3 * Copyright (c) 2000,2001 Fabrice Bellard
4 * Copyright (c) 2002-2004 Michael Niedermayer
6 * new motion estimation (X1/EPZS) by Michael Niedermayer <michaelni@gmx.at>
8 * This file is part of FFmpeg.
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.
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.
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
36 #include "mpegutils.h"
37 #include "mpegvideo.h"
44 #define P_TOPRIGHT P[3]
48 static int sad_hpel_motion_search(MpegEncContext * s,
49 int *mx_ptr, int *my_ptr, int dmin,
50 int src_index, int ref_index,
53 static inline unsigned update_map_generation(MotionEstContext *c)
55 c->map_generation+= 1<<(ME_MAP_MV_BITS*2);
56 if(c->map_generation==0){
57 c->map_generation= 1<<(ME_MAP_MV_BITS*2);
58 memset(c->map, 0, sizeof(uint32_t)*ME_MAP_SIZE);
60 return c->map_generation;
63 /* shape adaptive search stuff */
64 typedef struct Minima{
70 static int minima_cmp(const void *a, const void *b){
71 const Minima *da = (const Minima *) a;
72 const Minima *db = (const Minima *) b;
74 return da->height - db->height;
77 #define FLAG_QPEL 1 //must be 1
81 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){
82 const int offset[3]= {
84 ((y*c->uvstride + x)>>1),
85 ((y*c->uvstride + x)>>1),
89 c->src[0][i]= src [i] + offset[i];
90 c->ref[0][i]= ref [i] + offset[i];
94 c->ref[ref_index][i]= ref2[i] + offset[i];
99 static int get_flags(MotionEstContext *c, int direct, int chroma){
100 return ((c->avctx->flags&CODEC_FLAG_QPEL) ? FLAG_QPEL : 0)
101 + (direct ? FLAG_DIRECT : 0)
102 + (chroma ? FLAG_CHROMA : 0);
105 static av_always_inline int cmp_direct_inline(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
106 const int size, const int h, int ref_index, int src_index,
107 me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, int qpel){
108 MotionEstContext * const c= &s->me;
109 const int stride= c->stride;
110 const int hx= subx + (x<<(1+qpel));
111 const int hy= suby + (y<<(1+qpel));
112 uint8_t * const * const ref= c->ref[ref_index];
113 uint8_t * const * const src= c->src[src_index];
115 //FIXME check chroma 4mv, (no crashes ...)
116 av_assert2(x >= c->xmin && hx <= c->xmax<<(qpel+1) && y >= c->ymin && hy <= c->ymax<<(qpel+1));
117 if(x >= c->xmin && hx <= c->xmax<<(qpel+1) && y >= c->ymin && hy <= c->ymax<<(qpel+1)){
118 const int time_pp= s->pp_time;
119 const int time_pb= s->pb_time;
120 const int mask= 2*qpel+1;
121 if(s->mv_type==MV_TYPE_8X8){
124 int fx = c->direct_basis_mv[i][0] + hx;
125 int fy = c->direct_basis_mv[i][1] + hy;
126 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));
127 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));
128 int fxy= (fx&mask) + ((fy&mask)<<(qpel+1));
129 int bxy= (bx&mask) + ((by&mask)<<(qpel+1));
131 uint8_t *dst= c->temp + 8*(i&1) + 8*stride*(i>>1);
133 c->qpel_put[1][fxy](dst, ref[0] + (fx>>2) + (fy>>2)*stride, stride);
134 c->qpel_avg[1][bxy](dst, ref[8] + (bx>>2) + (by>>2)*stride, stride);
136 c->hpel_put[1][fxy](dst, ref[0] + (fx>>1) + (fy>>1)*stride, stride, 8);
137 c->hpel_avg[1][bxy](dst, ref[8] + (bx>>1) + (by>>1)*stride, stride, 8);
141 int fx = c->direct_basis_mv[0][0] + hx;
142 int fy = c->direct_basis_mv[0][1] + hy;
143 int bx = hx ? fx - c->co_located_mv[0][0] : (c->co_located_mv[0][0]*(time_pb - time_pp)/time_pp);
144 int by = hy ? fy - c->co_located_mv[0][1] : (c->co_located_mv[0][1]*(time_pb - time_pp)/time_pp);
145 int fxy= (fx&mask) + ((fy&mask)<<(qpel+1));
146 int bxy= (bx&mask) + ((by&mask)<<(qpel+1));
149 c->qpel_put[1][fxy](c->temp , ref[0] + (fx>>2) + (fy>>2)*stride , stride);
150 c->qpel_put[1][fxy](c->temp + 8 , ref[0] + (fx>>2) + (fy>>2)*stride + 8 , stride);
151 c->qpel_put[1][fxy](c->temp + 8*stride, ref[0] + (fx>>2) + (fy>>2)*stride + 8*stride, stride);
152 c->qpel_put[1][fxy](c->temp + 8 + 8*stride, ref[0] + (fx>>2) + (fy>>2)*stride + 8 + 8*stride, stride);
153 c->qpel_avg[1][bxy](c->temp , ref[8] + (bx>>2) + (by>>2)*stride , stride);
154 c->qpel_avg[1][bxy](c->temp + 8 , ref[8] + (bx>>2) + (by>>2)*stride + 8 , stride);
155 c->qpel_avg[1][bxy](c->temp + 8*stride, ref[8] + (bx>>2) + (by>>2)*stride + 8*stride, stride);
156 c->qpel_avg[1][bxy](c->temp + 8 + 8*stride, ref[8] + (bx>>2) + (by>>2)*stride + 8 + 8*stride, stride);
158 av_assert2((fx>>1) + 16*s->mb_x >= -16);
159 av_assert2((fy>>1) + 16*s->mb_y >= -16);
160 av_assert2((fx>>1) + 16*s->mb_x <= s->width);
161 av_assert2((fy>>1) + 16*s->mb_y <= s->height);
162 av_assert2((bx>>1) + 16*s->mb_x >= -16);
163 av_assert2((by>>1) + 16*s->mb_y >= -16);
164 av_assert2((bx>>1) + 16*s->mb_x <= s->width);
165 av_assert2((by>>1) + 16*s->mb_y <= s->height);
167 c->hpel_put[0][fxy](c->temp, ref[0] + (fx>>1) + (fy>>1)*stride, stride, 16);
168 c->hpel_avg[0][bxy](c->temp, ref[8] + (bx>>1) + (by>>1)*stride, stride, 16);
171 d = cmp_func(s, c->temp, src[0], stride, 16);
177 static av_always_inline int cmp_inline(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
178 const int size, const int h, int ref_index, int src_index,
179 me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, int qpel, int chroma){
180 MotionEstContext * const c= &s->me;
181 const int stride= c->stride;
182 const int uvstride= c->uvstride;
183 const int dxy= subx + (suby<<(1+qpel)); //FIXME log2_subpel?
184 const int hx= subx + (x<<(1+qpel));
185 const int hy= suby + (y<<(1+qpel));
186 uint8_t * const * const ref= c->ref[ref_index];
187 uint8_t * const * const src= c->src[src_index];
189 //FIXME check chroma 4mv, (no crashes ...)
190 int uvdxy; /* no, it might not be used uninitialized */
193 c->qpel_put[size][dxy](c->temp, ref[0] + x + y*stride, stride); //FIXME prototype (add h)
199 uvdxy= (cx&1) + 2*(cy&1);
200 //FIXME x/y wrong, but mpeg4 qpel is sick anyway, we should drop as much of it as possible in favor for h264
203 c->hpel_put[size][dxy](c->temp, ref[0] + x + y*stride, stride, h);
205 uvdxy= dxy | (x&1) | (2*(y&1));
207 d = cmp_func(s, c->temp, src[0], stride, h);
209 d = cmp_func(s, src[0], ref[0] + x + y*stride, stride, h);
211 uvdxy= (x&1) + 2*(y&1);
214 uint8_t * const uvtemp= c->temp + 16*stride;
215 c->hpel_put[size+1][uvdxy](uvtemp , ref[1] + (x>>1) + (y>>1)*uvstride, uvstride, h>>1);
216 c->hpel_put[size+1][uvdxy](uvtemp+8, ref[2] + (x>>1) + (y>>1)*uvstride, uvstride, h>>1);
217 d += chroma_cmp_func(s, uvtemp , src[1], uvstride, h>>1);
218 d += chroma_cmp_func(s, uvtemp+8, src[2], uvstride, h>>1);
223 static int cmp_simple(MpegEncContext *s, const int x, const int y,
224 int ref_index, int src_index,
225 me_cmp_func cmp_func, me_cmp_func chroma_cmp_func){
226 return cmp_inline(s,x,y,0,0,0,16,ref_index,src_index, cmp_func, chroma_cmp_func, 0, 0);
229 static int cmp_fpel_internal(MpegEncContext *s, const int x, const int y,
230 const int size, const int h, int ref_index, int src_index,
231 me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
232 if(flags&FLAG_DIRECT){
233 return cmp_direct_inline(s,x,y,0,0,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, flags&FLAG_QPEL);
235 return cmp_inline(s,x,y,0,0,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 0, flags&FLAG_CHROMA);
239 static int cmp_internal(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
240 const int size, const int h, int ref_index, int src_index,
241 me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
242 if(flags&FLAG_DIRECT){
243 return cmp_direct_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, flags&FLAG_QPEL);
245 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);
249 /** @brief compares a block (either a full macroblock or a partition thereof)
250 against a proposed motion-compensated prediction of that block
252 static av_always_inline int cmp(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
253 const int size, const int h, int ref_index, int src_index,
254 me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
255 if(av_builtin_constant_p(flags) && av_builtin_constant_p(h) && av_builtin_constant_p(size)
256 && av_builtin_constant_p(subx) && av_builtin_constant_p(suby)
257 && flags==0 && h==16 && size==0 && subx==0 && suby==0){
258 return cmp_simple(s,x,y,ref_index,src_index, cmp_func, chroma_cmp_func);
259 }else if(av_builtin_constant_p(subx) && av_builtin_constant_p(suby)
260 && subx==0 && suby==0){
261 return cmp_fpel_internal(s,x,y,size,h,ref_index,src_index, cmp_func, chroma_cmp_func,flags);
263 return cmp_internal(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, flags);
267 static int cmp_hpel(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
268 const int size, const int h, int ref_index, int src_index,
269 me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
270 if(flags&FLAG_DIRECT){
271 return cmp_direct_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 0);
273 return cmp_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 0, flags&FLAG_CHROMA);
277 static int cmp_qpel(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
278 const int size, const int h, int ref_index, int src_index,
279 me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
280 if(flags&FLAG_DIRECT){
281 return cmp_direct_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 1);
283 return cmp_inline(s,x,y,subx,suby,size,h,ref_index,src_index, cmp_func, chroma_cmp_func, 1, flags&FLAG_CHROMA);
287 #include "motion_est_template.c"
289 static int zero_cmp(MpegEncContext *s, uint8_t *a, uint8_t *b,
295 static void zero_hpel(uint8_t *a, const uint8_t *b, ptrdiff_t stride, int h){
298 int ff_init_me(MpegEncContext *s){
299 MotionEstContext * const c= &s->me;
300 int cache_size= FFMIN(ME_MAP_SIZE>>ME_MAP_SHIFT, 1<<ME_MAP_SHIFT);
301 int dia_size= FFMAX(FFABS(s->avctx->dia_size)&255, FFABS(s->avctx->pre_dia_size)&255);
303 if(FFMIN(s->avctx->dia_size, s->avctx->pre_dia_size) < -FFMIN(ME_MAP_SIZE, MAX_SAB_SIZE)){
304 av_log(s->avctx, AV_LOG_ERROR, "ME_MAP size is too small for SAB diamond\n");
307 //special case of snow is needed because snow uses its own iterative ME code
308 if(s->me_method!=ME_ZERO && s->me_method!=ME_EPZS && s->me_method!=ME_X1 && s->avctx->codec_id != AV_CODEC_ID_SNOW){
309 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");
315 if(cache_size < 2*dia_size && !c->stride){
316 av_log(s->avctx, AV_LOG_INFO, "ME_MAP size may be a little small for the selected diamond size\n");
319 ff_set_cmp(&s->dsp, s->dsp.me_pre_cmp, c->avctx->me_pre_cmp);
320 ff_set_cmp(&s->dsp, s->dsp.me_cmp, c->avctx->me_cmp);
321 ff_set_cmp(&s->dsp, s->dsp.me_sub_cmp, c->avctx->me_sub_cmp);
322 ff_set_cmp(&s->dsp, s->dsp.mb_cmp, c->avctx->mb_cmp);
324 c->flags = get_flags(c, 0, c->avctx->me_cmp &FF_CMP_CHROMA);
325 c->sub_flags= get_flags(c, 0, c->avctx->me_sub_cmp&FF_CMP_CHROMA);
326 c->mb_flags = get_flags(c, 0, c->avctx->mb_cmp &FF_CMP_CHROMA);
328 /*FIXME s->no_rounding b_type*/
329 if(s->flags&CODEC_FLAG_QPEL){
330 c->sub_motion_search= qpel_motion_search;
331 c->qpel_avg = s->qdsp.avg_qpel_pixels_tab;
333 c->qpel_put = s->qdsp.put_no_rnd_qpel_pixels_tab;
335 c->qpel_put = s->qdsp.put_qpel_pixels_tab;
337 if(c->avctx->me_sub_cmp&FF_CMP_CHROMA)
338 c->sub_motion_search= hpel_motion_search;
339 else if( c->avctx->me_sub_cmp == FF_CMP_SAD
340 && c->avctx-> me_cmp == FF_CMP_SAD
341 && c->avctx-> mb_cmp == FF_CMP_SAD)
342 c->sub_motion_search= sad_hpel_motion_search; // 2050 vs. 2450 cycles
344 c->sub_motion_search= hpel_motion_search;
346 c->hpel_avg = s->hdsp.avg_pixels_tab;
348 c->hpel_put = s->hdsp.put_no_rnd_pixels_tab;
350 c->hpel_put = s->hdsp.put_pixels_tab;
353 c->stride = s->linesize;
354 c->uvstride= s->uvlinesize;
356 c->stride = 16*s->mb_width + 32;
357 c->uvstride= 8*s->mb_width + 16;
360 /* 8x8 fullpel search would need a 4x4 chroma compare, which we do
361 * not have yet, and even if we had, the motion estimation code
362 * does not expect it. */
363 if(s->codec_id != AV_CODEC_ID_SNOW){
364 if((c->avctx->me_cmp&FF_CMP_CHROMA)/* && !s->dsp.me_cmp[2]*/){
365 s->dsp.me_cmp[2]= zero_cmp;
367 if((c->avctx->me_sub_cmp&FF_CMP_CHROMA) && !s->dsp.me_sub_cmp[2]){
368 s->dsp.me_sub_cmp[2]= zero_cmp;
370 c->hpel_put[2][0]= c->hpel_put[2][1]=
371 c->hpel_put[2][2]= c->hpel_put[2][3]= zero_hpel;
374 if(s->codec_id == AV_CODEC_ID_H261){
375 c->sub_motion_search= no_sub_motion_search;
381 #define CHECK_SAD_HALF_MV(suffix, x, y) \
383 d= s->dsp.pix_abs[size][(x?1:0)+(y?2:0)](NULL, pix, ptr+((x)>>1), stride, h);\
384 d += (mv_penalty[pen_x + x] + mv_penalty[pen_y + y])*penalty_factor;\
385 COPY3_IF_LT(dminh, d, dx, x, dy, y)\
388 static int sad_hpel_motion_search(MpegEncContext * s,
389 int *mx_ptr, int *my_ptr, int dmin,
390 int src_index, int ref_index,
393 MotionEstContext * const c= &s->me;
394 const int penalty_factor= c->sub_penalty_factor;
397 int stride= c->stride;
400 av_assert2(c->sub_flags == 0);
408 pix = c->src[src_index][0];
412 ptr = c->ref[ref_index][0] + (my * stride) + mx;
416 if (mx > xmin && mx < xmax &&
417 my > ymin && my < ymax) {
420 const int index= (my<<ME_MAP_SHIFT) + mx;
421 const int t= score_map[(index-(1<<ME_MAP_SHIFT))&(ME_MAP_SIZE-1)];
422 const int l= score_map[(index- 1 )&(ME_MAP_SIZE-1)];
423 const int r= score_map[(index+ 1 )&(ME_MAP_SIZE-1)];
424 const int b= score_map[(index+(1<<ME_MAP_SHIFT))&(ME_MAP_SIZE-1)];
434 CHECK_SAD_HALF_MV(y2 , 0, -1)
436 CHECK_SAD_HALF_MV(xy2, -1, -1)
438 CHECK_SAD_HALF_MV(xy2, +1, -1)
442 CHECK_SAD_HALF_MV(xy2, -1, +1)
444 CHECK_SAD_HALF_MV(x2 , -1, 0)
446 CHECK_SAD_HALF_MV(xy2, +1, -1)
448 CHECK_SAD_HALF_MV(xy2, -1, -1)
452 CHECK_SAD_HALF_MV(xy2, +1, +1)
454 CHECK_SAD_HALF_MV(x2 , +1, 0)
459 CHECK_SAD_HALF_MV(xy2, -1, -1)
463 CHECK_SAD_HALF_MV(xy2, +1, +1)
465 CHECK_SAD_HALF_MV(x2 , -1, 0)
466 CHECK_SAD_HALF_MV(xy2, -1, +1)
469 CHECK_SAD_HALF_MV(xy2, +1, -1)
473 CHECK_SAD_HALF_MV(xy2, -1, +1)
475 CHECK_SAD_HALF_MV(x2 , +1, 0)
476 CHECK_SAD_HALF_MV(xy2, +1, +1)
478 CHECK_SAD_HALF_MV(y2 , 0, +1)
493 static inline void set_p_mv_tables(MpegEncContext * s, int mx, int my, int mv4)
495 const int xy= s->mb_x + s->mb_y*s->mb_stride;
497 s->p_mv_table[xy][0] = mx;
498 s->p_mv_table[xy][1] = my;
500 /* has already been set to the 4 MV if 4MV is done */
502 int mot_xy= s->block_index[0];
504 s->current_picture.motion_val[0][mot_xy ][0] = mx;
505 s->current_picture.motion_val[0][mot_xy ][1] = my;
506 s->current_picture.motion_val[0][mot_xy + 1][0] = mx;
507 s->current_picture.motion_val[0][mot_xy + 1][1] = my;
509 mot_xy += s->b8_stride;
510 s->current_picture.motion_val[0][mot_xy ][0] = mx;
511 s->current_picture.motion_val[0][mot_xy ][1] = my;
512 s->current_picture.motion_val[0][mot_xy + 1][0] = mx;
513 s->current_picture.motion_val[0][mot_xy + 1][1] = my;
518 * get fullpel ME search limits.
520 static inline void get_limits(MpegEncContext *s, int x, int y)
522 MotionEstContext * const c= &s->me;
523 int range= c->avctx->me_range >> (1 + !!(c->flags&FLAG_QPEL));
524 int max_range = MAX_MV >> (1 + !!(c->flags&FLAG_QPEL));
526 if(c->avctx->me_range) c->range= c->avctx->me_range >> 1;
529 if (s->unrestricted_mv) {
532 c->xmax = - x + s->width;
533 c->ymax = - y + s->height;
534 } else if (s->out_format == FMT_H261){
535 // Search range of H261 is different from other codec standards
536 c->xmin = (x > 15) ? - 15 : 0;
537 c->ymin = (y > 15) ? - 15 : 0;
538 c->xmax = (x < s->mb_width * 16 - 16) ? 15 : 0;
539 c->ymax = (y < s->mb_height * 16 - 16) ? 15 : 0;
543 c->xmax = - x + s->mb_width *16 - 16;
544 c->ymax = - y + s->mb_height*16 - 16;
546 if(!range || range > max_range)
549 c->xmin = FFMAX(c->xmin,-range);
550 c->xmax = FFMIN(c->xmax, range);
551 c->ymin = FFMAX(c->ymin,-range);
552 c->ymax = FFMIN(c->ymax, range);
556 static inline void init_mv4_ref(MotionEstContext *c){
557 const int stride= c->stride;
559 c->ref[1][0] = c->ref[0][0] + 8;
560 c->ref[2][0] = c->ref[0][0] + 8*stride;
561 c->ref[3][0] = c->ref[2][0] + 8;
562 c->src[1][0] = c->src[0][0] + 8;
563 c->src[2][0] = c->src[0][0] + 8*stride;
564 c->src[3][0] = c->src[2][0] + 8;
567 static inline int h263_mv4_search(MpegEncContext *s, int mx, int my, int shift)
569 MotionEstContext * const c= &s->me;
574 int dmin_sum=0, mx4_sum=0, my4_sum=0, i;
576 const int stride= c->stride;
577 uint8_t *mv_penalty= c->current_mv_penalty;
578 int saftey_cliping= s->unrestricted_mv && (s->width&15) && (s->height&15);
582 for(block=0; block<4; block++){
584 int pred_x4, pred_y4;
586 static const int off[4]= {2, 1, 1, -1};
587 const int mot_stride = s->b8_stride;
588 const int mot_xy = s->block_index[block];
591 c->xmax = - 16*s->mb_x + s->width - 8*(block &1);
592 c->ymax = - 16*s->mb_y + s->height - 8*(block>>1);
595 P_LEFT[0] = s->current_picture.motion_val[0][mot_xy - 1][0];
596 P_LEFT[1] = s->current_picture.motion_val[0][mot_xy - 1][1];
598 if(P_LEFT[0] > (c->xmax<<shift)) P_LEFT[0] = (c->xmax<<shift);
600 /* special case for first line */
601 if (s->first_slice_line && block<2) {
602 c->pred_x= pred_x4= P_LEFT[0];
603 c->pred_y= pred_y4= P_LEFT[1];
605 P_TOP[0] = s->current_picture.motion_val[0][mot_xy - mot_stride ][0];
606 P_TOP[1] = s->current_picture.motion_val[0][mot_xy - mot_stride ][1];
607 P_TOPRIGHT[0] = s->current_picture.motion_val[0][mot_xy - mot_stride + off[block]][0];
608 P_TOPRIGHT[1] = s->current_picture.motion_val[0][mot_xy - mot_stride + off[block]][1];
609 if(P_TOP[1] > (c->ymax<<shift)) P_TOP[1] = (c->ymax<<shift);
610 if(P_TOPRIGHT[0] < (c->xmin<<shift)) P_TOPRIGHT[0]= (c->xmin<<shift);
611 if(P_TOPRIGHT[0] > (c->xmax<<shift)) P_TOPRIGHT[0]= (c->xmax<<shift);
612 if(P_TOPRIGHT[1] > (c->ymax<<shift)) P_TOPRIGHT[1]= (c->ymax<<shift);
614 P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
615 P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
617 c->pred_x= pred_x4 = P_MEDIAN[0];
618 c->pred_y= pred_y4 = P_MEDIAN[1];
624 if(P[i][0] > (c->xmax<<shift)) P[i][0]= (c->xmax<<shift);
625 if(P[i][1] > (c->ymax<<shift)) P[i][1]= (c->ymax<<shift);
628 dmin4 = epzs_motion_search4(s, &mx4, &my4, P, block, block, s->p_mv_table, (1<<16)>>shift);
630 dmin4= c->sub_motion_search(s, &mx4, &my4, dmin4, block, block, size, h);
632 if(s->dsp.me_sub_cmp[0] != s->dsp.mb_cmp[0]){
634 const int offset= ((block&1) + (block>>1)*stride)*8;
635 uint8_t *dest_y = c->scratchpad + offset;
636 if(s->quarter_sample){
637 uint8_t *ref= c->ref[block][0] + (mx4>>2) + (my4>>2)*stride;
638 dxy = ((my4 & 3) << 2) | (mx4 & 3);
641 s->qdsp.put_no_rnd_qpel_pixels_tab[1][dxy](dest_y, ref, stride);
643 s->qdsp.put_qpel_pixels_tab[1][dxy](dest_y, ref, stride);
645 uint8_t *ref= c->ref[block][0] + (mx4>>1) + (my4>>1)*stride;
646 dxy = ((my4 & 1) << 1) | (mx4 & 1);
649 s->hdsp.put_no_rnd_pixels_tab[1][dxy](dest_y , ref , stride, h);
651 s->hdsp.put_pixels_tab [1][dxy](dest_y , ref , stride, h);
653 dmin_sum+= (mv_penalty[mx4-pred_x4] + mv_penalty[my4-pred_y4])*c->mb_penalty_factor;
657 if(s->quarter_sample){
665 s->current_picture.motion_val[0][s->block_index[block]][0] = mx4;
666 s->current_picture.motion_val[0][s->block_index[block]][1] = my4;
668 if(mx4 != mx || my4 != my) same=0;
674 if(s->dsp.me_sub_cmp[0] != s->dsp.mb_cmp[0]){
675 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);
678 if(c->avctx->mb_cmp&FF_CMP_CHROMA){
683 mx= ff_h263_round_chroma(mx4_sum);
684 my= ff_h263_round_chroma(my4_sum);
685 dxy = ((my & 1) << 1) | (mx & 1);
687 offset= (s->mb_x*8 + (mx>>1)) + (s->mb_y*8 + (my>>1))*s->uvlinesize;
690 s->hdsp.put_no_rnd_pixels_tab[1][dxy](c->scratchpad , s->last_picture.f->data[1] + offset, s->uvlinesize, 8);
691 s->hdsp.put_no_rnd_pixels_tab[1][dxy](c->scratchpad + 8, s->last_picture.f->data[2] + offset, s->uvlinesize, 8);
693 s->hdsp.put_pixels_tab [1][dxy](c->scratchpad , s->last_picture.f->data[1] + offset, s->uvlinesize, 8);
694 s->hdsp.put_pixels_tab [1][dxy](c->scratchpad + 8, s->last_picture.f->data[2] + offset, s->uvlinesize, 8);
697 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);
698 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);
704 switch(c->avctx->mb_cmp&0xFF){
706 return dmin_sum+ 32*s->qscale*s->qscale;*/
710 return dmin_sum+ 11*c->mb_penalty_factor;
714 static inline void init_interlaced_ref(MpegEncContext *s, int ref_index){
715 MotionEstContext * const c= &s->me;
717 c->ref[1+ref_index][0] = c->ref[0+ref_index][0] + s->linesize;
718 c->src[1][0] = c->src[0][0] + s->linesize;
719 if(c->flags & FLAG_CHROMA){
720 c->ref[1+ref_index][1] = c->ref[0+ref_index][1] + s->uvlinesize;
721 c->ref[1+ref_index][2] = c->ref[0+ref_index][2] + s->uvlinesize;
722 c->src[1][1] = c->src[0][1] + s->uvlinesize;
723 c->src[1][2] = c->src[0][2] + s->uvlinesize;
727 static int interlaced_search(MpegEncContext *s, int ref_index,
728 int16_t (*mv_tables[2][2])[2], uint8_t *field_select_tables[2], int mx, int my, int user_field_select)
730 MotionEstContext * const c= &s->me;
735 uint8_t * const mv_penalty= c->current_mv_penalty;
737 const int stride= 2*s->linesize;
739 const int mot_stride= s->mb_stride;
740 const int xy= s->mb_x + s->mb_y*mot_stride;
746 init_interlaced_ref(s, ref_index);
748 for(block=0; block<2; block++){
750 int best_dmin= INT_MAX;
753 for(field_select=0; field_select<2; field_select++){
754 int dmin, mx_i, my_i;
755 int16_t (*mv_table)[2]= mv_tables[block][field_select];
757 if(user_field_select){
758 av_assert1(field_select==0 || field_select==1);
759 av_assert1(field_select_tables[block][xy]==0 || field_select_tables[block][xy]==1);
760 if(field_select_tables[block][xy] != field_select)
764 P_LEFT[0] = mv_table[xy - 1][0];
765 P_LEFT[1] = mv_table[xy - 1][1];
766 if(P_LEFT[0] > (c->xmax<<1)) P_LEFT[0] = (c->xmax<<1);
768 c->pred_x= P_LEFT[0];
769 c->pred_y= P_LEFT[1];
771 if(!s->first_slice_line){
772 P_TOP[0] = mv_table[xy - mot_stride][0];
773 P_TOP[1] = mv_table[xy - mot_stride][1];
774 P_TOPRIGHT[0] = mv_table[xy - mot_stride + 1][0];
775 P_TOPRIGHT[1] = mv_table[xy - mot_stride + 1][1];
776 if(P_TOP[1] > (c->ymax<<1)) P_TOP[1] = (c->ymax<<1);
777 if(P_TOPRIGHT[0] < (c->xmin<<1)) P_TOPRIGHT[0]= (c->xmin<<1);
778 if(P_TOPRIGHT[0] > (c->xmax<<1)) P_TOPRIGHT[0]= (c->xmax<<1);
779 if(P_TOPRIGHT[1] > (c->ymax<<1)) P_TOPRIGHT[1]= (c->ymax<<1);
781 P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
782 P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
784 P_MV1[0]= mx; //FIXME not correct if block != field_select
787 dmin = epzs_motion_search2(s, &mx_i, &my_i, P, block, field_select+ref_index, mv_table, (1<<16)>>1);
789 dmin= c->sub_motion_search(s, &mx_i, &my_i, dmin, block, field_select+ref_index, size, h);
791 mv_table[xy][0]= mx_i;
792 mv_table[xy][1]= my_i;
794 if(s->dsp.me_sub_cmp[0] != s->dsp.mb_cmp[0]){
798 uint8_t *ref= c->ref[field_select+ref_index][0] + (mx_i>>1) + (my_i>>1)*stride;
799 dxy = ((my_i & 1) << 1) | (mx_i & 1);
802 s->hdsp.put_no_rnd_pixels_tab[size][dxy](c->scratchpad, ref , stride, h);
804 s->hdsp.put_pixels_tab [size][dxy](c->scratchpad, ref , stride, h);
806 dmin= s->dsp.mb_cmp[size](s, c->src[block][0], c->scratchpad, stride, h);
807 dmin+= (mv_penalty[mx_i-c->pred_x] + mv_penalty[my_i-c->pred_y] + 1)*c->mb_penalty_factor;
809 dmin+= c->mb_penalty_factor; //field_select bits
811 dmin += field_select != block; //slightly prefer same field
813 if(dmin < best_dmin){
815 best_field= field_select;
819 int16_t (*mv_table)[2]= mv_tables[block][best_field];
821 if(mv_table[xy][0] != mx) same=0; //FIXME check if these checks work and are any good at all
822 if(mv_table[xy][1]&1) same=0;
823 if(mv_table[xy][1]*2 != my) same=0;
824 if(best_field != block) same=0;
827 field_select_tables[block][xy]= best_field;
828 dmin_sum += best_dmin;
839 switch(c->avctx->mb_cmp&0xFF){
841 return dmin_sum+ 32*s->qscale*s->qscale;*/
845 return dmin_sum+ 11*c->mb_penalty_factor;
849 static inline int get_penalty_factor(int lambda, int lambda2, int type){
853 return lambda>>FF_LAMBDA_SHIFT;
855 return (3*lambda)>>(FF_LAMBDA_SHIFT+1);
857 return (4*lambda)>>(FF_LAMBDA_SHIFT);
859 return (2*lambda)>>(FF_LAMBDA_SHIFT);
862 return (2*lambda)>>FF_LAMBDA_SHIFT;
867 return lambda2>>FF_LAMBDA_SHIFT;
873 void ff_estimate_p_frame_motion(MpegEncContext * s,
876 MotionEstContext * const c= &s->me;
878 int sum, mx, my, dmin;
879 int varc; ///< the variance of the block (sum of squared (p[y][x]-average))
880 int vard; ///< sum of squared differences with the estimated motion vector
882 const int shift= 1+s->quarter_sample;
884 Picture * const pic= &s->current_picture;
886 init_ref(c, s->new_picture.f->data, s->last_picture.f->data, NULL, 16*mb_x, 16*mb_y, 0);
888 av_assert0(s->quarter_sample==0 || s->quarter_sample==1);
889 av_assert0(s->linesize == c->stride);
890 av_assert0(s->uvlinesize == c->uvstride);
892 c->penalty_factor = get_penalty_factor(s->lambda, s->lambda2, c->avctx->me_cmp);
893 c->sub_penalty_factor= get_penalty_factor(s->lambda, s->lambda2, c->avctx->me_sub_cmp);
894 c->mb_penalty_factor = get_penalty_factor(s->lambda, s->lambda2, c->avctx->mb_cmp);
895 c->current_mv_penalty= c->mv_penalty[s->f_code] + MAX_MV;
897 get_limits(s, 16*mb_x, 16*mb_y);
900 /* intra / predictive decision */
902 sum = s->dsp.pix_sum(pix, s->linesize);
903 varc = s->dsp.pix_norm1(pix, s->linesize) - (((unsigned)sum*sum)>>8) + 500;
905 pic->mb_mean[s->mb_stride * mb_y + mb_x] = (sum+128)>>8;
906 pic->mb_var [s->mb_stride * mb_y + mb_x] = (varc+128)>>8;
907 c->mb_var_sum_temp += (varc+128)>>8;
909 switch(s->me_method) {
919 const int mot_stride = s->b8_stride;
920 const int mot_xy = s->block_index[0];
922 P_LEFT[0] = s->current_picture.motion_val[0][mot_xy - 1][0];
923 P_LEFT[1] = s->current_picture.motion_val[0][mot_xy - 1][1];
925 if(P_LEFT[0] > (c->xmax<<shift)) P_LEFT[0] = (c->xmax<<shift);
927 if(!s->first_slice_line) {
928 P_TOP[0] = s->current_picture.motion_val[0][mot_xy - mot_stride ][0];
929 P_TOP[1] = s->current_picture.motion_val[0][mot_xy - mot_stride ][1];
930 P_TOPRIGHT[0] = s->current_picture.motion_val[0][mot_xy - mot_stride + 2][0];
931 P_TOPRIGHT[1] = s->current_picture.motion_val[0][mot_xy - mot_stride + 2][1];
932 if(P_TOP[1] > (c->ymax<<shift)) P_TOP[1] = (c->ymax<<shift);
933 if(P_TOPRIGHT[0] < (c->xmin<<shift)) P_TOPRIGHT[0]= (c->xmin<<shift);
934 if(P_TOPRIGHT[1] > (c->ymax<<shift)) P_TOPRIGHT[1]= (c->ymax<<shift);
936 P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
937 P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
939 if(s->out_format == FMT_H263){
940 c->pred_x = P_MEDIAN[0];
941 c->pred_y = P_MEDIAN[1];
942 }else { /* mpeg1 at least */
943 c->pred_x= P_LEFT[0];
944 c->pred_y= P_LEFT[1];
947 c->pred_x= P_LEFT[0];
948 c->pred_y= P_LEFT[1];
952 dmin = ff_epzs_motion_search(s, &mx, &my, P, 0, 0, s->p_mv_table, (1<<16)>>shift, 0, 16);
957 /* At this point (mx,my) are full-pell and the relative displacement */
958 ppix = c->ref[0][0] + (my * s->linesize) + mx;
960 vard = s->dsp.sse[0](NULL, pix, ppix, s->linesize, 16);
962 pic->mc_mb_var[s->mb_stride * mb_y + mb_x] = (vard+128)>>8;
963 c->mc_mb_var_sum_temp += (vard+128)>>8;
966 int p_score= FFMIN(vard, varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*100);
967 int i_score= varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*20;
968 c->scene_change_score+= ff_sqrt(p_score) - ff_sqrt(i_score);
970 if(mb_type == CANDIDATE_MB_TYPE_INTER){
971 c->sub_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
972 set_p_mv_tables(s, mx, my, 1);
977 if(mb_type == CANDIDATE_MB_TYPE_INTER4V){
978 h263_mv4_search(s, mx, my, shift);
980 set_p_mv_tables(s, mx, my, 0);
982 if(mb_type == CANDIDATE_MB_TYPE_INTER_I){
983 interlaced_search(s, 0, s->p_field_mv_table, s->p_field_select_table, mx, my, 1);
985 }else if(c->avctx->mb_decision > FF_MB_DECISION_SIMPLE){
986 int p_score= FFMIN(vard, varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*100);
987 int i_score= varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*20;
988 c->scene_change_score+= ff_sqrt(p_score) - ff_sqrt(i_score);
990 if (vard*2 + 200*256 > varc)
991 mb_type|= CANDIDATE_MB_TYPE_INTRA;
992 if (varc*2 + 200*256 > vard || s->qscale > 24){
993 // if (varc*2 + 200*256 + 50*(s->lambda2>>FF_LAMBDA_SHIFT) > vard){
994 mb_type|= CANDIDATE_MB_TYPE_INTER;
995 c->sub_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
996 if (s->mpv_flags & FF_MPV_FLAG_MV0)
998 mb_type |= CANDIDATE_MB_TYPE_SKIPPED; //FIXME check difference
1003 if((s->flags&CODEC_FLAG_4MV)
1004 && !c->skip && varc>50<<8 && vard>10<<8){
1005 if(h263_mv4_search(s, mx, my, shift) < INT_MAX)
1006 mb_type|=CANDIDATE_MB_TYPE_INTER4V;
1008 set_p_mv_tables(s, mx, my, 0);
1010 set_p_mv_tables(s, mx, my, 1);
1011 if((s->flags&CODEC_FLAG_INTERLACED_ME)
1012 && !c->skip){ //FIXME varc/d checks
1013 if(interlaced_search(s, 0, s->p_field_mv_table, s->p_field_select_table, mx, my, 0) < INT_MAX)
1014 mb_type |= CANDIDATE_MB_TYPE_INTER_I;
1018 mb_type= CANDIDATE_MB_TYPE_INTER;
1020 dmin= c->sub_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
1021 if(c->avctx->me_sub_cmp != c->avctx->mb_cmp && !c->skip)
1022 dmin= get_mb_score(s, mx, my, 0, 0, 0, 16, 1);
1024 if((s->flags&CODEC_FLAG_4MV)
1025 && !c->skip && varc>50<<8 && vard>10<<8){
1026 int dmin4= h263_mv4_search(s, mx, my, shift);
1028 mb_type= CANDIDATE_MB_TYPE_INTER4V;
1032 if((s->flags&CODEC_FLAG_INTERLACED_ME)
1033 && !c->skip){ //FIXME varc/d checks
1034 int dmin_i= interlaced_search(s, 0, s->p_field_mv_table, s->p_field_select_table, mx, my, 0);
1036 mb_type = CANDIDATE_MB_TYPE_INTER_I;
1041 set_p_mv_tables(s, mx, my, mb_type!=CANDIDATE_MB_TYPE_INTER4V);
1043 /* get intra luma score */
1044 if((c->avctx->mb_cmp&0xFF)==FF_CMP_SSE){
1045 intra_score= varc - 500;
1047 unsigned mean = (sum+128)>>8;
1050 for(i=0; i<16; i++){
1051 *(uint32_t*)(&c->scratchpad[i*s->linesize+ 0]) = mean;
1052 *(uint32_t*)(&c->scratchpad[i*s->linesize+ 4]) = mean;
1053 *(uint32_t*)(&c->scratchpad[i*s->linesize+ 8]) = mean;
1054 *(uint32_t*)(&c->scratchpad[i*s->linesize+12]) = mean;
1057 intra_score= s->dsp.mb_cmp[0](s, c->scratchpad, pix, s->linesize, 16);
1059 intra_score += c->mb_penalty_factor*16;
1061 if(intra_score < dmin){
1062 mb_type= CANDIDATE_MB_TYPE_INTRA;
1063 s->current_picture.mb_type[mb_y*s->mb_stride + mb_x] = CANDIDATE_MB_TYPE_INTRA; //FIXME cleanup
1065 s->current_picture.mb_type[mb_y*s->mb_stride + mb_x] = 0;
1068 int p_score= FFMIN(vard, varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*100);
1069 int i_score= varc-500+(s->lambda2>>FF_LAMBDA_SHIFT)*20;
1070 c->scene_change_score+= ff_sqrt(p_score) - ff_sqrt(i_score);
1074 s->mb_type[mb_y*s->mb_stride + mb_x]= mb_type;
1077 int ff_pre_estimate_p_frame_motion(MpegEncContext * s,
1080 MotionEstContext * const c= &s->me;
1083 const int shift= 1+s->quarter_sample;
1084 const int xy= mb_x + mb_y*s->mb_stride;
1085 init_ref(c, s->new_picture.f->data, s->last_picture.f->data, NULL, 16*mb_x, 16*mb_y, 0);
1087 av_assert0(s->quarter_sample==0 || s->quarter_sample==1);
1089 c->pre_penalty_factor = get_penalty_factor(s->lambda, s->lambda2, c->avctx->me_pre_cmp);
1090 c->current_mv_penalty= c->mv_penalty[s->f_code] + MAX_MV;
1092 get_limits(s, 16*mb_x, 16*mb_y);
1095 P_LEFT[0] = s->p_mv_table[xy + 1][0];
1096 P_LEFT[1] = s->p_mv_table[xy + 1][1];
1098 if(P_LEFT[0] < (c->xmin<<shift)) P_LEFT[0] = (c->xmin<<shift);
1100 /* special case for first line */
1101 if (s->first_slice_line) {
1102 c->pred_x= P_LEFT[0];
1103 c->pred_y= P_LEFT[1];
1104 P_TOP[0]= P_TOPRIGHT[0]= P_MEDIAN[0]=
1105 P_TOP[1]= P_TOPRIGHT[1]= P_MEDIAN[1]= 0; //FIXME
1107 P_TOP[0] = s->p_mv_table[xy + s->mb_stride ][0];
1108 P_TOP[1] = s->p_mv_table[xy + s->mb_stride ][1];
1109 P_TOPRIGHT[0] = s->p_mv_table[xy + s->mb_stride - 1][0];
1110 P_TOPRIGHT[1] = s->p_mv_table[xy + s->mb_stride - 1][1];
1111 if(P_TOP[1] < (c->ymin<<shift)) P_TOP[1] = (c->ymin<<shift);
1112 if(P_TOPRIGHT[0] > (c->xmax<<shift)) P_TOPRIGHT[0]= (c->xmax<<shift);
1113 if(P_TOPRIGHT[1] < (c->ymin<<shift)) P_TOPRIGHT[1]= (c->ymin<<shift);
1115 P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
1116 P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
1118 c->pred_x = P_MEDIAN[0];
1119 c->pred_y = P_MEDIAN[1];
1122 dmin = ff_epzs_motion_search(s, &mx, &my, P, 0, 0, s->p_mv_table, (1<<16)>>shift, 0, 16);
1124 s->p_mv_table[xy][0] = mx<<shift;
1125 s->p_mv_table[xy][1] = my<<shift;
1130 static int estimate_motion_b(MpegEncContext *s, int mb_x, int mb_y,
1131 int16_t (*mv_table)[2], int ref_index, int f_code)
1133 MotionEstContext * const c= &s->me;
1136 const int shift= 1+s->quarter_sample;
1137 const int mot_stride = s->mb_stride;
1138 const int mot_xy = mb_y*mot_stride + mb_x;
1139 uint8_t * const mv_penalty= c->mv_penalty[f_code] + MAX_MV;
1142 c->penalty_factor = get_penalty_factor(s->lambda, s->lambda2, c->avctx->me_cmp);
1143 c->sub_penalty_factor= get_penalty_factor(s->lambda, s->lambda2, c->avctx->me_sub_cmp);
1144 c->mb_penalty_factor = get_penalty_factor(s->lambda, s->lambda2, c->avctx->mb_cmp);
1145 c->current_mv_penalty= mv_penalty;
1147 get_limits(s, 16*mb_x, 16*mb_y);
1149 switch(s->me_method) {
1158 P_LEFT[0] = mv_table[mot_xy - 1][0];
1159 P_LEFT[1] = mv_table[mot_xy - 1][1];
1161 if (P_LEFT[0] > (c->xmax << shift)) P_LEFT[0] = (c->xmax << shift);
1163 /* special case for first line */
1164 if (!s->first_slice_line) {
1165 P_TOP[0] = mv_table[mot_xy - mot_stride ][0];
1166 P_TOP[1] = mv_table[mot_xy - mot_stride ][1];
1167 P_TOPRIGHT[0] = mv_table[mot_xy - mot_stride + 1][0];
1168 P_TOPRIGHT[1] = mv_table[mot_xy - mot_stride + 1][1];
1169 if (P_TOP[1] > (c->ymax << shift)) P_TOP[1] = (c->ymax << shift);
1170 if (P_TOPRIGHT[0] < (c->xmin << shift)) P_TOPRIGHT[0] = (c->xmin << shift);
1171 if (P_TOPRIGHT[1] > (c->ymax << shift)) P_TOPRIGHT[1] = (c->ymax << shift);
1173 P_MEDIAN[0] = mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
1174 P_MEDIAN[1] = mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
1176 c->pred_x = P_LEFT[0];
1177 c->pred_y = P_LEFT[1];
1179 if(mv_table == s->b_forw_mv_table){
1180 mv_scale= (s->pb_time<<16) / (s->pp_time<<shift);
1182 mv_scale= ((s->pb_time - s->pp_time)<<16) / (s->pp_time<<shift);
1185 dmin = ff_epzs_motion_search(s, &mx, &my, P, 0, ref_index, s->p_mv_table, mv_scale, 0, 16);
1190 dmin= c->sub_motion_search(s, &mx, &my, dmin, 0, ref_index, 0, 16);
1192 if(c->avctx->me_sub_cmp != c->avctx->mb_cmp && !c->skip)
1193 dmin= get_mb_score(s, mx, my, 0, ref_index, 0, 16, 1);
1195 // s->mb_type[mb_y*s->mb_width + mb_x]= mb_type;
1196 mv_table[mot_xy][0]= mx;
1197 mv_table[mot_xy][1]= my;
1202 static inline int check_bidir_mv(MpegEncContext * s,
1203 int motion_fx, int motion_fy,
1204 int motion_bx, int motion_by,
1205 int pred_fx, int pred_fy,
1206 int pred_bx, int pred_by,
1210 //FIXME better f_code prediction (max mv & distance)
1212 MotionEstContext * const c= &s->me;
1213 uint8_t * const mv_penalty_f= c->mv_penalty[s->f_code] + MAX_MV; // f_code of the prev frame
1214 uint8_t * const mv_penalty_b= c->mv_penalty[s->b_code] + MAX_MV; // f_code of the prev frame
1215 int stride= c->stride;
1216 uint8_t *dest_y = c->scratchpad;
1221 uint8_t **src_data= c->src[0];
1222 uint8_t **ref_data= c->ref[0];
1223 uint8_t **ref2_data= c->ref[2];
1225 if(s->quarter_sample){
1226 dxy = ((motion_fy & 3) << 2) | (motion_fx & 3);
1227 src_x = motion_fx >> 2;
1228 src_y = motion_fy >> 2;
1230 ptr = ref_data[0] + (src_y * stride) + src_x;
1231 s->qdsp.put_qpel_pixels_tab[0][dxy](dest_y, ptr, stride);
1233 dxy = ((motion_by & 3) << 2) | (motion_bx & 3);
1234 src_x = motion_bx >> 2;
1235 src_y = motion_by >> 2;
1237 ptr = ref2_data[0] + (src_y * stride) + src_x;
1238 s->qdsp.avg_qpel_pixels_tab[size][dxy](dest_y, ptr, stride);
1240 dxy = ((motion_fy & 1) << 1) | (motion_fx & 1);
1241 src_x = motion_fx >> 1;
1242 src_y = motion_fy >> 1;
1244 ptr = ref_data[0] + (src_y * stride) + src_x;
1245 s->hdsp.put_pixels_tab[size][dxy](dest_y , ptr , stride, h);
1247 dxy = ((motion_by & 1) << 1) | (motion_bx & 1);
1248 src_x = motion_bx >> 1;
1249 src_y = motion_by >> 1;
1251 ptr = ref2_data[0] + (src_y * stride) + src_x;
1252 s->hdsp.avg_pixels_tab[size][dxy](dest_y , ptr , stride, h);
1255 fbmin = (mv_penalty_f[motion_fx-pred_fx] + mv_penalty_f[motion_fy-pred_fy])*c->mb_penalty_factor
1256 +(mv_penalty_b[motion_bx-pred_bx] + mv_penalty_b[motion_by-pred_by])*c->mb_penalty_factor
1257 + s->dsp.mb_cmp[size](s, src_data[0], dest_y, stride, h); //FIXME new_pic
1259 if(c->avctx->mb_cmp&FF_CMP_CHROMA){
1266 /* refine the bidir vectors in hq mode and return the score in both lq & hq mode*/
1267 static inline int bidir_refine(MpegEncContext * s, int mb_x, int mb_y)
1269 MotionEstContext * const c= &s->me;
1270 const int mot_stride = s->mb_stride;
1271 const int xy = mb_y *mot_stride + mb_x;
1273 int pred_fx= s->b_bidir_forw_mv_table[xy-1][0];
1274 int pred_fy= s->b_bidir_forw_mv_table[xy-1][1];
1275 int pred_bx= s->b_bidir_back_mv_table[xy-1][0];
1276 int pred_by= s->b_bidir_back_mv_table[xy-1][1];
1277 int motion_fx= s->b_bidir_forw_mv_table[xy][0]= s->b_forw_mv_table[xy][0];
1278 int motion_fy= s->b_bidir_forw_mv_table[xy][1]= s->b_forw_mv_table[xy][1];
1279 int motion_bx= s->b_bidir_back_mv_table[xy][0]= s->b_back_mv_table[xy][0];
1280 int motion_by= s->b_bidir_back_mv_table[xy][1]= s->b_back_mv_table[xy][1];
1281 const int flags= c->sub_flags;
1282 const int qpel= flags&FLAG_QPEL;
1283 const int shift= 1+qpel;
1284 const int xmin= c->xmin<<shift;
1285 const int ymin= c->ymin<<shift;
1286 const int xmax= c->xmax<<shift;
1287 const int ymax= c->ymax<<shift;
1288 #define HASH(fx,fy,bx,by) ((fx)+17*(fy)+63*(bx)+117*(by))
1289 #define HASH8(fx,fy,bx,by) ((uint8_t)HASH(fx,fy,bx,by))
1290 int hashidx= HASH(motion_fx,motion_fy, motion_bx, motion_by);
1291 uint8_t map[256] = { 0 };
1293 map[hashidx&255] = 1;
1295 fbmin= check_bidir_mv(s, motion_fx, motion_fy,
1296 motion_bx, motion_by,
1301 if(s->avctx->bidir_refine){
1303 static const uint8_t limittab[5]={0,8,32,64,80};
1304 const int limit= limittab[s->avctx->bidir_refine];
1305 static const int8_t vect[][4]={
1306 { 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},
1308 { 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},
1309 { 0, 1, 0, 1}, { 0,-1, 0,-1}, { 1, 0, 1, 0}, {-1, 0,-1, 0},
1310 { 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},
1311 { 0,-1, 0, 1}, { 0, 1, 0,-1}, {-1, 0, 1, 0}, { 1, 0,-1, 0},
1313 { 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},
1314 { 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},
1315 { 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},
1316 { 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},
1318 { 1, 1, 1, 1}, {-1,-1,-1,-1},
1319 { 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},
1320 { 1, 1,-1,-1}, {-1,-1, 1, 1}, { 1,-1,-1, 1}, {-1, 1, 1,-1}, { 1,-1, 1,-1}, {-1, 1,-1, 1},
1322 static const uint8_t hash[]={
1323 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),
1325 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),
1326 HASH8( 0, 1, 0, 1), HASH8( 0,-1, 0,-1), HASH8( 1, 0, 1, 0), HASH8(-1, 0,-1, 0),
1327 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),
1328 HASH8( 0,-1, 0, 1), HASH8( 0, 1, 0,-1), HASH8(-1, 0, 1, 0), HASH8( 1, 0,-1, 0),
1330 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),
1331 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),
1332 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),
1333 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),
1335 HASH8( 1, 1, 1, 1), HASH8(-1,-1,-1,-1),
1336 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),
1337 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),
1340 #define CHECK_BIDIR(fx,fy,bx,by)\
1341 if( !map[(hashidx+HASH(fx,fy,bx,by))&255]\
1342 &&(fx<=0 || motion_fx+fx<=xmax) && (fy<=0 || motion_fy+fy<=ymax) && (bx<=0 || motion_bx+bx<=xmax) && (by<=0 || motion_by+by<=ymax)\
1343 &&(fx>=0 || motion_fx+fx>=xmin) && (fy>=0 || motion_fy+fy>=ymin) && (bx>=0 || motion_bx+bx>=xmin) && (by>=0 || motion_by+by>=ymin)){\
1345 map[(hashidx+HASH(fx,fy,bx,by))&255] = 1;\
1346 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);\
1348 hashidx += HASH(fx,fy,bx,by);\
1357 #define CHECK_BIDIR2(a,b,c,d)\
1358 CHECK_BIDIR(a,b,c,d)\
1359 CHECK_BIDIR(-(a),-(b),-(c),-(d))
1366 CHECK_BIDIR2(0,0,0,1)
1367 CHECK_BIDIR2(0,0,1,0)
1368 CHECK_BIDIR2(0,1,0,0)
1369 CHECK_BIDIR2(1,0,0,0)
1371 for(i=8; i<limit; i++){
1372 int fx= motion_fx+vect[i][0];
1373 int fy= motion_fy+vect[i][1];
1374 int bx= motion_bx+vect[i][2];
1375 int by= motion_by+vect[i][3];
1377 int a= (xmax - FFMAX(fx,bx))|(FFMIN(fx,bx) - xmin);
1378 int b= (ymax - FFMAX(fy,by))|(FFMIN(fy,by) - ymin);
1380 map[(hashidx+hash[i])&255] = 1;
1382 if(!map[(hashidx+hash[i])&255]){
1384 map[(hashidx+hash[i])&255] = 1;
1385 score= check_bidir_mv(s, fx, fy, bx, by, pred_fx, pred_fy, pred_bx, pred_by, 0, 16);
1396 int a= FFMIN(xmax - FFMAX(fx,bx), FFMIN(fx,bx) - xmin);
1397 int b= FFMIN(ymax - FFMAX(fy,by), FFMIN(fy,by) - ymin);
1398 borderdist= FFMIN(a,b);
1406 s->b_bidir_forw_mv_table[xy][0]= motion_fx;
1407 s->b_bidir_forw_mv_table[xy][1]= motion_fy;
1408 s->b_bidir_back_mv_table[xy][0]= motion_bx;
1409 s->b_bidir_back_mv_table[xy][1]= motion_by;
1414 static inline int direct_search(MpegEncContext * s, int mb_x, int mb_y)
1416 MotionEstContext * const c= &s->me;
1418 const int mot_stride = s->mb_stride;
1419 const int mot_xy = mb_y*mot_stride + mb_x;
1420 const int shift= 1+s->quarter_sample;
1422 const int time_pp= s->pp_time;
1423 const int time_pb= s->pb_time;
1424 int mx, my, xmin, xmax, ymin, ymax;
1425 int16_t (*mv_table)[2]= s->b_direct_mv_table;
1427 c->current_mv_penalty= c->mv_penalty[1] + MAX_MV;
1428 ymin= xmin=(-32)>>shift;
1429 ymax= xmax= 31>>shift;
1431 if (IS_8X8(s->next_picture.mb_type[mot_xy])) {
1432 s->mv_type= MV_TYPE_8X8;
1434 s->mv_type= MV_TYPE_16X16;
1438 int index= s->block_index[i];
1441 c->co_located_mv[i][0] = s->next_picture.motion_val[0][index][0];
1442 c->co_located_mv[i][1] = s->next_picture.motion_val[0][index][1];
1443 c->direct_basis_mv[i][0]= c->co_located_mv[i][0]*time_pb/time_pp + ((i& 1)<<(shift+3));
1444 c->direct_basis_mv[i][1]= c->co_located_mv[i][1]*time_pb/time_pp + ((i>>1)<<(shift+3));
1445 // c->direct_basis_mv[1][i][0]= c->co_located_mv[i][0]*(time_pb - time_pp)/time_pp + ((i &1)<<(shift+3);
1446 // c->direct_basis_mv[1][i][1]= c->co_located_mv[i][1]*(time_pb - time_pp)/time_pp + ((i>>1)<<(shift+3);
1448 max= FFMAX(c->direct_basis_mv[i][0], c->direct_basis_mv[i][0] - c->co_located_mv[i][0])>>shift;
1449 min= FFMIN(c->direct_basis_mv[i][0], c->direct_basis_mv[i][0] - c->co_located_mv[i][0])>>shift;
1450 max+= 16*mb_x + 1; // +-1 is for the simpler rounding
1452 xmax= FFMIN(xmax, s->width - max);
1453 xmin= FFMAX(xmin, - 16 - min);
1455 max= FFMAX(c->direct_basis_mv[i][1], c->direct_basis_mv[i][1] - c->co_located_mv[i][1])>>shift;
1456 min= FFMIN(c->direct_basis_mv[i][1], c->direct_basis_mv[i][1] - c->co_located_mv[i][1])>>shift;
1457 max+= 16*mb_y + 1; // +-1 is for the simpler rounding
1459 ymax= FFMIN(ymax, s->height - max);
1460 ymin= FFMAX(ymin, - 16 - min);
1462 if(s->mv_type == MV_TYPE_16X16) break;
1465 av_assert2(xmax <= 15 && ymax <= 15 && xmin >= -16 && ymin >= -16);
1467 if(xmax < 0 || xmin >0 || ymax < 0 || ymin > 0){
1468 s->b_direct_mv_table[mot_xy][0]= 0;
1469 s->b_direct_mv_table[mot_xy][1]= 0;
1471 return 256*256*256*64;
1478 c->flags |= FLAG_DIRECT;
1479 c->sub_flags |= FLAG_DIRECT;
1483 P_LEFT[0] = av_clip(mv_table[mot_xy - 1][0], xmin<<shift, xmax<<shift);
1484 P_LEFT[1] = av_clip(mv_table[mot_xy - 1][1], ymin<<shift, ymax<<shift);
1486 /* special case for first line */
1487 if (!s->first_slice_line) { //FIXME maybe allow this over thread boundary as it is clipped
1488 P_TOP[0] = av_clip(mv_table[mot_xy - mot_stride ][0], xmin<<shift, xmax<<shift);
1489 P_TOP[1] = av_clip(mv_table[mot_xy - mot_stride ][1], ymin<<shift, ymax<<shift);
1490 P_TOPRIGHT[0] = av_clip(mv_table[mot_xy - mot_stride + 1 ][0], xmin<<shift, xmax<<shift);
1491 P_TOPRIGHT[1] = av_clip(mv_table[mot_xy - mot_stride + 1 ][1], ymin<<shift, ymax<<shift);
1493 P_MEDIAN[0]= mid_pred(P_LEFT[0], P_TOP[0], P_TOPRIGHT[0]);
1494 P_MEDIAN[1]= mid_pred(P_LEFT[1], P_TOP[1], P_TOPRIGHT[1]);
1497 dmin = ff_epzs_motion_search(s, &mx, &my, P, 0, 0, mv_table, 1<<(16-shift), 0, 16);
1498 if(c->sub_flags&FLAG_QPEL)
1499 dmin = qpel_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
1501 dmin = hpel_motion_search(s, &mx, &my, dmin, 0, 0, 0, 16);
1503 if(c->avctx->me_sub_cmp != c->avctx->mb_cmp && !c->skip)
1504 dmin= get_mb_score(s, mx, my, 0, 0, 0, 16, 1);
1506 get_limits(s, 16*mb_x, 16*mb_y); //restore c->?min/max, maybe not needed
1508 mv_table[mot_xy][0]= mx;
1509 mv_table[mot_xy][1]= my;
1510 c->flags &= ~FLAG_DIRECT;
1511 c->sub_flags &= ~FLAG_DIRECT;
1516 void ff_estimate_b_frame_motion(MpegEncContext * s,
1519 MotionEstContext * const c= &s->me;
1520 const int penalty_factor= c->mb_penalty_factor;
1521 int fmin, bmin, dmin, fbmin, bimin, fimin;
1523 const int xy = mb_y*s->mb_stride + mb_x;
1524 init_ref(c, s->new_picture.f->data, s->last_picture.f->data,
1525 s->next_picture.f->data, 16 * mb_x, 16 * mb_y, 2);
1527 get_limits(s, 16*mb_x, 16*mb_y);
1531 if (s->codec_id == AV_CODEC_ID_MPEG4 && s->next_picture.mbskip_table[xy]) {
1532 int score= direct_search(s, mb_x, mb_y); //FIXME just check 0,0
1534 score= ((unsigned)(score*score + 128*256))>>16;
1535 c->mc_mb_var_sum_temp += score;
1536 s->current_picture.mc_mb_var[mb_y*s->mb_stride + mb_x] = score; //FIXME use SSE
1537 s->mb_type[mb_y*s->mb_stride + mb_x]= CANDIDATE_MB_TYPE_DIRECT0;
1542 if (s->codec_id == AV_CODEC_ID_MPEG4)
1543 dmin= direct_search(s, mb_x, mb_y);
1546 //FIXME penalty stuff for non mpeg4
1548 fmin = estimate_motion_b(s, mb_x, mb_y, s->b_forw_mv_table, 0, s->f_code) +
1552 bmin = estimate_motion_b(s, mb_x, mb_y, s->b_back_mv_table, 2, s->b_code) +
1554 av_dlog(s, " %d %d ", s->b_forw_mv_table[xy][0], s->b_forw_mv_table[xy][1]);
1557 fbmin= bidir_refine(s, mb_x, mb_y) + penalty_factor;
1558 av_dlog(s, "%d %d %d %d\n", dmin, fmin, bmin, fbmin);
1560 if(s->flags & CODEC_FLAG_INTERLACED_ME){
1561 //FIXME mb type penalty
1563 c->current_mv_penalty= c->mv_penalty[s->f_code] + MAX_MV;
1564 fimin= interlaced_search(s, 0,
1565 s->b_field_mv_table[0], s->b_field_select_table[0],
1566 s->b_forw_mv_table[xy][0], s->b_forw_mv_table[xy][1], 0);
1567 c->current_mv_penalty= c->mv_penalty[s->b_code] + MAX_MV;
1568 bimin= interlaced_search(s, 2,
1569 s->b_field_mv_table[1], s->b_field_select_table[1],
1570 s->b_back_mv_table[xy][0], s->b_back_mv_table[xy][1], 0);
1572 fimin= bimin= INT_MAX;
1576 type = CANDIDATE_MB_TYPE_FORWARD;
1580 type = CANDIDATE_MB_TYPE_DIRECT;
1584 type= CANDIDATE_MB_TYPE_BACKWARD;
1588 type= CANDIDATE_MB_TYPE_BIDIR;
1592 type= CANDIDATE_MB_TYPE_FORWARD_I;
1596 type= CANDIDATE_MB_TYPE_BACKWARD_I;
1599 score= ((unsigned)(score*score + 128*256))>>16;
1600 c->mc_mb_var_sum_temp += score;
1601 s->current_picture.mc_mb_var[mb_y*s->mb_stride + mb_x] = score; //FIXME use SSE
1604 if(c->avctx->mb_decision > FF_MB_DECISION_SIMPLE){
1605 type= CANDIDATE_MB_TYPE_FORWARD | CANDIDATE_MB_TYPE_BACKWARD | CANDIDATE_MB_TYPE_BIDIR | CANDIDATE_MB_TYPE_DIRECT;
1607 type |= CANDIDATE_MB_TYPE_FORWARD_I;
1609 type |= CANDIDATE_MB_TYPE_BACKWARD_I;
1610 if(fimin < INT_MAX && bimin < INT_MAX){
1611 type |= CANDIDATE_MB_TYPE_BIDIR_I;
1613 //FIXME something smarter
1614 if(dmin>256*256*16) type&= ~CANDIDATE_MB_TYPE_DIRECT; //do not try direct mode if it is invalid for this MB
1615 if (s->codec_id == AV_CODEC_ID_MPEG4 && type&CANDIDATE_MB_TYPE_DIRECT &&
1616 s->mpv_flags & FF_MPV_FLAG_MV0 && *(uint32_t*)s->b_direct_mv_table[xy])
1617 type |= CANDIDATE_MB_TYPE_DIRECT0;
1620 s->mb_type[mb_y*s->mb_stride + mb_x]= type;
1623 /* find best f_code for ME which do unlimited searches */
1624 int ff_get_best_fcode(MpegEncContext * s, int16_t (*mv_table)[2], int type)
1626 if(s->me_method>=ME_EPZS){
1628 int i, y, range= s->avctx->me_range ? s->avctx->me_range : (INT_MAX/2);
1629 uint8_t * fcode_tab= s->fcode_tab;
1631 int best_score=-10000000;
1633 if(s->msmpeg4_version)
1634 range= FFMIN(range, 16);
1635 else if(s->codec_id == AV_CODEC_ID_MPEG2VIDEO && s->avctx->strict_std_compliance >= FF_COMPLIANCE_NORMAL)
1636 range= FFMIN(range, 256);
1638 for(i=0; i<8; i++) score[i]= s->mb_num*(8-i);
1640 for(y=0; y<s->mb_height; y++){
1642 int xy= y*s->mb_stride;
1643 for(x=0; x<s->mb_width; x++){
1644 if(s->mb_type[xy] & type){
1645 int mx= mv_table[xy][0];
1646 int my= mv_table[xy][1];
1647 int fcode= FFMAX(fcode_tab[mx + MAX_MV],
1648 fcode_tab[my + MAX_MV]);
1651 if(mx >= range || mx < -range ||
1652 my >= range || my < -range)
1655 for(j=0; j<fcode && j<8; j++){
1656 if(s->pict_type==AV_PICTURE_TYPE_B || s->current_picture.mc_mb_var[xy] < s->current_picture.mb_var[xy])
1665 if(score[i] > best_score){
1666 best_score= score[i];
1677 void ff_fix_long_p_mvs(MpegEncContext * s)
1679 MotionEstContext * const c= &s->me;
1680 const int f_code= s->f_code;
1682 av_assert0(s->pict_type==AV_PICTURE_TYPE_P);
1684 range = (((s->out_format == FMT_MPEG1 || s->msmpeg4_version) ? 8 : 16) << f_code);
1686 av_assert0(range <= 16 || !s->msmpeg4_version);
1687 av_assert0(range <=256 || !(s->codec_id == AV_CODEC_ID_MPEG2VIDEO && s->avctx->strict_std_compliance >= FF_COMPLIANCE_NORMAL));
1689 if(c->avctx->me_range && range > c->avctx->me_range) range= c->avctx->me_range;
1691 if(s->flags&CODEC_FLAG_4MV){
1692 const int wrap= s->b8_stride;
1694 /* clip / convert to intra 8x8 type MVs */
1695 for(y=0; y<s->mb_height; y++){
1697 int i= y*s->mb_stride;
1700 for(x=0; x<s->mb_width; x++){
1701 if(s->mb_type[i]&CANDIDATE_MB_TYPE_INTER4V){
1703 for(block=0; block<4; block++){
1704 int off= (block& 1) + (block>>1)*wrap;
1705 int mx = s->current_picture.motion_val[0][ xy + off ][0];
1706 int my = s->current_picture.motion_val[0][ xy + off ][1];
1708 if( mx >=range || mx <-range
1709 || my >=range || my <-range){
1710 s->mb_type[i] &= ~CANDIDATE_MB_TYPE_INTER4V;
1711 s->mb_type[i] |= CANDIDATE_MB_TYPE_INTRA;
1712 s->current_picture.mb_type[i] = CANDIDATE_MB_TYPE_INTRA;
1725 * @param truncate 1 for truncation, 0 for using intra
1727 void ff_fix_long_mvs(MpegEncContext * s, uint8_t *field_select_table, int field_select,
1728 int16_t (*mv_table)[2], int f_code, int type, int truncate)
1730 MotionEstContext * const c= &s->me;
1731 int y, h_range, v_range;
1733 // RAL: 8 in MPEG-1, 16 in MPEG-4
1734 int range = (((s->out_format == FMT_MPEG1 || s->msmpeg4_version) ? 8 : 16) << f_code);
1736 if(c->avctx->me_range && range > c->avctx->me_range) range= c->avctx->me_range;
1739 v_range= field_select_table ? range>>1 : range;
1741 /* clip / convert to intra 16x16 type MVs */
1742 for(y=0; y<s->mb_height; y++){
1744 int xy= y*s->mb_stride;
1745 for(x=0; x<s->mb_width; x++){
1746 if (s->mb_type[xy] & type){ // RAL: "type" test added...
1747 if(field_select_table==NULL || field_select_table[xy] == field_select){
1748 if( mv_table[xy][0] >=h_range || mv_table[xy][0] <-h_range
1749 || mv_table[xy][1] >=v_range || mv_table[xy][1] <-v_range){
1752 if (mv_table[xy][0] > h_range-1) mv_table[xy][0]= h_range-1;
1753 else if(mv_table[xy][0] < -h_range ) mv_table[xy][0]= -h_range;
1754 if (mv_table[xy][1] > v_range-1) mv_table[xy][1]= v_range-1;
1755 else if(mv_table[xy][1] < -v_range ) mv_table[xy][1]= -v_range;
1757 s->mb_type[xy] &= ~type;
1758 s->mb_type[xy] |= CANDIDATE_MB_TYPE_INTRA;