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 "internal.h"
36 #include "mathops.h"
37 #include "motion_est.h"
38 #include "mpegutils.h"
39 #include "mpegvideo.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 #define ME_MAP_SHIFT 3
48 #define ME_MAP_MV_BITS 11
49 
51  int *mx_ptr, int *my_ptr, int dmin,
52  int src_index, int ref_index,
53  int size, int h);
54 
55 static inline unsigned update_map_generation(MotionEstContext *c)
56 {
57  c->map_generation+= 1<<(ME_MAP_MV_BITS*2);
58  if(c->map_generation==0){
59  c->map_generation= 1<<(ME_MAP_MV_BITS*2);
60  memset(c->map, 0, sizeof(uint32_t)*ME_MAP_SIZE);
61  }
62  return c->map_generation;
63 }
64 
65 /* shape adaptive search stuff */
66 typedef struct Minima{
67  int height;
68  int x, y;
69  int checked;
70 }Minima;
71 
72 static int minima_cmp(const void *a, const void *b){
73  const Minima *da = (const Minima *) a;
74  const Minima *db = (const Minima *) b;
75 
76  return da->height - db->height;
77 }
78 
79 #define FLAG_QPEL 1 //must be 1
80 #define FLAG_CHROMA 2
81 #define FLAG_DIRECT 4
82 
83 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){
84  const int offset[3]= {
85  y*c-> stride + x,
86  ((y*c->uvstride + x)>>1),
87  ((y*c->uvstride + x)>>1),
88  };
89  int i;
90  for(i=0; i<3; i++){
91  c->src[0][i]= src [i] + offset[i];
92  c->ref[0][i]= ref [i] + offset[i];
93  }
94  if(ref_index){
95  for(i=0; i<3; i++){
96  c->ref[ref_index][i]= ref2[i] + offset[i];
97  }
98  }
99 }
100 
101 static int get_flags(MotionEstContext *c, int direct, int chroma){
102  return ((c->avctx->flags&AV_CODEC_FLAG_QPEL) ? FLAG_QPEL : 0)
103  + (direct ? FLAG_DIRECT : 0)
104  + (chroma ? FLAG_CHROMA : 0);
105 }
106 
107 static av_always_inline int cmp_direct_inline(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
108  const int size, const int h, int ref_index, int src_index,
109  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, int qpel){
110  MotionEstContext * const c= &s->me;
111  const int stride= c->stride;
112  const int hx= subx + (x<<(1+qpel));
113  const int hy= suby + (y<<(1+qpel));
114  uint8_t * const * const ref= c->ref[ref_index];
115  uint8_t * const * const src= c->src[src_index];
116  int d;
117  //FIXME check chroma 4mv, (no crashes ...)
118  av_assert2(x >= c->xmin && hx <= c->xmax<<(qpel+1) && y >= c->ymin && hy <= c->ymax<<(qpel+1));
119  if(x >= c->xmin && hx <= c->xmax<<(qpel+1) && y >= c->ymin && hy <= c->ymax<<(qpel+1)){
120  const int time_pp= s->pp_time;
121  const int time_pb= s->pb_time;
122  const int mask= 2*qpel+1;
123  if(s->mv_type==MV_TYPE_8X8){
124  int i;
125  for(i=0; i<4; i++){
126  int fx = c->direct_basis_mv[i][0] + hx;
127  int fy = c->direct_basis_mv[i][1] + hy;
128  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));
129  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));
130  int fxy= (fx&mask) + ((fy&mask)<<(qpel+1));
131  int bxy= (bx&mask) + ((by&mask)<<(qpel+1));
132 
133  uint8_t *dst= c->temp + 8*(i&1) + 8*stride*(i>>1);
134  if(qpel){
135  c->qpel_put[1][fxy](dst, ref[0] + (fx>>2) + (fy>>2)*stride, stride);
136  c->qpel_avg[1][bxy](dst, ref[8] + (bx>>2) + (by>>2)*stride, stride);
137  }else{
138  c->hpel_put[1][fxy](dst, ref[0] + (fx>>1) + (fy>>1)*stride, stride, 8);
139  c->hpel_avg[1][bxy](dst, ref[8] + (bx>>1) + (by>>1)*stride, stride, 8);
140  }
141  }
142  }else{
143  int fx = c->direct_basis_mv[0][0] + hx;
144  int fy = c->direct_basis_mv[0][1] + hy;
145  int bx = hx ? fx - c->co_located_mv[0][0] : (c->co_located_mv[0][0]*(time_pb - time_pp)/time_pp);
146  int by = hy ? fy - c->co_located_mv[0][1] : (c->co_located_mv[0][1]*(time_pb - time_pp)/time_pp);
147  int fxy= (fx&mask) + ((fy&mask)<<(qpel+1));
148  int bxy= (bx&mask) + ((by&mask)<<(qpel+1));
149 
150  if(qpel){
151  c->qpel_put[1][fxy](c->temp , ref[0] + (fx>>2) + (fy>>2)*stride , stride);
152  c->qpel_put[1][fxy](c->temp + 8 , ref[0] + (fx>>2) + (fy>>2)*stride + 8 , stride);
153  c->qpel_put[1][fxy](c->temp + 8*stride, ref[0] + (fx>>2) + (fy>>2)*stride + 8*stride, stride);
154  c->qpel_put[1][fxy](c->temp + 8 + 8*stride, ref[0] + (fx>>2) + (fy>>2)*stride + 8 + 8*stride, stride);
155  c->qpel_avg[1][bxy](c->temp , ref[8] + (bx>>2) + (by>>2)*stride , stride);
156  c->qpel_avg[1][bxy](c->temp + 8 , ref[8] + (bx>>2) + (by>>2)*stride + 8 , stride);
157  c->qpel_avg[1][bxy](c->temp + 8*stride, ref[8] + (bx>>2) + (by>>2)*stride + 8*stride, stride);
158  c->qpel_avg[1][bxy](c->temp + 8 + 8*stride, ref[8] + (bx>>2) + (by>>2)*stride + 8 + 8*stride, stride);
159  }else{
160  av_assert2((fx>>1) + 16*s->mb_x >= -16);
161  av_assert2((fy>>1) + 16*s->mb_y >= -16);
162  av_assert2((fx>>1) + 16*s->mb_x <= s->width);
163  av_assert2((fy>>1) + 16*s->mb_y <= s->height);
164  av_assert2((bx>>1) + 16*s->mb_x >= -16);
165  av_assert2((by>>1) + 16*s->mb_y >= -16);
166  av_assert2((bx>>1) + 16*s->mb_x <= s->width);
167  av_assert2((by>>1) + 16*s->mb_y <= s->height);
168 
169  c->hpel_put[0][fxy](c->temp, ref[0] + (fx>>1) + (fy>>1)*stride, stride, 16);
170  c->hpel_avg[0][bxy](c->temp, ref[8] + (bx>>1) + (by>>1)*stride, stride, 16);
171  }
172  }
173  d = cmp_func(s, c->temp, src[0], stride, 16);
174  }else
175  d= 256*256*256*32;
176  return d;
177 }
178 
179 static av_always_inline int cmp_inline(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
180  const int size, const int h, int ref_index, int src_index,
181  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, int qpel, int chroma){
182  MotionEstContext * const c= &s->me;
183  const int stride= c->stride;
184  const int uvstride= c->uvstride;
185  const int dxy= subx + (suby<<(1+qpel)); //FIXME log2_subpel?
186  const int hx= subx + x*(1<<(1+qpel));
187  const int hy= suby + y*(1<<(1+qpel));
188  uint8_t * const * const ref= c->ref[ref_index];
189  uint8_t * const * const src= c->src[src_index];
190  int d;
191  //FIXME check chroma 4mv, (no crashes ...)
192  int uvdxy; /* no, it might not be used uninitialized */
193  if(dxy){
194  if(qpel){
195  if (h << size == 16) {
196  c->qpel_put[size][dxy](c->temp, ref[0] + x + y*stride, stride); //FIXME prototype (add h)
197  } else if (size == 0 && h == 8) {
198  c->qpel_put[1][dxy](c->temp , ref[0] + x + y*stride , stride);
199  c->qpel_put[1][dxy](c->temp + 8, ref[0] + x + y*stride + 8, stride);
200  } else
201  av_assert2(0);
202  if(chroma){
203  int cx= hx/2;
204  int cy= hy/2;
205  cx= (cx>>1)|(cx&1);
206  cy= (cy>>1)|(cy&1);
207  uvdxy= (cx&1) + 2*(cy&1);
208  // FIXME x/y wrong, but MPEG-4 qpel is sick anyway, we should drop as much of it as possible in favor for H.264
209  }
210  }else{
211  c->hpel_put[size][dxy](c->temp, ref[0] + x + y*stride, stride, h);
212  if(chroma)
213  uvdxy= dxy | (x&1) | (2*(y&1));
214  }
215  d = cmp_func(s, c->temp, src[0], stride, h);
216  }else{
217  d = cmp_func(s, src[0], ref[0] + x + y*stride, stride, h);
218  if(chroma)
219  uvdxy= (x&1) + 2*(y&1);
220  }
221  if(chroma){
222  uint8_t * const uvtemp= c->temp + 16*stride;
223  c->hpel_put[size+1][uvdxy](uvtemp , ref[1] + (x>>1) + (y>>1)*uvstride, uvstride, h>>1);
224  c->hpel_put[size+1][uvdxy](uvtemp+8, ref[2] + (x>>1) + (y>>1)*uvstride, uvstride, h>>1);
225  d += chroma_cmp_func(s, uvtemp , src[1], uvstride, h>>1);
226  d += chroma_cmp_func(s, uvtemp+8, src[2], uvstride, h>>1);
227  }
228  return d;
229 }
230 
231 static int cmp_simple(MpegEncContext *s, const int x, const int y,
232  int ref_index, int src_index,
233  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func){
234  return cmp_inline(s,x,y,0,0,0,16,ref_index,src_index, cmp_func, chroma_cmp_func, 0, 0);
235 }
236 
237 static int cmp_fpel_internal(MpegEncContext *s, const int x, const int y,
238  const int size, const int h, int ref_index, int src_index,
239  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
240  if(flags&FLAG_DIRECT){
241  return cmp_direct_inline(s,x,y,0,0,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, flags&FLAG_QPEL);
242  }else{
243  return cmp_inline(s,x,y,0,0,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 0, flags&FLAG_CHROMA);
244  }
245 }
246 
247 static int cmp_internal(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
248  const int size, const int h, int ref_index, int src_index,
249  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
250  if(flags&FLAG_DIRECT){
251  return cmp_direct_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, flags&FLAG_QPEL);
252  }else{
253  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);
254  }
255 }
256 
257 /** @brief compares a block (either a full macroblock or a partition thereof)
258  against a proposed motion-compensated prediction of that block
259  */
260 static av_always_inline int cmp(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
261  const int size, const int h, int ref_index, int src_index,
262  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
265  && flags==0 && h==16 && size==0 && subx==0 && suby==0){
266  return cmp_simple(s,x,y,ref_index,src_index, cmp_func, chroma_cmp_func);
267  }else if(av_builtin_constant_p(subx) && av_builtin_constant_p(suby)
268  && subx==0 && suby==0){
269  return cmp_fpel_internal(s,x,y,size,h,ref_index,src_index, cmp_func, chroma_cmp_func,flags);
270  }else{
271  return cmp_internal(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, flags);
272  }
273 }
274 
275 static int cmp_hpel(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
276  const int size, const int h, int ref_index, int src_index,
277  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
278  if(flags&FLAG_DIRECT){
279  return cmp_direct_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 0);
280  }else{
281  return cmp_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 0, flags&FLAG_CHROMA);
282  }
283 }
284 
285 static int cmp_qpel(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
286  const int size, const int h, int ref_index, int src_index,
287  me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
288  if(flags&FLAG_DIRECT){
289  return cmp_direct_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 1);
290  }else{
291  return cmp_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 1, flags&FLAG_CHROMA);
292  }
293 }
294 
295 #include "motion_est_template.c"
296 
298  ptrdiff_t stride, int h)
299 {
300  return 0;
301 }
302 
303 static void zero_hpel(uint8_t *a, const uint8_t *b, ptrdiff_t stride, int h){
304 }
305 
307  MotionEstContext * const c= &s->me;
308  int cache_size= FFMIN(ME_MAP_SIZE>>ME_MAP_SHIFT, 1<<ME_MAP_SHIFT);
309  int dia_size= FFMAX(FFABS(s->avctx->dia_size)&255, FFABS(s->avctx->pre_dia_size)&255);
310 
312  av_log(s->avctx, AV_LOG_ERROR, "ME_MAP size is too small for SAB diamond\n");
313  return -1;
314  }
315 
316 #if FF_API_MOTION_EST
317  //special case of snow is needed because snow uses its own iterative ME code
319  if (s->motion_est == FF_ME_EPZS) {
320  if (s->me_method == ME_ZERO)
321  s->motion_est = FF_ME_ZERO;
322  else if (s->me_method == ME_EPZS)
323  s->motion_est = FF_ME_EPZS;
324  else if (s->me_method == ME_X1)
325  s->motion_est = FF_ME_XONE;
326  else if (s->avctx->codec_id != AV_CODEC_ID_SNOW) {
328  "me_method is only allowed to be set to zero and epzs; "
329  "for hex,umh,full and others see dia_size\n");
330  return -1;
331  }
332  }
334 #endif
335 
336  c->avctx= s->avctx;
337 
338  if(s->codec_id == AV_CODEC_ID_H261)
339  c->avctx->me_sub_cmp = c->avctx->me_cmp;
340 
341  if(cache_size < 2*dia_size && !c->stride){
342  av_log(s->avctx, AV_LOG_INFO, "ME_MAP size may be a little small for the selected diamond size\n");
343  }
344 
346  ff_set_cmp(&s->mecc, s->mecc.me_cmp, c->avctx->me_cmp);
348  ff_set_cmp(&s->mecc, s->mecc.mb_cmp, c->avctx->mb_cmp);
349 
350  c->flags = get_flags(c, 0, c->avctx->me_cmp &FF_CMP_CHROMA);
352  c->mb_flags = get_flags(c, 0, c->avctx->mb_cmp &FF_CMP_CHROMA);
353 
354 /*FIXME s->no_rounding b_type*/
355  if (s->avctx->flags & AV_CODEC_FLAG_QPEL) {
358  if (s->no_rounding)
360  else
362  }else{
365  else if( c->avctx->me_sub_cmp == FF_CMP_SAD
366  && c->avctx-> me_cmp == FF_CMP_SAD
367  && c->avctx-> mb_cmp == FF_CMP_SAD)
368  c->sub_motion_search= sad_hpel_motion_search; // 2050 vs. 2450 cycles
369  else
371  }
372  c->hpel_avg = s->hdsp.avg_pixels_tab;
373  if (s->no_rounding)
375  else
376  c->hpel_put = s->hdsp.put_pixels_tab;
377 
378  if(s->linesize){
379  c->stride = s->linesize;
380  c->uvstride= s->uvlinesize;
381  }else{
382  c->stride = 16*s->mb_width + 32;
383  c->uvstride= 8*s->mb_width + 16;
384  }
385 
386  /* 8x8 fullpel search would need a 4x4 chroma compare, which we do
387  * not have yet, and even if we had, the motion estimation code
388  * does not expect it. */
389  if (s->codec_id != AV_CODEC_ID_SNOW) {
390  if ((c->avctx->me_cmp & FF_CMP_CHROMA) /* && !s->mecc.me_cmp[2] */)
391  s->mecc.me_cmp[2] = zero_cmp;
392  if ((c->avctx->me_sub_cmp & FF_CMP_CHROMA) && !s->mecc.me_sub_cmp[2])
393  s->mecc.me_sub_cmp[2] = zero_cmp;
394  c->hpel_put[2][0]= c->hpel_put[2][1]=
395  c->hpel_put[2][2]= c->hpel_put[2][3]= zero_hpel;
396  }
397 
398  if(s->codec_id == AV_CODEC_ID_H261){
400  }
401 
402  return 0;
403 }
404 
405 #define CHECK_SAD_HALF_MV(suffix, x, y) \
406 {\
407  d = s->mecc.pix_abs[size][(x ? 1 : 0) + (y ? 2 : 0)](NULL, pix, ptr + ((x) >> 1), stride, h); \
408  d += (mv_penalty[pen_x + x] + mv_penalty[pen_y + y])*penalty_factor;\
409  COPY3_IF_LT(dminh, d, dx, x, dy, y)\
410 }
411 
413  int *mx_ptr, int *my_ptr, int dmin,
414  int src_index, int ref_index,
415  int size, int h)
416 {
417  MotionEstContext * const c= &s->me;
418  const int penalty_factor= c->sub_penalty_factor;
419  int mx, my, dminh;
420  uint8_t *pix, *ptr;
421  int stride= c->stride;
423 
424  av_assert2(c->sub_flags == 0);
425 
426  if(c->skip){
427  *mx_ptr = 0;
428  *my_ptr = 0;
429  return dmin;
430  }
431 
432  pix = c->src[src_index][0];
433 
434  mx = *mx_ptr;
435  my = *my_ptr;
436  ptr = c->ref[ref_index][0] + (my * stride) + mx;
437 
438  dminh = dmin;
439 
440  if (mx > xmin && mx < xmax &&
441  my > ymin && my < ymax) {
442  int dx=0, dy=0;
443  int d, pen_x, pen_y;
444  const int index= my*(1<<ME_MAP_SHIFT) + mx;
445  const int t= score_map[(index-(1<<ME_MAP_SHIFT))&(ME_MAP_SIZE-1)];
446  const int l= score_map[(index- 1 )&(ME_MAP_SIZE-1)];
447  const int r= score_map[(index+ 1 )&(ME_MAP_SIZE-1)];
448  const int b= score_map[(index+(1<<ME_MAP_SHIFT))&(ME_MAP_SIZE-1)];
449  mx += mx;
450  my += my;
451 
452 
453  pen_x= pred_x + mx;
454  pen_y= pred_y + my;
455 
456  ptr-= stride;
457  if(t<=b){
458  CHECK_SAD_HALF_MV(y2 , 0, -1)
459  if(l<=r){
460  CHECK_SAD_HALF_MV(xy2, -1, -1)
461  if(t+r<=b+l){
462  CHECK_SAD_HALF_MV(xy2, +1, -1)
463  ptr+= stride;
464  }else{
465  ptr+= stride;
466  CHECK_SAD_HALF_MV(xy2, -1, +1)
467  }
468  CHECK_SAD_HALF_MV(x2 , -1, 0)
469  }else{
470  CHECK_SAD_HALF_MV(xy2, +1, -1)
471  if(t+l<=b+r){
472  CHECK_SAD_HALF_MV(xy2, -1, -1)
473  ptr+= stride;
474  }else{
475  ptr+= stride;
476  CHECK_SAD_HALF_MV(xy2, +1, +1)
477  }
478  CHECK_SAD_HALF_MV(x2 , +1, 0)
479  }
480  }else{
481  if(l<=r){
482  if(t+l<=b+r){
483  CHECK_SAD_HALF_MV(xy2, -1, -1)
484  ptr+= stride;
485  }else{
486  ptr+= stride;
487  CHECK_SAD_HALF_MV(xy2, +1, +1)
488  }
489  CHECK_SAD_HALF_MV(x2 , -1, 0)
490  CHECK_SAD_HALF_MV(xy2, -1, +1)
491  }else{
492  if(t+r<=b+l){
493  CHECK_SAD_HALF_MV(xy2, +1, -1)
494  ptr+= stride;
495  }else{
496  ptr+= stride;
497  CHECK_SAD_HALF_MV(xy2, -1, +1)
498  }
499  CHECK_SAD_HALF_MV(x2 , +1, 0)
500  CHECK_SAD_HALF_MV(xy2, +1, +1)
501  }
502  CHECK_SAD_HALF_MV(y2 , 0, +1)
503  }
504  mx+=dx;
505  my+=dy;
506 
507  }else{
508  mx += mx;
509  my += my;
510  }
511 
512  *mx_ptr = mx;
513  *my_ptr = my;
514  return dminh;
515 }
516 
517 static inline void set_p_mv_tables(MpegEncContext * s, int mx, int my, int mv4)
518 {
519  const int xy= s->mb_x + s->mb_y*s->mb_stride;
520 
521  s->p_mv_table[xy][0] = mx;
522  s->p_mv_table[xy][1] = my;
523 
524  /* has already been set to the 4 MV if 4MV is done */
525  if(mv4){
526  int mot_xy= s->block_index[0];
527 
528  s->current_picture.motion_val[0][mot_xy ][0] = mx;
529  s->current_picture.motion_val[0][mot_xy ][1] = my;
530  s->current_picture.motion_val[0][mot_xy + 1][0] = mx;
531  s->current_picture.motion_val[0][mot_xy + 1][1] = my;
532 
533  mot_xy += s->b8_stride;
534  s->current_picture.motion_val[0][mot_xy ][0] = mx;
535  s->current_picture.motion_val[0][mot_xy ][1] = my;
536  s->current_picture.motion_val[0][mot_xy + 1][0] = mx;
537  s->current_picture.motion_val[0][mot_xy + 1][1] = my;
538  }
539 }
540 
541 /**
542  * get fullpel ME search limits.
543  */
544 static inline void get_limits(MpegEncContext *s, int x, int y)
545 {
546  MotionEstContext * const c= &s->me;
547  int range= c->avctx->me_range >> (1 + !!(c->flags&FLAG_QPEL));
548  int max_range = MAX_MV >> (1 + !!(c->flags&FLAG_QPEL));
549 /*
550  if(c->avctx->me_range) c->range= c->avctx->me_range >> 1;
551  else c->range= 16;
552 */
553  if (s->unrestricted_mv) {
554  c->xmin = - x - 16;
555  c->ymin = - y - 16;
556  c->xmax = - x + s->width;
557  c->ymax = - y + s->height;
558  } else if (s->out_format == FMT_H261){
559  // Search range of H.261 is different from other codec standards
560  c->xmin = (x > 15) ? - 15 : 0;
561  c->ymin = (y > 15) ? - 15 : 0;
562  c->xmax = (x < s->mb_width * 16 - 16) ? 15 : 0;
563  c->ymax = (y < s->mb_height * 16 - 16) ? 15 : 0;
564  } else {
565  c->xmin = - x;
566  c->ymin = - y;
567  c->xmax = - x + s->mb_width *16 - 16;
568  c->ymax = - y + s->mb_height*16 - 16;
569  }
570  if(!range || range > max_range)
571  range = max_range;
572  if(range){
573  c->xmin = FFMAX(c->xmin,-range);
574  c->xmax = FFMIN(c->xmax, range);
575  c->ymin = FFMAX(c->ymin,-range);
576  c->ymax = FFMIN(c->ymax, range);
577  }
578 }
579 
580 static inline void init_mv4_ref(MotionEstContext *c){
581  const int stride= c->stride;
582 
583  c->ref[1][0] = c->ref[0][0] + 8;
584  c->ref[2][0] = c->ref[0][0] + 8*stride;
585  c->ref[3][0] = c->ref[2][0] + 8;
586  c->src[1][0] = c->src[0][0] + 8;
587  c->src[2][0] = c->src[0][0] + 8*stride;
588  c->src[3][0] = c->src[2][0] + 8;
589 }
590 
591 static inline int h263_mv4_search(MpegEncContext *s, int mx, int my, int shift)
592 {
593  MotionEstContext * const c= &s->me;
594  const int size= 1;
595  const int h=8;
596  int block;
597  int P[10][2];
598  int dmin_sum=0, mx4_sum=0, my4_sum=0, i;
599  int same=1;
600  const int stride= c->stride;
602  int safety_clipping= s->unrestricted_mv && (s->width&15) && (s->height&15);
603 
604  init_mv4_ref(c);
605 
606  for(block=0; block<4; block++){
607  int mx4, my4;
608  int pred_x4, pred_y4;
609  int dmin4;
610  static const int off[4]= {2, 1, 1, -1};
611  const int mot_stride = s->b8_stride;
612  const int mot_xy = s->block_index[block];
613 
614  if(safety_clipping){
615  c->xmax = - 16*s->mb_x + s->width - 8*(block &1);
616  c->ymax = - 16*s->mb_y + s->height - 8*(block>>1);
617  }
618 
619  P_LEFT[0] = s->current_picture.motion_val[0][mot_xy - 1][0];
620  P_LEFT[1] = s->current_picture.motion_val[0][mot_xy - 1][1];
621 
622  if(P_LEFT[0] > (c->xmax<<shift)) P_LEFT[0] = (c->xmax<<shift);
623 
624  /* special case for first line */
625  if (s->first_slice_line && block<2) {
626  c->pred_x= pred_x4= P_LEFT[0];
627  c->pred_y= pred_y4= P_LEFT[1];
628  } else {
629  P_TOP[0] = s->current_picture.motion_val[0][mot_xy - mot_stride ][0];
630  P_TOP[1] = s->current_picture.motion_val[0][mot_xy - mot_stride ][1];
631  P_TOPRIGHT[0] = s->current_picture.motion_val[0][mot_xy - mot_stride + off[block]][0];
632  P_TOPRIGHT[1] = s->current_picture.motion_val[0][mot_xy - mot_stride + off[block]][1];
633  if(P_TOP[1] > (c->ymax<<shift)) P_TOP[1] = (c->ymax<<shift);
634  if(P_TOPRIGHT[0] < (c->xmin<<shift)) P_TOPRIGHT[0]= (c->xmin<<shift);
635  if(P_TOPRIGHT[0] > (c->xmax<<shift)) P_TOPRIGHT[0]= (c->xmax<<shift);
636  if(P_TOPRIGHT[1] > (c->ymax<<shift)) P_TOPRIGHT[1]= (c->ymax<<shift);
637 
638  P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
639  P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
640 
641  c->pred_x= pred_x4 = P_MEDIAN[0];
642  c->pred_y= pred_y4 = P_MEDIAN[1];
643  }
644  P_MV1[0]= mx;
645  P_MV1[1]= my;
646  if(safety_clipping)
647  for(i=1; i<10; i++){
648  if (s->first_slice_line && block<2 && i>1 && i<9)
649  continue;
650  if (i>4 && i<9)
651  continue;
652  if(P[i][0] > (c->xmax<<shift)) P[i][0]= (c->xmax<<shift);
653  if(P[i][1] > (c->ymax<<shift)) P[i][1]= (c->ymax<<shift);
654  }
655 
656  dmin4 = epzs_motion_search4(s, &mx4, &my4, P, block, block, s->p_mv_table, (1<<16)>>shift);
657 
658  dmin4= c->sub_motion_search(s, &mx4, &my4, dmin4, block, block, size, h);
659 
660  if (s->mecc.me_sub_cmp[0] != s->mecc.mb_cmp[0]) {
661  int dxy;
662  const int offset= ((block&1) + (block>>1)*stride)*8;
663  uint8_t *dest_y = c->scratchpad + offset;
664  if(s->quarter_sample){
665  uint8_t *ref= c->ref[block][0] + (mx4>>2) + (my4>>2)*stride;
666  dxy = ((my4 & 3) << 2) | (mx4 & 3);
667 
668  if(s->no_rounding)
669  s->qdsp.put_no_rnd_qpel_pixels_tab[1][dxy](dest_y, ref, stride);
670  else
671  s->qdsp.put_qpel_pixels_tab[1][dxy](dest_y, ref, stride);
672  }else{
673  uint8_t *ref= c->ref[block][0] + (mx4>>1) + (my4>>1)*stride;
674  dxy = ((my4 & 1) << 1) | (mx4 & 1);
675 
676  if(s->no_rounding)
677  s->hdsp.put_no_rnd_pixels_tab[1][dxy](dest_y , ref , stride, h);
678  else
679  s->hdsp.put_pixels_tab [1][dxy](dest_y , ref , stride, h);
680  }
681  dmin_sum+= (mv_penalty[mx4-pred_x4] + mv_penalty[my4-pred_y4])*c->mb_penalty_factor;
682  }else
683  dmin_sum+= dmin4;
684 
685  if(s->quarter_sample){
686  mx4_sum+= mx4/2;
687  my4_sum+= my4/2;
688  }else{
689  mx4_sum+= mx4;
690  my4_sum+= my4;
691  }
692 
693  s->current_picture.motion_val[0][s->block_index[block]][0] = mx4;
694  s->current_picture.motion_val[0][s->block_index[block]][1] = my4;
695 
696  if(mx4 != mx || my4 != my) same=0;
697  }
698 
699  if(same)
700  return INT_MAX;
701 
702  if (s->mecc.me_sub_cmp[0] != s->mecc.mb_cmp[0]) {
703  dmin_sum += s->mecc.mb_cmp[0](s,
704  s->new_picture.f->data[0] +
705  s->mb_x * 16 + s->mb_y * 16 * stride,
706  c->scratchpad, stride, 16);
707  }
708 
709  if(c->avctx->mb_cmp&FF_CMP_CHROMA){
710  int dxy;
711  int mx, my;
712  int offset;
713 
714  mx= ff_h263_round_chroma(mx4_sum);
715  my= ff_h263_round_chroma(my4_sum);
716  dxy = ((my & 1) << 1) | (mx & 1);
717 
718  offset= (s->mb_x*8 + (mx>>1)) + (s->mb_y*8 + (my>>1))*s->uvlinesize;
719 
720  if(s->no_rounding){
721  s->hdsp.put_no_rnd_pixels_tab[1][dxy](c->scratchpad , s->last_picture.f->data[1] + offset, s->uvlinesize, 8);
722  s->hdsp.put_no_rnd_pixels_tab[1][dxy](c->scratchpad + 8, s->last_picture.f->data[2] + offset, s->uvlinesize, 8);
723  }else{
724  s->hdsp.put_pixels_tab [1][dxy](c->scratchpad , s->last_picture.f->data[1] + offset, s->uvlinesize, 8);
725  s->hdsp.put_pixels_tab [1][dxy](c->scratchpad + 8, s->last_picture.f->data[2] + offset, s->uvlinesize, 8);
726  }
727 
728  dmin_sum += s->mecc.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);
729  dmin_sum += s->mecc.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);
730  }
731 
732  c->pred_x= mx;
733  c->pred_y= my;
734 
735  switch(c->avctx->mb_cmp&0xFF){
736  /*case FF_CMP_SSE:
737  return dmin_sum+ 32*s->qscale*s->qscale;*/
738  case FF_CMP_RD:
739  return dmin_sum;
740  default:
741  return dmin_sum+ 11*c->mb_penalty_factor;
742  }
743 }
744 
745 static inline void init_interlaced_ref(MpegEncContext *s, int ref_index){
746  MotionEstContext * const c= &s->me;
747 
748  c->ref[1+ref_index][0] = c->ref[0+ref_index][0] + s->linesize;
749  c->src[1][0] = c->src[0][0] + s->linesize;
750  if(c->flags & FLAG_CHROMA){
751  c->ref[1+ref_index][1] = c->ref[0+ref_index][1] + s->uvlinesize;
752  c->ref[1+ref_index][2] = c->ref[0+ref_index][2] + s->uvlinesize;
753  c->src[1][1] = c->src[0][1] + s->uvlinesize;
754  c->src[1][2] = c->src[0][2] + s->uvlinesize;
755  }
756 }
757 
758 static int interlaced_search(MpegEncContext *s, int ref_index,
759  int16_t (*mv_tables[2][2])[2], uint8_t *field_select_tables[2], int mx, int my, int user_field_select)
760 {
761  MotionEstContext * const c= &s->me;
762  const int size=0;
763  const int h=8;
764  int block;
765  int P[10][2];
767  int same=1;
768  const int stride= 2*s->linesize;
769  int dmin_sum= 0;
770  const int mot_stride= s->mb_stride;
771  const int xy= s->mb_x + s->mb_y*mot_stride;
772 
773  c->ymin>>=1;
774  c->ymax>>=1;
775  c->stride<<=1;
776  c->uvstride<<=1;
777  init_interlaced_ref(s, ref_index);
778 
779  for(block=0; block<2; block++){
780  int field_select;
781  int best_dmin= INT_MAX;
782  int best_field= -1;
783 
784  for(field_select=0; field_select<2; field_select++){
785  int dmin, mx_i, my_i;
786  int16_t (*mv_table)[2]= mv_tables[block][field_select];
787 
788  if(user_field_select){
789  av_assert1(field_select==0 || field_select==1);
790  av_assert1(field_select_tables[block][xy]==0 || field_select_tables[block][xy]==1);
791  if(field_select_tables[block][xy] != field_select)
792  continue;
793  }
794 
795  P_LEFT[0] = mv_table[xy - 1][0];
796  P_LEFT[1] = mv_table[xy - 1][1];
797  if(P_LEFT[0] > (c->xmax<<1)) P_LEFT[0] = (c->xmax<<1);
798 
799  c->pred_x= P_LEFT[0];
800  c->pred_y= P_LEFT[1];
801 
802  if(!s->first_slice_line){
803  P_TOP[0] = mv_table[xy - mot_stride][0];
804  P_TOP[1] = mv_table[xy - mot_stride][1];
805  P_TOPRIGHT[0] = mv_table[xy - mot_stride + 1][0];
806  P_TOPRIGHT[1] = mv_table[xy - mot_stride + 1][1];
807  if(P_TOP[1] > (c->ymax<<1)) P_TOP[1] = (c->ymax<<1);
808  if(P_TOPRIGHT[0] < (c->xmin<<1)) P_TOPRIGHT[0]= (c->xmin<<1);
809  if(P_TOPRIGHT[0] > (c->xmax<<1)) P_TOPRIGHT[0]= (c->xmax<<1);
810  if(P_TOPRIGHT[1] > (c->ymax<<1)) P_TOPRIGHT[1]= (c->ymax<<1);
811 
812  P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
813  P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
814  }
815  P_MV1[0]= mx; //FIXME not correct if block != field_select
816  P_MV1[1]= my / 2;
817 
818  dmin = epzs_motion_search2(s, &mx_i, &my_i, P, block, field_select+ref_index, mv_table, (1<<16)>>1);
819 
820  dmin= c->sub_motion_search(s, &mx_i, &my_i, dmin, block, field_select+ref_index, size, h);
821 
822  mv_table[xy][0]= mx_i;
823  mv_table[xy][1]= my_i;
824 
825  if (s->mecc.me_sub_cmp[0] != s->mecc.mb_cmp[0]) {
826  int dxy;
827 
828  //FIXME chroma ME
829  uint8_t *ref= c->ref[field_select+ref_index][0] + (mx_i>>1) + (my_i>>1)*stride;
830  dxy = ((my_i & 1) << 1) | (mx_i & 1);
831 
832  if(s->no_rounding){
834  }else{
835  s->hdsp.put_pixels_tab [size][dxy](c->scratchpad, ref , stride, h);
836  }
837  dmin = s->mecc.mb_cmp[size](s, c->src[block][0], c->scratchpad, stride, h);
838  dmin+= (mv_penalty[mx_i-c->pred_x] + mv_penalty[my_i-c->pred_y] + 1)*c->mb_penalty_factor;
839  }else
840  dmin+= c->mb_penalty_factor; //field_select bits
841 
842  dmin += field_select != block; //slightly prefer same field
843 
844  if(dmin < best_dmin){
845  best_dmin= dmin;
846  best_field= field_select;
847  }
848  }
849  {
850  int16_t (*mv_table)[2]= mv_tables[block][best_field];
851 
852  if(mv_table[xy][0] != mx) same=0; //FIXME check if these checks work and are any good at all
853  if(mv_table[xy][1]&1) same=0;
854  if(mv_table[xy][1]*2 != my) same=0;
855  if(best_field != block) same=0;
856  }
857 
858  field_select_tables[block][xy]= best_field;
859  dmin_sum += best_dmin;
860  }
861 
862  c->ymin<<=1;
863  c->ymax<<=1;
864  c->stride>>=1;
865  c->uvstride>>=1;
866 
867  if(same)
868  return INT_MAX;
869 
870  switch(c->avctx->mb_cmp&0xFF){
871  /*case FF_CMP_SSE:
872  return dmin_sum+ 32*s->qscale*s->qscale;*/
873  case FF_CMP_RD:
874  return dmin_sum;
875  default:
876  return dmin_sum+ 11*c->mb_penalty_factor;
877  }
878 }
879 
880 static inline int get_penalty_factor(int lambda, int lambda2, int type){
881  switch(type&0xFF){
882  default:
883  case FF_CMP_SAD:
884  return lambda>>FF_LAMBDA_SHIFT;
885  case FF_CMP_DCT:
886  return (3*lambda)>>(FF_LAMBDA_SHIFT+1);
887  case FF_CMP_W53:
888  return (4*lambda)>>(FF_LAMBDA_SHIFT);
889  case FF_CMP_W97:
890  return (2*lambda)>>(FF_LAMBDA_SHIFT);
891  case FF_CMP_SATD:
892  case FF_CMP_DCT264:
893  return (2*lambda)>>FF_LAMBDA_SHIFT;
894  case FF_CMP_RD:
895  case FF_CMP_PSNR:
896  case FF_CMP_SSE:
897  case FF_CMP_NSSE:
898  return lambda2>>FF_LAMBDA_SHIFT;
899  case FF_CMP_BIT:
900  return 1;
901  }
902 }
903 
905  int mb_x, int mb_y)
906 {
907  MotionEstContext * const c= &s->me;
908  uint8_t *pix, *ppix;
909  int sum, mx = 0, my = 0, dmin = 0;
910  int varc; ///< the variance of the block (sum of squared (p[y][x]-average))
911  int vard; ///< sum of squared differences with the estimated motion vector
912  int P[10][2];
913  const int shift= 1+s->quarter_sample;
914  int mb_type=0;
915  Picture * const pic= &s->current_picture;
916 
917  init_ref(c, s->new_picture.f->data, s->last_picture.f->data, NULL, 16*mb_x, 16*mb_y, 0);
918 
919  av_assert0(s->quarter_sample==0 || s->quarter_sample==1);
920  av_assert0(s->linesize == c->stride);
921  av_assert0(s->uvlinesize == c->uvstride);
922 
927 
928  get_limits(s, 16*mb_x, 16*mb_y);
929  c->skip=0;
930 
931  /* intra / predictive decision */
932  pix = c->src[0][0];
933  sum = s->mpvencdsp.pix_sum(pix, s->linesize);
934  varc = s->mpvencdsp.pix_norm1(pix, s->linesize) -
935  (((unsigned) sum * sum) >> 8) + 500;
936 
937  pic->mb_mean[s->mb_stride * mb_y + mb_x] = (sum+128)>>8;
938  pic->mb_var [s->mb_stride * mb_y + mb_x] = (varc+128)>>8;
939  c->mb_var_sum_temp += (varc+128)>>8;
940 
941  if (s->motion_est != FF_ME_ZERO) {
942  const int mot_stride = s->b8_stride;
943  const int mot_xy = s->block_index[0];
944 
945  P_LEFT[0] = s->current_picture.motion_val[0][mot_xy - 1][0];
946  P_LEFT[1] = s->current_picture.motion_val[0][mot_xy - 1][1];
947 
948  if (P_LEFT[0] > (c->xmax << shift))
949  P_LEFT[0] = c->xmax << shift;
950 
951  if (!s->first_slice_line) {
952  P_TOP[0] = s->current_picture.motion_val[0][mot_xy - mot_stride ][0];
953  P_TOP[1] = s->current_picture.motion_val[0][mot_xy - mot_stride ][1];
954  P_TOPRIGHT[0] = s->current_picture.motion_val[0][mot_xy - mot_stride + 2][0];
955  P_TOPRIGHT[1] = s->current_picture.motion_val[0][mot_xy - mot_stride + 2][1];
956  if (P_TOP[1] > (c->ymax << shift))
957  P_TOP[1] = c->ymax << shift;
958  if (P_TOPRIGHT[0] < (c->xmin << shift))
959  P_TOPRIGHT[0] = c->xmin << shift;
960  if (P_TOPRIGHT[1] > (c->ymax << shift))
961  P_TOPRIGHT[1] = c->ymax << shift;
962 
963  P_MEDIAN[0] = mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
964  P_MEDIAN[1] = mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
965 
966  if (s->out_format == FMT_H263) {
967  c->pred_x = P_MEDIAN[0];
968  c->pred_y = P_MEDIAN[1];
969  } else { /* MPEG-1 at least */
970  c->pred_x = P_LEFT[0];
971  c->pred_y = P_LEFT[1];
972  }
973  } else {
974  c->pred_x = P_LEFT[0];
975  c->pred_y = P_LEFT[1];
976  }
977  dmin = ff_epzs_motion_search(s, &mx, &my, P, 0, 0, s->p_mv_table, (1<<16)>>shift, 0, 16);
978  }
979 
980  /* At this point (mx,my) are full-pell and the relative displacement */
981  ppix = c->ref[0][0] + (my * s->linesize) + mx;
982 
983  vard = s->mecc.sse[0](NULL, pix, ppix, s->linesize, 16);
984 
985  pic->mc_mb_var[s->mb_stride * mb_y + mb_x] = (vard+128)>>8;
986  c->mc_mb_var_sum_temp += (vard+128)>>8;
987 
989  int p_score= FFMIN(vard, varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*100);
990  int i_score= varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*20;
991  c->scene_change_score+= ff_sqrt(p_score) - ff_sqrt(i_score);
992 
993  if (vard*2 + 200*256 > varc)
994  mb_type|= CANDIDATE_MB_TYPE_INTRA;
995  if (varc*2 + 200*256 > vard || s->qscale > 24){
996 // if (varc*2 + 200*256 + 50*(s->lambda2>>FF_LAMBDA_SHIFT) > vard){
997  mb_type|= CANDIDATE_MB_TYPE_INTER;
998  c->sub_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
999  if (s->mpv_flags & FF_MPV_FLAG_MV0)
1000  if(mx || my)
1001  mb_type |= CANDIDATE_MB_TYPE_SKIPPED; //FIXME check difference
1002  }else{
1003  mx <<=shift;
1004  my <<=shift;
1005  }
1006  if ((s->avctx->flags & AV_CODEC_FLAG_4MV)
1007  && !c->skip && varc>50<<8 && vard>10<<8){
1008  if(h263_mv4_search(s, mx, my, shift) < INT_MAX)
1009  mb_type|=CANDIDATE_MB_TYPE_INTER4V;
1010 
1011  set_p_mv_tables(s, mx, my, 0);
1012  }else
1013  set_p_mv_tables(s, mx, my, 1);
1015  && !c->skip){ //FIXME varc/d checks
1016  if(interlaced_search(s, 0, s->p_field_mv_table, s->p_field_select_table, mx, my, 0) < INT_MAX)
1017  mb_type |= CANDIDATE_MB_TYPE_INTER_I;
1018  }
1019  }else{
1020  int intra_score, i;
1021  mb_type= CANDIDATE_MB_TYPE_INTER;
1022 
1023  dmin= c->sub_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
1024  if(c->avctx->me_sub_cmp != c->avctx->mb_cmp && !c->skip)
1025  dmin= get_mb_score(s, mx, my, 0, 0, 0, 16, 1);
1026 
1027  if ((s->avctx->flags & AV_CODEC_FLAG_4MV)
1028  && !c->skip && varc>50<<8 && vard>10<<8){
1029  int dmin4= h263_mv4_search(s, mx, my, shift);
1030  if(dmin4 < dmin){
1031  mb_type= CANDIDATE_MB_TYPE_INTER4V;
1032  dmin=dmin4;
1033  }
1034  }
1036  && !c->skip){ //FIXME varc/d checks
1037  int dmin_i= interlaced_search(s, 0, s->p_field_mv_table, s->p_field_select_table, mx, my, 0);
1038  if(dmin_i < dmin){
1039  mb_type = CANDIDATE_MB_TYPE_INTER_I;
1040  dmin= dmin_i;
1041  }
1042  }
1043 
1044  set_p_mv_tables(s, mx, my, mb_type!=CANDIDATE_MB_TYPE_INTER4V);
1045 
1046  /* get intra luma score */
1047  if((c->avctx->mb_cmp&0xFF)==FF_CMP_SSE){
1048  intra_score= varc - 500;
1049  }else{
1050  unsigned mean = (sum+128)>>8;
1051  mean*= 0x01010101;
1052 
1053  for(i=0; i<16; i++){
1054  *(uint32_t*)(&c->scratchpad[i*s->linesize+ 0]) = mean;
1055  *(uint32_t*)(&c->scratchpad[i*s->linesize+ 4]) = mean;
1056  *(uint32_t*)(&c->scratchpad[i*s->linesize+ 8]) = mean;
1057  *(uint32_t*)(&c->scratchpad[i*s->linesize+12]) = mean;
1058  }
1059 
1060  intra_score= s->mecc.mb_cmp[0](s, c->scratchpad, pix, s->linesize, 16);
1061  }
1062  intra_score += c->mb_penalty_factor*16;
1063 
1064  if(intra_score < dmin){
1065  mb_type= CANDIDATE_MB_TYPE_INTRA;
1066  s->current_picture.mb_type[mb_y*s->mb_stride + mb_x] = CANDIDATE_MB_TYPE_INTRA; //FIXME cleanup
1067  }else
1068  s->current_picture.mb_type[mb_y*s->mb_stride + mb_x] = 0;
1069 
1070  {
1071  int p_score= FFMIN(vard, varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*100);
1072  int i_score= varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*20;
1073  c->scene_change_score+= ff_sqrt(p_score) - ff_sqrt(i_score);
1074  }
1075  }
1076 
1077  s->mb_type[mb_y*s->mb_stride + mb_x]= mb_type;
1078 }
1079 
1081  int mb_x, int mb_y)
1082 {
1083  MotionEstContext * const c= &s->me;
1084  int mx, my, dmin;
1085  int P[10][2];
1086  const int shift= 1+s->quarter_sample;
1087  const int xy= mb_x + mb_y*s->mb_stride;
1088  init_ref(c, s->new_picture.f->data, s->last_picture.f->data, NULL, 16*mb_x, 16*mb_y, 0);
1089 
1090  av_assert0(s->quarter_sample==0 || s->quarter_sample==1);
1091 
1094 
1095  get_limits(s, 16*mb_x, 16*mb_y);
1096  c->skip=0;
1097 
1098  P_LEFT[0] = s->p_mv_table[xy + 1][0];
1099  P_LEFT[1] = s->p_mv_table[xy + 1][1];
1100 
1101  if(P_LEFT[0] < (c->xmin<<shift)) P_LEFT[0] = (c->xmin<<shift);
1102 
1103  /* special case for first line */
1104  if (s->first_slice_line) {
1105  c->pred_x= P_LEFT[0];
1106  c->pred_y= P_LEFT[1];
1107  P_TOP[0]= P_TOPRIGHT[0]= P_MEDIAN[0]=
1108  P_TOP[1]= P_TOPRIGHT[1]= P_MEDIAN[1]= 0; //FIXME
1109  } else {
1110  P_TOP[0] = s->p_mv_table[xy + s->mb_stride ][0];
1111  P_TOP[1] = s->p_mv_table[xy + s->mb_stride ][1];
1112  P_TOPRIGHT[0] = s->p_mv_table[xy + s->mb_stride - 1][0];
1113  P_TOPRIGHT[1] = s->p_mv_table[xy + s->mb_stride - 1][1];
1114  if(P_TOP[1] < (c->ymin<<shift)) P_TOP[1] = (c->ymin<<shift);
1115  if(P_TOPRIGHT[0] > (c->xmax<<shift)) P_TOPRIGHT[0]= (c->xmax<<shift);
1116  if(P_TOPRIGHT[1] < (c->ymin<<shift)) P_TOPRIGHT[1]= (c->ymin<<shift);
1117 
1118  P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
1119  P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
1120 
1121  c->pred_x = P_MEDIAN[0];
1122  c->pred_y = P_MEDIAN[1];
1123  }
1124 
1125  dmin = ff_epzs_motion_search(s, &mx, &my, P, 0, 0, s->p_mv_table, (1<<16)>>shift, 0, 16);
1126 
1127  s->p_mv_table[xy][0] = mx<<shift;
1128  s->p_mv_table[xy][1] = my<<shift;
1129 
1130  return dmin;
1131 }
1132 
1133 static int estimate_motion_b(MpegEncContext *s, int mb_x, int mb_y,
1134  int16_t (*mv_table)[2], int ref_index, int f_code)
1135 {
1136  MotionEstContext * const c= &s->me;
1137  int mx = 0, my = 0, dmin = 0;
1138  int P[10][2];
1139  const int shift= 1+s->quarter_sample;
1140  const int mot_stride = s->mb_stride;
1141  const int mot_xy = mb_y*mot_stride + mb_x;
1142  uint8_t * const mv_penalty= c->mv_penalty[f_code] + MAX_DMV;
1143  int mv_scale;
1144 
1149 
1150  get_limits(s, 16*mb_x, 16*mb_y);
1151 
1152  if (s->motion_est != FF_ME_ZERO) {
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 
1183  dmin= c->sub_motion_search(s, &mx, &my, dmin, 0, ref_index, 0, 16);
1184 
1185  if(c->avctx->me_sub_cmp != c->avctx->mb_cmp && !c->skip)
1186  dmin= get_mb_score(s, mx, my, 0, ref_index, 0, 16, 1);
1187 
1188 // s->mb_type[mb_y*s->mb_width + mb_x]= mb_type;
1189  mv_table[mot_xy][0]= mx;
1190  mv_table[mot_xy][1]= my;
1191 
1192  return dmin;
1193 }
1194 
1195 static inline int check_bidir_mv(MpegEncContext * s,
1196  int motion_fx, int motion_fy,
1197  int motion_bx, int motion_by,
1198  int pred_fx, int pred_fy,
1199  int pred_bx, int pred_by,
1200  int size, int h)
1201 {
1202  //FIXME optimize?
1203  //FIXME better f_code prediction (max mv & distance)
1204  //FIXME pointers
1205  MotionEstContext * const c= &s->me;
1206  uint8_t * const mv_penalty_f= c->mv_penalty[s->f_code] + MAX_DMV; // f_code of the prev frame
1207  uint8_t * const mv_penalty_b= c->mv_penalty[s->b_code] + MAX_DMV; // f_code of the prev frame
1208  int stride= c->stride;
1209  uint8_t *dest_y = c->scratchpad;
1210  uint8_t *ptr;
1211  int dxy;
1212  int src_x, src_y;
1213  int fbmin;
1214  uint8_t **src_data= c->src[0];
1215  uint8_t **ref_data= c->ref[0];
1216  uint8_t **ref2_data= c->ref[2];
1217 
1218  if(s->quarter_sample){
1219  dxy = ((motion_fy & 3) << 2) | (motion_fx & 3);
1220  src_x = motion_fx >> 2;
1221  src_y = motion_fy >> 2;
1222 
1223  ptr = ref_data[0] + (src_y * stride) + src_x;
1224  s->qdsp.put_qpel_pixels_tab[0][dxy](dest_y, ptr, stride);
1225 
1226  dxy = ((motion_by & 3) << 2) | (motion_bx & 3);
1227  src_x = motion_bx >> 2;
1228  src_y = motion_by >> 2;
1229 
1230  ptr = ref2_data[0] + (src_y * stride) + src_x;
1231  s->qdsp.avg_qpel_pixels_tab[size][dxy](dest_y, ptr, stride);
1232  }else{
1233  dxy = ((motion_fy & 1) << 1) | (motion_fx & 1);
1234  src_x = motion_fx >> 1;
1235  src_y = motion_fy >> 1;
1236 
1237  ptr = ref_data[0] + (src_y * stride) + src_x;
1238  s->hdsp.put_pixels_tab[size][dxy](dest_y , ptr , stride, h);
1239 
1240  dxy = ((motion_by & 1) << 1) | (motion_bx & 1);
1241  src_x = motion_bx >> 1;
1242  src_y = motion_by >> 1;
1243 
1244  ptr = ref2_data[0] + (src_y * stride) + src_x;
1245  s->hdsp.avg_pixels_tab[size][dxy](dest_y , ptr , stride, h);
1246  }
1247 
1248  fbmin = (mv_penalty_f[motion_fx-pred_fx] + mv_penalty_f[motion_fy-pred_fy])*c->mb_penalty_factor
1249  +(mv_penalty_b[motion_bx-pred_bx] + mv_penalty_b[motion_by-pred_by])*c->mb_penalty_factor
1250  + s->mecc.mb_cmp[size](s, src_data[0], dest_y, stride, h); // FIXME new_pic
1251 
1252  if(c->avctx->mb_cmp&FF_CMP_CHROMA){
1253  }
1254  //FIXME CHROMA !!!
1255 
1256  return fbmin;
1257 }
1258 
1259 /* refine the bidir vectors in hq mode and return the score in both lq & hq mode*/
1260 static inline int bidir_refine(MpegEncContext * s, int mb_x, int mb_y)
1261 {
1262  MotionEstContext * const c= &s->me;
1263  const int mot_stride = s->mb_stride;
1264  const int xy = mb_y *mot_stride + mb_x;
1265  int fbmin;
1266  int pred_fx= s->b_bidir_forw_mv_table[xy-1][0];
1267  int pred_fy= s->b_bidir_forw_mv_table[xy-1][1];
1268  int pred_bx= s->b_bidir_back_mv_table[xy-1][0];
1269  int pred_by= s->b_bidir_back_mv_table[xy-1][1];
1270  int motion_fx= s->b_bidir_forw_mv_table[xy][0]= s->b_forw_mv_table[xy][0];
1271  int motion_fy= s->b_bidir_forw_mv_table[xy][1]= s->b_forw_mv_table[xy][1];
1272  int motion_bx= s->b_bidir_back_mv_table[xy][0]= s->b_back_mv_table[xy][0];
1273  int motion_by= s->b_bidir_back_mv_table[xy][1]= s->b_back_mv_table[xy][1];
1274  const int flags= c->sub_flags;
1275  const int qpel= flags&FLAG_QPEL;
1276  const int shift= 1+qpel;
1277  const int xmin= c->xmin<<shift;
1278  const int ymin= c->ymin<<shift;
1279  const int xmax= c->xmax<<shift;
1280  const int ymax= c->ymax<<shift;
1281 #define HASH(fx,fy,bx,by) ((fx)+17*(fy)+63*(bx)+117*(by))
1282 #define HASH8(fx,fy,bx,by) ((uint8_t)HASH(fx,fy,bx,by))
1283  int hashidx= HASH(motion_fx,motion_fy, motion_bx, motion_by);
1284  uint8_t map[256] = { 0 };
1285 
1286  map[hashidx&255] = 1;
1287 
1288  fbmin= check_bidir_mv(s, motion_fx, motion_fy,
1289  motion_bx, motion_by,
1290  pred_fx, pred_fy,
1291  pred_bx, pred_by,
1292  0, 16);
1293 
1294  if(s->avctx->bidir_refine){
1295  int end;
1296  static const uint8_t limittab[5]={0,8,32,64,80};
1297  const int limit= limittab[s->avctx->bidir_refine];
1298  static const int8_t vect[][4]={
1299 { 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},
1300 
1301 { 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},
1302 { 0, 1, 0, 1}, { 0,-1, 0,-1}, { 1, 0, 1, 0}, {-1, 0,-1, 0},
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 
1306 { 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},
1307 { 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},
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 
1311 { 1, 1, 1, 1}, {-1,-1,-1,-1},
1312 { 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},
1313 { 1, 1,-1,-1}, {-1,-1, 1, 1}, { 1,-1,-1, 1}, {-1, 1, 1,-1}, { 1,-1, 1,-1}, {-1, 1,-1, 1},
1314  };
1315  static const uint8_t hash[]={
1316 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),
1317 
1318 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),
1319 HASH8( 0, 1, 0, 1), HASH8( 0,-1, 0,-1), HASH8( 1, 0, 1, 0), HASH8(-1, 0,-1, 0),
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 
1323 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),
1324 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),
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 
1328 HASH8( 1, 1, 1, 1), HASH8(-1,-1,-1,-1),
1329 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),
1330 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),
1331 };
1332 
1333 #define CHECK_BIDIR(fx,fy,bx,by)\
1334  if( !map[(hashidx+HASH(fx,fy,bx,by))&255]\
1335  &&(fx<=0 || motion_fx+fx<=xmax) && (fy<=0 || motion_fy+fy<=ymax) && (bx<=0 || motion_bx+bx<=xmax) && (by<=0 || motion_by+by<=ymax)\
1336  &&(fx>=0 || motion_fx+fx>=xmin) && (fy>=0 || motion_fy+fy>=ymin) && (bx>=0 || motion_bx+bx>=xmin) && (by>=0 || motion_by+by>=ymin)){\
1337  int score;\
1338  map[(hashidx+HASH(fx,fy,bx,by))&255] = 1;\
1339  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);\
1340  if(score < fbmin){\
1341  hashidx += HASH(fx,fy,bx,by);\
1342  fbmin= score;\
1343  motion_fx+=fx;\
1344  motion_fy+=fy;\
1345  motion_bx+=bx;\
1346  motion_by+=by;\
1347  end=0;\
1348  }\
1349  }
1350 #define CHECK_BIDIR2(a,b,c,d)\
1351 CHECK_BIDIR(a,b,c,d)\
1352 CHECK_BIDIR(-(a),-(b),-(c),-(d))
1353 
1354  do{
1355  int i;
1356  int borderdist=0;
1357  end=1;
1358 
1359  CHECK_BIDIR2(0,0,0,1)
1360  CHECK_BIDIR2(0,0,1,0)
1361  CHECK_BIDIR2(0,1,0,0)
1362  CHECK_BIDIR2(1,0,0,0)
1363 
1364  for(i=8; i<limit; i++){
1365  int fx= motion_fx+vect[i][0];
1366  int fy= motion_fy+vect[i][1];
1367  int bx= motion_bx+vect[i][2];
1368  int by= motion_by+vect[i][3];
1369  if(borderdist<=0){
1370  int a= (xmax - FFMAX(fx,bx))|(FFMIN(fx,bx) - xmin);
1371  int b= (ymax - FFMAX(fy,by))|(FFMIN(fy,by) - ymin);
1372  if((a|b) < 0)
1373  map[(hashidx+hash[i])&255] = 1;
1374  }
1375  if(!map[(hashidx+hash[i])&255]){
1376  int score;
1377  map[(hashidx+hash[i])&255] = 1;
1378  score= check_bidir_mv(s, fx, fy, bx, by, pred_fx, pred_fy, pred_bx, pred_by, 0, 16);
1379  if(score < fbmin){
1380  hashidx += hash[i];
1381  fbmin= score;
1382  motion_fx=fx;
1383  motion_fy=fy;
1384  motion_bx=bx;
1385  motion_by=by;
1386  end=0;
1387  borderdist--;
1388  if(borderdist<=0){
1389  int a= FFMIN(xmax - FFMAX(fx,bx), FFMIN(fx,bx) - xmin);
1390  int b= FFMIN(ymax - FFMAX(fy,by), FFMIN(fy,by) - ymin);
1391  borderdist= FFMIN(a,b);
1392  }
1393  }
1394  }
1395  }
1396  }while(!end);
1397  }
1398 
1399  s->b_bidir_forw_mv_table[xy][0]= motion_fx;
1400  s->b_bidir_forw_mv_table[xy][1]= motion_fy;
1401  s->b_bidir_back_mv_table[xy][0]= motion_bx;
1402  s->b_bidir_back_mv_table[xy][1]= motion_by;
1403 
1404  return fbmin;
1405 }
1406 
1407 static inline int direct_search(MpegEncContext * s, int mb_x, int mb_y)
1408 {
1409  MotionEstContext * const c= &s->me;
1410  int P[10][2];
1411  const int mot_stride = s->mb_stride;
1412  const int mot_xy = mb_y*mot_stride + mb_x;
1413  const int shift= 1+s->quarter_sample;
1414  int dmin, i;
1415  const int time_pp= s->pp_time;
1416  const int time_pb= s->pb_time;
1417  int mx, my, xmin, xmax, ymin, ymax;
1418  int16_t (*mv_table)[2]= s->b_direct_mv_table;
1419 
1421  ymin= xmin=(-32)>>shift;
1422  ymax= xmax= 31>>shift;
1423 
1424  if (IS_8X8(s->next_picture.mb_type[mot_xy])) {
1425  s->mv_type= MV_TYPE_8X8;
1426  }else{
1427  s->mv_type= MV_TYPE_16X16;
1428  }
1429 
1430  for(i=0; i<4; i++){
1431  int index= s->block_index[i];
1432  int min, max;
1433 
1434  c->co_located_mv[i][0] = s->next_picture.motion_val[0][index][0];
1435  c->co_located_mv[i][1] = s->next_picture.motion_val[0][index][1];
1436  c->direct_basis_mv[i][0]= c->co_located_mv[i][0]*time_pb/time_pp + ((i& 1)<<(shift+3));
1437  c->direct_basis_mv[i][1]= c->co_located_mv[i][1]*time_pb/time_pp + ((i>>1)<<(shift+3));
1438 // c->direct_basis_mv[1][i][0]= c->co_located_mv[i][0]*(time_pb - time_pp)/time_pp + ((i &1)<<(shift+3);
1439 // c->direct_basis_mv[1][i][1]= c->co_located_mv[i][1]*(time_pb - time_pp)/time_pp + ((i>>1)<<(shift+3);
1440 
1441  max= FFMAX(c->direct_basis_mv[i][0], c->direct_basis_mv[i][0] - c->co_located_mv[i][0])>>shift;
1442  min= FFMIN(c->direct_basis_mv[i][0], c->direct_basis_mv[i][0] - c->co_located_mv[i][0])>>shift;
1443  max+= 16*mb_x + 1; // +-1 is for the simpler rounding
1444  min+= 16*mb_x - 1;
1445  xmax= FFMIN(xmax, s->width - max);
1446  xmin= FFMAX(xmin, - 16 - min);
1447 
1448  max= FFMAX(c->direct_basis_mv[i][1], c->direct_basis_mv[i][1] - c->co_located_mv[i][1])>>shift;
1449  min= FFMIN(c->direct_basis_mv[i][1], c->direct_basis_mv[i][1] - c->co_located_mv[i][1])>>shift;
1450  max+= 16*mb_y + 1; // +-1 is for the simpler rounding
1451  min+= 16*mb_y - 1;
1452  ymax= FFMIN(ymax, s->height - max);
1453  ymin= FFMAX(ymin, - 16 - min);
1454 
1455  if(s->mv_type == MV_TYPE_16X16) break;
1456  }
1457 
1458  av_assert2(xmax <= 15 && ymax <= 15 && xmin >= -16 && ymin >= -16);
1459 
1460  if(xmax < 0 || xmin >0 || ymax < 0 || ymin > 0){
1461  s->b_direct_mv_table[mot_xy][0]= 0;
1462  s->b_direct_mv_table[mot_xy][1]= 0;
1463 
1464  return 256*256*256*64;
1465  }
1466 
1467  c->xmin= xmin;
1468  c->ymin= ymin;
1469  c->xmax= xmax;
1470  c->ymax= ymax;
1471  c->flags |= FLAG_DIRECT;
1472  c->sub_flags |= FLAG_DIRECT;
1473  c->pred_x=0;
1474  c->pred_y=0;
1475 
1476  P_LEFT[0] = av_clip(mv_table[mot_xy - 1][0], xmin<<shift, xmax<<shift);
1477  P_LEFT[1] = av_clip(mv_table[mot_xy - 1][1], ymin<<shift, ymax<<shift);
1478 
1479  /* special case for first line */
1480  if (!s->first_slice_line) { //FIXME maybe allow this over thread boundary as it is clipped
1481  P_TOP[0] = av_clip(mv_table[mot_xy - mot_stride ][0], xmin<<shift, xmax<<shift);
1482  P_TOP[1] = av_clip(mv_table[mot_xy - mot_stride ][1], ymin<<shift, ymax<<shift);
1483  P_TOPRIGHT[0] = av_clip(mv_table[mot_xy - mot_stride + 1 ][0], xmin<<shift, xmax<<shift);
1484  P_TOPRIGHT[1] = av_clip(mv_table[mot_xy - mot_stride + 1 ][1], ymin<<shift, ymax<<shift);
1485 
1486  P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
1487  P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
1488  }
1489 
1490  dmin = ff_epzs_motion_search(s, &mx, &my, P, 0, 0, mv_table, 1<<(16-shift), 0, 16);
1491  if(c->sub_flags&FLAG_QPEL)
1492  dmin = qpel_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
1493  else
1494  dmin = hpel_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
1495 
1496  if(c->avctx->me_sub_cmp != c->avctx->mb_cmp && !c->skip)
1497  dmin= get_mb_score(s, mx, my, 0, 0, 0, 16, 1);
1498 
1499  get_limits(s, 16*mb_x, 16*mb_y); //restore c->?min/max, maybe not needed
1500 
1501  mv_table[mot_xy][0]= mx;
1502  mv_table[mot_xy][1]= my;
1503  c->flags &= ~FLAG_DIRECT;
1504  c->sub_flags &= ~FLAG_DIRECT;
1505 
1506  return dmin;
1507 }
1508 
1510  int mb_x, int mb_y)
1511 {
1512  MotionEstContext * const c= &s->me;
1513  const int penalty_factor= c->mb_penalty_factor;
1514  int fmin, bmin, dmin, fbmin, bimin, fimin;
1515  int type=0;
1516  const int xy = mb_y*s->mb_stride + mb_x;
1518  s->next_picture.f->data, 16 * mb_x, 16 * mb_y, 2);
1519 
1520  get_limits(s, 16*mb_x, 16*mb_y);
1521 
1522  c->skip=0;
1523 
1524  if (s->codec_id == AV_CODEC_ID_MPEG4 && s->next_picture.mbskip_table[xy]) {
1525  int score= direct_search(s, mb_x, mb_y); //FIXME just check 0,0
1526 
1527  score= ((unsigned)(score*score + 128*256))>>16;
1528  c->mc_mb_var_sum_temp += score;
1529  s->current_picture.mc_mb_var[mb_y*s->mb_stride + mb_x] = score; //FIXME use SSE
1530  s->mb_type[mb_y*s->mb_stride + mb_x]= CANDIDATE_MB_TYPE_DIRECT0;
1531 
1532  return;
1533  }
1534 
1535  if (s->codec_id == AV_CODEC_ID_MPEG4)
1536  dmin= direct_search(s, mb_x, mb_y);
1537  else
1538  dmin= INT_MAX;
1539 // FIXME penalty stuff for non-MPEG-4
1540  c->skip=0;
1541  fmin = estimate_motion_b(s, mb_x, mb_y, s->b_forw_mv_table, 0, s->f_code) +
1542  3 * penalty_factor;
1543 
1544  c->skip=0;
1545  bmin = estimate_motion_b(s, mb_x, mb_y, s->b_back_mv_table, 2, s->b_code) +
1546  2 * penalty_factor;
1547  ff_dlog(s, " %d %d ", s->b_forw_mv_table[xy][0], s->b_forw_mv_table[xy][1]);
1548 
1549  c->skip=0;
1550  fbmin= bidir_refine(s, mb_x, mb_y) + penalty_factor;
1551  ff_dlog(s, "%d %d %d %d\n", dmin, fmin, bmin, fbmin);
1552 
1554 //FIXME mb type penalty
1555  c->skip=0;
1557  fimin= interlaced_search(s, 0,
1559  s->b_forw_mv_table[xy][0], s->b_forw_mv_table[xy][1], 0);
1561  bimin= interlaced_search(s, 2,
1563  s->b_back_mv_table[xy][0], s->b_back_mv_table[xy][1], 0);
1564  }else
1565  fimin= bimin= INT_MAX;
1566 
1567  {
1568  int score= fmin;
1570 
1571  if (dmin <= score){
1572  score = dmin;
1573  type = CANDIDATE_MB_TYPE_DIRECT;
1574  }
1575  if(bmin<score){
1576  score=bmin;
1578  }
1579  if(fbmin<score){
1580  score=fbmin;
1582  }
1583  if(fimin<score){
1584  score=fimin;
1586  }
1587  if(bimin<score){
1588  score=bimin;
1590  }
1591 
1592  score= ((unsigned)(score*score + 128*256))>>16;
1593  c->mc_mb_var_sum_temp += score;
1594  s->current_picture.mc_mb_var[mb_y*s->mb_stride + mb_x] = score; //FIXME use SSE
1595  }
1596 
1599  if(fimin < INT_MAX)
1601  if(bimin < INT_MAX)
1603  if(fimin < INT_MAX && bimin < INT_MAX){
1604  type |= CANDIDATE_MB_TYPE_BIDIR_I;
1605  }
1606  //FIXME something smarter
1607  if(dmin>256*256*16) type&= ~CANDIDATE_MB_TYPE_DIRECT; //do not try direct mode if it is invalid for this MB
1609  s->mpv_flags & FF_MPV_FLAG_MV0 && *(uint32_t*)s->b_direct_mv_table[xy])
1610  type |= CANDIDATE_MB_TYPE_DIRECT0;
1611  }
1612 
1613  s->mb_type[mb_y*s->mb_stride + mb_x]= type;
1614 }
1615 
1616 /* find best f_code for ME which do unlimited searches */
1617 int ff_get_best_fcode(MpegEncContext * s, int16_t (*mv_table)[2], int type)
1618 {
1619  if (s->motion_est != FF_ME_ZERO) {
1620  int score[8];
1621  int i, y, range= s->avctx->me_range ? s->avctx->me_range : (INT_MAX/2);
1622  uint8_t * fcode_tab= s->fcode_tab;
1623  int best_fcode=-1;
1624  int best_score=-10000000;
1625 
1626  if(s->msmpeg4_version)
1627  range= FFMIN(range, 16);
1629  range= FFMIN(range, 256);
1630 
1631  for(i=0; i<8; i++) score[i]= s->mb_num*(8-i);
1632 
1633  for(y=0; y<s->mb_height; y++){
1634  int x;
1635  int xy= y*s->mb_stride;
1636  for(x=0; x<s->mb_width; x++){
1637  if(s->mb_type[xy] & type){
1638  int mx= mv_table[xy][0];
1639  int my= mv_table[xy][1];
1640  int fcode= FFMAX(fcode_tab[mx + MAX_MV],
1641  fcode_tab[my + MAX_MV]);
1642  int j;
1643 
1644  if(mx >= range || mx < -range ||
1645  my >= range || my < -range)
1646  continue;
1647 
1648  for(j=0; j<fcode && j<8; j++){
1650  score[j]-= 170;
1651  }
1652  }
1653  xy++;
1654  }
1655  }
1656 
1657  for(i=1; i<8; i++){
1658  if(score[i] > best_score){
1659  best_score= score[i];
1660  best_fcode= i;
1661  }
1662  }
1663 
1664  return best_fcode;
1665  }else{
1666  return 1;
1667  }
1668 }
1669 
1671 {
1672  MotionEstContext * const c= &s->me;
1673  const int f_code= s->f_code;
1674  int y, range;
1676 
1677  range = (((s->out_format == FMT_MPEG1 || s->msmpeg4_version) ? 8 : 16) << f_code);
1678 
1679  av_assert0(range <= 16 || !s->msmpeg4_version);
1681 
1682  if(c->avctx->me_range && range > c->avctx->me_range) range= c->avctx->me_range;
1683 
1684  if (s->avctx->flags & AV_CODEC_FLAG_4MV) {
1685  const int wrap= s->b8_stride;
1686 
1687  /* clip / convert to intra 8x8 type MVs */
1688  for(y=0; y<s->mb_height; y++){
1689  int xy= y*2*wrap;
1690  int i= y*s->mb_stride;
1691  int x;
1692 
1693  for(x=0; x<s->mb_width; x++){
1695  int block;
1696  for(block=0; block<4; block++){
1697  int off= (block& 1) + (block>>1)*wrap;
1698  int mx = s->current_picture.motion_val[0][ xy + off ][0];
1699  int my = s->current_picture.motion_val[0][ xy + off ][1];
1700 
1701  if( mx >=range || mx <-range
1702  || my >=range || my <-range){
1703  s->mb_type[i] &= ~CANDIDATE_MB_TYPE_INTER4V;
1706  }
1707  }
1708  }
1709  xy+=2;
1710  i++;
1711  }
1712  }
1713  }
1714 }
1715 
1716 /**
1717  * @param truncate 1 for truncation, 0 for using intra
1718  */
1719 void ff_fix_long_mvs(MpegEncContext * s, uint8_t *field_select_table, int field_select,
1720  int16_t (*mv_table)[2], int f_code, int type, int truncate)
1721 {
1722  MotionEstContext * const c= &s->me;
1723  int y, h_range, v_range;
1724 
1725  // RAL: 8 in MPEG-1, 16 in MPEG-4
1726  int range = (((s->out_format == FMT_MPEG1 || s->msmpeg4_version) ? 8 : 16) << f_code);
1727 
1728  if(c->avctx->me_range && range > c->avctx->me_range) range= c->avctx->me_range;
1729 
1730  h_range= range;
1731  v_range= field_select_table ? range>>1 : range;
1732 
1733  /* clip / convert to intra 16x16 type MVs */
1734  for(y=0; y<s->mb_height; y++){
1735  int x;
1736  int xy= y*s->mb_stride;
1737  for(x=0; x<s->mb_width; x++){
1738  if (s->mb_type[xy] & type){ // RAL: "type" test added...
1739  if (!field_select_table || field_select_table[xy] == field_select) {
1740  if( mv_table[xy][0] >=h_range || mv_table[xy][0] <-h_range
1741  || mv_table[xy][1] >=v_range || mv_table[xy][1] <-v_range){
1742 
1743  if(truncate){
1744  if (mv_table[xy][0] > h_range-1) mv_table[xy][0]= h_range-1;
1745  else if(mv_table[xy][0] < -h_range ) mv_table[xy][0]= -h_range;
1746  if (mv_table[xy][1] > v_range-1) mv_table[xy][1]= v_range-1;
1747  else if(mv_table[xy][1] < -v_range ) mv_table[xy][1]= -v_range;
1748  }else{
1749  s->mb_type[xy] &= ~type;
1750  s->mb_type[xy] |= CANDIDATE_MB_TYPE_INTRA;
1751  mv_table[xy][0]=
1752  mv_table[xy][1]= 0;
1753  }
1754  }
1755  }
1756  }
1757  xy++;
1758  }
1759  }
1760 }
uint8_t * scratchpad
data area for the ME algo, so that the ME does not need to malloc/free.
Definition: motion_est.h:52
static uint8_t mv_penalty[MAX_FCODE+1][MAX_DMV *2+1]
Table of number of bits a motion vector component needs.
Definition: ituh263enc.c:48
#define AV_CODEC_FLAG_INTERLACED_ME
interlaced motion estimation
Definition: avcodec.h:892
static int minima_cmp(const void *a, const void *b)
Definition: motion_est.c:72
static int cmp_qpel(MpegEncContext *s, const int x, const int y, const int subx, const int suby, const int size, const int h, int ref_index, int src_index, me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags)
Definition: motion_est.c:285
#define NULL
Definition: coverity.c:32
static unsigned update_map_generation(MotionEstContext *c)
Definition: motion_est.c:55
void ff_estimate_b_frame_motion(MpegEncContext *s, int mb_x, int mb_y)
Definition: motion_est.c:1509
qpel_mc_func avg_qpel_pixels_tab[2][16]
Definition: qpeldsp.h:74
const char * s
Definition: avisynth_c.h:631
#define P
#define CANDIDATE_MB_TYPE_SKIPPED
Definition: mpegutils.h:109
static int shift(int a, int b)
Definition: sonic.c:82
static av_always_inline void mv_scale(Mv *dst, Mv *src, int td, int tb)
Definition: hevc_mvs.c:114
static int epzs_motion_search2(MpegEncContext *s, int *mx_ptr, int *my_ptr, int P[10][2], int src_index, int ref_index, int16_t(*last_mv)[2], int ref_mv_scale)
static av_always_inline void chroma(WaveformContext *s, AVFrame *in, AVFrame *out, int component, int intensity, int offset_y, int offset_x, int column, int mirror)
Definition: vf_waveform.c:1318
#define P_TOPRIGHT
Definition: motion_est.c:43
static int check_bidir_mv(MpegEncContext *s, int motion_fx, int motion_fy, int motion_bx, int motion_by, int pred_fx, int pred_fy, int pred_bx, int pred_by, int size, int h)
Definition: motion_est.c:1195
int skip
set if ME is skipped for the current MB
Definition: motion_est.h:49
int16_t(* p_mv_table)[2]
MV table (1MV per MB) P-frame encoding.
Definition: mpegvideo.h:245
static void get_limits(MpegEncContext *s, int x, int y)
get fullpel ME search limits.
Definition: motion_est.c:544
uint8_t * fcode_tab
smallest fcode needed for each MV
Definition: mpegvideo.h:279
void ff_estimate_p_frame_motion(MpegEncContext *s, int mb_x, int mb_y)
Definition: motion_est.c:904
uint8_t * mb_mean
Table for MB luminance.
Definition: mpegpicture.h:74
op_pixels_func avg_pixels_tab[4][4]
Halfpel motion compensation with rounding (a+b+1)>>1.
Definition: hpeldsp.h:68
static void init_mv4_ref(MotionEstContext *c)
Definition: motion_est.c:580
qpel_mc_func put_no_rnd_qpel_pixels_tab[2][16]
Definition: qpeldsp.h:75
#define CANDIDATE_MB_TYPE_INTER_I
Definition: mpegutils.h:116
#define P_LEFT
Definition: motion_est.c:41
int ff_epzs_motion_search(struct MpegEncContext *s, int *mx_ptr, int *my_ptr, int P[10][2], int src_index, int ref_index, int16_t(*last_mv)[2], int ref_mv_scale, int size, int h)
uint16_t * mb_var
Table for MB variances.
Definition: mpegpicture.h:65
#define MAX_MV
Definition: motion_est.h:35
uint8_t * current_mv_penalty
Definition: motion_est.h:94
int msmpeg4_version
0=not msmpeg4, 1=mp41, 2=mp42, 3=mp43/divx3 4=wmv1/7 5=wmv2/8
Definition: mpegvideo.h:435
#define CANDIDATE_MB_TYPE_BIDIR
Definition: mpegutils.h:114
enum AVCodecID codec_id
Definition: mpegvideo.h:109
const char * b
Definition: vf_curves.c:109
enhanced predictive zonal search
Definition: avcodec.h:763
static int cmp_simple(MpegEncContext *s, const int x, const int y, int ref_index, int src_index, me_cmp_func cmp_func, me_cmp_func chroma_cmp_func)
Definition: motion_est.c:231
int sub_penalty_factor
Definition: motion_est.h:67
int16_t(*[2][2] p_field_mv_table)[2]
MV table (2MV per MB) interlaced P-frame encoding.
Definition: mpegvideo.h:251
static int bidir_refine(MpegEncContext *s, int mb_x, int mb_y)
Definition: motion_est.c:1260
#define FF_CMP_W97
Definition: avcodec.h:2098
mpegvideo header.
int pre_penalty_factor
Definition: motion_est.h:61
int scene_change_score
Definition: motion_est.h:87
int mpv_flags
flags set by private options
Definition: mpegvideo.h:525
int mb_num
number of MBs of a picture
Definition: mpegvideo.h:130
#define FF_LAMBDA_SHIFT
Definition: avutil.h:217
QpelDSPContext qdsp
Definition: mpegvideo.h:232
uint8_t(* mv_penalty)[MAX_DMV *2+1]
bit amount needed to encode a MV
Definition: motion_est.h:93
me_cmp_func me_pre_cmp[6]
Definition: me_cmp.h:71
int qscale
QP.
Definition: mpegvideo.h:201
int16_t(* b_back_mv_table)[2]
MV table (1MV per MB) backward mode B-frame encoding.
Definition: mpegvideo.h:247
#define P_MV1
Definition: motion_est.c:45
uint8_t * ref[4][4]
Definition: motion_est.h:81
#define FF_ME_ZERO
Definition: motion_est.h:40
static int cmp_internal(MpegEncContext *s, const int x, const int y, const int subx, const int suby, const int size, const int h, int ref_index, int src_index, me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags)
Definition: motion_est.c:247
static int16_t block[64]
Definition: dct.c:113
#define CANDIDATE_MB_TYPE_INTER
Definition: mpegutils.h:107
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
static int zero_cmp(MpegEncContext *s, uint8_t *a, uint8_t *b, ptrdiff_t stride, int h)
Definition: motion_est.c:297
int y
Definition: motion_est.c:68
op_pixels_func(* hpel_put)[4]
Definition: motion_est.h:89
uint8_t
#define ME_MAP_SIZE
Definition: motion_est.h:38
#define av_assert2(cond)
assert() equivalent, that does lie in speed critical code.
Definition: avassert.h:63
me_cmp_func mb_cmp[6]
Definition: me_cmp.h:74
#define CANDIDATE_MB_TYPE_INTER4V
Definition: mpegutils.h:108
enum OutputFormat out_format
output format
Definition: mpegvideo.h:101
int me_range
maximum motion estimation search range in subpel units If 0 then no limit.
Definition: avcodec.h:2171
#define CANDIDATE_MB_TYPE_FORWARD_I
Definition: mpegutils.h:117
#define FLAG_DIRECT
Definition: motion_est.c:81
static av_cold int end(AVCodecContext *avctx)
Definition: avrndec.c:90
int pre_dia_size
ME prepass diamond size & shape.
Definition: avcodec.h:2135
Motion estimation context.
Definition: motion_est.h:47
qpel_mc_func(* qpel_put)[16]
Definition: motion_est.h:91
int no_rounding
apply no rounding to motion compensation (MPEG-4, msmpeg4, ...) for B-frames rounding mode is always ...
Definition: mpegvideo.h:284
#define CANDIDATE_MB_TYPE_BACKWARD_I
Definition: mpegutils.h:118
int me_cmp
motion estimation comparison function
Definition: avcodec.h:2067
Picture current_picture
copy of the current picture structure.
Definition: mpegvideo.h:177
int16_t(* b_bidir_forw_mv_table)[2]
MV table (1MV per MB) bidir mode B-frame encoding.
Definition: mpegvideo.h:248
static int get_flags(MotionEstContext *c, int direct, int chroma)
Definition: motion_est.c:101
#define ff_dlog(a,...)
uint16_t pp_time
time distance between the last 2 p,s,i frames
Definition: mpegvideo.h:390
op_pixels_func(* hpel_avg)[4]
Definition: motion_est.h:90
int mb_height
number of MBs horizontally & vertically
Definition: mpegvideo.h:126
ptrdiff_t size
Definition: opengl_enc.c:101
#define FF_CMP_CHROMA
Definition: avcodec.h:2101
#define FF_CMP_SSE
Definition: avcodec.h:2087
#define CHECK_BIDIR2(a, b, c, d)
#define av_log(a,...)
#define ff_sqrt
Definition: mathops.h:215
uint8_t hash[HASH_SIZE]
Definition: movenc.c:57
#define FF_MPV_FLAG_MV0
Definition: mpegvideo.h:575
no search, that is use 0,0 vector whenever one is needed
Definition: avcodec.h:759
#define MAX_DMV
Definition: motion_est.h:37
reserved for experiments
Definition: avcodec.h:764
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
uint8_t * mbskip_table
Definition: mpegpicture.h:59
int height
Definition: motion_est.c:67
#define AV_CODEC_FLAG_4MV
4 MV per MB allowed / advanced prediction for H.263.
Definition: avcodec.h:834
static const uint16_t mask[17]
Definition: lzw.c:38
static int epzs_motion_search4(MpegEncContext *s, int *mx_ptr, int *my_ptr, int P[10][2], int src_index, int ref_index, int16_t(*last_mv)[2], int ref_mv_scale)
static int no_sub_motion_search(MpegEncContext *s, int *mx_ptr, int *my_ptr, int dmin, int src_index, int ref_index, int size, int h)
#define LOAD_COMMON
int me_sub_cmp
subpixel motion estimation comparison function
Definition: avcodec.h:2073
static uint8_t fcode_tab[MAX_MV *2+1]
Minimal fcode that a motion vector component would need.
Definition: ituh263enc.c:53
int unrestricted_mv
mv can point outside of the coded picture
Definition: mpegvideo.h:220
const char * r
Definition: vf_curves.c:107
static int get_penalty_factor(int lambda, int lambda2, int type)
Definition: motion_est.c:880
#define P_MEDIAN
Definition: motion_est.c:44
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:1744
#define wrap(func)
Definition: neontest.h:65
static void init_interlaced_ref(MpegEncContext *s, int ref_index)
Definition: motion_est.c:745
MpegvideoEncDSPContext mpvencdsp
Definition: mpegvideo.h:230
int quarter_sample
1->qpel, 0->half pel ME/MC
Definition: mpegvideo.h:399
uint16_t * mb_type
Table for candidate MB types for encoding (defines in mpegutils.h)
Definition: mpegvideo.h:291
qpel_mc_func put_qpel_pixels_tab[2][16]
Definition: qpeldsp.h:73
uint8_t *[2][2] b_field_select_table
Definition: mpegvideo.h:254
static const uint8_t offset[127][2]
Definition: vf_spp.c:92
#define FFMAX(a, b)
Definition: common.h:94
int64_t mb_var_sum_temp
Definition: motion_est.h:86
int checked
Definition: motion_est.c:69
uint8_t * src[4][4]
Definition: motion_est.h:80
#define FF_CMP_BIT
Definition: avcodec.h:2091
#define FLAG_CHROMA
Definition: motion_est.c:80
Motion estimation template.
#define av_assert1(cond)
assert() equivalent, that does not lie in speed critical code.
Definition: avassert.h:53
#define FLAG_QPEL
Definition: motion_est.c:79
#define FFMIN(a, b)
Definition: common.h:96
#define ME_MAP_SHIFT
Definition: motion_est.c:47
int me_method
ME algorithm.
Definition: mpegvideo.h:256
static int cmp_fpel_internal(MpegEncContext *s, const int x, const int y, const int size, const int h, int ref_index, int src_index, me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags)
Definition: motion_est.c:237
Picture new_picture
copy of the source picture structure for encoding.
Definition: mpegvideo.h:171
static int hpel_motion_search(MpegEncContext *s, int *mx_ptr, int *my_ptr, int dmin, int src_index, int ref_index, int size, int h)
#define P_TOP
Definition: motion_est.c:42
int(* pix_sum)(uint8_t *pix, int line_size)
int16_t(*[2] motion_val)[2]
Definition: mpegpicture.h:53
Picture.
Definition: mpegpicture.h:45
unsigned map_generation
Definition: motion_est.h:60
#define FF_MB_DECISION_SIMPLE
uses mb_cmp
Definition: avcodec.h:2212
#define FFABS(a)
Absolute value, Note, INT_MIN / INT64_MIN result in undefined behavior as they are not representable ...
Definition: common.h:72
static int interlaced_search(MpegEncContext *s, int ref_index, int16_t(*mv_tables[2][2])[2], uint8_t *field_select_tables[2], int mx, int my, int user_field_select)
Definition: motion_est.c:758
static av_always_inline int cmp(MpegEncContext *s, const int x, const int y, const int subx, const int suby, const int size, const int h, int ref_index, int src_index, me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags)
compares a block (either a full macroblock or a partition thereof) against a proposed motion-compensa...
Definition: motion_est.c:260
MotionEstContext me
Definition: mpegvideo.h:282
int(* me_cmp_func)(struct MpegEncContext *c, uint8_t *blk1, uint8_t *blk2, ptrdiff_t stride, int h)
Definition: me_cmp.h:48
int mb_decision
macroblock decision mode
Definition: avcodec.h:2211
#define src
Definition: vp9dsp.c:530
static int get_mb_score(MpegEncContext *s, int mx, int my, int src_index, int ref_index, int size, int h, int add_rate)
preferred ID for MPEG-1/2 video decoding
Definition: avcodec.h:194
#define FF_CMP_SAD
Definition: avcodec.h:2086
int block_index[6]
index to current MB in block based arrays with edges
Definition: mpegvideo.h:293
static int qpel_motion_search(MpegEncContext *s, int *mx_ptr, int *my_ptr, int dmin, int src_index, int ref_index, int size, int h)
int penalty_factor
an estimate of the bits required to code a given mv value, e.g.
Definition: motion_est.h:62
static int estimate_motion_b(MpegEncContext *s, int mb_x, int mb_y, int16_t(*mv_table)[2], int ref_index, int f_code)
Definition: motion_est.c:1133
#define HASH8(fx, fy, bx, by)
#define MV_TYPE_16X16
1 vector for the whole mb
Definition: mpegvideo.h:266
int first_slice_line
used in MPEG-4 too to handle resync markers
Definition: mpegvideo.h:433
uint16_t * mc_mb_var
Table for motion compensated MB variances.
Definition: mpegpicture.h:68
int bidir_refine
Definition: avcodec.h:2309
#define AV_LOG_INFO
Standard information.
Definition: log.h:187
#define FF_ME_XONE
Definition: motion_est.h:42
unsigned int lambda2
(lambda*lambda) >> FF_LAMBDA_SHIFT
Definition: mpegvideo.h:204
Libavcodec external API header.
#define FF_CMP_RD
Definition: avcodec.h:2092
ptrdiff_t linesize
line size, in bytes, may be different from width
Definition: mpegvideo.h:131
AVCodecContext * avctx
Definition: motion_est.h:48
enum AVCodecID codec_id
Definition: avcodec.h:1666
void ff_set_cmp(MECmpContext *c, me_cmp_func *cmp, int type)
Definition: me_cmp.c:370
#define CHECK_SAD_HALF_MV(suffix, x, y)
Definition: motion_est.c:405
int(* cmp_func)(const void *, const void *)
int height
picture size. must be a multiple of 16
Definition: mpegvideo.h:97
#define FF_CMP_NSSE
Definition: avcodec.h:2096
#define FF_ME_EPZS
Definition: motion_est.h:41
#define FF_CMP_SATD
Definition: avcodec.h:2088
op_pixels_func put_pixels_tab[4][4]
Halfpel motion compensation with rounding (a+b+1)>>1.
Definition: hpeldsp.h:56
static int direct_search(MpegEncContext *s, int mb_x, int mb_y)
Definition: motion_est.c:1407
#define FF_CMP_DCT
Definition: avcodec.h:2089
GLint GLenum type
Definition: opengl_enc.c:105
void ff_fix_long_p_mvs(MpegEncContext *s)
Definition: motion_est.c:1670
op_pixels_func put_no_rnd_pixels_tab[4][4]
Halfpel motion compensation with no rounding (a+b)>>1.
Definition: hpeldsp.h:80
int16_t(*[2][2][2] b_field_mv_table)[2]
MV table (4MV per MB) interlaced B-frame encoding.
Definition: mpegvideo.h:252
int(* pix_norm1)(uint8_t *pix, int line_size)
#define FF_COMPLIANCE_NORMAL
Definition: avcodec.h:2869
int index
Definition: gxfenc.c:89
#define CANDIDATE_MB_TYPE_DIRECT
Definition: mpegutils.h:111
struct AVFrame * f
Definition: mpegpicture.h:46
static av_always_inline int cmp_direct_inline(MpegEncContext *s, const int x, const int y, const int subx, const int suby, const int size, const int h, int ref_index, int src_index, me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, int qpel)
Definition: motion_est.c:107
#define mid_pred
Definition: mathops.h:96
ptrdiff_t uvlinesize
line size, for chroma in bytes, may be different from width
Definition: mpegvideo.h:132
#define MAX_SAB_SIZE
#define FF_CMP_PSNR
Definition: avcodec.h:2090
const VDPAUPixFmtMap * map
static int ff_h263_round_chroma(int x)
Definition: motion_est.h:101
#define FF_CMP_W53
Definition: avcodec.h:2097
#define CANDIDATE_MB_TYPE_BIDIR_I
Definition: mpegutils.h:119
int f_code
forward MV resolution
Definition: mpegvideo.h:235
int ff_pre_estimate_p_frame_motion(MpegEncContext *s, int mb_x, int mb_y)
Definition: motion_est.c:1080
#define CANDIDATE_MB_TYPE_DIRECT0
Definition: mpegutils.h:121
#define ME_MAP_MV_BITS
Definition: motion_est.c:48
#define FF_CMP_DCT264
Definition: avcodec.h:2100
int pict_type
AV_PICTURE_TYPE_I, AV_PICTURE_TYPE_P, AV_PICTURE_TYPE_B, ...
Definition: mpegvideo.h:209
static void set_p_mv_tables(MpegEncContext *s, int mx, int my, int mv4)
Definition: motion_est.c:517
int motion_est
ME algorithm.
Definition: mpegvideo.h:258
static int sad_hpel_motion_search(MpegEncContext *s, int *mx_ptr, int *my_ptr, int dmin, int src_index, int ref_index, int size, int h)
Definition: motion_est.c:412
int16_t(* b_bidir_back_mv_table)[2]
MV table (1MV per MB) bidir mode B-frame encoding.
Definition: mpegvideo.h:249
me_cmp_func me_cmp[6]
Definition: me_cmp.h:72
int ff_init_me(MpegEncContext *s)
Definition: motion_est.c:306
static int flags
Definition: cpu.c:47
uint8_t *[2] p_field_select_table
Definition: mpegvideo.h:253
int16_t(* b_direct_mv_table)[2]
MV table (1MV per MB) direct mode B-frame encoding.
Definition: mpegvideo.h:250
#define AV_CODEC_FLAG_QPEL
Use qpel MC.
Definition: avcodec.h:842
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:198
#define av_builtin_constant_p
Definition: attributes.h:155
qpel_mc_func(* qpel_avg)[16]
Definition: motion_est.h:92
int64_t mc_mb_var_sum_temp
Definition: motion_est.h:85
int16_t(* b_forw_mv_table)[2]
MV table (1MV per MB) forward mode B-frame encoding.
Definition: mpegvideo.h:246
static void zero_hpel(uint8_t *a, const uint8_t *b, ptrdiff_t stride, int h)
Definition: motion_est.c:303
int b8_stride
2*mb_width+1 used for some 8x8 block arrays to allow simple addressing
Definition: mpegvideo.h:128
me_cmp_func sse[6]
Definition: me_cmp.h:57
MpegEncContext.
Definition: mpegvideo.h:78
struct AVCodecContext * avctx
Definition: mpegvideo.h:95
GLint GLenum GLboolean GLsizei stride
Definition: opengl_enc.c:105
int mb_cmp
macroblock comparison function (not supported yet)
Definition: avcodec.h:2079
MECmpContext mecc
Definition: mpegvideo.h:228
int direct_basis_mv[4][2]
Definition: motion_est.h:51
#define FF_DISABLE_DEPRECATION_WARNINGS
Definition: internal.h:80
common internal api header.
int mb_stride
mb_width+1 used for some arrays to allow simple addressing of left & top MBs without sig11 ...
Definition: mpegvideo.h:127
if(ret< 0)
Definition: vf_mcdeint.c:282
static int ref[MAX_W *MAX_W]
Definition: jpeg2000dwt.c:107
#define CANDIDATE_MB_TYPE_FORWARD
Definition: mpegutils.h:112
static double c[64]
Picture last_picture
copy of the previous picture structure.
Definition: mpegvideo.h:159
Bi-dir predicted.
Definition: avutil.h:268
int co_located_mv[4][2]
mv from last P-frame for direct mode ME
Definition: motion_est.h:50
me_cmp_func me_sub_cmp[6]
Definition: me_cmp.h:73
uint32_t * map
map to avoid duplicate evaluations
Definition: motion_est.h:58
static int h263_mv4_search(MpegEncContext *s, int mx, int my, int shift)
Definition: motion_est.c:591
#define CANDIDATE_MB_TYPE_INTRA
Definition: mpegutils.h:106
int dia_size
ME diamond size & shape.
Definition: avcodec.h:2108
#define FF_ENABLE_DEPRECATION_WARNINGS
Definition: internal.h:81
#define IS_8X8(a)
Definition: mpegutils.h:91
static void init_ref(MotionEstContext *c, uint8_t *src[3], uint8_t *ref[3], uint8_t *ref2[3], int x, int y, int ref_index)
Definition: motion_est.c:83
static int cmp_hpel(MpegEncContext *s, const int x, const int y, const int subx, const int suby, const int size, const int h, int ref_index, int src_index, me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags)
Definition: motion_est.c:275
int ff_get_best_fcode(MpegEncContext *s, int16_t(*mv_table)[2], int type)
Definition: motion_est.c:1617
Picture next_picture
copy of the next picture structure.
Definition: mpegvideo.h:165
void ff_fix_long_mvs(MpegEncContext *s, uint8_t *field_select_table, int field_select, int16_t(*mv_table)[2], int f_code, int type, int truncate)
Definition: motion_est.c:1719
uint32_t * mb_type
types and macros are defined in mpegutils.h
Definition: mpegpicture.h:56
#define av_always_inline
Definition: attributes.h:39
uint8_t * temp
Definition: motion_est.h:56
static av_always_inline int cmp_inline(MpegEncContext *s, const int x, const int y, const int subx, const int suby, const int size, const int h, int ref_index, int src_index, me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, int qpel, int chroma)
Definition: motion_est.c:179
#define CANDIDATE_MB_TYPE_BACKWARD
Definition: mpegutils.h:113
#define stride
#define MV_TYPE_8X8
4 vectors (H.263, MPEG-4 4MV)
Definition: mpegvideo.h:267
int b_code
backward MV resolution for B-frames (MPEG-4)
Definition: mpegvideo.h:236
int me_pre_cmp
motion estimation prepass comparison function
Definition: avcodec.h:2128
float min
int(* sub_motion_search)(struct MpegEncContext *s, int *mx_ptr, int *my_ptr, int dmin, int src_index, int ref_index, int size, int h)
Definition: motion_est.h:95
int x
Definition: motion_est.c:68
int strict_std_compliance
strictly follow the standard (MPEG-4, ...).
Definition: avcodec.h:2866
Predicted.
Definition: avutil.h:267
unsigned int lambda
Lagrange multiplier used in rate distortion.
Definition: mpegvideo.h:203
#define HASH(fx, fy, bx, by)
uint16_t pb_time
time distance between the last b and p,s,i frame
Definition: mpegvideo.h:391
HpelDSPContext hdsp
Definition: mpegvideo.h:226