContext * const c= &s->me;
1412  int P[10][2];
1413  const int mot_stride = s->mb_stride;
1414  const int mot_xy = mb_y*mot_stride + mb_x;
1415  const int shift= 1+s->quarter_sample;
1416  int dmin, i;
1417  const int time_pp= s->pp_time;
1418  const int time_pb= s->pb_time;
1419  int mx, my, xmin, xmax, ymin, ymax;
1420  int16_t (*mv_table)[2]= s->b_direct_mv_table;
1421 
1422  c->current_mv_penalty= c->mv_penalty[1] + MAX_MV;
1423  ymin= xmin=(-32)>>shift;
1424  ymax= xmax= 31>>shift;
1425 
1426  if (IS_8X8(s->next_picture.mb_type[mot_xy])) {
1427  s->mv_type= MV_TYPE_8X8;
1428  }else{
1429  s->mv_type= MV_TYPE_16X16;
1430  }
1431 
1432  for(i=0; i<4; i++){
1433  int index= s->block_index[i];
1434  int min, max;
1435 
1436  c->co_located_mv[i][0] = s->next_picture.motion_val[0][index][0];
1437  c->co_located_mv[i][1] = s->next_picture.motion_val[0][index][1];
1438  c->direct_basis_mv[i][0]= c->co_located_mv[i][0]*time_pb/time_pp + ((i& 1)<<(shift+3));
1439  c->direct_basis_mv[i][1]= c->co_located_mv[i][1]*time_pb/time_pp + ((i>>1)<<(shift+3));
1440 // c->direct_basis_mv[1][i][0]= c->co_located_mv[i][0]*(time_pb - time_pp)/time_pp + ((i &1)<<(shift+3);
1441 // c->direct_basis_mv[1][i][1]= c->co_located_mv[i][1]*(time_pb - time_pp)/time_pp + ((i>>1)<<(shift+3);
1442 
1443  max= FFMAX(c->direct_basis_mv[i][0], c->direct_basis_mv[i][0] - c->co_located_mv[i][0])>>shift;
1444  min= FFMIN(c->direct_basis_mv[i][0], c->direct_basis_mv[i][0] - c->co_located_mv[i][0])>>shift;
1445  max+= 16*mb_x + 1; // +-1 is for the simpler rounding
1446  min+= 16*mb_x - 1;
1447  xmax= FFMIN(xmax, s->width - max);
1448  xmin= FFMAX(xmin, - 16 - min);
1449 
1450  max= FFMAX(c->direct_basis_mv[i][1], c->direct_basis_mv[i][1] - c->co_located_mv[i][1])>>shift;
1451  min= FFMIN(c->direct_basis_mv[i][1], c->direct_basis_mv[i][1] - c->co_located_mv[i][1])>>shift;
1452  max+= 16*mb_y + 1; // +-1 is for the simpler rounding
1453  min+= 16*mb_y - 1;
1454  ymax= FFMIN(ymax, s->height - max);
1455  ymin= FFMAX(ymin, - 16 - min);
1456 
1457  if(s->mv_type == MV_TYPE_16X16) break;
1458  }
1459 
1460  av_assert2(xmax <= 15 && ymax <= 15 && xmin >= -16 && ymin >= -16);
1461 
1462  if(xmax < 0 || xmin >0 || ymax < 0 || ymin > 0){
1463  s->b_direct_mv_table[mot_xy][0]= 0;
1464  s->b_direct_mv_table[mot_xy][1]= 0;
1465 
1466  return 256*256*256*64;
1467  }
1468 
1469  c->xmin= xmin;
1470  c->ymin= ymin;
1471  c->xmax= xmax;
1472  c->ymax= ymax;
1473  c->flags |= FLAG_DIRECT;
1474  c->sub_flags |= FLAG_DIRECT;
1475  c->pred_x=0;
1476  c->pred_y=0;
1477 
1478  P_LEFT[0] = av_clip(mv_table[mot_xy - 1][0], xmin<<shift, xmax<<shift);
1479  P_LEFT[1] = av_clip(mv_table[mot_xy - 1][1], ymin<<shift, ymax<<shift);
1480 
1481  /* special case for first line */
1482  if (!s->first_slice_line) { //FIXME maybe allow this over thread boundary as it is clipped
1483  P_TOP[0] = av_clip(mv_table[mot_xy - mot_stride ][0], xmin<<shift, xmax<<shift);
1484  P_TOP[1] = av_clip(mv_table[mot_xy - mot_stride ][1], ymin<<shift, ymax<<shift);
1485  P_TOPRIGHT[0] = av_clip(mv_table[mot_xy - mot_stride + 1 ][0], xmin<<shift, xmax<<shift);
1486  P_TOPRIGHT[1] = av_clip(mv_table[mot_xy - mot_stride + 1 ][1], ymin<<shift, ymax<<shift);
1487 
1488  P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
1489  P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
1490  }
1491 
1492  dmin = ff_epzs_motion_search(s, &mx, &my, P, 0, 0, mv_table, 1<<(16-shift), 0, 16);
1493  if(c->sub_flags&FLAG_QPEL)
1494  dmin = qpel_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
1495  else
1496  dmin = hpel_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
1497 
1498  if(c->avctx->me_sub_cmp != c->avctx->mb_cmp && !c->skip)
1499  dmin= get_mb_score(s, mx, my, 0, 0, 0, 16, 1);
1500 
1501  get_limits(s, 16*mb_x, 16*mb_y); //restore c->?min/max, maybe not needed
1502 
1503  mv_table[mot_xy][0]= mx;
1504  mv_table[mot_xy][1]= my;
1505  c->flags &= ~FLAG_DIRECT;
1506  c->sub_flags &= ~FLAG_DIRECT;
1507 
1508  return dmin;
1509 }
1510 
1511 void ff_estimate_b_frame_motion(MpegEncContext * s,
1512  int mb_x, int mb_y)
1513 {
1514  MotionEstContext * const c= &s->me;
1515  const int penalty_factor= c->mb_penalty_factor;
1516  int fmin, bmin, dmin, fbmin, bimin, fimin;
1517  int type=0;
1518  const int xy = mb_y*s->mb_stride + mb_x;
1519  init_ref(c, s->new_picture.f.data, s->last_picture.f.data,
1520  s->next_picture.f.data, 16 * mb_x, 16 * mb_y, 2);
1521 
1522  get_limits(s, 16*mb_x, 16*mb_y);
1523 
1524  c->skip=0;
1525 
1526  if (s->codec_id == AV_CODEC_ID_MPEG4 && s->next_picture.mbskip_table[xy]) {
1527  int score= direct_search(s, mb_x, mb_y); //FIXME just check 0,0
1528 
1529  score= ((unsigned)(score*score + 128*256))>>16;
1530  c->mc_mb_var_sum_temp += score;
1531  s->current_picture.mc_mb_var[mb_y*s->mb_stride + mb_x] = score; //FIXME use SSE
1532  s->mb_type[mb_y*s->mb_stride + mb_x]= CANDIDATE_MB_TYPE_DIRECT0;
1533 
1534  return;
1535  }
1536 
1537  if (s->codec_id == AV_CODEC_ID_MPEG4)
1538  dmin= direct_search(s, mb_x, mb_y);
1539  else
1540  dmin= INT_MAX;
1541 //FIXME penalty stuff for non mpeg4
1542  c->skip=0;
1543  fmin = estimate_motion_b(s, mb_x, mb_y, s->b_forw_mv_table, 0, s->f_code) +
1544  3 * penalty_factor;
1545 
1546  c->skip=0;
1547  bmin = estimate_motion_b(s, mb_x, mb_y, s->b_back_mv_table, 2, s->b_code) +
1548  2 * penalty_factor;
1549  av_dlog(s, " %d %d ", s->b_forw_mv_table[xy][0], s->b_forw_mv_table[xy][1]);
1550 
1551  c->skip=0;
1552  fbmin= bidir_refine(s, mb_x, mb_y) + penalty_factor;
1553  av_dlog(s, "%d %d %d %d\n", dmin, fmin, bmin, fbmin);
1554 
1555  if(s->flags & CODEC_FLAG_INTERLACED_ME){
1556 //FIXME mb type penalty
1557  c->skip=0;
1558  c->current_mv_penalty= c->mv_penalty[s->f_code] + MAX_MV;
1559  fimin= interlaced_search(s, 0,
1560  s->b_field_mv_table[0], s->b_field_select_table[0],
1561  s->b_forw_mv_table[xy][0], s->b_forw_mv_table[xy][1], 0);
1562  c->current_mv_penalty= c->mv_penalty[s->b_code] + MAX_MV;
1563  bimin= interlaced_search(s, 2,
1564  s->b_field_mv_table[1], s->b_field_select_table[1],
1565  s->b_back_mv_table[xy][0], s->b_back_mv_table[xy][1], 0);
1566  }else
1567  fimin= bimin= INT_MAX;
1568 
1569  {
1570  int score= fmin;
1571  type = CANDIDATE_MB_TYPE_FORWARD;
1572 
1573  if (dmin <= score){
1574  score = dmin;
1575  type = CANDIDATE_MB_TYPE_DIRECT;
1576  }
1577  if(bmin<score){
1578  score=bmin;
1579  type= CANDIDATE_MB_TYPE_BACKWARD;
1580  }
1581  if(fbmin<score){
1582  score=fbmin;
1583  type= CANDIDATE_MB_TYPE_BIDIR;
1584  }
1585  if(fimin<score){
1586  score=fimin;
1587  type= CANDIDATE_MB_TYPE_FORWARD_I;
1588  }
1589  if(bimin<score){
1590  score=bimin;
1591  type= CANDIDATE_MB_TYPE_BACKWARD_I;
1592  }
1593 
1594  score= ((unsigned)(score*score + 128*256))>>16;
1595  c->mc_mb_var_sum_temp += score;
1596  s->current_picture.mc_mb_var[mb_y*s->mb_stride + mb_x] = score; //FIXME use SSE
1597  }
1598 
1599  if(c->avctx->mb_decision > FF_MB_DECISION_SIMPLE){
1600  type= CANDIDATE_MB_TYPE_FORWARD | CANDIDATE_MB_TYPE_BACKWARD | CANDIDATE_MB_TYPE_BIDIR | CANDIDATE_MB_TYPE_DIRECT;
1601  if(fimin < INT_MAX)
1602  type |= CANDIDATE_MB_TYPE_FORWARD_I;
1603  if(bimin < INT_MAX)
1604  type |= CANDIDATE_MB_TYPE_BACKWARD_I;
1605  if(fimin < INT_MAX && bimin < INT_MAX){
1606  type |= CANDIDATE_MB_TYPE_BIDIR_I;
1607  }
1608  //FIXME something smarter
1609  if(dmin>256*256*16) type&= ~CANDIDATE_MB_TYPE_DIRECT; //do not try direct mode if it is invalid for this MB
1610  if(s->codec_id == AV_CODEC_ID_MPEG4 && type&CANDIDATE_MB_TYPE_DIRECT && s->flags&CODEC_FLAG_MV0 && *(uint32_t*)s->b_direct_mv_table[xy])
1611  type |= CANDIDATE_MB_TYPE_DIRECT0;
1612  }
1613 
1614  s->mb_type[mb_y*s->mb_stride + mb_x]= type;
1615 }
1616 
1617 /* find best f_code for ME which do unlimited searches */
1618 int ff_get_best_fcode(MpegEncContext * s, int16_t (*mv_table)[2], int type)
1619 {
1620  if(s->me_method>=ME_EPZS){
1621  int score[8];
1622  int i, y, range= s->avctx->me_range ? s->avctx->me_range : (INT_MAX/2);
1623  uint8_t * fcode_tab= s->fcode_tab;
1624  int best_fcode=-1;
1625  int best_score=-10000000;
1626 
1627  if(s->msmpeg4_version)
1628  range= FFMIN(range, 16);
1629  else if(s->codec_id == AV_CODEC_ID_MPEG2VIDEO && s->avctx->strict_std_compliance >= FF_COMPLIANCE_NORMAL)
1630  range= FFMIN(range, 256);
1631 
1632  for(i=0; i<8; i++) score[i]= s->mb_num*(8-i);
1633 
1634  for(y=0; y<s->mb_height; y++){
1635  int x;
1636  int xy= y*s->mb_stride;
1637  for(x=0; x<s->mb_width; x++){
1638  if(s->mb_type[xy] & type){
1639  int mx= mv_table[xy][0];
1640  int my= mv_table[xy][1];
1641  int fcode= FFMAX(fcode_tab[mx + MAX_MV],
1642  fcode_tab[my + MAX_MV]);
1643  int j;
1644 
1645  if(mx >= range || mx < -range ||
1646  my >= range || my < -range)
1647  continue;
1648 
1649  for(j=0; j<fcode && j<8; j++){
1650  if(s->pict_type==AV_PICTURE_TYPE_B || s->current_picture.mc_mb_var[xy] < s->current_picture.mb_var[xy])
1651  score[j]-= 170;
1652  }
1653  }
1654  xy++;
1655  }
1656  }
1657 
1658  for(i=1; i<8; i++){
1659  if(score[i] > best_score){
1660  best_score= score[i];
1661  best_fcode= i;
1662  }
1663  }
1664 
1665  return best_fcode;
1666  }else{
1667  return 1;
1668  }
1669 }
1670 
1671 void ff_fix_long_p_mvs(MpegEncContext * s)
1672 {
1673  MotionEstContext * const c= &s->me;
1674  const int f_code= s->f_code;
1675  int y, range;
1676  av_assert0(s->pict_type==AV_PICTURE_TYPE_P);
1677 
1678  range = (((s->out_format == FMT_MPEG1 || s->msmpeg4_version) ? 8 : 16) << f_code);
1679 
1680  av_assert0(range <= 16 || !s->msmpeg4_version);
1681  av_assert0(range <=256 || !(s->codec_id == AV_CODEC_ID_MPEG2VIDEO && s->avctx->strict_std_compliance >= FF_COMPLIANCE_NORMAL));
1682 
1683  if(c->avctx->me_range && range > c->avctx->me_range) range= c->avctx->me_range;
1684 
1685  if(s->flags&CODEC_FLAG_4MV){
1686  const int wrap= s->b8_stride;
1687 
1688  /* clip / convert to intra 8x8 type MVs */
1689  for(y=0; y<s->mb_height; y++){
1690  int xy= y*2*wrap;
1691  int i= y*s->mb_stride;
1692  int x;
1693 
1694  for(x=0; x<s->mb_width; x++){
1695  if(s->mb_type[i]&CANDIDATE_MB_TYPE_INTER4V){
1696  int block;
1697  for(block=0; block<4; block++){
1698  int off= (block& 1) + (block>>1)*wrap;
1699  int mx = s->current_picture.motion_val[0][ xy + off ][0];
1700  int my = s->current_picture.motion_val[0][ xy + off ][1];
1701 
1702  if( mx >=range || mx <-range
1703  || my >=range || my <-range){
1704  s->mb_type[i] &= ~CANDIDATE_MB_TYPE_INTER4V;
1705  s->mb_type[i] |= CANDIDATE_MB_TYPE_INTRA;
1706  s->current_picture.mb_type[i] = CANDIDATE_MB_TYPE_INTRA;
1707  }
1708  }
1709  }
1710  xy+=2;
1711  i++;
1712  }
1713  }
1714  }
1715 }
1716 
1717 /**
1718  *
1719  * @param truncate 1 for truncation, 0 for using intra
1720  */
1721 void ff_fix_long_mvs(MpegEncContext * s, uint8_t *field_select_table, int field_select,
1722  int16_t (*mv_table)[2], int f_code, int type, int truncate)
1723 {
1724  MotionEstContext * const c= &s->me;
1725  int y, h_range, v_range;
1726 
1727  // RAL: 8 in MPEG-1, 16 in MPEG-4
1728  int range = (((s->out_format == FMT_MPEG1 || s->msmpeg4_version) ? 8 : 16) << f_code);
1729 
1730  if(c->avctx->me_range && range > c->avctx->me_range) range= c->avctx->me_range;
1731 
1732  h_range= range;
1733  v_range= field_select_table ? range>>1 : range;
1734 
1735  /* clip / convert to intra 16x16 type MVs */
1736  for(y=0; y<s->mb_height; y++){
1737  int x;
1738  int xy= y*s->mb_stride;
1739  for(x=0; x<s->mb_width; x++){
1740  if (s->mb_type[xy] & type){ // RAL: "type" test added...
1741  if(field_select_table==NULL || field_select_table[xy] == field_select){
1742  if( mv_table[xy][0] >=h_range || mv_table[xy][0] <-h_range
1743  || mv_table[xy][1] >=v_range || mv_table[xy][1] <-v_range){
1744 
1745  if(truncate){
1746  if (mv_table[xy][0] > h_range-1) mv_table[xy][0]= h_range-1;
1747  else if(mv_table[xy][0] < -h_range ) mv_table[xy][0]= -h_range;
1748  if (mv_table[xy][1] > v_range-1) mv_table[xy][1]= v_range-1;
1749  else if(mv_table[xy][1] < -v_range ) mv_table[xy][1]= -v_range;
1750  }else{
1751  s->mb_type[xy] &= ~type;
1752  s->mb_type[xy] |= CANDIDATE_MB_TYPE_INTRA;
1753  mv_table[xy][0]=
1754  mv_table[xy][1]= 0;
1755  }
1756  }
1757  }
1758  }
1759  xy++;
1760  }
1761  }
1762 }
Generated on Sun Mar 23 2014 23:49:59 for FFmpeg by   doxygen 1.8.2