2 * The simplest mpeg encoder (well, it was the simplest!)
3 * Copyright (c) 2000,2001 Fabrice Bellard.
4 * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
6 * This file is part of FFmpeg.
8 * FFmpeg is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
13 * FFmpeg is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
22 * 4MV & hq & b-frame encoding stuff by Michael Niedermayer <michaelni@gmx.at>
27 * The simplest mpeg encoder (well, it was the simplest!).
32 #include "mpegvideo.h"
37 #include "libvo/fastmemcpy.h"
43 #ifdef CONFIG_ENCODERS
44 static int encode_picture(MpegEncContext *s, int picture_number);
45 #endif //CONFIG_ENCODERS
46 static void dct_unquantize_mpeg1_intra_c(MpegEncContext *s,
47 DCTELEM *block, int n, int qscale);
48 static void dct_unquantize_mpeg1_inter_c(MpegEncContext *s,
49 DCTELEM *block, int n, int qscale);
50 static void dct_unquantize_mpeg2_intra_c(MpegEncContext *s,
51 DCTELEM *block, int n, int qscale);
52 static void dct_unquantize_mpeg2_intra_bitexact(MpegEncContext *s,
53 DCTELEM *block, int n, int qscale);
54 static void dct_unquantize_mpeg2_inter_c(MpegEncContext *s,
55 DCTELEM *block, int n, int qscale);
56 static void dct_unquantize_h263_intra_c(MpegEncContext *s,
57 DCTELEM *block, int n, int qscale);
58 static void dct_unquantize_h263_inter_c(MpegEncContext *s,
59 DCTELEM *block, int n, int qscale);
60 static void draw_edges_c(uint8_t *buf, int wrap, int width, int height, int w);
61 #ifdef CONFIG_ENCODERS
62 static int dct_quantize_c(MpegEncContext *s, DCTELEM *block, int n, int qscale, int *overflow);
63 static int dct_quantize_trellis_c(MpegEncContext *s, DCTELEM *block, int n, int qscale, int *overflow);
64 static int dct_quantize_refine(MpegEncContext *s, DCTELEM *block, int16_t *weight, DCTELEM *orig, int n, int qscale);
65 static int sse_mb(MpegEncContext *s);
66 static void denoise_dct_c(MpegEncContext *s, DCTELEM *block);
67 #endif //CONFIG_ENCODERS
70 extern int XVMC_field_start(MpegEncContext*s, AVCodecContext *avctx);
71 extern void XVMC_field_end(MpegEncContext *s);
72 extern void XVMC_decode_mb(MpegEncContext *s);
75 void (*draw_edges)(uint8_t *buf, int wrap, int width, int height, int w)= draw_edges_c;
78 /* enable all paranoid tests for rounding, overflows, etc... */
84 /* for jpeg fast DCT */
87 static const uint16_t aanscales[64] = {
88 /* precomputed values scaled up by 14 bits */
89 16384, 22725, 21407, 19266, 16384, 12873, 8867, 4520,
90 22725, 31521, 29692, 26722, 22725, 17855, 12299, 6270,
91 21407, 29692, 27969, 25172, 21407, 16819, 11585, 5906,
92 19266, 26722, 25172, 22654, 19266, 15137, 10426, 5315,
93 16384, 22725, 21407, 19266, 16384, 12873, 8867, 4520,
94 12873, 17855, 16819, 15137, 12873, 10114, 6967, 3552,
95 8867 , 12299, 11585, 10426, 8867, 6967, 4799, 2446,
96 4520 , 6270, 5906, 5315, 4520, 3552, 2446, 1247
99 static const uint8_t h263_chroma_roundtab[16] = {
100 // 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
101 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2,
104 static const uint8_t ff_default_chroma_qscale_table[32]={
105 // 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
106 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31
109 #ifdef CONFIG_ENCODERS
110 static uint8_t (*default_mv_penalty)[MAX_MV*2+1]=NULL;
111 static uint8_t default_fcode_tab[MAX_MV*2+1];
113 enum PixelFormat ff_yuv420p_list[2]= {PIX_FMT_YUV420P, -1};
115 static void convert_matrix(DSPContext *dsp, int (*qmat)[64], uint16_t (*qmat16)[2][64],
116 const uint16_t *quant_matrix, int bias, int qmin, int qmax, int intra)
121 for(qscale=qmin; qscale<=qmax; qscale++){
123 if (dsp->fdct == ff_jpeg_fdct_islow
124 #ifdef FAAN_POSTSCALE
125 || dsp->fdct == ff_faandct
129 const int j= dsp->idct_permutation[i];
130 /* 16 <= qscale * quant_matrix[i] <= 7905 */
131 /* 19952 <= aanscales[i] * qscale * quant_matrix[i] <= 249205026 */
132 /* (1<<36)/19952 >= (1<<36)/(aanscales[i] * qscale * quant_matrix[i]) >= (1<<36)/249205026 */
133 /* 3444240 >= (1<<36)/(aanscales[i] * qscale * quant_matrix[i]) >= 275 */
135 qmat[qscale][i] = (int)((UINT64_C(1) << QMAT_SHIFT) /
136 (qscale * quant_matrix[j]));
138 } else if (dsp->fdct == fdct_ifast
139 #ifndef FAAN_POSTSCALE
140 || dsp->fdct == ff_faandct
144 const int j= dsp->idct_permutation[i];
145 /* 16 <= qscale * quant_matrix[i] <= 7905 */
146 /* 19952 <= aanscales[i] * qscale * quant_matrix[i] <= 249205026 */
147 /* (1<<36)/19952 >= (1<<36)/(aanscales[i] * qscale * quant_matrix[i]) >= (1<<36)/249205026 */
148 /* 3444240 >= (1<<36)/(aanscales[i] * qscale * quant_matrix[i]) >= 275 */
150 qmat[qscale][i] = (int)((UINT64_C(1) << (QMAT_SHIFT + 14)) /
151 (aanscales[i] * qscale * quant_matrix[j]));
155 const int j= dsp->idct_permutation[i];
156 /* We can safely suppose that 16 <= quant_matrix[i] <= 255
157 So 16 <= qscale * quant_matrix[i] <= 7905
158 so (1<<19) / 16 >= (1<<19) / (qscale * quant_matrix[i]) >= (1<<19) / 7905
159 so 32768 >= (1<<19) / (qscale * quant_matrix[i]) >= 67
161 qmat[qscale][i] = (int)((UINT64_C(1) << QMAT_SHIFT) / (qscale * quant_matrix[j]));
162 // qmat [qscale][i] = (1 << QMAT_SHIFT_MMX) / (qscale * quant_matrix[i]);
163 qmat16[qscale][0][i] = (1 << QMAT_SHIFT_MMX) / (qscale * quant_matrix[j]);
165 if(qmat16[qscale][0][i]==0 || qmat16[qscale][0][i]==128*256) qmat16[qscale][0][i]=128*256-1;
166 qmat16[qscale][1][i]= ROUNDED_DIV(bias<<(16-QUANT_BIAS_SHIFT), qmat16[qscale][0][i]);
170 for(i=intra; i<64; i++){
172 if (dsp->fdct == fdct_ifast
173 #ifndef FAAN_POSTSCALE
174 || dsp->fdct == ff_faandct
177 max= (8191LL*aanscales[i]) >> 14;
179 while(((max * qmat[qscale][i]) >> shift) > INT_MAX){
185 av_log(NULL, AV_LOG_INFO, "Warning, QMAT_SHIFT is larger then %d, overflows possible\n", QMAT_SHIFT - shift);
189 static inline void update_qscale(MpegEncContext *s){
190 s->qscale= (s->lambda*139 + FF_LAMBDA_SCALE*64) >> (FF_LAMBDA_SHIFT + 7);
191 s->qscale= clip(s->qscale, s->avctx->qmin, s->avctx->qmax);
193 s->lambda2= (s->lambda*s->lambda + FF_LAMBDA_SCALE/2) >> FF_LAMBDA_SHIFT;
195 #endif //CONFIG_ENCODERS
197 void ff_init_scantable(uint8_t *permutation, ScanTable *st, const uint8_t *src_scantable){
201 st->scantable= src_scantable;
205 j = src_scantable[i];
206 st->permutated[i] = permutation[j];
215 j = st->permutated[i];
217 st->raster_end[i]= end;
221 #ifdef CONFIG_ENCODERS
222 void ff_write_quant_matrix(PutBitContext *pb, uint16_t *matrix){
228 put_bits(pb, 8, matrix[ ff_zigzag_direct[i] ]);
233 #endif //CONFIG_ENCODERS
235 const uint8_t *ff_find_start_code(const uint8_t * restrict p, const uint8_t *end, uint32_t * restrict state){
243 uint32_t tmp= *state << 8;
244 *state= tmp + *(p++);
245 if(tmp == 0x100 || p==end)
250 if (p[-1] > 1 ) p+= 3;
251 else if(p[-2] ) p+= 2;
252 else if(p[-3]|(p[-1]-1)) p++;
260 *state= be2me_32(unaligned32(p));
265 /* init common dct for both encoder and decoder */
266 int DCT_common_init(MpegEncContext *s)
268 s->dct_unquantize_h263_intra = dct_unquantize_h263_intra_c;
269 s->dct_unquantize_h263_inter = dct_unquantize_h263_inter_c;
270 s->dct_unquantize_mpeg1_intra = dct_unquantize_mpeg1_intra_c;
271 s->dct_unquantize_mpeg1_inter = dct_unquantize_mpeg1_inter_c;
272 s->dct_unquantize_mpeg2_intra = dct_unquantize_mpeg2_intra_c;
273 if(s->flags & CODEC_FLAG_BITEXACT)
274 s->dct_unquantize_mpeg2_intra = dct_unquantize_mpeg2_intra_bitexact;
275 s->dct_unquantize_mpeg2_inter = dct_unquantize_mpeg2_inter_c;
277 #ifdef CONFIG_ENCODERS
278 s->dct_quantize= dct_quantize_c;
279 s->denoise_dct= denoise_dct_c;
280 #endif //CONFIG_ENCODERS
283 MPV_common_init_mmx(s);
286 MPV_common_init_axp(s);
289 MPV_common_init_mlib(s);
292 MPV_common_init_mmi(s);
295 MPV_common_init_armv4l(s);
298 MPV_common_init_ppc(s);
301 #ifdef CONFIG_ENCODERS
302 s->fast_dct_quantize= s->dct_quantize;
304 if(s->flags&CODEC_FLAG_TRELLIS_QUANT){
305 s->dct_quantize= dct_quantize_trellis_c; //move before MPV_common_init_*
308 #endif //CONFIG_ENCODERS
310 /* load & permutate scantables
311 note: only wmv uses different ones
313 if(s->alternate_scan){
314 ff_init_scantable(s->dsp.idct_permutation, &s->inter_scantable , ff_alternate_vertical_scan);
315 ff_init_scantable(s->dsp.idct_permutation, &s->intra_scantable , ff_alternate_vertical_scan);
317 ff_init_scantable(s->dsp.idct_permutation, &s->inter_scantable , ff_zigzag_direct);
318 ff_init_scantable(s->dsp.idct_permutation, &s->intra_scantable , ff_zigzag_direct);
320 ff_init_scantable(s->dsp.idct_permutation, &s->intra_h_scantable, ff_alternate_horizontal_scan);
321 ff_init_scantable(s->dsp.idct_permutation, &s->intra_v_scantable, ff_alternate_vertical_scan);
326 static void copy_picture(Picture *dst, Picture *src){
328 dst->type= FF_BUFFER_TYPE_COPY;
331 #ifdef CONFIG_ENCODERS
332 static void copy_picture_attributes(MpegEncContext *s, AVFrame *dst, AVFrame *src){
335 dst->pict_type = src->pict_type;
336 dst->quality = src->quality;
337 dst->coded_picture_number = src->coded_picture_number;
338 dst->display_picture_number = src->display_picture_number;
339 // dst->reference = src->reference;
341 dst->interlaced_frame = src->interlaced_frame;
342 dst->top_field_first = src->top_field_first;
344 if(s->avctx->me_threshold){
345 if(!src->motion_val[0])
346 av_log(s->avctx, AV_LOG_ERROR, "AVFrame.motion_val not set!\n");
348 av_log(s->avctx, AV_LOG_ERROR, "AVFrame.mb_type not set!\n");
349 if(!src->ref_index[0])
350 av_log(s->avctx, AV_LOG_ERROR, "AVFrame.ref_index not set!\n");
351 if(src->motion_subsample_log2 != dst->motion_subsample_log2)
352 av_log(s->avctx, AV_LOG_ERROR, "AVFrame.motion_subsample_log2 doesn't match! (%d!=%d)\n",
353 src->motion_subsample_log2, dst->motion_subsample_log2);
355 memcpy(dst->mb_type, src->mb_type, s->mb_stride * s->mb_height * sizeof(dst->mb_type[0]));
358 int stride= ((16*s->mb_width )>>src->motion_subsample_log2) + 1;
359 int height= ((16*s->mb_height)>>src->motion_subsample_log2);
361 if(src->motion_val[i] && src->motion_val[i] != dst->motion_val[i]){
362 memcpy(dst->motion_val[i], src->motion_val[i], 2*stride*height*sizeof(int16_t));
364 if(src->ref_index[i] && src->ref_index[i] != dst->ref_index[i]){
365 memcpy(dst->ref_index[i], src->ref_index[i], s->b8_stride*2*s->mb_height*sizeof(int8_t));
373 * allocates a Picture
374 * The pixels are allocated/set by calling get_buffer() if shared=0
376 static int alloc_picture(MpegEncContext *s, Picture *pic, int shared){
377 const int big_mb_num= s->mb_stride*(s->mb_height+1) + 1; //the +1 is needed so memset(,,stride*height) doesnt sig11
378 const int mb_array_size= s->mb_stride*s->mb_height;
379 const int b8_array_size= s->b8_stride*s->mb_height*2;
380 const int b4_array_size= s->b4_stride*s->mb_height*4;
384 assert(pic->data[0]);
385 assert(pic->type == 0 || pic->type == FF_BUFFER_TYPE_SHARED);
386 pic->type= FF_BUFFER_TYPE_SHARED;
390 assert(!pic->data[0]);
392 r= s->avctx->get_buffer(s->avctx, (AVFrame*)pic);
394 if(r<0 || !pic->age || !pic->type || !pic->data[0]){
395 av_log(s->avctx, AV_LOG_ERROR, "get_buffer() failed (%d %d %d %p)\n", r, pic->age, pic->type, pic->data[0]);
399 if(s->linesize && (s->linesize != pic->linesize[0] || s->uvlinesize != pic->linesize[1])){
400 av_log(s->avctx, AV_LOG_ERROR, "get_buffer() failed (stride changed)\n");
404 if(pic->linesize[1] != pic->linesize[2]){
405 av_log(s->avctx, AV_LOG_ERROR, "get_buffer() failed (uv stride mismatch)\n");
409 s->linesize = pic->linesize[0];
410 s->uvlinesize= pic->linesize[1];
413 if(pic->qscale_table==NULL){
415 CHECKED_ALLOCZ(pic->mb_var , mb_array_size * sizeof(int16_t))
416 CHECKED_ALLOCZ(pic->mc_mb_var, mb_array_size * sizeof(int16_t))
417 CHECKED_ALLOCZ(pic->mb_mean , mb_array_size * sizeof(int8_t))
420 CHECKED_ALLOCZ(pic->mbskip_table , mb_array_size * sizeof(uint8_t)+2) //the +2 is for the slice end check
421 CHECKED_ALLOCZ(pic->qscale_table , mb_array_size * sizeof(uint8_t))
422 CHECKED_ALLOCZ(pic->mb_type_base , big_mb_num * sizeof(uint32_t))
423 pic->mb_type= pic->mb_type_base + s->mb_stride+1;
424 if(s->out_format == FMT_H264){
426 CHECKED_ALLOCZ(pic->motion_val_base[i], 2 * (b4_array_size+4) * sizeof(int16_t))
427 pic->motion_val[i]= pic->motion_val_base[i]+4;
428 CHECKED_ALLOCZ(pic->ref_index[i], b8_array_size * sizeof(uint8_t))
430 pic->motion_subsample_log2= 2;
431 }else if(s->out_format == FMT_H263 || s->encoding || (s->avctx->debug&FF_DEBUG_MV) || (s->avctx->debug_mv)){
433 CHECKED_ALLOCZ(pic->motion_val_base[i], 2 * (b8_array_size+4) * sizeof(int16_t))
434 pic->motion_val[i]= pic->motion_val_base[i]+4;
435 CHECKED_ALLOCZ(pic->ref_index[i], b8_array_size * sizeof(uint8_t))
437 pic->motion_subsample_log2= 3;
439 if(s->avctx->debug&FF_DEBUG_DCT_COEFF) {
440 CHECKED_ALLOCZ(pic->dct_coeff, 64 * mb_array_size * sizeof(DCTELEM)*6)
442 pic->qstride= s->mb_stride;
443 CHECKED_ALLOCZ(pic->pan_scan , 1 * sizeof(AVPanScan))
446 //it might be nicer if the application would keep track of these but it would require a API change
447 memmove(s->prev_pict_types+1, s->prev_pict_types, PREV_PICT_TYPES_BUFFER_SIZE-1);
448 s->prev_pict_types[0]= s->pict_type;
449 if(pic->age < PREV_PICT_TYPES_BUFFER_SIZE && s->prev_pict_types[pic->age] == B_TYPE)
450 pic->age= INT_MAX; // skipped MBs in b frames are quite rare in mpeg1/2 and its a bit tricky to skip them anyway
453 fail: //for the CHECKED_ALLOCZ macro
458 * deallocates a picture
460 static void free_picture(MpegEncContext *s, Picture *pic){
463 if(pic->data[0] && pic->type!=FF_BUFFER_TYPE_SHARED){
464 s->avctx->release_buffer(s->avctx, (AVFrame*)pic);
467 av_freep(&pic->mb_var);
468 av_freep(&pic->mc_mb_var);
469 av_freep(&pic->mb_mean);
470 av_freep(&pic->mbskip_table);
471 av_freep(&pic->qscale_table);
472 av_freep(&pic->mb_type_base);
473 av_freep(&pic->dct_coeff);
474 av_freep(&pic->pan_scan);
477 av_freep(&pic->motion_val_base[i]);
478 av_freep(&pic->ref_index[i]);
481 if(pic->type == FF_BUFFER_TYPE_SHARED){
490 static int init_duplicate_context(MpegEncContext *s, MpegEncContext *base){
493 // edge emu needs blocksize + filter length - 1 (=17x17 for halfpel / 21x21 for h264)
494 CHECKED_ALLOCZ(s->allocated_edge_emu_buffer, (s->width+64)*2*21*2); //(width + edge + align)*interlaced*MBsize*tolerance
495 s->edge_emu_buffer= s->allocated_edge_emu_buffer + (s->width+64)*2*21;
497 //FIXME should be linesize instead of s->width*2 but that isnt known before get_buffer()
498 CHECKED_ALLOCZ(s->me.scratchpad, (s->width+64)*4*16*2*sizeof(uint8_t))
499 s->rd_scratchpad= s->me.scratchpad;
500 s->b_scratchpad= s->me.scratchpad;
501 s->obmc_scratchpad= s->me.scratchpad + 16;
503 CHECKED_ALLOCZ(s->me.map , ME_MAP_SIZE*sizeof(uint32_t))
504 CHECKED_ALLOCZ(s->me.score_map, ME_MAP_SIZE*sizeof(uint32_t))
505 if(s->avctx->noise_reduction){
506 CHECKED_ALLOCZ(s->dct_error_sum, 2 * 64 * sizeof(int))
509 CHECKED_ALLOCZ(s->blocks, 64*12*2 * sizeof(DCTELEM))
510 s->block= s->blocks[0];
513 s->pblocks[i] = (short *)(&s->block[i]);
517 return -1; //free() through MPV_common_end()
520 static void free_duplicate_context(MpegEncContext *s){
523 av_freep(&s->allocated_edge_emu_buffer); s->edge_emu_buffer= NULL;
524 av_freep(&s->me.scratchpad);
527 s->obmc_scratchpad= NULL;
529 av_freep(&s->dct_error_sum);
530 av_freep(&s->me.map);
531 av_freep(&s->me.score_map);
532 av_freep(&s->blocks);
536 static void backup_duplicate_context(MpegEncContext *bak, MpegEncContext *src){
537 #define COPY(a) bak->a= src->a
538 COPY(allocated_edge_emu_buffer);
539 COPY(edge_emu_buffer);
543 COPY(obmc_scratchpad);
550 COPY(me.map_generation);
558 void ff_update_duplicate_context(MpegEncContext *dst, MpegEncContext *src){
561 //FIXME copy only needed parts
563 backup_duplicate_context(&bak, dst);
564 memcpy(dst, src, sizeof(MpegEncContext));
565 backup_duplicate_context(dst, &bak);
567 dst->pblocks[i] = (short *)(&dst->block[i]);
569 //STOP_TIMER("update_duplicate_context") //about 10k cycles / 0.01 sec for 1000frames on 1ghz with 2 threads
572 #ifdef CONFIG_ENCODERS
573 static void update_duplicate_context_after_me(MpegEncContext *dst, MpegEncContext *src){
574 #define COPY(a) dst->a= src->a
576 COPY(current_picture);
582 COPY(picture_in_gop_number);
583 COPY(gop_picture_number);
584 COPY(frame_pred_frame_dct); //FIXME don't set in encode_header
585 COPY(progressive_frame); //FIXME don't set in encode_header
586 COPY(partitioned_frame); //FIXME don't set in encode_header
592 * sets the given MpegEncContext to common defaults (same for encoding and decoding).
593 * the changed fields will not depend upon the prior state of the MpegEncContext.
595 static void MPV_common_defaults(MpegEncContext *s){
597 s->c_dc_scale_table= ff_mpeg1_dc_scale_table;
598 s->chroma_qscale_table= ff_default_chroma_qscale_table;
599 s->progressive_frame= 1;
600 s->progressive_sequence= 1;
601 s->picture_structure= PICT_FRAME;
603 s->coded_picture_number = 0;
604 s->picture_number = 0;
605 s->input_picture_number = 0;
607 s->picture_in_gop_number = 0;
614 * sets the given MpegEncContext to defaults for decoding.
615 * the changed fields will not depend upon the prior state of the MpegEncContext.
617 void MPV_decode_defaults(MpegEncContext *s){
618 MPV_common_defaults(s);
622 * sets the given MpegEncContext to defaults for encoding.
623 * the changed fields will not depend upon the prior state of the MpegEncContext.
626 #ifdef CONFIG_ENCODERS
627 static void MPV_encode_defaults(MpegEncContext *s){
630 MPV_common_defaults(s);
636 default_mv_penalty= av_mallocz( sizeof(uint8_t)*(MAX_FCODE+1)*(2*MAX_MV+1) );
637 memset(default_fcode_tab , 0, sizeof(uint8_t)*(2*MAX_MV+1));
639 for(i=-16; i<16; i++){
640 default_fcode_tab[i + MAX_MV]= 1;
643 s->me.mv_penalty= default_mv_penalty;
644 s->fcode_tab= default_fcode_tab;
646 #endif //CONFIG_ENCODERS
649 * init common structure for both encoder and decoder.
650 * this assumes that some variables like width/height are already set
652 int MPV_common_init(MpegEncContext *s)
654 int y_size, c_size, yc_size, i, mb_array_size, mv_table_size, x, y;
656 s->mb_height = (s->height + 15) / 16;
658 if(s->avctx->thread_count > MAX_THREADS || (s->avctx->thread_count > s->mb_height && s->mb_height)){
659 av_log(s->avctx, AV_LOG_ERROR, "too many threads\n");
663 if((s->width || s->height) && avcodec_check_dimensions(s->avctx, s->width, s->height))
666 dsputil_init(&s->dsp, s->avctx);
669 s->flags= s->avctx->flags;
670 s->flags2= s->avctx->flags2;
672 s->mb_width = (s->width + 15) / 16;
673 s->mb_stride = s->mb_width + 1;
674 s->b8_stride = s->mb_width*2 + 1;
675 s->b4_stride = s->mb_width*4 + 1;
676 mb_array_size= s->mb_height * s->mb_stride;
677 mv_table_size= (s->mb_height+2) * s->mb_stride + 1;
679 /* set chroma shifts */
680 avcodec_get_chroma_sub_sample(s->avctx->pix_fmt,&(s->chroma_x_shift),
681 &(s->chroma_y_shift) );
683 /* set default edge pos, will be overriden in decode_header if needed */
684 s->h_edge_pos= s->mb_width*16;
685 s->v_edge_pos= s->mb_height*16;
687 s->mb_num = s->mb_width * s->mb_height;
692 s->block_wrap[3]= s->b8_stride;
694 s->block_wrap[5]= s->mb_stride;
696 y_size = s->b8_stride * (2 * s->mb_height + 1);
697 c_size = s->mb_stride * (s->mb_height + 1);
698 yc_size = y_size + 2 * c_size;
700 /* convert fourcc to upper case */
701 s->codec_tag= toupper( s->avctx->codec_tag &0xFF)
702 + (toupper((s->avctx->codec_tag>>8 )&0xFF)<<8 )
703 + (toupper((s->avctx->codec_tag>>16)&0xFF)<<16)
704 + (toupper((s->avctx->codec_tag>>24)&0xFF)<<24);
706 s->stream_codec_tag= toupper( s->avctx->stream_codec_tag &0xFF)
707 + (toupper((s->avctx->stream_codec_tag>>8 )&0xFF)<<8 )
708 + (toupper((s->avctx->stream_codec_tag>>16)&0xFF)<<16)
709 + (toupper((s->avctx->stream_codec_tag>>24)&0xFF)<<24);
711 s->avctx->coded_frame= (AVFrame*)&s->current_picture;
713 CHECKED_ALLOCZ(s->mb_index2xy, (s->mb_num+1)*sizeof(int)) //error ressilience code looks cleaner with this
714 for(y=0; y<s->mb_height; y++){
715 for(x=0; x<s->mb_width; x++){
716 s->mb_index2xy[ x + y*s->mb_width ] = x + y*s->mb_stride;
719 s->mb_index2xy[ s->mb_height*s->mb_width ] = (s->mb_height-1)*s->mb_stride + s->mb_width; //FIXME really needed?
722 /* Allocate MV tables */
723 CHECKED_ALLOCZ(s->p_mv_table_base , mv_table_size * 2 * sizeof(int16_t))
724 CHECKED_ALLOCZ(s->b_forw_mv_table_base , mv_table_size * 2 * sizeof(int16_t))
725 CHECKED_ALLOCZ(s->b_back_mv_table_base , mv_table_size * 2 * sizeof(int16_t))
726 CHECKED_ALLOCZ(s->b_bidir_forw_mv_table_base , mv_table_size * 2 * sizeof(int16_t))
727 CHECKED_ALLOCZ(s->b_bidir_back_mv_table_base , mv_table_size * 2 * sizeof(int16_t))
728 CHECKED_ALLOCZ(s->b_direct_mv_table_base , mv_table_size * 2 * sizeof(int16_t))
729 s->p_mv_table = s->p_mv_table_base + s->mb_stride + 1;
730 s->b_forw_mv_table = s->b_forw_mv_table_base + s->mb_stride + 1;
731 s->b_back_mv_table = s->b_back_mv_table_base + s->mb_stride + 1;
732 s->b_bidir_forw_mv_table= s->b_bidir_forw_mv_table_base + s->mb_stride + 1;
733 s->b_bidir_back_mv_table= s->b_bidir_back_mv_table_base + s->mb_stride + 1;
734 s->b_direct_mv_table = s->b_direct_mv_table_base + s->mb_stride + 1;
736 if(s->msmpeg4_version){
737 CHECKED_ALLOCZ(s->ac_stats, 2*2*(MAX_LEVEL+1)*(MAX_RUN+1)*2*sizeof(int));
739 CHECKED_ALLOCZ(s->avctx->stats_out, 256);
741 /* Allocate MB type table */
742 CHECKED_ALLOCZ(s->mb_type , mb_array_size * sizeof(uint16_t)) //needed for encoding
744 CHECKED_ALLOCZ(s->lambda_table, mb_array_size * sizeof(int))
746 CHECKED_ALLOCZ(s->q_intra_matrix, 64*32 * sizeof(int))
747 CHECKED_ALLOCZ(s->q_inter_matrix, 64*32 * sizeof(int))
748 CHECKED_ALLOCZ(s->q_intra_matrix16, 64*32*2 * sizeof(uint16_t))
749 CHECKED_ALLOCZ(s->q_inter_matrix16, 64*32*2 * sizeof(uint16_t))
750 CHECKED_ALLOCZ(s->input_picture, MAX_PICTURE_COUNT * sizeof(Picture*))
751 CHECKED_ALLOCZ(s->reordered_input_picture, MAX_PICTURE_COUNT * sizeof(Picture*))
753 if(s->avctx->noise_reduction){
754 CHECKED_ALLOCZ(s->dct_offset, 2 * 64 * sizeof(uint16_t))
757 CHECKED_ALLOCZ(s->picture, MAX_PICTURE_COUNT * sizeof(Picture))
759 CHECKED_ALLOCZ(s->error_status_table, mb_array_size*sizeof(uint8_t))
761 if(s->codec_id==CODEC_ID_MPEG4 || (s->flags & CODEC_FLAG_INTERLACED_ME)){
762 /* interlaced direct mode decoding tables */
767 CHECKED_ALLOCZ(s->b_field_mv_table_base[i][j][k] , mv_table_size * 2 * sizeof(int16_t))
768 s->b_field_mv_table[i][j][k] = s->b_field_mv_table_base[i][j][k] + s->mb_stride + 1;
770 CHECKED_ALLOCZ(s->b_field_select_table[i][j] , mb_array_size * 2 * sizeof(uint8_t))
771 CHECKED_ALLOCZ(s->p_field_mv_table_base[i][j] , mv_table_size * 2 * sizeof(int16_t))
772 s->p_field_mv_table[i][j] = s->p_field_mv_table_base[i][j] + s->mb_stride + 1;
774 CHECKED_ALLOCZ(s->p_field_select_table[i] , mb_array_size * 2 * sizeof(uint8_t))
777 if (s->out_format == FMT_H263) {
779 CHECKED_ALLOCZ(s->ac_val_base, yc_size * sizeof(int16_t) * 16);
780 s->ac_val[0] = s->ac_val_base + s->b8_stride + 1;
781 s->ac_val[1] = s->ac_val_base + y_size + s->mb_stride + 1;
782 s->ac_val[2] = s->ac_val[1] + c_size;
785 CHECKED_ALLOCZ(s->coded_block_base, y_size);
786 s->coded_block= s->coded_block_base + s->b8_stride + 1;
788 /* cbp, ac_pred, pred_dir */
789 CHECKED_ALLOCZ(s->cbp_table , mb_array_size * sizeof(uint8_t))
790 CHECKED_ALLOCZ(s->pred_dir_table, mb_array_size * sizeof(uint8_t))
793 if (s->h263_pred || s->h263_plus || !s->encoding) {
795 //MN: we need these for error resilience of intra-frames
796 CHECKED_ALLOCZ(s->dc_val_base, yc_size * sizeof(int16_t));
797 s->dc_val[0] = s->dc_val_base + s->b8_stride + 1;
798 s->dc_val[1] = s->dc_val_base + y_size + s->mb_stride + 1;
799 s->dc_val[2] = s->dc_val[1] + c_size;
800 for(i=0;i<yc_size;i++)
801 s->dc_val_base[i] = 1024;
804 /* which mb is a intra block */
805 CHECKED_ALLOCZ(s->mbintra_table, mb_array_size);
806 memset(s->mbintra_table, 1, mb_array_size);
808 /* init macroblock skip table */
809 CHECKED_ALLOCZ(s->mbskip_table, mb_array_size+2);
810 //Note the +1 is for a quicker mpeg4 slice_end detection
811 CHECKED_ALLOCZ(s->prev_pict_types, PREV_PICT_TYPES_BUFFER_SIZE);
813 s->parse_context.state= -1;
814 if((s->avctx->debug&(FF_DEBUG_VIS_QP|FF_DEBUG_VIS_MB_TYPE)) || (s->avctx->debug_mv)){
815 s->visualization_buffer[0] = av_malloc((s->mb_width*16 + 2*EDGE_WIDTH) * s->mb_height*16 + 2*EDGE_WIDTH);
816 s->visualization_buffer[1] = av_malloc((s->mb_width*8 + EDGE_WIDTH) * s->mb_height*8 + EDGE_WIDTH);
817 s->visualization_buffer[2] = av_malloc((s->mb_width*8 + EDGE_WIDTH) * s->mb_height*8 + EDGE_WIDTH);
820 s->context_initialized = 1;
822 s->thread_context[0]= s;
823 for(i=1; i<s->avctx->thread_count; i++){
824 s->thread_context[i]= av_malloc(sizeof(MpegEncContext));
825 memcpy(s->thread_context[i], s, sizeof(MpegEncContext));
828 for(i=0; i<s->avctx->thread_count; i++){
829 if(init_duplicate_context(s->thread_context[i], s) < 0)
831 s->thread_context[i]->start_mb_y= (s->mb_height*(i ) + s->avctx->thread_count/2) / s->avctx->thread_count;
832 s->thread_context[i]->end_mb_y = (s->mb_height*(i+1) + s->avctx->thread_count/2) / s->avctx->thread_count;
841 /* init common structure for both encoder and decoder */
842 void MPV_common_end(MpegEncContext *s)
846 for(i=0; i<s->avctx->thread_count; i++){
847 free_duplicate_context(s->thread_context[i]);
849 for(i=1; i<s->avctx->thread_count; i++){
850 av_freep(&s->thread_context[i]);
853 av_freep(&s->parse_context.buffer);
854 s->parse_context.buffer_size=0;
856 av_freep(&s->mb_type);
857 av_freep(&s->p_mv_table_base);
858 av_freep(&s->b_forw_mv_table_base);
859 av_freep(&s->b_back_mv_table_base);
860 av_freep(&s->b_bidir_forw_mv_table_base);
861 av_freep(&s->b_bidir_back_mv_table_base);
862 av_freep(&s->b_direct_mv_table_base);
864 s->b_forw_mv_table= NULL;
865 s->b_back_mv_table= NULL;
866 s->b_bidir_forw_mv_table= NULL;
867 s->b_bidir_back_mv_table= NULL;
868 s->b_direct_mv_table= NULL;
872 av_freep(&s->b_field_mv_table_base[i][j][k]);
873 s->b_field_mv_table[i][j][k]=NULL;
875 av_freep(&s->b_field_select_table[i][j]);
876 av_freep(&s->p_field_mv_table_base[i][j]);
877 s->p_field_mv_table[i][j]=NULL;
879 av_freep(&s->p_field_select_table[i]);
882 av_freep(&s->dc_val_base);
883 av_freep(&s->ac_val_base);
884 av_freep(&s->coded_block_base);
885 av_freep(&s->mbintra_table);
886 av_freep(&s->cbp_table);
887 av_freep(&s->pred_dir_table);
889 av_freep(&s->mbskip_table);
890 av_freep(&s->prev_pict_types);
891 av_freep(&s->bitstream_buffer);
892 s->allocated_bitstream_buffer_size=0;
894 av_freep(&s->avctx->stats_out);
895 av_freep(&s->ac_stats);
896 av_freep(&s->error_status_table);
897 av_freep(&s->mb_index2xy);
898 av_freep(&s->lambda_table);
899 av_freep(&s->q_intra_matrix);
900 av_freep(&s->q_inter_matrix);
901 av_freep(&s->q_intra_matrix16);
902 av_freep(&s->q_inter_matrix16);
903 av_freep(&s->input_picture);
904 av_freep(&s->reordered_input_picture);
905 av_freep(&s->dct_offset);
908 for(i=0; i<MAX_PICTURE_COUNT; i++){
909 free_picture(s, &s->picture[i]);
912 av_freep(&s->picture);
913 s->context_initialized = 0;
916 s->current_picture_ptr= NULL;
917 s->linesize= s->uvlinesize= 0;
920 av_freep(&s->visualization_buffer[i]);
922 avcodec_default_free_buffers(s->avctx);
925 #ifdef CONFIG_ENCODERS
927 /* init video encoder */
928 int MPV_encode_init(AVCodecContext *avctx)
930 MpegEncContext *s = avctx->priv_data;
932 int chroma_h_shift, chroma_v_shift;
934 MPV_encode_defaults(s);
936 switch (avctx->codec_id) {
937 case CODEC_ID_MPEG2VIDEO:
938 if(avctx->pix_fmt != PIX_FMT_YUV420P && avctx->pix_fmt != PIX_FMT_YUV422P){
939 av_log(avctx, AV_LOG_ERROR, "only YUV420 and YUV422 are supported\n");
945 if(avctx->pix_fmt != PIX_FMT_YUVJ420P && avctx->pix_fmt != PIX_FMT_YUVJ422P &&
946 ((avctx->pix_fmt != PIX_FMT_YUV420P && avctx->pix_fmt != PIX_FMT_YUV422P) || avctx->strict_std_compliance>FF_COMPLIANCE_INOFFICIAL)){
947 av_log(avctx, AV_LOG_ERROR, "colorspace not supported in jpeg\n");
952 if(avctx->pix_fmt != PIX_FMT_YUV420P){
953 av_log(avctx, AV_LOG_ERROR, "only YUV420 is supported\n");
958 switch (avctx->pix_fmt) {
959 case PIX_FMT_YUVJ422P:
960 case PIX_FMT_YUV422P:
961 s->chroma_format = CHROMA_422;
963 case PIX_FMT_YUVJ420P:
964 case PIX_FMT_YUV420P:
966 s->chroma_format = CHROMA_420;
970 s->bit_rate = avctx->bit_rate;
971 s->width = avctx->width;
972 s->height = avctx->height;
973 if(avctx->gop_size > 600 && avctx->strict_std_compliance>FF_COMPLIANCE_EXPERIMENTAL){
974 av_log(avctx, AV_LOG_ERROR, "Warning keyframe interval too large! reducing it ...\n");
977 s->gop_size = avctx->gop_size;
979 s->flags= avctx->flags;
980 s->flags2= avctx->flags2;
981 s->max_b_frames= avctx->max_b_frames;
982 s->codec_id= avctx->codec->id;
983 s->luma_elim_threshold = avctx->luma_elim_threshold;
984 s->chroma_elim_threshold= avctx->chroma_elim_threshold;
985 s->strict_std_compliance= avctx->strict_std_compliance;
986 s->data_partitioning= avctx->flags & CODEC_FLAG_PART;
987 s->quarter_sample= (avctx->flags & CODEC_FLAG_QPEL)!=0;
988 s->mpeg_quant= avctx->mpeg_quant;
989 s->rtp_mode= !!avctx->rtp_payload_size;
990 s->intra_dc_precision= avctx->intra_dc_precision;
991 s->user_specified_pts = AV_NOPTS_VALUE;
993 if (s->gop_size <= 1) {
1000 s->me_method = avctx->me_method;
1003 s->fixed_qscale = !!(avctx->flags & CODEC_FLAG_QSCALE);
1005 s->adaptive_quant= ( s->avctx->lumi_masking
1006 || s->avctx->dark_masking
1007 || s->avctx->temporal_cplx_masking
1008 || s->avctx->spatial_cplx_masking
1009 || s->avctx->p_masking
1010 || s->avctx->border_masking
1011 || (s->flags&CODEC_FLAG_QP_RD))
1012 && !s->fixed_qscale;
1014 s->obmc= !!(s->flags & CODEC_FLAG_OBMC);
1015 s->loop_filter= !!(s->flags & CODEC_FLAG_LOOP_FILTER);
1016 s->alternate_scan= !!(s->flags & CODEC_FLAG_ALT_SCAN);
1017 s->intra_vlc_format= !!(s->flags2 & CODEC_FLAG2_INTRA_VLC);
1019 if(avctx->rc_max_rate && !avctx->rc_buffer_size){
1020 av_log(avctx, AV_LOG_ERROR, "a vbv buffer size is needed, for encoding with a maximum bitrate\n");
1024 if(avctx->rc_min_rate && avctx->rc_max_rate != avctx->rc_min_rate){
1025 av_log(avctx, AV_LOG_INFO, "Warning min_rate > 0 but min_rate != max_rate isn't recommended!\n");
1028 if(avctx->rc_min_rate && avctx->rc_min_rate > avctx->bit_rate){
1029 av_log(avctx, AV_LOG_INFO, "bitrate below min bitrate\n");
1033 if(avctx->rc_max_rate && avctx->rc_max_rate < avctx->bit_rate){
1034 av_log(avctx, AV_LOG_INFO, "bitrate above max bitrate\n");
1038 if( s->avctx->rc_max_rate && s->avctx->rc_min_rate == s->avctx->rc_max_rate
1039 && (s->codec_id == CODEC_ID_MPEG1VIDEO || s->codec_id == CODEC_ID_MPEG2VIDEO)
1040 && 90000LL * (avctx->rc_buffer_size-1) > s->avctx->rc_max_rate*0xFFFFLL){
1042 av_log(avctx, AV_LOG_INFO, "Warning vbv_delay will be set to 0xFFFF (=VBR) as the specified vbv buffer is too large for the given bitrate!\n");
1045 if((s->flags & CODEC_FLAG_4MV) && s->codec_id != CODEC_ID_MPEG4
1046 && s->codec_id != CODEC_ID_H263 && s->codec_id != CODEC_ID_H263P && s->codec_id != CODEC_ID_FLV1){
1047 av_log(avctx, AV_LOG_ERROR, "4MV not supported by codec\n");
1051 if(s->obmc && s->avctx->mb_decision != FF_MB_DECISION_SIMPLE){
1052 av_log(avctx, AV_LOG_ERROR, "OBMC is only supported with simple mb decision\n");
1056 if(s->obmc && s->codec_id != CODEC_ID_H263 && s->codec_id != CODEC_ID_H263P){
1057 av_log(avctx, AV_LOG_ERROR, "OBMC is only supported with H263(+)\n");
1061 if(s->quarter_sample && s->codec_id != CODEC_ID_MPEG4){
1062 av_log(avctx, AV_LOG_ERROR, "qpel not supported by codec\n");
1066 if(s->data_partitioning && s->codec_id != CODEC_ID_MPEG4){
1067 av_log(avctx, AV_LOG_ERROR, "data partitioning not supported by codec\n");
1071 if(s->max_b_frames && s->codec_id != CODEC_ID_MPEG4 && s->codec_id != CODEC_ID_MPEG1VIDEO && s->codec_id != CODEC_ID_MPEG2VIDEO){
1072 av_log(avctx, AV_LOG_ERROR, "b frames not supported by codec\n");
1076 if((s->flags & (CODEC_FLAG_INTERLACED_DCT|CODEC_FLAG_INTERLACED_ME|CODEC_FLAG_ALT_SCAN))
1077 && s->codec_id != CODEC_ID_MPEG4 && s->codec_id != CODEC_ID_MPEG2VIDEO){
1078 av_log(avctx, AV_LOG_ERROR, "interlacing not supported by codec\n");
1082 if(s->mpeg_quant && s->codec_id != CODEC_ID_MPEG4){ //FIXME mpeg2 uses that too
1083 av_log(avctx, AV_LOG_ERROR, "mpeg2 style quantization not supported by codec\n");
1087 if((s->flags & CODEC_FLAG_CBP_RD) && !(s->flags & CODEC_FLAG_TRELLIS_QUANT)){
1088 av_log(avctx, AV_LOG_ERROR, "CBP RD needs trellis quant\n");
1092 if((s->flags & CODEC_FLAG_QP_RD) && s->avctx->mb_decision != FF_MB_DECISION_RD){
1093 av_log(avctx, AV_LOG_ERROR, "QP RD needs mbd=2\n");
1097 if(s->avctx->scenechange_threshold < 1000000000 && (s->flags & CODEC_FLAG_CLOSED_GOP)){
1098 av_log(avctx, AV_LOG_ERROR, "closed gop with scene change detection arent supported yet\n");
1102 if((s->flags2 & CODEC_FLAG2_INTRA_VLC) && s->codec_id != CODEC_ID_MPEG2VIDEO){
1103 av_log(avctx, AV_LOG_ERROR, "intra vlc table not supported by codec\n");
1107 if(s->flags & CODEC_FLAG_LOW_DELAY){
1108 if (s->codec_id != CODEC_ID_MPEG2VIDEO && s->codec_id != CODEC_ID_MPEG1VIDEO){
1109 av_log(avctx, AV_LOG_ERROR, "low delay forcing is only available for mpeg1/2\n");
1112 if (s->max_b_frames != 0){
1113 av_log(avctx, AV_LOG_ERROR, "b frames cannot be used with low delay\n");
1118 if(s->avctx->thread_count > 1 && s->codec_id != CODEC_ID_MPEG4
1119 && s->codec_id != CODEC_ID_MPEG1VIDEO && s->codec_id != CODEC_ID_MPEG2VIDEO
1120 && (s->codec_id != CODEC_ID_H263P || !(s->flags & CODEC_FLAG_H263P_SLICE_STRUCT))){
1121 av_log(avctx, AV_LOG_ERROR, "multi threaded encoding not supported by codec\n");
1125 if(s->avctx->thread_count > 1)
1128 if(!avctx->time_base.den || !avctx->time_base.num){
1129 av_log(avctx, AV_LOG_ERROR, "framerate not set\n");
1133 i= (INT_MAX/2+128)>>8;
1134 if(avctx->me_threshold >= i){
1135 av_log(avctx, AV_LOG_ERROR, "me_threshold too large, max is %d\n", i - 1);
1138 if(avctx->mb_threshold >= i){
1139 av_log(avctx, AV_LOG_ERROR, "mb_threshold too large, max is %d\n", i - 1);
1143 if(avctx->b_frame_strategy && (avctx->flags&CODEC_FLAG_PASS2)){
1144 av_log(avctx, AV_LOG_INFO, "notice: b_frame_strategy only affects the first pass\n");
1145 avctx->b_frame_strategy = 0;
1148 i= ff_gcd(avctx->time_base.den, avctx->time_base.num);
1150 av_log(avctx, AV_LOG_INFO, "removing common factors from framerate\n");
1151 avctx->time_base.den /= i;
1152 avctx->time_base.num /= i;
1156 if(s->codec_id==CODEC_ID_MJPEG){
1157 s->intra_quant_bias= 1<<(QUANT_BIAS_SHIFT-1); //(a + x/2)/x
1158 s->inter_quant_bias= 0;
1159 }else if(s->mpeg_quant || s->codec_id==CODEC_ID_MPEG1VIDEO || s->codec_id==CODEC_ID_MPEG2VIDEO){
1160 s->intra_quant_bias= 3<<(QUANT_BIAS_SHIFT-3); //(a + x*3/8)/x
1161 s->inter_quant_bias= 0;
1163 s->intra_quant_bias=0;
1164 s->inter_quant_bias=-(1<<(QUANT_BIAS_SHIFT-2)); //(a - x/4)/x
1167 if(avctx->intra_quant_bias != FF_DEFAULT_QUANT_BIAS)
1168 s->intra_quant_bias= avctx->intra_quant_bias;
1169 if(avctx->inter_quant_bias != FF_DEFAULT_QUANT_BIAS)
1170 s->inter_quant_bias= avctx->inter_quant_bias;
1172 avcodec_get_chroma_sub_sample(avctx->pix_fmt, &chroma_h_shift, &chroma_v_shift);
1174 if(avctx->codec_id == CODEC_ID_MPEG4 && s->avctx->time_base.den > (1<<16)-1){
1175 av_log(avctx, AV_LOG_ERROR, "timebase not supported by mpeg 4 standard\n");
1178 s->time_increment_bits = av_log2(s->avctx->time_base.den - 1) + 1;
1180 switch(avctx->codec->id) {
1181 case CODEC_ID_MPEG1VIDEO:
1182 s->out_format = FMT_MPEG1;
1183 s->low_delay= !!(s->flags & CODEC_FLAG_LOW_DELAY);
1184 avctx->delay= s->low_delay ? 0 : (s->max_b_frames + 1);
1186 case CODEC_ID_MPEG2VIDEO:
1187 s->out_format = FMT_MPEG1;
1188 s->low_delay= !!(s->flags & CODEC_FLAG_LOW_DELAY);
1189 avctx->delay= s->low_delay ? 0 : (s->max_b_frames + 1);
1192 case CODEC_ID_LJPEG:
1193 case CODEC_ID_JPEGLS:
1194 case CODEC_ID_MJPEG:
1195 s->out_format = FMT_MJPEG;
1196 s->intra_only = 1; /* force intra only for jpeg */
1197 s->mjpeg_write_tables = avctx->codec->id != CODEC_ID_JPEGLS;
1198 s->mjpeg_data_only_frames = 0; /* write all the needed headers */
1199 s->mjpeg_vsample[0] = 2;
1200 s->mjpeg_vsample[1] = 2>>chroma_v_shift;
1201 s->mjpeg_vsample[2] = 2>>chroma_v_shift;
1202 s->mjpeg_hsample[0] = 2;
1203 s->mjpeg_hsample[1] = 2>>chroma_h_shift;
1204 s->mjpeg_hsample[2] = 2>>chroma_h_shift;
1205 if (mjpeg_init(s) < 0)
1210 #ifdef CONFIG_H261_ENCODER
1212 if (ff_h261_get_picture_format(s->width, s->height) < 0) {
1213 av_log(avctx, AV_LOG_ERROR, "The specified picture size of %dx%d is not valid for the H.261 codec.\nValid sizes are 176x144, 352x288\n", s->width, s->height);
1216 s->out_format = FMT_H261;
1222 if (h263_get_picture_format(s->width, s->height) == 7) {
1223 av_log(avctx, AV_LOG_INFO, "The specified picture size of %dx%d is not valid for the H.263 codec.\nValid sizes are 128x96, 176x144, 352x288, 704x576, and 1408x1152. Try H.263+.\n", s->width, s->height);
1226 s->out_format = FMT_H263;
1227 s->obmc= (avctx->flags & CODEC_FLAG_OBMC) ? 1:0;
1231 case CODEC_ID_H263P:
1232 s->out_format = FMT_H263;
1235 s->umvplus = (avctx->flags & CODEC_FLAG_H263P_UMV) ? 1:0;
1236 s->h263_aic= (avctx->flags & CODEC_FLAG_H263P_AIC) ? 1:0;
1237 s->modified_quant= s->h263_aic;
1238 s->alt_inter_vlc= (avctx->flags & CODEC_FLAG_H263P_AIV) ? 1:0;
1239 s->obmc= (avctx->flags & CODEC_FLAG_OBMC) ? 1:0;
1240 s->loop_filter= (avctx->flags & CODEC_FLAG_LOOP_FILTER) ? 1:0;
1241 s->unrestricted_mv= s->obmc || s->loop_filter || s->umvplus;
1242 s->h263_slice_structured= (s->flags & CODEC_FLAG_H263P_SLICE_STRUCT) ? 1:0;
1245 /* These are just to be sure */
1250 s->out_format = FMT_H263;
1251 s->h263_flv = 2; /* format = 1; 11-bit codes */
1252 s->unrestricted_mv = 1;
1253 s->rtp_mode=0; /* don't allow GOB */
1258 s->out_format = FMT_H263;
1263 s->out_format = FMT_H263;
1266 s->modified_quant=1;
1270 s->unrestricted_mv= s->obmc || s->loop_filter || s->umvplus;
1272 case CODEC_ID_MPEG4:
1273 s->out_format = FMT_H263;
1275 s->unrestricted_mv = 1;
1276 s->low_delay= s->max_b_frames ? 0 : 1;
1277 avctx->delay= s->low_delay ? 0 : (s->max_b_frames + 1);
1279 case CODEC_ID_MSMPEG4V1:
1280 s->out_format = FMT_H263;
1281 s->h263_msmpeg4 = 1;
1283 s->unrestricted_mv = 1;
1284 s->msmpeg4_version= 1;
1288 case CODEC_ID_MSMPEG4V2:
1289 s->out_format = FMT_H263;
1290 s->h263_msmpeg4 = 1;
1292 s->unrestricted_mv = 1;
1293 s->msmpeg4_version= 2;
1297 case CODEC_ID_MSMPEG4V3:
1298 s->out_format = FMT_H263;
1299 s->h263_msmpeg4 = 1;
1301 s->unrestricted_mv = 1;
1302 s->msmpeg4_version= 3;
1303 s->flipflop_rounding=1;
1308 s->out_format = FMT_H263;
1309 s->h263_msmpeg4 = 1;
1311 s->unrestricted_mv = 1;
1312 s->msmpeg4_version= 4;
1313 s->flipflop_rounding=1;
1318 s->out_format = FMT_H263;
1319 s->h263_msmpeg4 = 1;
1321 s->unrestricted_mv = 1;
1322 s->msmpeg4_version= 5;
1323 s->flipflop_rounding=1;
1331 avctx->has_b_frames= !s->low_delay;
1336 if (MPV_common_init(s) < 0)
1339 if(s->modified_quant)
1340 s->chroma_qscale_table= ff_h263_chroma_qscale_table;
1341 s->progressive_frame=
1342 s->progressive_sequence= !(avctx->flags & (CODEC_FLAG_INTERLACED_DCT|CODEC_FLAG_INTERLACED_ME|CODEC_FLAG_ALT_SCAN));
1343 s->quant_precision=5;
1345 ff_set_cmp(&s->dsp, s->dsp.ildct_cmp, s->avctx->ildct_cmp);
1346 ff_set_cmp(&s->dsp, s->dsp.frame_skip_cmp, s->avctx->frame_skip_cmp);
1348 #ifdef CONFIG_H261_ENCODER
1349 if (s->out_format == FMT_H261)
1350 ff_h261_encode_init(s);
1352 if (s->out_format == FMT_H263)
1353 h263_encode_init(s);
1354 if(s->msmpeg4_version)
1355 ff_msmpeg4_encode_init(s);
1356 if (s->out_format == FMT_MPEG1)
1357 ff_mpeg1_encode_init(s);
1361 int j= s->dsp.idct_permutation[i];
1362 if(s->codec_id==CODEC_ID_MPEG4 && s->mpeg_quant){
1363 s->intra_matrix[j] = ff_mpeg4_default_intra_matrix[i];
1364 s->inter_matrix[j] = ff_mpeg4_default_non_intra_matrix[i];
1365 }else if(s->out_format == FMT_H263 || s->out_format == FMT_H261){
1366 s->intra_matrix[j] =
1367 s->inter_matrix[j] = ff_mpeg1_default_non_intra_matrix[i];
1370 s->intra_matrix[j] = ff_mpeg1_default_intra_matrix[i];
1371 s->inter_matrix[j] = ff_mpeg1_default_non_intra_matrix[i];
1373 if(s->avctx->intra_matrix)
1374 s->intra_matrix[j] = s->avctx->intra_matrix[i];
1375 if(s->avctx->inter_matrix)
1376 s->inter_matrix[j] = s->avctx->inter_matrix[i];
1379 /* precompute matrix */
1380 /* for mjpeg, we do include qscale in the matrix */
1381 if (s->out_format != FMT_MJPEG) {
1382 convert_matrix(&s->dsp, s->q_intra_matrix, s->q_intra_matrix16,
1383 s->intra_matrix, s->intra_quant_bias, avctx->qmin, 31, 1);
1384 convert_matrix(&s->dsp, s->q_inter_matrix, s->q_inter_matrix16,
1385 s->inter_matrix, s->inter_quant_bias, avctx->qmin, 31, 0);
1388 if(ff_rate_control_init(s) < 0)
1394 int MPV_encode_end(AVCodecContext *avctx)
1396 MpegEncContext *s = avctx->priv_data;
1398 ff_rate_control_uninit(s);
1401 if (s->out_format == FMT_MJPEG)
1404 av_freep(&avctx->extradata);
1409 #endif //CONFIG_ENCODERS
1411 void init_rl(RLTable *rl, int use_static)
1413 int8_t max_level[MAX_RUN+1], max_run[MAX_LEVEL+1];
1414 uint8_t index_run[MAX_RUN+1];
1415 int last, run, level, start, end, i;
1417 /* If table is static, we can quit if rl->max_level[0] is not NULL */
1418 if(use_static && rl->max_level[0])
1421 /* compute max_level[], max_run[] and index_run[] */
1422 for(last=0;last<2;last++) {
1431 memset(max_level, 0, MAX_RUN + 1);
1432 memset(max_run, 0, MAX_LEVEL + 1);
1433 memset(index_run, rl->n, MAX_RUN + 1);
1434 for(i=start;i<end;i++) {
1435 run = rl->table_run[i];
1436 level = rl->table_level[i];
1437 if (index_run[run] == rl->n)
1439 if (level > max_level[run])
1440 max_level[run] = level;
1441 if (run > max_run[level])
1442 max_run[level] = run;
1445 rl->max_level[last] = av_mallocz_static(MAX_RUN + 1);
1447 rl->max_level[last] = av_malloc(MAX_RUN + 1);
1448 memcpy(rl->max_level[last], max_level, MAX_RUN + 1);
1450 rl->max_run[last] = av_mallocz_static(MAX_LEVEL + 1);
1452 rl->max_run[last] = av_malloc(MAX_LEVEL + 1);
1453 memcpy(rl->max_run[last], max_run, MAX_LEVEL + 1);
1455 rl->index_run[last] = av_mallocz_static(MAX_RUN + 1);
1457 rl->index_run[last] = av_malloc(MAX_RUN + 1);
1458 memcpy(rl->index_run[last], index_run, MAX_RUN + 1);
1462 /* draw the edges of width 'w' of an image of size width, height */
1463 //FIXME check that this is ok for mpeg4 interlaced
1464 static void draw_edges_c(uint8_t *buf, int wrap, int width, int height, int w)
1466 uint8_t *ptr, *last_line;
1469 last_line = buf + (height - 1) * wrap;
1471 /* top and bottom */
1472 memcpy(buf - (i + 1) * wrap, buf, width);
1473 memcpy(last_line + (i + 1) * wrap, last_line, width);
1475 /* left and right */
1477 for(i=0;i<height;i++) {
1478 memset(ptr - w, ptr[0], w);
1479 memset(ptr + width, ptr[width-1], w);
1484 memset(buf - (i + 1) * wrap - w, buf[0], w); /* top left */
1485 memset(buf - (i + 1) * wrap + width, buf[width-1], w); /* top right */
1486 memset(last_line + (i + 1) * wrap - w, last_line[0], w); /* top left */
1487 memset(last_line + (i + 1) * wrap + width, last_line[width-1], w); /* top right */
1491 int ff_find_unused_picture(MpegEncContext *s, int shared){
1495 for(i=0; i<MAX_PICTURE_COUNT; i++){
1496 if(s->picture[i].data[0]==NULL && s->picture[i].type==0) return i;
1499 for(i=0; i<MAX_PICTURE_COUNT; i++){
1500 if(s->picture[i].data[0]==NULL && s->picture[i].type!=0) return i; //FIXME
1502 for(i=0; i<MAX_PICTURE_COUNT; i++){
1503 if(s->picture[i].data[0]==NULL) return i;
1511 static void update_noise_reduction(MpegEncContext *s){
1514 for(intra=0; intra<2; intra++){
1515 if(s->dct_count[intra] > (1<<16)){
1516 for(i=0; i<64; i++){
1517 s->dct_error_sum[intra][i] >>=1;
1519 s->dct_count[intra] >>= 1;
1522 for(i=0; i<64; i++){
1523 s->dct_offset[intra][i]= (s->avctx->noise_reduction * s->dct_count[intra] + s->dct_error_sum[intra][i]/2) / (s->dct_error_sum[intra][i]+1);
1529 * generic function for encode/decode called after coding/decoding the header and before a frame is coded/decoded
1531 int MPV_frame_start(MpegEncContext *s, AVCodecContext *avctx)
1537 assert(s->last_picture_ptr==NULL || s->out_format != FMT_H264 || s->codec_id == CODEC_ID_SVQ3);
1539 /* mark&release old frames */
1540 if (s->pict_type != B_TYPE && s->last_picture_ptr && s->last_picture_ptr != s->next_picture_ptr && s->last_picture_ptr->data[0]) {
1541 avctx->release_buffer(avctx, (AVFrame*)s->last_picture_ptr);
1543 /* release forgotten pictures */
1544 /* if(mpeg124/h263) */
1546 for(i=0; i<MAX_PICTURE_COUNT; i++){
1547 if(s->picture[i].data[0] && &s->picture[i] != s->next_picture_ptr && s->picture[i].reference){
1548 av_log(avctx, AV_LOG_ERROR, "releasing zombie picture\n");
1549 avctx->release_buffer(avctx, (AVFrame*)&s->picture[i]);
1556 /* release non reference frames */
1557 for(i=0; i<MAX_PICTURE_COUNT; i++){
1558 if(s->picture[i].data[0] && !s->picture[i].reference /*&& s->picture[i].type!=FF_BUFFER_TYPE_SHARED*/){
1559 s->avctx->release_buffer(s->avctx, (AVFrame*)&s->picture[i]);
1563 if(s->current_picture_ptr && s->current_picture_ptr->data[0]==NULL)
1564 pic= (AVFrame*)s->current_picture_ptr; //we allready have a unused image (maybe it was set before reading the header)
1566 i= ff_find_unused_picture(s, 0);
1567 pic= (AVFrame*)&s->picture[i];
1570 pic->reference= (s->pict_type != B_TYPE || s->codec_id == CODEC_ID_H264)
1571 && !s->dropable ? 3 : 0;
1573 pic->coded_picture_number= s->coded_picture_number++;
1575 if( alloc_picture(s, (Picture*)pic, 0) < 0)
1578 s->current_picture_ptr= (Picture*)pic;
1579 s->current_picture_ptr->top_field_first= s->top_field_first; //FIXME use only the vars from current_pic
1580 s->current_picture_ptr->interlaced_frame= !s->progressive_frame && !s->progressive_sequence;
1583 s->current_picture_ptr->pict_type= s->pict_type;
1584 // if(s->flags && CODEC_FLAG_QSCALE)
1585 // s->current_picture_ptr->quality= s->new_picture_ptr->quality;
1586 s->current_picture_ptr->key_frame= s->pict_type == I_TYPE;
1588 copy_picture(&s->current_picture, s->current_picture_ptr);
1590 if(s->out_format != FMT_H264 || s->codec_id == CODEC_ID_SVQ3){
1591 if (s->pict_type != B_TYPE) {
1592 s->last_picture_ptr= s->next_picture_ptr;
1594 s->next_picture_ptr= s->current_picture_ptr;
1596 /* av_log(s->avctx, AV_LOG_DEBUG, "L%p N%p C%p L%p N%p C%p type:%d drop:%d\n", s->last_picture_ptr, s->next_picture_ptr,s->current_picture_ptr,
1597 s->last_picture_ptr ? s->last_picture_ptr->data[0] : NULL,
1598 s->next_picture_ptr ? s->next_picture_ptr->data[0] : NULL,
1599 s->current_picture_ptr ? s->current_picture_ptr->data[0] : NULL,
1600 s->pict_type, s->dropable);*/
1602 if(s->last_picture_ptr) copy_picture(&s->last_picture, s->last_picture_ptr);
1603 if(s->next_picture_ptr) copy_picture(&s->next_picture, s->next_picture_ptr);
1605 if(s->pict_type != I_TYPE && (s->last_picture_ptr==NULL || s->last_picture_ptr->data[0]==NULL)){
1606 av_log(avctx, AV_LOG_ERROR, "warning: first frame is no keyframe\n");
1607 assert(s->pict_type != B_TYPE); //these should have been dropped if we don't have a reference
1611 assert(s->pict_type == I_TYPE || (s->last_picture_ptr && s->last_picture_ptr->data[0]));
1613 if(s->picture_structure!=PICT_FRAME){
1616 if(s->picture_structure == PICT_BOTTOM_FIELD){
1617 s->current_picture.data[i] += s->current_picture.linesize[i];
1619 s->current_picture.linesize[i] *= 2;
1620 s->last_picture.linesize[i] *=2;
1621 s->next_picture.linesize[i] *=2;
1626 s->hurry_up= s->avctx->hurry_up;
1627 s->error_resilience= avctx->error_resilience;
1629 /* set dequantizer, we can't do it during init as it might change for mpeg4
1630 and we can't do it in the header decode as init isnt called for mpeg4 there yet */
1631 if(s->mpeg_quant || s->codec_id == CODEC_ID_MPEG2VIDEO){
1632 s->dct_unquantize_intra = s->dct_unquantize_mpeg2_intra;
1633 s->dct_unquantize_inter = s->dct_unquantize_mpeg2_inter;
1634 }else if(s->out_format == FMT_H263 || s->out_format == FMT_H261){
1635 s->dct_unquantize_intra = s->dct_unquantize_h263_intra;
1636 s->dct_unquantize_inter = s->dct_unquantize_h263_inter;
1638 s->dct_unquantize_intra = s->dct_unquantize_mpeg1_intra;
1639 s->dct_unquantize_inter = s->dct_unquantize_mpeg1_inter;
1642 if(s->dct_error_sum){
1643 assert(s->avctx->noise_reduction && s->encoding);
1645 update_noise_reduction(s);
1649 if(s->avctx->xvmc_acceleration)
1650 return XVMC_field_start(s, avctx);
1655 /* generic function for encode/decode called after a frame has been coded/decoded */
1656 void MPV_frame_end(MpegEncContext *s)
1659 /* draw edge for correct motion prediction if outside */
1661 //just to make sure that all data is rendered.
1662 if(s->avctx->xvmc_acceleration){
1666 if(s->unrestricted_mv && s->current_picture.reference && !s->intra_only && !(s->flags&CODEC_FLAG_EMU_EDGE)) {
1667 draw_edges(s->current_picture.data[0], s->linesize , s->h_edge_pos , s->v_edge_pos , EDGE_WIDTH );
1668 draw_edges(s->current_picture.data[1], s->uvlinesize, s->h_edge_pos>>1, s->v_edge_pos>>1, EDGE_WIDTH/2);
1669 draw_edges(s->current_picture.data[2], s->uvlinesize, s->h_edge_pos>>1, s->v_edge_pos>>1, EDGE_WIDTH/2);
1673 s->last_pict_type = s->pict_type;
1674 s->last_lambda_for[s->pict_type]= s->current_picture_ptr->quality;
1675 if(s->pict_type!=B_TYPE){
1676 s->last_non_b_pict_type= s->pict_type;
1679 /* copy back current_picture variables */
1680 for(i=0; i<MAX_PICTURE_COUNT; i++){
1681 if(s->picture[i].data[0] == s->current_picture.data[0]){
1682 s->picture[i]= s->current_picture;
1686 assert(i<MAX_PICTURE_COUNT);
1690 /* release non-reference frames */
1691 for(i=0; i<MAX_PICTURE_COUNT; i++){
1692 if(s->picture[i].data[0] && !s->picture[i].reference /*&& s->picture[i].type!=FF_BUFFER_TYPE_SHARED*/){
1693 s->avctx->release_buffer(s->avctx, (AVFrame*)&s->picture[i]);
1697 // clear copies, to avoid confusion
1699 memset(&s->last_picture, 0, sizeof(Picture));
1700 memset(&s->next_picture, 0, sizeof(Picture));
1701 memset(&s->current_picture, 0, sizeof(Picture));
1703 s->avctx->coded_frame= (AVFrame*)s->current_picture_ptr;
1707 * draws an line from (ex, ey) -> (sx, sy).
1708 * @param w width of the image
1709 * @param h height of the image
1710 * @param stride stride/linesize of the image
1711 * @param color color of the arrow
1713 static void draw_line(uint8_t *buf, int sx, int sy, int ex, int ey, int w, int h, int stride, int color){
1716 sx= clip(sx, 0, w-1);
1717 sy= clip(sy, 0, h-1);
1718 ex= clip(ex, 0, w-1);
1719 ey= clip(ey, 0, h-1);
1721 buf[sy*stride + sx]+= color;
1723 if(FFABS(ex - sx) > FFABS(ey - sy)){
1725 FFSWAP(int, sx, ex);
1726 FFSWAP(int, sy, ey);
1728 buf+= sx + sy*stride;
1730 f= ((ey-sy)<<16)/ex;
1731 for(x= 0; x <= ex; x++){
1734 buf[ y *stride + x]+= (color*(0x10000-fr))>>16;
1735 buf[(y+1)*stride + x]+= (color* fr )>>16;
1739 FFSWAP(int, sx, ex);
1740 FFSWAP(int, sy, ey);
1742 buf+= sx + sy*stride;
1744 if(ey) f= ((ex-sx)<<16)/ey;
1746 for(y= 0; y <= ey; y++){
1749 buf[y*stride + x ]+= (color*(0x10000-fr))>>16;;
1750 buf[y*stride + x+1]+= (color* fr )>>16;;
1756 * draws an arrow from (ex, ey) -> (sx, sy).
1757 * @param w width of the image
1758 * @param h height of the image
1759 * @param stride stride/linesize of the image
1760 * @param color color of the arrow
1762 static void draw_arrow(uint8_t *buf, int sx, int sy, int ex, int ey, int w, int h, int stride, int color){
1765 sx= clip(sx, -100, w+100);
1766 sy= clip(sy, -100, h+100);
1767 ex= clip(ex, -100, w+100);
1768 ey= clip(ey, -100, h+100);
1773 if(dx*dx + dy*dy > 3*3){
1776 int length= ff_sqrt((rx*rx + ry*ry)<<8);
1778 //FIXME subpixel accuracy
1779 rx= ROUNDED_DIV(rx*3<<4, length);
1780 ry= ROUNDED_DIV(ry*3<<4, length);
1782 draw_line(buf, sx, sy, sx + rx, sy + ry, w, h, stride, color);
1783 draw_line(buf, sx, sy, sx - ry, sy + rx, w, h, stride, color);
1785 draw_line(buf, sx, sy, ex, ey, w, h, stride, color);
1789 * prints debuging info for the given picture.
1791 void ff_print_debug_info(MpegEncContext *s, AVFrame *pict){
1793 if(!pict || !pict->mb_type) return;
1795 if(s->avctx->debug&(FF_DEBUG_SKIP | FF_DEBUG_QP | FF_DEBUG_MB_TYPE)){
1798 av_log(s->avctx,AV_LOG_DEBUG,"New frame, type: ");
1799 switch (pict->pict_type) {
1800 case FF_I_TYPE: av_log(s->avctx,AV_LOG_DEBUG,"I\n"); break;
1801 case FF_P_TYPE: av_log(s->avctx,AV_LOG_DEBUG,"P\n"); break;
1802 case FF_B_TYPE: av_log(s->avctx,AV_LOG_DEBUG,"B\n"); break;
1803 case FF_S_TYPE: av_log(s->avctx,AV_LOG_DEBUG,"S\n"); break;
1804 case FF_SI_TYPE: av_log(s->avctx,AV_LOG_DEBUG,"SI\n"); break;
1805 case FF_SP_TYPE: av_log(s->avctx,AV_LOG_DEBUG,"SP\n"); break;
1807 for(y=0; y<s->mb_height; y++){
1808 for(x=0; x<s->mb_width; x++){
1809 if(s->avctx->debug&FF_DEBUG_SKIP){
1810 int count= s->mbskip_table[x + y*s->mb_stride];
1811 if(count>9) count=9;
1812 av_log(s->avctx, AV_LOG_DEBUG, "%1d", count);
1814 if(s->avctx->debug&FF_DEBUG_QP){
1815 av_log(s->avctx, AV_LOG_DEBUG, "%2d", pict->qscale_table[x + y*s->mb_stride]);
1817 if(s->avctx->debug&FF_DEBUG_MB_TYPE){
1818 int mb_type= pict->mb_type[x + y*s->mb_stride];
1819 //Type & MV direction
1821 av_log(s->avctx, AV_LOG_DEBUG, "P");
1822 else if(IS_INTRA(mb_type) && IS_ACPRED(mb_type))
1823 av_log(s->avctx, AV_LOG_DEBUG, "A");
1824 else if(IS_INTRA4x4(mb_type))
1825 av_log(s->avctx, AV_LOG_DEBUG, "i");
1826 else if(IS_INTRA16x16(mb_type))
1827 av_log(s->avctx, AV_LOG_DEBUG, "I");
1828 else if(IS_DIRECT(mb_type) && IS_SKIP(mb_type))
1829 av_log(s->avctx, AV_LOG_DEBUG, "d");
1830 else if(IS_DIRECT(mb_type))
1831 av_log(s->avctx, AV_LOG_DEBUG, "D");
1832 else if(IS_GMC(mb_type) && IS_SKIP(mb_type))
1833 av_log(s->avctx, AV_LOG_DEBUG, "g");
1834 else if(IS_GMC(mb_type))
1835 av_log(s->avctx, AV_LOG_DEBUG, "G");
1836 else if(IS_SKIP(mb_type))
1837 av_log(s->avctx, AV_LOG_DEBUG, "S");
1838 else if(!USES_LIST(mb_type, 1))
1839 av_log(s->avctx, AV_LOG_DEBUG, ">");
1840 else if(!USES_LIST(mb_type, 0))
1841 av_log(s->avctx, AV_LOG_DEBUG, "<");
1843 assert(USES_LIST(mb_type, 0) && USES_LIST(mb_type, 1));
1844 av_log(s->avctx, AV_LOG_DEBUG, "X");
1849 av_log(s->avctx, AV_LOG_DEBUG, "+");
1850 else if(IS_16X8(mb_type))
1851 av_log(s->avctx, AV_LOG_DEBUG, "-");
1852 else if(IS_8X16(mb_type))
1853 av_log(s->avctx, AV_LOG_DEBUG, "|");
1854 else if(IS_INTRA(mb_type) || IS_16X16(mb_type))
1855 av_log(s->avctx, AV_LOG_DEBUG, " ");
1857 av_log(s->avctx, AV_LOG_DEBUG, "?");
1860 if(IS_INTERLACED(mb_type) && s->codec_id == CODEC_ID_H264)
1861 av_log(s->avctx, AV_LOG_DEBUG, "=");
1863 av_log(s->avctx, AV_LOG_DEBUG, " ");
1865 // av_log(s->avctx, AV_LOG_DEBUG, " ");
1867 av_log(s->avctx, AV_LOG_DEBUG, "\n");
1871 if((s->avctx->debug&(FF_DEBUG_VIS_QP|FF_DEBUG_VIS_MB_TYPE)) || (s->avctx->debug_mv)){
1872 const int shift= 1 + s->quarter_sample;
1876 int h_chroma_shift, v_chroma_shift;
1877 const int width = s->avctx->width;
1878 const int height= s->avctx->height;
1879 const int mv_sample_log2= 4 - pict->motion_subsample_log2;
1880 const int mv_stride= (s->mb_width << mv_sample_log2) + (s->codec_id == CODEC_ID_H264 ? 0 : 1);
1881 s->low_delay=0; //needed to see the vectors without trashing the buffers
1883 avcodec_get_chroma_sub_sample(s->avctx->pix_fmt, &h_chroma_shift, &v_chroma_shift);
1885 memcpy(s->visualization_buffer[i], pict->data[i], (i==0) ? pict->linesize[i]*height:pict->linesize[i]*height >> v_chroma_shift);
1886 pict->data[i]= s->visualization_buffer[i];
1888 pict->type= FF_BUFFER_TYPE_COPY;
1891 for(mb_y=0; mb_y<s->mb_height; mb_y++){
1893 for(mb_x=0; mb_x<s->mb_width; mb_x++){
1894 const int mb_index= mb_x + mb_y*s->mb_stride;
1895 if((s->avctx->debug_mv) && pict->motion_val){
1897 for(type=0; type<3; type++){
1900 case 0: if ((!(s->avctx->debug_mv&FF_DEBUG_VIS_MV_P_FOR)) || (pict->pict_type!=FF_P_TYPE))
1904 case 1: if ((!(s->avctx->debug_mv&FF_DEBUG_VIS_MV_B_FOR)) || (pict->pict_type!=FF_B_TYPE))
1908 case 2: if ((!(s->avctx->debug_mv&FF_DEBUG_VIS_MV_B_BACK)) || (pict->pict_type!=FF_B_TYPE))
1913 if(!USES_LIST(pict->mb_type[mb_index], direction))
1916 if(IS_8X8(pict->mb_type[mb_index])){
1919 int sx= mb_x*16 + 4 + 8*(i&1);
1920 int sy= mb_y*16 + 4 + 8*(i>>1);
1921 int xy= (mb_x*2 + (i&1) + (mb_y*2 + (i>>1))*mv_stride) << (mv_sample_log2-1);
1922 int mx= (pict->motion_val[direction][xy][0]>>shift) + sx;
1923 int my= (pict->motion_val[direction][xy][1]>>shift) + sy;
1924 draw_arrow(ptr, sx, sy, mx, my, width, height, s->linesize, 100);
1926 }else if(IS_16X8(pict->mb_type[mb_index])){
1930 int sy=mb_y*16 + 4 + 8*i;
1931 int xy= (mb_x*2 + (mb_y*2 + i)*mv_stride) << (mv_sample_log2-1);
1932 int mx=(pict->motion_val[direction][xy][0]>>shift);
1933 int my=(pict->motion_val[direction][xy][1]>>shift);
1935 if(IS_INTERLACED(pict->mb_type[mb_index]))
1938 draw_arrow(ptr, sx, sy, mx+sx, my+sy, width, height, s->linesize, 100);
1940 }else if(IS_8X16(pict->mb_type[mb_index])){
1943 int sx=mb_x*16 + 4 + 8*i;
1945 int xy= (mb_x*2 + i + mb_y*2*mv_stride) << (mv_sample_log2-1);
1946 int mx=(pict->motion_val[direction][xy][0]>>shift);
1947 int my=(pict->motion_val[direction][xy][1]>>shift);
1949 if(IS_INTERLACED(pict->mb_type[mb_index]))
1952 draw_arrow(ptr, sx, sy, mx+sx, my+sy, width, height, s->linesize, 100);
1955 int sx= mb_x*16 + 8;
1956 int sy= mb_y*16 + 8;
1957 int xy= (mb_x + mb_y*mv_stride) << mv_sample_log2;
1958 int mx= (pict->motion_val[direction][xy][0]>>shift) + sx;
1959 int my= (pict->motion_val[direction][xy][1]>>shift) + sy;
1960 draw_arrow(ptr, sx, sy, mx, my, width, height, s->linesize, 100);
1964 if((s->avctx->debug&FF_DEBUG_VIS_QP) && pict->motion_val){
1965 uint64_t c= (pict->qscale_table[mb_index]*128/31) * 0x0101010101010101ULL;
1968 *(uint64_t*)(pict->data[1] + 8*mb_x + (8*mb_y + y)*pict->linesize[1])= c;
1969 *(uint64_t*)(pict->data[2] + 8*mb_x + (8*mb_y + y)*pict->linesize[2])= c;
1972 if((s->avctx->debug&FF_DEBUG_VIS_MB_TYPE) && pict->motion_val){
1973 int mb_type= pict->mb_type[mb_index];
1976 #define COLOR(theta, r)\
1977 u= (int)(128 + r*cos(theta*3.141592/180));\
1978 v= (int)(128 + r*sin(theta*3.141592/180));
1982 if(IS_PCM(mb_type)){
1984 }else if((IS_INTRA(mb_type) && IS_ACPRED(mb_type)) || IS_INTRA16x16(mb_type)){
1986 }else if(IS_INTRA4x4(mb_type)){
1988 }else if(IS_DIRECT(mb_type) && IS_SKIP(mb_type)){
1990 }else if(IS_DIRECT(mb_type)){
1992 }else if(IS_GMC(mb_type) && IS_SKIP(mb_type)){
1994 }else if(IS_GMC(mb_type)){
1996 }else if(IS_SKIP(mb_type)){
1998 }else if(!USES_LIST(mb_type, 1)){
2000 }else if(!USES_LIST(mb_type, 0)){
2003 assert(USES_LIST(mb_type, 0) && USES_LIST(mb_type, 1));
2007 u*= 0x0101010101010101ULL;
2008 v*= 0x0101010101010101ULL;
2010 *(uint64_t*)(pict->data[1] + 8*mb_x + (8*mb_y + y)*pict->linesize[1])= u;
2011 *(uint64_t*)(pict->data[2] + 8*mb_x + (8*mb_y + y)*pict->linesize[2])= v;
2015 if(IS_8X8(mb_type) || IS_16X8(mb_type)){
2016 *(uint64_t*)(pict->data[0] + 16*mb_x + 0 + (16*mb_y + 8)*pict->linesize[0])^= 0x8080808080808080ULL;
2017 *(uint64_t*)(pict->data[0] + 16*mb_x + 8 + (16*mb_y + 8)*pict->linesize[0])^= 0x8080808080808080ULL;
2019 if(IS_8X8(mb_type) || IS_8X16(mb_type)){
2021 pict->data[0][16*mb_x + 8 + (16*mb_y + y)*pict->linesize[0]]^= 0x80;
2023 if(IS_8X8(mb_type) && mv_sample_log2 >= 2){
2024 int dm= 1 << (mv_sample_log2-2);
2026 int sx= mb_x*16 + 8*(i&1);
2027 int sy= mb_y*16 + 8*(i>>1);
2028 int xy= (mb_x*2 + (i&1) + (mb_y*2 + (i>>1))*mv_stride) << (mv_sample_log2-1);
2030 int32_t *mv = (int32_t*)&pict->motion_val[0][xy];
2031 if(mv[0] != mv[dm] || mv[dm*mv_stride] != mv[dm*(mv_stride+1)])
2033 pict->data[0][sx + 4 + (sy + y)*pict->linesize[0]]^= 0x80;
2034 if(mv[0] != mv[dm*mv_stride] || mv[dm] != mv[dm*(mv_stride+1)])
2035 *(uint64_t*)(pict->data[0] + sx + (sy + 4)*pict->linesize[0])^= 0x8080808080808080ULL;
2039 if(IS_INTERLACED(mb_type) && s->codec_id == CODEC_ID_H264){
2043 s->mbskip_table[mb_index]=0;
2049 #ifdef CONFIG_ENCODERS
2051 static int get_sae(uint8_t *src, int ref, int stride){
2055 for(y=0; y<16; y++){
2056 for(x=0; x<16; x++){
2057 acc+= FFABS(src[x+y*stride] - ref);
2064 static int get_intra_count(MpegEncContext *s, uint8_t *src, uint8_t *ref, int stride){
2071 for(y=0; y<h; y+=16){
2072 for(x=0; x<w; x+=16){
2073 int offset= x + y*stride;
2074 int sad = s->dsp.sad[0](NULL, src + offset, ref + offset, stride, 16);
2075 int mean= (s->dsp.pix_sum(src + offset, stride) + 128)>>8;
2076 int sae = get_sae(src + offset, mean, stride);
2078 acc+= sae + 500 < sad;
2085 static int load_input_picture(MpegEncContext *s, AVFrame *pic_arg){
2089 const int encoding_delay= s->max_b_frames;
2094 pic_arg->display_picture_number= s->input_picture_number++;
2096 if(pts != AV_NOPTS_VALUE){
2097 if(s->user_specified_pts != AV_NOPTS_VALUE){
2099 int64_t last= s->user_specified_pts;
2102 av_log(s->avctx, AV_LOG_ERROR, "Error, Invalid timestamp=%"PRId64", last=%"PRId64"\n", pts, s->user_specified_pts);
2106 s->user_specified_pts= pts;
2108 if(s->user_specified_pts != AV_NOPTS_VALUE){
2109 s->user_specified_pts=
2110 pts= s->user_specified_pts + 1;
2111 av_log(s->avctx, AV_LOG_INFO, "Warning: AVFrame.pts=? trying to guess (%"PRId64")\n", pts);
2113 pts= pic_arg->display_picture_number;
2119 if(encoding_delay && !(s->flags&CODEC_FLAG_INPUT_PRESERVED)) direct=0;
2120 if(pic_arg->linesize[0] != s->linesize) direct=0;
2121 if(pic_arg->linesize[1] != s->uvlinesize) direct=0;
2122 if(pic_arg->linesize[2] != s->uvlinesize) direct=0;
2124 // av_log(AV_LOG_DEBUG, "%d %d %d %d\n",pic_arg->linesize[0], pic_arg->linesize[1], s->linesize, s->uvlinesize);
2127 i= ff_find_unused_picture(s, 1);
2129 pic= (AVFrame*)&s->picture[i];
2133 pic->data[i]= pic_arg->data[i];
2134 pic->linesize[i]= pic_arg->linesize[i];
2136 alloc_picture(s, (Picture*)pic, 1);
2138 i= ff_find_unused_picture(s, 0);
2140 pic= (AVFrame*)&s->picture[i];
2143 alloc_picture(s, (Picture*)pic, 0);
2145 if( pic->data[0] + INPLACE_OFFSET == pic_arg->data[0]
2146 && pic->data[1] + INPLACE_OFFSET == pic_arg->data[1]
2147 && pic->data[2] + INPLACE_OFFSET == pic_arg->data[2]){
2150 int h_chroma_shift, v_chroma_shift;
2151 avcodec_get_chroma_sub_sample(s->avctx->pix_fmt, &h_chroma_shift, &v_chroma_shift);
2154 int src_stride= pic_arg->linesize[i];
2155 int dst_stride= i ? s->uvlinesize : s->linesize;
2156 int h_shift= i ? h_chroma_shift : 0;
2157 int v_shift= i ? v_chroma_shift : 0;
2158 int w= s->width >>h_shift;
2159 int h= s->height>>v_shift;
2160 uint8_t *src= pic_arg->data[i];
2161 uint8_t *dst= pic->data[i];
2163 if(!s->avctx->rc_buffer_size)
2164 dst +=INPLACE_OFFSET;
2166 if(src_stride==dst_stride)
2167 memcpy(dst, src, src_stride*h);
2170 memcpy(dst, src, w);
2178 copy_picture_attributes(s, pic, pic_arg);
2179 pic->pts= pts; //we set this here to avoid modifiying pic_arg
2182 /* shift buffer entries */
2183 for(i=1; i<MAX_PICTURE_COUNT /*s->encoding_delay+1*/; i++)
2184 s->input_picture[i-1]= s->input_picture[i];
2186 s->input_picture[encoding_delay]= (Picture*)pic;
2191 static int skip_check(MpegEncContext *s, Picture *p, Picture *ref){
2196 for(plane=0; plane<3; plane++){
2197 const int stride= p->linesize[plane];
2198 const int bw= plane ? 1 : 2;
2199 for(y=0; y<s->mb_height*bw; y++){
2200 for(x=0; x<s->mb_width*bw; x++){
2201 int off= p->type == FF_BUFFER_TYPE_SHARED ? 0: 16;
2202 int v= s->dsp.frame_skip_cmp[1](s, p->data[plane] + 8*(x + y*stride)+off, ref->data[plane] + 8*(x + y*stride), stride, 8);
2204 switch(s->avctx->frame_skip_exp){
2205 case 0: score= FFMAX(score, v); break;
2206 case 1: score+= FFABS(v);break;
2207 case 2: score+= v*v;break;
2208 case 3: score64+= FFABS(v*v*(int64_t)v);break;
2209 case 4: score64+= v*v*(int64_t)(v*v);break;
2215 if(score) score64= score;
2217 if(score64 < s->avctx->frame_skip_threshold)
2219 if(score64 < ((s->avctx->frame_skip_factor * (int64_t)s->lambda)>>8))
2224 static int estimate_best_b_count(MpegEncContext *s){
2225 AVCodec *codec= avcodec_find_encoder(s->avctx->codec_id);
2226 AVCodecContext *c= avcodec_alloc_context();
2227 AVFrame input[FF_MAX_B_FRAMES+2];
2228 const int scale= s->avctx->brd_scale;
2229 int i, j, out_size, p_lambda, b_lambda, lambda2;
2230 int outbuf_size= s->width * s->height; //FIXME
2231 uint8_t *outbuf= av_malloc(outbuf_size);
2232 int64_t best_rd= INT64_MAX;
2233 int best_b_count= -1;
2235 assert(scale>=0 && scale <=3);
2238 p_lambda= s->last_lambda_for[P_TYPE]; //s->next_picture_ptr->quality;
2239 b_lambda= s->last_lambda_for[B_TYPE]; //p_lambda *FFABS(s->avctx->b_quant_factor) + s->avctx->b_quant_offset;
2240 if(!b_lambda) b_lambda= p_lambda; //FIXME we should do this somewhere else
2241 lambda2= (b_lambda*b_lambda + (1<<FF_LAMBDA_SHIFT)/2 ) >> FF_LAMBDA_SHIFT;
2243 c->width = s->width >> scale;
2244 c->height= s->height>> scale;
2245 c->flags= CODEC_FLAG_QSCALE | CODEC_FLAG_PSNR | CODEC_FLAG_INPUT_PRESERVED /*| CODEC_FLAG_EMU_EDGE*/;
2246 c->flags|= s->avctx->flags & CODEC_FLAG_QPEL;
2247 c->mb_decision= s->avctx->mb_decision;
2248 c->me_cmp= s->avctx->me_cmp;
2249 c->mb_cmp= s->avctx->mb_cmp;
2250 c->me_sub_cmp= s->avctx->me_sub_cmp;
2251 c->pix_fmt = PIX_FMT_YUV420P;
2252 c->time_base= s->avctx->time_base;
2253 c->max_b_frames= s->max_b_frames;
2255 if (avcodec_open(c, codec) < 0)
2258 for(i=0; i<s->max_b_frames+2; i++){
2259 int ysize= c->width*c->height;
2260 int csize= (c->width/2)*(c->height/2);
2261 Picture pre_input, *pre_input_ptr= i ? s->input_picture[i-1] : s->next_picture_ptr;
2264 pre_input= *pre_input_ptr;
2266 if(pre_input.type != FF_BUFFER_TYPE_SHARED && i){
2267 pre_input.data[0]+=INPLACE_OFFSET;
2268 pre_input.data[1]+=INPLACE_OFFSET;
2269 pre_input.data[2]+=INPLACE_OFFSET;
2272 avcodec_get_frame_defaults(&input[i]);
2273 input[i].data[0]= av_malloc(ysize + 2*csize);
2274 input[i].data[1]= input[i].data[0] + ysize;
2275 input[i].data[2]= input[i].data[1] + csize;
2276 input[i].linesize[0]= c->width;
2277 input[i].linesize[1]=
2278 input[i].linesize[2]= c->width/2;
2280 if(!i || s->input_picture[i-1]){
2281 s->dsp.shrink[scale](input[i].data[0], input[i].linesize[0], pre_input.data[0], pre_input.linesize[0], c->width, c->height);
2282 s->dsp.shrink[scale](input[i].data[1], input[i].linesize[1], pre_input.data[1], pre_input.linesize[1], c->width>>1, c->height>>1);
2283 s->dsp.shrink[scale](input[i].data[2], input[i].linesize[2], pre_input.data[2], pre_input.linesize[2], c->width>>1, c->height>>1);
2287 for(j=0; j<s->max_b_frames+1; j++){
2290 if(!s->input_picture[j])
2293 c->error[0]= c->error[1]= c->error[2]= 0;
2295 input[0].pict_type= I_TYPE;
2296 input[0].quality= 1 * FF_QP2LAMBDA;
2297 out_size = avcodec_encode_video(c, outbuf, outbuf_size, &input[0]);
2298 // rd += (out_size * lambda2) >> FF_LAMBDA_SHIFT;
2300 for(i=0; i<s->max_b_frames+1; i++){
2301 int is_p= i % (j+1) == j || i==s->max_b_frames;
2303 input[i+1].pict_type= is_p ? P_TYPE : B_TYPE;
2304 input[i+1].quality= is_p ? p_lambda : b_lambda;
2305 out_size = avcodec_encode_video(c, outbuf, outbuf_size, &input[i+1]);
2306 rd += (out_size * lambda2) >> (FF_LAMBDA_SHIFT - 3);
2309 /* get the delayed frames */
2311 out_size = avcodec_encode_video(c, outbuf, outbuf_size, NULL);
2312 rd += (out_size * lambda2) >> (FF_LAMBDA_SHIFT - 3);
2315 rd += c->error[0] + c->error[1] + c->error[2];
2327 for(i=0; i<s->max_b_frames+2; i++){
2328 av_freep(&input[i].data[0]);
2331 return best_b_count;
2334 static void select_input_picture(MpegEncContext *s){
2337 for(i=1; i<MAX_PICTURE_COUNT; i++)
2338 s->reordered_input_picture[i-1]= s->reordered_input_picture[i];
2339 s->reordered_input_picture[MAX_PICTURE_COUNT-1]= NULL;
2341 /* set next picture type & ordering */
2342 if(s->reordered_input_picture[0]==NULL && s->input_picture[0]){
2343 if(/*s->picture_in_gop_number >= s->gop_size ||*/ s->next_picture_ptr==NULL || s->intra_only){
2344 s->reordered_input_picture[0]= s->input_picture[0];
2345 s->reordered_input_picture[0]->pict_type= I_TYPE;
2346 s->reordered_input_picture[0]->coded_picture_number= s->coded_picture_number++;
2350 if(s->avctx->frame_skip_threshold || s->avctx->frame_skip_factor){
2351 if(s->picture_in_gop_number < s->gop_size && skip_check(s, s->input_picture[0], s->next_picture_ptr)){
2352 //FIXME check that te gop check above is +-1 correct
2353 //av_log(NULL, AV_LOG_DEBUG, "skip %p %"PRId64"\n", s->input_picture[0]->data[0], s->input_picture[0]->pts);
2355 if(s->input_picture[0]->type == FF_BUFFER_TYPE_SHARED){
2357 s->input_picture[0]->data[i]= NULL;
2358 s->input_picture[0]->type= 0;
2360 assert( s->input_picture[0]->type==FF_BUFFER_TYPE_USER
2361 || s->input_picture[0]->type==FF_BUFFER_TYPE_INTERNAL);
2363 s->avctx->release_buffer(s->avctx, (AVFrame*)s->input_picture[0]);
2367 ff_vbv_update(s, 0);
2373 if(s->flags&CODEC_FLAG_PASS2){
2374 for(i=0; i<s->max_b_frames+1; i++){
2375 int pict_num= s->input_picture[0]->display_picture_number + i;
2377 if(pict_num >= s->rc_context.num_entries)
2379 if(!s->input_picture[i]){
2380 s->rc_context.entry[pict_num-1].new_pict_type = P_TYPE;
2384 s->input_picture[i]->pict_type=
2385 s->rc_context.entry[pict_num].new_pict_type;
2389 if(s->avctx->b_frame_strategy==0){
2390 b_frames= s->max_b_frames;
2391 while(b_frames && !s->input_picture[b_frames]) b_frames--;
2392 }else if(s->avctx->b_frame_strategy==1){
2393 for(i=1; i<s->max_b_frames+1; i++){
2394 if(s->input_picture[i] && s->input_picture[i]->b_frame_score==0){
2395 s->input_picture[i]->b_frame_score=
2396 get_intra_count(s, s->input_picture[i ]->data[0],
2397 s->input_picture[i-1]->data[0], s->linesize) + 1;
2400 for(i=0; i<s->max_b_frames+1; i++){
2401 if(s->input_picture[i]==NULL || s->input_picture[i]->b_frame_score - 1 > s->mb_num/s->avctx->b_sensitivity) break;
2404 b_frames= FFMAX(0, i-1);
2407 for(i=0; i<b_frames+1; i++){
2408 s->input_picture[i]->b_frame_score=0;
2410 }else if(s->avctx->b_frame_strategy==2){
2411 b_frames= estimate_best_b_count(s);
2413 av_log(s->avctx, AV_LOG_ERROR, "illegal b frame strategy\n");
2418 //static int b_count=0;
2419 //b_count+= b_frames;
2420 //av_log(s->avctx, AV_LOG_DEBUG, "b_frames: %d\n", b_count);
2422 for(i= b_frames - 1; i>=0; i--){
2423 int type= s->input_picture[i]->pict_type;
2424 if(type && type != B_TYPE)
2427 if(s->input_picture[b_frames]->pict_type == B_TYPE && b_frames == s->max_b_frames){
2428 av_log(s->avctx, AV_LOG_ERROR, "warning, too many b frames in a row\n");
2431 if(s->picture_in_gop_number + b_frames >= s->gop_size){
2432 if((s->flags2 & CODEC_FLAG2_STRICT_GOP) && s->gop_size > s->picture_in_gop_number){
2433 b_frames= s->gop_size - s->picture_in_gop_number - 1;
2435 if(s->flags & CODEC_FLAG_CLOSED_GOP)
2437 s->input_picture[b_frames]->pict_type= I_TYPE;
2441 if( (s->flags & CODEC_FLAG_CLOSED_GOP)
2443 && s->input_picture[b_frames]->pict_type== I_TYPE)
2446 s->reordered_input_picture[0]= s->input_picture[b_frames];
2447 if(s->reordered_input_picture[0]->pict_type != I_TYPE)
2448 s->reordered_input_picture[0]->pict_type= P_TYPE;
2449 s->reordered_input_picture[0]->coded_picture_number= s->coded_picture_number++;
2450 for(i=0; i<b_frames; i++){
2451 s->reordered_input_picture[i+1]= s->input_picture[i];
2452 s->reordered_input_picture[i+1]->pict_type= B_TYPE;
2453 s->reordered_input_picture[i+1]->coded_picture_number= s->coded_picture_number++;
2458 if(s->reordered_input_picture[0]){
2459 s->reordered_input_picture[0]->reference= s->reordered_input_picture[0]->pict_type!=B_TYPE ? 3 : 0;
2461 copy_picture(&s->new_picture, s->reordered_input_picture[0]);
2463 if(s->reordered_input_picture[0]->type == FF_BUFFER_TYPE_SHARED || s->avctx->rc_buffer_size){
2464 // input is a shared pix, so we can't modifiy it -> alloc a new one & ensure that the shared one is reuseable
2466 int i= ff_find_unused_picture(s, 0);
2467 Picture *pic= &s->picture[i];
2469 pic->reference = s->reordered_input_picture[0]->reference;
2470 alloc_picture(s, pic, 0);
2472 /* mark us unused / free shared pic */
2473 if(s->reordered_input_picture[0]->type == FF_BUFFER_TYPE_INTERNAL)
2474 s->avctx->release_buffer(s->avctx, (AVFrame*)s->reordered_input_picture[0]);
2476 s->reordered_input_picture[0]->data[i]= NULL;
2477 s->reordered_input_picture[0]->type= 0;
2479 copy_picture_attributes(s, (AVFrame*)pic, (AVFrame*)s->reordered_input_picture[0]);
2481 s->current_picture_ptr= pic;
2483 // input is not a shared pix -> reuse buffer for current_pix
2485 assert( s->reordered_input_picture[0]->type==FF_BUFFER_TYPE_USER
2486 || s->reordered_input_picture[0]->type==FF_BUFFER_TYPE_INTERNAL);
2488 s->current_picture_ptr= s->reordered_input_picture[0];
2490 s->new_picture.data[i]+= INPLACE_OFFSET;
2493 copy_picture(&s->current_picture, s->current_picture_ptr);
2495 s->picture_number= s->new_picture.display_picture_number;
2496 //printf("dpn:%d\n", s->picture_number);
2498 memset(&s->new_picture, 0, sizeof(Picture));
2502 int MPV_encode_picture(AVCodecContext *avctx,
2503 unsigned char *buf, int buf_size, void *data)
2505 MpegEncContext *s = avctx->priv_data;
2506 AVFrame *pic_arg = data;
2507 int i, stuffing_count;
2509 for(i=0; i<avctx->thread_count; i++){
2510 int start_y= s->thread_context[i]->start_mb_y;
2511 int end_y= s->thread_context[i]-> end_mb_y;
2512 int h= s->mb_height;
2513 uint8_t *start= buf + (size_t)(((int64_t) buf_size)*start_y/h);
2514 uint8_t *end = buf + (size_t)(((int64_t) buf_size)* end_y/h);
2516 init_put_bits(&s->thread_context[i]->pb, start, end - start);
2519 s->picture_in_gop_number++;
2521 if(load_input_picture(s, pic_arg) < 0)
2524 select_input_picture(s);
2527 if(s->new_picture.data[0]){
2528 s->pict_type= s->new_picture.pict_type;
2530 //printf("qs:%f %f %d\n", s->new_picture.quality, s->current_picture.quality, s->qscale);
2531 MPV_frame_start(s, avctx);
2533 if (encode_picture(s, s->picture_number) < 0)
2536 avctx->real_pict_num = s->picture_number;
2537 avctx->header_bits = s->header_bits;
2538 avctx->mv_bits = s->mv_bits;
2539 avctx->misc_bits = s->misc_bits;
2540 avctx->i_tex_bits = s->i_tex_bits;
2541 avctx->p_tex_bits = s->p_tex_bits;
2542 avctx->i_count = s->i_count;
2543 avctx->p_count = s->mb_num - s->i_count - s->skip_count; //FIXME f/b_count in avctx
2544 avctx->skip_count = s->skip_count;
2548 if (s->out_format == FMT_MJPEG)
2549 mjpeg_picture_trailer(s);
2551 if(avctx->rc_buffer_size){
2552 RateControlContext *rcc= &s->rc_context;
2553 int max_size= rcc->buffer_index/3;
2555 if(put_bits_count(&s->pb) > max_size && s->lambda < s->avctx->lmax){
2556 s->next_lambda= FFMAX(s->lambda+1, s->lambda*(s->qscale+1) / s->qscale);
2557 if(s->adaptive_quant){
2559 for(i=0; i<s->mb_height*s->mb_stride; i++)
2560 s->lambda_table[i]= FFMAX(s->lambda_table[i]+1, s->lambda_table[i]*(s->qscale+1) / s->qscale);
2562 s->mb_skipped = 0; //done in MPV_frame_start()
2563 if(s->pict_type==P_TYPE){ //done in encode_picture() so we must undo it
2564 if(s->flipflop_rounding || s->codec_id == CODEC_ID_H263P || s->codec_id == CODEC_ID_MPEG4)
2565 s->no_rounding ^= 1;
2567 if(s->pict_type!=B_TYPE){
2568 s->time_base= s->last_time_base;
2569 s->last_non_b_time= s->time - s->pp_time;
2571 // av_log(NULL, AV_LOG_ERROR, "R:%d ", s->next_lambda);
2572 for(i=0; i<avctx->thread_count; i++){
2573 PutBitContext *pb= &s->thread_context[i]->pb;
2574 init_put_bits(pb, pb->buf, pb->buf_end - pb->buf);
2579 assert(s->avctx->rc_max_rate);
2582 if(s->flags&CODEC_FLAG_PASS1)
2583 ff_write_pass1_stats(s);
2586 s->current_picture_ptr->error[i]= s->current_picture.error[i];
2587 avctx->error[i] += s->current_picture_ptr->error[i];
2590 if(s->flags&CODEC_FLAG_PASS1)
2591 assert(avctx->header_bits + avctx->mv_bits + avctx->misc_bits + avctx->i_tex_bits + avctx->p_tex_bits == put_bits_count(&s->pb));
2592 flush_put_bits(&s->pb);
2593 s->frame_bits = put_bits_count(&s->pb);
2595 stuffing_count= ff_vbv_update(s, s->frame_bits);
2597 if(s->pb.buf_end - s->pb.buf - (put_bits_count(&s->pb)>>3) < stuffing_count + 50){
2598 av_log(s->avctx, AV_LOG_ERROR, "stuffing too large\n");
2602 switch(s->codec_id){
2603 case CODEC_ID_MPEG1VIDEO:
2604 case CODEC_ID_MPEG2VIDEO:
2605 while(stuffing_count--){
2606 put_bits(&s->pb, 8, 0);
2609 case CODEC_ID_MPEG4:
2610 put_bits(&s->pb, 16, 0);
2611 put_bits(&s->pb, 16, 0x1C3);
2612 stuffing_count -= 4;
2613 while(stuffing_count--){
2614 put_bits(&s->pb, 8, 0xFF);
2618 av_log(s->avctx, AV_LOG_ERROR, "vbv buffer overflow\n");
2620 flush_put_bits(&s->pb);
2621 s->frame_bits = put_bits_count(&s->pb);
2624 /* update mpeg1/2 vbv_delay for CBR */
2625 if(s->avctx->rc_max_rate && s->avctx->rc_min_rate == s->avctx->rc_max_rate && s->out_format == FMT_MPEG1
2626 && 90000LL * (avctx->rc_buffer_size-1) <= s->avctx->rc_max_rate*0xFFFFLL){
2629 assert(s->repeat_first_field==0);
2631 vbv_delay= lrintf(90000 * s->rc_context.buffer_index / s->avctx->rc_max_rate);
2632 assert(vbv_delay < 0xFFFF);
2634 s->vbv_delay_ptr[0] &= 0xF8;
2635 s->vbv_delay_ptr[0] |= vbv_delay>>13;
2636 s->vbv_delay_ptr[1] = vbv_delay>>5;
2637 s->vbv_delay_ptr[2] &= 0x07;
2638 s->vbv_delay_ptr[2] |= vbv_delay<<3;
2640 s->total_bits += s->frame_bits;
2641 avctx->frame_bits = s->frame_bits;
2643 assert((pbBufPtr(&s->pb) == s->pb.buf));
2646 assert((s->frame_bits&7)==0);
2648 return s->frame_bits/8;
2651 #endif //CONFIG_ENCODERS
2653 static inline void gmc1_motion(MpegEncContext *s,
2654 uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
2655 uint8_t **ref_picture)
2658 int offset, src_x, src_y, linesize, uvlinesize;
2659 int motion_x, motion_y;
2662 motion_x= s->sprite_offset[0][0];
2663 motion_y= s->sprite_offset[0][1];
2664 src_x = s->mb_x * 16 + (motion_x >> (s->sprite_warping_accuracy+1));
2665 src_y = s->mb_y * 16 + (motion_y >> (s->sprite_warping_accuracy+1));
2666 motion_x<<=(3-s->sprite_warping_accuracy);
2667 motion_y<<=(3-s->sprite_warping_accuracy);
2668 src_x = clip(src_x, -16, s->width);
2669 if (src_x == s->width)
2671 src_y = clip(src_y, -16, s->height);
2672 if (src_y == s->height)
2675 linesize = s->linesize;
2676 uvlinesize = s->uvlinesize;
2678 ptr = ref_picture[0] + (src_y * linesize) + src_x;
2680 if(s->flags&CODEC_FLAG_EMU_EDGE){
2681 if( (unsigned)src_x >= s->h_edge_pos - 17
2682 || (unsigned)src_y >= s->v_edge_pos - 17){
2683 ff_emulated_edge_mc(s->edge_emu_buffer, ptr, linesize, 17, 17, src_x, src_y, s->h_edge_pos, s->v_edge_pos);
2684 ptr= s->edge_emu_buffer;
2688 if((motion_x|motion_y)&7){
2689 s->dsp.gmc1(dest_y , ptr , linesize, 16, motion_x&15, motion_y&15, 128 - s->no_rounding);
2690 s->dsp.gmc1(dest_y+8, ptr+8, linesize, 16, motion_x&15, motion_y&15, 128 - s->no_rounding);
2694 dxy= ((motion_x>>3)&1) | ((motion_y>>2)&2);
2695 if (s->no_rounding){
2696 s->dsp.put_no_rnd_pixels_tab[0][dxy](dest_y, ptr, linesize, 16);
2698 s->dsp.put_pixels_tab [0][dxy](dest_y, ptr, linesize, 16);
2702 if(s->flags&CODEC_FLAG_GRAY) return;
2704 motion_x= s->sprite_offset[1][0];
2705 motion_y= s->sprite_offset[1][1];
2706 src_x = s->mb_x * 8 + (motion_x >> (s->sprite_warping_accuracy+1));
2707 src_y = s->mb_y * 8 + (motion_y >> (s->sprite_warping_accuracy+1));
2708 motion_x<<=(3-s->sprite_warping_accuracy);
2709 motion_y<<=(3-s->sprite_warping_accuracy);
2710 src_x = clip(src_x, -8, s->width>>1);
2711 if (src_x == s->width>>1)
2713 src_y = clip(src_y, -8, s->height>>1);
2714 if (src_y == s->height>>1)
2717 offset = (src_y * uvlinesize) + src_x;
2718 ptr = ref_picture[1] + offset;
2719 if(s->flags&CODEC_FLAG_EMU_EDGE){
2720 if( (unsigned)src_x >= (s->h_edge_pos>>1) - 9
2721 || (unsigned)src_y >= (s->v_edge_pos>>1) - 9){
2722 ff_emulated_edge_mc(s->edge_emu_buffer, ptr, uvlinesize, 9, 9, src_x, src_y, s->h_edge_pos>>1, s->v_edge_pos>>1);
2723 ptr= s->edge_emu_buffer;
2727 s->dsp.gmc1(dest_cb, ptr, uvlinesize, 8, motion_x&15, motion_y&15, 128 - s->no_rounding);
2729 ptr = ref_picture[2] + offset;
2731 ff_emulated_edge_mc(s->edge_emu_buffer, ptr, uvlinesize, 9, 9, src_x, src_y, s->h_edge_pos>>1, s->v_edge_pos>>1);
2732 ptr= s->edge_emu_buffer;
2734 s->dsp.gmc1(dest_cr, ptr, uvlinesize, 8, motion_x&15, motion_y&15, 128 - s->no_rounding);
2739 static inline void gmc_motion(MpegEncContext *s,
2740 uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
2741 uint8_t **ref_picture)
2744 int linesize, uvlinesize;
2745 const int a= s->sprite_warping_accuracy;
2748 linesize = s->linesize;
2749 uvlinesize = s->uvlinesize;
2751 ptr = ref_picture[0];
2753 ox= s->sprite_offset[0][0] + s->sprite_delta[0][0]*s->mb_x*16 + s->sprite_delta[0][1]*s->mb_y*16;
2754 oy= s->sprite_offset[0][1] + s->sprite_delta[1][0]*s->mb_x*16 + s->sprite_delta[1][1]*s->mb_y*16;
2756 s->dsp.gmc(dest_y, ptr, linesize, 16,
2759 s->sprite_delta[0][0], s->sprite_delta[0][1],
2760 s->sprite_delta[1][0], s->sprite_delta[1][1],
2761 a+1, (1<<(2*a+1)) - s->no_rounding,
2762 s->h_edge_pos, s->v_edge_pos);
2763 s->dsp.gmc(dest_y+8, ptr, linesize, 16,
2764 ox + s->sprite_delta[0][0]*8,
2765 oy + s->sprite_delta[1][0]*8,
2766 s->sprite_delta[0][0], s->sprite_delta[0][1],
2767 s->sprite_delta[1][0], s->sprite_delta[1][1],
2768 a+1, (1<<(2*a+1)) - s->no_rounding,
2769 s->h_edge_pos, s->v_edge_pos);
2771 if(s->flags&CODEC_FLAG_GRAY) return;
2773 ox= s->sprite_offset[1][0] + s->sprite_delta[0][0]*s->mb_x*8 + s->sprite_delta[0][1]*s->mb_y*8;
2774 oy= s->sprite_offset[1][1] + s->sprite_delta[1][0]*s->mb_x*8 + s->sprite_delta[1][1]*s->mb_y*8;
2776 ptr = ref_picture[1];
2777 s->dsp.gmc(dest_cb, ptr, uvlinesize, 8,
2780 s->sprite_delta[0][0], s->sprite_delta[0][1],
2781 s->sprite_delta[1][0], s->sprite_delta[1][1],
2782 a+1, (1<<(2*a+1)) - s->no_rounding,
2783 s->h_edge_pos>>1, s->v_edge_pos>>1);
2785 ptr = ref_picture[2];
2786 s->dsp.gmc(dest_cr, ptr, uvlinesize, 8,
2789 s->sprite_delta[0][0], s->sprite_delta[0][1],
2790 s->sprite_delta[1][0], s->sprite_delta[1][1],
2791 a+1, (1<<(2*a+1)) - s->no_rounding,
2792 s->h_edge_pos>>1, s->v_edge_pos>>1);
2796 * Copies a rectangular area of samples to a temporary buffer and replicates the boarder samples.
2797 * @param buf destination buffer
2798 * @param src source buffer
2799 * @param linesize number of bytes between 2 vertically adjacent samples in both the source and destination buffers
2800 * @param block_w width of block
2801 * @param block_h height of block
2802 * @param src_x x coordinate of the top left sample of the block in the source buffer
2803 * @param src_y y coordinate of the top left sample of the block in the source buffer
2804 * @param w width of the source buffer
2805 * @param h height of the source buffer
2807 void ff_emulated_edge_mc(uint8_t *buf, uint8_t *src, int linesize, int block_w, int block_h,
2808 int src_x, int src_y, int w, int h){
2810 int start_y, start_x, end_y, end_x;
2813 src+= (h-1-src_y)*linesize;
2815 }else if(src_y<=-block_h){
2816 src+= (1-block_h-src_y)*linesize;
2822 }else if(src_x<=-block_w){
2823 src+= (1-block_w-src_x);
2827 start_y= FFMAX(0, -src_y);
2828 start_x= FFMAX(0, -src_x);
2829 end_y= FFMIN(block_h, h-src_y);
2830 end_x= FFMIN(block_w, w-src_x);
2832 // copy existing part
2833 for(y=start_y; y<end_y; y++){
2834 for(x=start_x; x<end_x; x++){
2835 buf[x + y*linesize]= src[x + y*linesize];
2840 for(y=0; y<start_y; y++){
2841 for(x=start_x; x<end_x; x++){
2842 buf[x + y*linesize]= buf[x + start_y*linesize];
2847 for(y=end_y; y<block_h; y++){
2848 for(x=start_x; x<end_x; x++){
2849 buf[x + y*linesize]= buf[x + (end_y-1)*linesize];
2853 for(y=0; y<block_h; y++){
2855 for(x=0; x<start_x; x++){
2856 buf[x + y*linesize]= buf[start_x + y*linesize];
2860 for(x=end_x; x<block_w; x++){
2861 buf[x + y*linesize]= buf[end_x - 1 + y*linesize];
2866 static inline int hpel_motion(MpegEncContext *s,
2867 uint8_t *dest, uint8_t *src,
2868 int field_based, int field_select,
2869 int src_x, int src_y,
2870 int width, int height, int stride,
2871 int h_edge_pos, int v_edge_pos,
2872 int w, int h, op_pixels_func *pix_op,
2873 int motion_x, int motion_y)
2878 dxy = ((motion_y & 1) << 1) | (motion_x & 1);
2879 src_x += motion_x >> 1;
2880 src_y += motion_y >> 1;
2882 /* WARNING: do no forget half pels */
2883 src_x = clip(src_x, -16, width); //FIXME unneeded for emu?
2886 src_y = clip(src_y, -16, height);
2887 if (src_y == height)
2889 src += src_y * stride + src_x;
2891 if(s->unrestricted_mv && (s->flags&CODEC_FLAG_EMU_EDGE)){
2892 if( (unsigned)src_x > h_edge_pos - (motion_x&1) - w
2893 || (unsigned)src_y > v_edge_pos - (motion_y&1) - h){
2894 ff_emulated_edge_mc(s->edge_emu_buffer, src, s->linesize, w+1, (h+1)<<field_based,
2895 src_x, src_y<<field_based, h_edge_pos, s->v_edge_pos);
2896 src= s->edge_emu_buffer;
2902 pix_op[dxy](dest, src, stride, h);
2906 static inline int hpel_motion_lowres(MpegEncContext *s,
2907 uint8_t *dest, uint8_t *src,
2908 int field_based, int field_select,
2909 int src_x, int src_y,
2910 int width, int height, int stride,
2911 int h_edge_pos, int v_edge_pos,
2912 int w, int h, h264_chroma_mc_func *pix_op,
2913 int motion_x, int motion_y)
2915 const int lowres= s->avctx->lowres;
2916 const int s_mask= (2<<lowres)-1;
2920 if(s->quarter_sample){
2925 sx= motion_x & s_mask;
2926 sy= motion_y & s_mask;
2927 src_x += motion_x >> (lowres+1);
2928 src_y += motion_y >> (lowres+1);
2930 src += src_y * stride + src_x;
2932 if( (unsigned)src_x > h_edge_pos - (!!sx) - w
2933 || (unsigned)src_y >(v_edge_pos >> field_based) - (!!sy) - h){
2934 ff_emulated_edge_mc(s->edge_emu_buffer, src, s->linesize, w+1, (h+1)<<field_based,
2935 src_x, src_y<<field_based, h_edge_pos, v_edge_pos);
2936 src= s->edge_emu_buffer;
2944 pix_op[lowres](dest, src, stride, h, sx, sy);
2948 /* apply one mpeg motion vector to the three components */
2949 static always_inline void mpeg_motion(MpegEncContext *s,
2950 uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
2951 int field_based, int bottom_field, int field_select,
2952 uint8_t **ref_picture, op_pixels_func (*pix_op)[4],
2953 int motion_x, int motion_y, int h)
2955 uint8_t *ptr_y, *ptr_cb, *ptr_cr;
2956 int dxy, uvdxy, mx, my, src_x, src_y, uvsrc_x, uvsrc_y, v_edge_pos, uvlinesize, linesize;
2959 if(s->quarter_sample)
2966 v_edge_pos = s->v_edge_pos >> field_based;
2967 linesize = s->current_picture.linesize[0] << field_based;
2968 uvlinesize = s->current_picture.linesize[1] << field_based;
2970 dxy = ((motion_y & 1) << 1) | (motion_x & 1);
2971 src_x = s->mb_x* 16 + (motion_x >> 1);
2972 src_y =(s->mb_y<<(4-field_based)) + (motion_y >> 1);
2974 if (s->out_format == FMT_H263) {
2975 if((s->workaround_bugs & FF_BUG_HPEL_CHROMA) && field_based){
2976 mx = (motion_x>>1)|(motion_x&1);
2978 uvdxy = ((my & 1) << 1) | (mx & 1);
2979 uvsrc_x = s->mb_x* 8 + (mx >> 1);
2980 uvsrc_y = (s->mb_y<<(3-field_based)) + (my >> 1);
2982 uvdxy = dxy | (motion_y & 2) | ((motion_x & 2) >> 1);
2986 }else if(s->out_format == FMT_H261){//even chroma mv's are full pel in H261
2990 uvsrc_x = s->mb_x*8 + mx;
2991 uvsrc_y = s->mb_y*8 + my;
2993 if(s->chroma_y_shift){
2996 uvdxy = ((my & 1) << 1) | (mx & 1);
2997 uvsrc_x = s->mb_x* 8 + (mx >> 1);
2998 uvsrc_y = (s->mb_y<<(3-field_based)) + (my >> 1);
3000 if(s->chroma_x_shift){
3003 uvdxy = ((motion_y & 1) << 1) | (mx & 1);
3004 uvsrc_x = s->mb_x* 8 + (mx >> 1);
3015 ptr_y = ref_picture[0] + src_y * linesize + src_x;
3016 ptr_cb = ref_picture[1] + uvsrc_y * uvlinesize + uvsrc_x;
3017 ptr_cr = ref_picture[2] + uvsrc_y * uvlinesize + uvsrc_x;
3019 if( (unsigned)src_x > s->h_edge_pos - (motion_x&1) - 16
3020 || (unsigned)src_y > v_edge_pos - (motion_y&1) - h){
3021 if(s->codec_id == CODEC_ID_MPEG2VIDEO ||
3022 s->codec_id == CODEC_ID_MPEG1VIDEO){
3023 av_log(s->avctx,AV_LOG_DEBUG,"MPEG motion vector out of boundary\n");
3026 ff_emulated_edge_mc(s->edge_emu_buffer, ptr_y, s->linesize, 17, 17+field_based,
3027 src_x, src_y<<field_based, s->h_edge_pos, s->v_edge_pos);
3028 ptr_y = s->edge_emu_buffer;
3029 if(!(s->flags&CODEC_FLAG_GRAY)){
3030 uint8_t *uvbuf= s->edge_emu_buffer+18*s->linesize;
3031 ff_emulated_edge_mc(uvbuf , ptr_cb, s->uvlinesize, 9, 9+field_based,
3032 uvsrc_x, uvsrc_y<<field_based, s->h_edge_pos>>1, s->v_edge_pos>>1);
3033 ff_emulated_edge_mc(uvbuf+16, ptr_cr, s->uvlinesize, 9, 9+field_based,
3034 uvsrc_x, uvsrc_y<<field_based, s->h_edge_pos>>1, s->v_edge_pos>>1);
3040 if(bottom_field){ //FIXME use this for field pix too instead of the obnoxious hack which changes picture.data
3041 dest_y += s->linesize;
3042 dest_cb+= s->uvlinesize;
3043 dest_cr+= s->uvlinesize;
3047 ptr_y += s->linesize;
3048 ptr_cb+= s->uvlinesize;
3049 ptr_cr+= s->uvlinesize;
3052 pix_op[0][dxy](dest_y, ptr_y, linesize, h);
3054 if(!(s->flags&CODEC_FLAG_GRAY)){
3055 pix_op[s->chroma_x_shift][uvdxy](dest_cb, ptr_cb, uvlinesize, h >> s->chroma_y_shift);
3056 pix_op[s->chroma_x_shift][uvdxy](dest_cr, ptr_cr, uvlinesize, h >> s->chroma_y_shift);
3058 #if defined(CONFIG_H261_ENCODER) || defined(CONFIG_H261_DECODER)
3059 if(s->out_format == FMT_H261){
3060 ff_h261_loop_filter(s);
3065 /* apply one mpeg motion vector to the three components */
3066 static always_inline void mpeg_motion_lowres(MpegEncContext *s,
3067 uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
3068 int field_based, int bottom_field, int field_select,
3069 uint8_t **ref_picture, h264_chroma_mc_func *pix_op,
3070 int motion_x, int motion_y, int h)
3072 uint8_t *ptr_y, *ptr_cb, *ptr_cr;
3073 int mx, my, src_x, src_y, uvsrc_x, uvsrc_y, uvlinesize, linesize, sx, sy, uvsx, uvsy;
3074 const int lowres= s->avctx->lowres;
3075 const int block_s= 8>>lowres;
3076 const int s_mask= (2<<lowres)-1;
3077 const int h_edge_pos = s->h_edge_pos >> lowres;
3078 const int v_edge_pos = s->v_edge_pos >> lowres;
3079 linesize = s->current_picture.linesize[0] << field_based;
3080 uvlinesize = s->current_picture.linesize[1] << field_based;
3082 if(s->quarter_sample){ //FIXME obviously not perfect but qpel wont work in lowres anyway
3088 motion_y += (bottom_field - field_select)*((1<<lowres)-1);
3091 sx= motion_x & s_mask;
3092 sy= motion_y & s_mask;
3093 src_x = s->mb_x*2*block_s + (motion_x >> (lowres+1));
3094 src_y =(s->mb_y*2*block_s>>field_based) + (motion_y >> (lowres+1));
3096 if (s->out_format == FMT_H263) {
3097 uvsx = ((motion_x>>1) & s_mask) | (sx&1);
3098 uvsy = ((motion_y>>1) & s_mask) | (sy&1);
3101 }else if(s->out_format == FMT_H261){//even chroma mv's are full pel in H261
3104 uvsx = (2*mx) & s_mask;
3105 uvsy = (2*my) & s_mask;
3106 uvsrc_x = s->mb_x*block_s + (mx >> lowres);
3107 uvsrc_y = s->mb_y*block_s + (my >> lowres);
3113 uvsrc_x = s->mb_x*block_s + (mx >> (lowres+1));
3114 uvsrc_y =(s->mb_y*block_s>>field_based) + (my >> (lowres+1));
3117 ptr_y = ref_picture[0] + src_y * linesize + src_x;
3118 ptr_cb = ref_picture[1] + uvsrc_y * uvlinesize + uvsrc_x;
3119 ptr_cr = ref_picture[2] + uvsrc_y * uvlinesize + uvsrc_x;
3121 if( (unsigned)src_x > h_edge_pos - (!!sx) - 2*block_s
3122 || (unsigned)src_y >(v_edge_pos >> field_based) - (!!sy) - h){
3123 ff_emulated_edge_mc(s->edge_emu_buffer, ptr_y, s->linesize, 17, 17+field_based,
3124 src_x, src_y<<field_based, h_edge_pos, v_edge_pos);
3125 ptr_y = s->edge_emu_buffer;
3126 if(!(s->flags&CODEC_FLAG_GRAY)){
3127 uint8_t *uvbuf= s->edge_emu_buffer+18*s->linesize;
3128 ff_emulated_edge_mc(uvbuf , ptr_cb, s->uvlinesize, 9, 9+field_based,
3129 uvsrc_x, uvsrc_y<<field_based, h_edge_pos>>1, v_edge_pos>>1);
3130 ff_emulated_edge_mc(uvbuf+16, ptr_cr, s->uvlinesize, 9, 9+field_based,
3131 uvsrc_x, uvsrc_y<<field_based, h_edge_pos>>1, v_edge_pos>>1);
3137 if(bottom_field){ //FIXME use this for field pix too instead of the obnoxious hack which changes picture.data
3138 dest_y += s->linesize;
3139 dest_cb+= s->uvlinesize;
3140 dest_cr+= s->uvlinesize;
3144 ptr_y += s->linesize;
3145 ptr_cb+= s->uvlinesize;
3146 ptr_cr+= s->uvlinesize;
3151 pix_op[lowres-1](dest_y, ptr_y, linesize, h, sx, sy);
3153 if(!(s->flags&CODEC_FLAG_GRAY)){
3154 uvsx <<= 2 - lowres;
3155 uvsy <<= 2 - lowres;
3156 pix_op[lowres](dest_cb, ptr_cb, uvlinesize, h >> s->chroma_y_shift, uvsx, uvsy);
3157 pix_op[lowres](dest_cr, ptr_cr, uvlinesize, h >> s->chroma_y_shift, uvsx, uvsy);
3159 //FIXME h261 lowres loop filter
3162 //FIXME move to dsputil, avg variant, 16x16 version
3163 static inline void put_obmc(uint8_t *dst, uint8_t *src[5], int stride){
3165 uint8_t * const top = src[1];
3166 uint8_t * const left = src[2];
3167 uint8_t * const mid = src[0];
3168 uint8_t * const right = src[3];
3169 uint8_t * const bottom= src[4];
3170 #define OBMC_FILTER(x, t, l, m, r, b)\
3171 dst[x]= (t*top[x] + l*left[x] + m*mid[x] + r*right[x] + b*bottom[x] + 4)>>3
3172 #define OBMC_FILTER4(x, t, l, m, r, b)\
3173 OBMC_FILTER(x , t, l, m, r, b);\
3174 OBMC_FILTER(x+1 , t, l, m, r, b);\
3175 OBMC_FILTER(x +stride, t, l, m, r, b);\
3176 OBMC_FILTER(x+1+stride, t, l, m, r, b);
3179 OBMC_FILTER (x , 2, 2, 4, 0, 0);
3180 OBMC_FILTER (x+1, 2, 1, 5, 0, 0);
3181 OBMC_FILTER4(x+2, 2, 1, 5, 0, 0);
3182 OBMC_FILTER4(x+4, 2, 0, 5, 1, 0);
3183 OBMC_FILTER (x+6, 2, 0, 5, 1, 0);
3184 OBMC_FILTER (x+7, 2, 0, 4, 2, 0);
3186 OBMC_FILTER (x , 1, 2, 5, 0, 0);
3187 OBMC_FILTER (x+1, 1, 2, 5, 0, 0);
3188 OBMC_FILTER (x+6, 1, 0, 5, 2, 0);
3189 OBMC_FILTER (x+7, 1, 0, 5, 2, 0);
3191 OBMC_FILTER4(x , 1, 2, 5, 0, 0);
3192 OBMC_FILTER4(x+2, 1, 1, 6, 0, 0);
3193 OBMC_FILTER4(x+4, 1, 0, 6, 1, 0);
3194 OBMC_FILTER4(x+6, 1, 0, 5, 2, 0);
3196 OBMC_FILTER4(x , 0, 2, 5, 0, 1);
3197 OBMC_FILTER4(x+2, 0, 1, 6, 0, 1);
3198 OBMC_FILTER4(x+4, 0, 0, 6, 1, 1);
3199 OBMC_FILTER4(x+6, 0, 0, 5, 2, 1);
3201 OBMC_FILTER (x , 0, 2, 5, 0, 1);
3202 OBMC_FILTER (x+1, 0, 2, 5, 0, 1);
3203 OBMC_FILTER4(x+2, 0, 1, 5, 0, 2);
3204 OBMC_FILTER4(x+4, 0, 0, 5, 1, 2);
3205 OBMC_FILTER (x+6, 0, 0, 5, 2, 1);
3206 OBMC_FILTER (x+7, 0, 0, 5, 2, 1);
3208 OBMC_FILTER (x , 0, 2, 4, 0, 2);
3209 OBMC_FILTER (x+1, 0, 1, 5, 0, 2);
3210 OBMC_FILTER (x+6, 0, 0, 5, 1, 2);
3211 OBMC_FILTER (x+7, 0, 0, 4, 2, 2);
3214 /* obmc for 1 8x8 luma block */
3215 static inline void obmc_motion(MpegEncContext *s,
3216 uint8_t *dest, uint8_t *src,
3217 int src_x, int src_y,
3218 op_pixels_func *pix_op,
3219 int16_t mv[5][2]/* mid top left right bottom*/)
3225 assert(s->quarter_sample==0);
3228 if(i && mv[i][0]==mv[MID][0] && mv[i][1]==mv[MID][1]){
3231 ptr[i]= s->obmc_scratchpad + 8*(i&1) + s->linesize*8*(i>>1);
3232 hpel_motion(s, ptr[i], src, 0, 0,
3234 s->width, s->height, s->linesize,
3235 s->h_edge_pos, s->v_edge_pos,
3237 mv[i][0], mv[i][1]);
3241 put_obmc(dest, ptr, s->linesize);
3244 static inline void qpel_motion(MpegEncContext *s,
3245 uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
3246 int field_based, int bottom_field, int field_select,
3247 uint8_t **ref_picture, op_pixels_func (*pix_op)[4],
3248 qpel_mc_func (*qpix_op)[16],
3249 int motion_x, int motion_y, int h)
3251 uint8_t *ptr_y, *ptr_cb, *ptr_cr;
3252 int dxy, uvdxy, mx, my, src_x, src_y, uvsrc_x, uvsrc_y, v_edge_pos, linesize, uvlinesize;
3254 dxy = ((motion_y & 3) << 2) | (motion_x & 3);
3255 src_x = s->mb_x * 16 + (motion_x >> 2);
3256 src_y = s->mb_y * (16 >> field_based) + (motion_y >> 2);
3258 v_edge_pos = s->v_edge_pos >> field_based;
3259 linesize = s->linesize << field_based;
3260 uvlinesize = s->uvlinesize << field_based;
3265 }else if(s->workaround_bugs&FF_BUG_QPEL_CHROMA2){
3266 static const int rtab[8]= {0,0,1,1,0,0,0,1};
3267 mx= (motion_x>>1) + rtab[motion_x&7];
3268 my= (motion_y>>1) + rtab[motion_y&7];
3269 }else if(s->workaround_bugs&FF_BUG_QPEL_CHROMA){
3270 mx= (motion_x>>1)|(motion_x&1);
3271 my= (motion_y>>1)|(motion_y&1);
3279 uvdxy= (mx&1) | ((my&1)<<1);
3283 uvsrc_x = s->mb_x * 8 + mx;
3284 uvsrc_y = s->mb_y * (8 >> field_based) + my;
3286 ptr_y = ref_picture[0] + src_y * linesize + src_x;
3287 ptr_cb = ref_picture[1] + uvsrc_y * uvlinesize + uvsrc_x;
3288 ptr_cr = ref_picture[2] + uvsrc_y * uvlinesize + uvsrc_x;
3290 if( (unsigned)src_x > s->h_edge_pos - (motion_x&3) - 16
3291 || (unsigned)src_y > v_edge_pos - (motion_y&3) - h ){
3292 ff_emulated_edge_mc(s->edge_emu_buffer, ptr_y, s->linesize, 17, 17+field_based,
3293 src_x, src_y<<field_based, s->h_edge_pos, s->v_edge_pos);
3294 ptr_y= s->edge_emu_buffer;
3295 if(!(s->flags&CODEC_FLAG_GRAY)){
3296 uint8_t *uvbuf= s->edge_emu_buffer + 18*s->linesize;
3297 ff_emulated_edge_mc(uvbuf, ptr_cb, s->uvlinesize, 9, 9 + field_based,
3298 uvsrc_x, uvsrc_y<<field_based, s->h_edge_pos>>1, s->v_edge_pos>>1);
3299 ff_emulated_edge_mc(uvbuf + 16, ptr_cr, s->uvlinesize, 9, 9 + field_based,
3300 uvsrc_x, uvsrc_y<<field_based, s->h_edge_pos>>1, s->v_edge_pos>>1);
3307 qpix_op[0][dxy](dest_y, ptr_y, linesize);
3310 dest_y += s->linesize;
3311 dest_cb+= s->uvlinesize;
3312 dest_cr+= s->uvlinesize;
3316 ptr_y += s->linesize;
3317 ptr_cb += s->uvlinesize;
3318 ptr_cr += s->uvlinesize;
3320 //damn interlaced mode
3321 //FIXME boundary mirroring is not exactly correct here
3322 qpix_op[1][dxy](dest_y , ptr_y , linesize);
3323 qpix_op[1][dxy](dest_y+8, ptr_y+8, linesize);
3325 if(!(s->flags&CODEC_FLAG_GRAY)){
3326 pix_op[1][uvdxy](dest_cr, ptr_cr, uvlinesize, h >> 1);
3327 pix_op[1][uvdxy](dest_cb, ptr_cb, uvlinesize, h >> 1);
3331 inline int ff_h263_round_chroma(int x){
3333 return (h263_chroma_roundtab[x & 0xf] + ((x >> 3) & ~1));
3336 return -(h263_chroma_roundtab[x & 0xf] + ((x >> 3) & ~1));
3341 * h263 chorma 4mv motion compensation.
3343 static inline void chroma_4mv_motion(MpegEncContext *s,
3344 uint8_t *dest_cb, uint8_t *dest_cr,
3345 uint8_t **ref_picture,
3346 op_pixels_func *pix_op,
3348 int dxy, emu=0, src_x, src_y, offset;
3351 /* In case of 8X8, we construct a single chroma motion vector
3352 with a special rounding */
3353 mx= ff_h263_round_chroma(mx);
3354 my= ff_h263_round_chroma(my);
3356 dxy = ((my & 1) << 1) | (mx & 1);
3360 src_x = s->mb_x * 8 + mx;
3361 src_y = s->mb_y * 8 + my;
3362 src_x = clip(src_x, -8, s->width/2);
3363 if (src_x == s->width/2)
3365 src_y = clip(src_y, -8, s->height/2);
3366 if (src_y == s->height/2)
3369 offset = (src_y * (s->uvlinesize)) + src_x;
3370 ptr = ref_picture[1] + offset;
3371 if(s->flags&CODEC_FLAG_EMU_EDGE){
3372 if( (unsigned)src_x > (s->h_edge_pos>>1) - (dxy &1) - 8
3373 || (unsigned)src_y > (s->v_edge_pos>>1) - (dxy>>1) - 8){
3374 ff_emulated_edge_mc(s->edge_emu_buffer, ptr, s->uvlinesize, 9, 9, src_x, src_y, s->h_edge_pos>>1, s->v_edge_pos>>1);
3375 ptr= s->edge_emu_buffer;
3379 pix_op[dxy](dest_cb, ptr, s->uvlinesize, 8);
3381 ptr = ref_picture[2] + offset;
3383 ff_emulated_edge_mc(s->edge_emu_buffer, ptr, s->uvlinesize, 9, 9, src_x, src_y, s->h_edge_pos>>1, s->v_edge_pos>>1);
3384 ptr= s->edge_emu_buffer;
3386 pix_op[dxy](dest_cr, ptr, s->uvlinesize, 8);
3389 static inline void chroma_4mv_motion_lowres(MpegEncContext *s,
3390 uint8_t *dest_cb, uint8_t *dest_cr,
3391 uint8_t **ref_picture,
3392 h264_chroma_mc_func *pix_op,
3394 const int lowres= s->avctx->lowres;
3395 const int block_s= 8>>lowres;
3396 const int s_mask= (2<<lowres)-1;
3397 const int h_edge_pos = s->h_edge_pos >> (lowres+1);
3398 const int v_edge_pos = s->v_edge_pos >> (lowres+1);
3399 int emu=0, src_x, src_y, offset, sx, sy;
3402 if(s->quarter_sample){
3407 /* In case of 8X8, we construct a single chroma motion vector
3408 with a special rounding */
3409 mx= ff_h263_round_chroma(mx);
3410 my= ff_h263_round_chroma(my);
3414 src_x = s->mb_x*block_s + (mx >> (lowres+1));
3415 src_y = s->mb_y*block_s + (my >> (lowres+1));
3417 offset = src_y * s->uvlinesize + src_x;
3418 ptr = ref_picture[1] + offset;
3419 if(s->flags&CODEC_FLAG_EMU_EDGE){
3420 if( (unsigned)src_x > h_edge_pos - (!!sx) - block_s
3421 || (unsigned)src_y > v_edge_pos - (!!sy) - block_s){
3422 ff_emulated_edge_mc(s->edge_emu_buffer, ptr, s->uvlinesize, 9, 9, src_x, src_y, h_edge_pos, v_edge_pos);
3423 ptr= s->edge_emu_buffer;
3429 pix_op[lowres](dest_cb, ptr, s->uvlinesize, block_s, sx, sy);
3431 ptr = ref_picture[2] + offset;
3433 ff_emulated_edge_mc(s->edge_emu_buffer, ptr, s->uvlinesize, 9, 9, src_x, src_y, h_edge_pos, v_edge_pos);
3434 ptr= s->edge_emu_buffer;
3436 pix_op[lowres](dest_cr, ptr, s->uvlinesize, block_s, sx, sy);
3439 static inline void prefetch_motion(MpegEncContext *s, uint8_t **pix, int dir){
3440 /* fetch pixels for estimated mv 4 macroblocks ahead
3441 * optimized for 64byte cache lines */
3442 const int shift = s->quarter_sample ? 2 : 1;
3443 const int mx= (s->mv[dir][0][0]>>shift) + 16*s->mb_x + 8;
3444 const int my= (s->mv[dir][0][1]>>shift) + 16*s->mb_y;
3445 int off= mx + (my + (s->mb_x&3)*4)*s->linesize + 64;
3446 s->dsp.prefetch(pix[0]+off, s->linesize, 4);
3447 off= (mx>>1) + ((my>>1) + (s->mb_x&7))*s->uvlinesize + 64;
3448 s->dsp.prefetch(pix[1]+off, pix[2]-pix[1], 2);
3452 * motion compensation of a single macroblock
3454 * @param dest_y luma destination pointer
3455 * @param dest_cb chroma cb/u destination pointer
3456 * @param dest_cr chroma cr/v destination pointer
3457 * @param dir direction (0->forward, 1->backward)
3458 * @param ref_picture array[3] of pointers to the 3 planes of the reference picture
3459 * @param pic_op halfpel motion compensation function (average or put normally)
3460 * @param pic_op qpel motion compensation function (average or put normally)
3461 * the motion vectors are taken from s->mv and the MV type from s->mv_type
3463 static inline void MPV_motion(MpegEncContext *s,
3464 uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
3465 int dir, uint8_t **ref_picture,
3466 op_pixels_func (*pix_op)[4], qpel_mc_func (*qpix_op)[16])
3468 int dxy, mx, my, src_x, src_y, motion_x, motion_y;
3470 uint8_t *ptr, *dest;
3475 prefetch_motion(s, ref_picture, dir);
3477 if(s->obmc && s->pict_type != B_TYPE){
3478 int16_t mv_cache[4][4][2];
3479 const int xy= s->mb_x + s->mb_y*s->mb_stride;
3480 const int mot_stride= s->b8_stride;
3481 const int mot_xy= mb_x*2 + mb_y*2*mot_stride;
3483 assert(!s->mb_skipped);
3485 memcpy(mv_cache[1][1], s->current_picture.motion_val[0][mot_xy ], sizeof(int16_t)*4);
3486 memcpy(mv_cache[2][1], s->current_picture.motion_val[0][mot_xy+mot_stride], sizeof(int16_t)*4);
3487 memcpy(mv_cache[3][1], s->current_picture.motion_val[0][mot_xy+mot_stride], sizeof(int16_t)*4);
3489 if(mb_y==0 || IS_INTRA(s->current_picture.mb_type[xy-s->mb_stride])){
3490 memcpy(mv_cache[0][1], mv_cache[1][1], sizeof(int16_t)*4);
3492 memcpy(mv_cache[0][1], s->current_picture.motion_val[0][mot_xy-mot_stride], sizeof(int16_t)*4);
3495 if(mb_x==0 || IS_INTRA(s->current_picture.mb_type[xy-1])){
3496 *(int32_t*)mv_cache[1][0]= *(int32_t*)mv_cache[1][1];
3497 *(int32_t*)mv_cache[2][0]= *(int32_t*)mv_cache[2][1];
3499 *(int32_t*)mv_cache[1][0]= *(int32_t*)s->current_picture.motion_val[0][mot_xy-1];
3500 *(int32_t*)mv_cache[2][0]= *(int32_t*)s->current_picture.motion_val[0][mot_xy-1+mot_stride];
3503 if(mb_x+1>=s->mb_width || IS_INTRA(s->current_picture.mb_type[xy+1])){
3504 *(int32_t*)mv_cache[1][3]= *(int32_t*)mv_cache[1][2];
3505 *(int32_t*)mv_cache[2][3]= *(int32_t*)mv_cache[2][2];
3507 *(int32_t*)mv_cache[1][3]= *(int32_t*)s->current_picture.motion_val[0][mot_xy+2];
3508 *(int32_t*)mv_cache[2][3]= *(int32_t*)s->current_picture.motion_val[0][mot_xy+2+mot_stride];
3514 const int x= (i&1)+1;
3515 const int y= (i>>1)+1;
3517 {mv_cache[y][x ][0], mv_cache[y][x ][1]},
3518 {mv_cache[y-1][x][0], mv_cache[y-1][x][1]},
3519 {mv_cache[y][x-1][0], mv_cache[y][x-1][1]},
3520 {mv_cache[y][x+1][0], mv_cache[y][x+1][1]},
3521 {mv_cache[y+1][x][0], mv_cache[y+1][x][1]}};
3523 obmc_motion(s, dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize,
3525 mb_x * 16 + (i & 1) * 8, mb_y * 16 + (i >>1) * 8,
3532 if(!(s->flags&CODEC_FLAG_GRAY))
3533 chroma_4mv_motion(s, dest_cb, dest_cr, ref_picture, pix_op[1], mx, my);
3538 switch(s->mv_type) {
3541 if(s->real_sprite_warping_points==1){
3542 gmc1_motion(s, dest_y, dest_cb, dest_cr,
3545 gmc_motion(s, dest_y, dest_cb, dest_cr,
3548 }else if(s->quarter_sample){
3549 qpel_motion(s, dest_y, dest_cb, dest_cr,
3551 ref_picture, pix_op, qpix_op,
3552 s->mv[dir][0][0], s->mv[dir][0][1], 16);
3554 ff_mspel_motion(s, dest_y, dest_cb, dest_cr,
3555 ref_picture, pix_op,
3556 s->mv[dir][0][0], s->mv[dir][0][1], 16);
3559 mpeg_motion(s, dest_y, dest_cb, dest_cr,
3561 ref_picture, pix_op,
3562 s->mv[dir][0][0], s->mv[dir][0][1], 16);
3568 if(s->quarter_sample){
3570 motion_x = s->mv[dir][i][0];
3571 motion_y = s->mv[dir][i][1];
3573 dxy = ((motion_y & 3) << 2) | (motion_x & 3);
3574 src_x = mb_x * 16 + (motion_x >> 2) + (i & 1) * 8;
3575 src_y = mb_y * 16 + (motion_y >> 2) + (i >>1) * 8;
3577 /* WARNING: do no forget half pels */
3578 src_x = clip(src_x, -16, s->width);
3579 if (src_x == s->width)
3581 src_y = clip(src_y, -16, s->height);
3582 if (src_y == s->height)
3585 ptr = ref_picture[0] + (src_y * s->linesize) + (src_x);
3586 if(s->flags&CODEC_FLAG_EMU_EDGE){
3587 if( (unsigned)src_x > s->h_edge_pos - (motion_x&3) - 8
3588 || (unsigned)src_y > s->v_edge_pos - (motion_y&3) - 8 ){
3589 ff_emulated_edge_mc(s->edge_emu_buffer, ptr, s->linesize, 9, 9, src_x, src_y, s->h_edge_pos, s->v_edge_pos);
3590 ptr= s->edge_emu_buffer;
3593 dest = dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize;
3594 qpix_op[1][dxy](dest, ptr, s->linesize);
3596 mx += s->mv[dir][i][0]/2;
3597 my += s->mv[dir][i][1]/2;
3601 hpel_motion(s, dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize,
3602 ref_picture[0], 0, 0,
3603 mb_x * 16 + (i & 1) * 8, mb_y * 16 + (i >>1) * 8,
3604 s->width, s->height, s->linesize,
3605 s->h_edge_pos, s->v_edge_pos,
3607 s->mv[dir][i][0], s->mv[dir][i][1]);
3609 mx += s->mv[dir][i][0];
3610 my += s->mv[dir][i][1];
3614 if(!(s->flags&CODEC_FLAG_GRAY))
3615 chroma_4mv_motion(s, dest_cb, dest_cr, ref_picture, pix_op[1], mx, my);
3618 if (s->picture_structure == PICT_FRAME) {
3619 if(s->quarter_sample){
3621 qpel_motion(s, dest_y, dest_cb, dest_cr,
3622 1, i, s->field_select[dir][i],
3623 ref_picture, pix_op, qpix_op,
3624 s->mv[dir][i][0], s->mv[dir][i][1], 8);
3628 mpeg_motion(s, dest_y, dest_cb, dest_cr,
3629 1, 0, s->field_select[dir][0],
3630 ref_picture, pix_op,
3631 s->mv[dir][0][0], s->mv[dir][0][1], 8);
3633 mpeg_motion(s, dest_y, dest_cb, dest_cr,
3634 1, 1, s->field_select[dir][1],
3635 ref_picture, pix_op,
3636 s->mv[dir][1][0], s->mv[dir][1][1], 8);
3639 if(s->picture_structure != s->field_select[dir][0] + 1 && s->pict_type != B_TYPE && !s->first_field){
3640 ref_picture= s->current_picture_ptr->data;
3643 mpeg_motion(s, dest_y, dest_cb, dest_cr,
3644 0, 0, s->field_select[dir][0],
3645 ref_picture, pix_op,
3646 s->mv[dir][0][0], s->mv[dir][0][1], 16);
3651 uint8_t ** ref2picture;
3653 if(s->picture_structure == s->field_select[dir][i] + 1 || s->pict_type == B_TYPE || s->first_field){
3654 ref2picture= ref_picture;
3656 ref2picture= s->current_picture_ptr->data;
3659 mpeg_motion(s, dest_y, dest_cb, dest_cr,
3660 0, 0, s->field_select[dir][i],
3661 ref2picture, pix_op,
3662 s->mv[dir][i][0], s->mv[dir][i][1] + 16*i, 8);
3664 dest_y += 16*s->linesize;
3665 dest_cb+= (16>>s->chroma_y_shift)*s->uvlinesize;
3666 dest_cr+= (16>>s->chroma_y_shift)*s->uvlinesize;
3670 if(s->picture_structure == PICT_FRAME){
3674 mpeg_motion(s, dest_y, dest_cb, dest_cr,
3676 ref_picture, pix_op,
3677 s->mv[dir][2*i + j][0], s->mv[dir][2*i + j][1], 8);
3679 pix_op = s->dsp.avg_pixels_tab;
3683 mpeg_motion(s, dest_y, dest_cb, dest_cr,
3684 0, 0, s->picture_structure != i+1,
3685 ref_picture, pix_op,
3686 s->mv[dir][2*i][0],s->mv[dir][2*i][1],16);
3688 // after put we make avg of the same block
3689 pix_op=s->dsp.avg_pixels_tab;
3691 //opposite parity is always in the same frame if this is second field
3692 if(!s->first_field){
3693 ref_picture = s->current_picture_ptr->data;
3703 * motion compensation of a single macroblock
3705 * @param dest_y luma destination pointer
3706 * @param dest_cb chroma cb/u destination pointer
3707 * @param dest_cr chroma cr/v destination pointer
3708 * @param dir direction (0->forward, 1->backward)
3709 * @param ref_picture array[3] of pointers to the 3 planes of the reference picture
3710 * @param pic_op halfpel motion compensation function (average or put normally)
3711 * the motion vectors are taken from s->mv and the MV type from s->mv_type
3713 static inline void MPV_motion_lowres(MpegEncContext *s,
3714 uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
3715 int dir, uint8_t **ref_picture,
3716 h264_chroma_mc_func *pix_op)
3720 const int lowres= s->avctx->lowres;
3721 const int block_s= 8>>lowres;
3726 switch(s->mv_type) {
3728 mpeg_motion_lowres(s, dest_y, dest_cb, dest_cr,
3730 ref_picture, pix_op,
3731 s->mv[dir][0][0], s->mv[dir][0][1], 2*block_s);
3737 hpel_motion_lowres(s, dest_y + ((i & 1) + (i >> 1) * s->linesize)*block_s,
3738 ref_picture[0], 0, 0,
3739 (2*mb_x + (i & 1))*block_s, (2*mb_y + (i >>1))*block_s,
3740 s->width, s->height, s->linesize,
3741 s->h_edge_pos >> lowres, s->v_edge_pos >> lowres,
3742 block_s, block_s, pix_op,
3743 s->mv[dir][i][0], s->mv[dir][i][1]);
3745 mx += s->mv[dir][i][0];
3746 my += s->mv[dir][i][1];
3749 if(!(s->flags&CODEC_FLAG_GRAY))
3750 chroma_4mv_motion_lowres(s, dest_cb, dest_cr, ref_picture, pix_op, mx, my);
3753 if (s->picture_structure == PICT_FRAME) {
3755 mpeg_motion_lowres(s, dest_y, dest_cb, dest_cr,
3756 1, 0, s->field_select[dir][0],
3757 ref_picture, pix_op,
3758 s->mv[dir][0][0], s->mv[dir][0][1], block_s);
3760 mpeg_motion_lowres(s, dest_y, dest_cb, dest_cr,
3761 1, 1, s->field_select[dir][1],
3762 ref_picture, pix_op,
3763 s->mv[dir][1][0], s->mv[dir][1][1], block_s);
3765 if(s->picture_structure != s->field_select[dir][0] + 1 && s->pict_type != B_TYPE && !s->first_field){
3766 ref_picture= s->current_picture_ptr->data;
3769 mpeg_motion_lowres(s, dest_y, dest_cb, dest_cr,
3770 0, 0, s->field_select[dir][0],
3771 ref_picture, pix_op,
3772 s->mv[dir][0][0], s->mv[dir][0][1], 2*block_s);
3777 uint8_t ** ref2picture;
3779 if(s->picture_structure == s->field_select[dir][i] + 1 || s->pict_type == B_TYPE || s->first_field){
3780 ref2picture= ref_picture;
3782 ref2picture= s->current_picture_ptr->data;
3785 mpeg_motion_lowres(s, dest_y, dest_cb, dest_cr,
3786 0, 0, s->field_select[dir][i],
3787 ref2picture, pix_op,
3788 s->mv[dir][i][0], s->mv[dir][i][1] + 2*block_s*i, block_s);
3790 dest_y += 2*block_s*s->linesize;
3791 dest_cb+= (2*block_s>>s->chroma_y_shift)*s->uvlinesize;
3792 dest_cr+= (2*block_s>>s->chroma_y_shift)*s->uvlinesize;
3796 if(s->picture_structure == PICT_FRAME){
3800 mpeg_motion_lowres(s, dest_y, dest_cb, dest_cr,
3802 ref_picture, pix_op,
3803 s->mv[dir][2*i + j][0], s->mv[dir][2*i + j][1], block_s);
3805 pix_op = s->dsp.avg_h264_chroma_pixels_tab;
3809 mpeg_motion_lowres(s, dest_y, dest_cb, dest_cr,
3810 0, 0, s->picture_structure != i+1,
3811 ref_picture, pix_op,
3812 s->mv[dir][2*i][0],s->mv[dir][2*i][1],2*block_s);
3814 // after put we make avg of the same block
3815 pix_op = s->dsp.avg_h264_chroma_pixels_tab;
3817 //opposite parity is always in the same frame if this is second field
3818 if(!s->first_field){
3819 ref_picture = s->current_picture_ptr->data;
3828 /* put block[] to dest[] */
3829 static inline void put_dct(MpegEncContext *s,
3830 DCTELEM *block, int i, uint8_t *dest, int line_size, int qscale)
3832 s->dct_unquantize_intra(s, block, i, qscale);
3833 s->dsp.idct_put (dest, line_size, block);
3836 /* add block[] to dest[] */
3837 static inline void add_dct(MpegEncContext *s,
3838 DCTELEM *block, int i, uint8_t *dest, int line_size)
3840 if (s->block_last_index[i] >= 0) {
3841 s->dsp.idct_add (dest, line_size, block);
3845 static inline void add_dequant_dct(MpegEncContext *s,
3846 DCTELEM *block, int i, uint8_t *dest, int line_size, int qscale)
3848 if (s->block_last_index[i] >= 0) {
3849 s->dct_unquantize_inter(s, block, i, qscale);
3851 s->dsp.idct_add (dest, line_size, block);
3856 * cleans dc, ac, coded_block for the current non intra MB
3858 void ff_clean_intra_table_entries(MpegEncContext *s)
3860 int wrap = s->b8_stride;
3861 int xy = s->block_index[0];
3864 s->dc_val[0][xy + 1 ] =
3865 s->dc_val[0][xy + wrap] =
3866 s->dc_val[0][xy + 1 + wrap] = 1024;
3868 memset(s->ac_val[0][xy ], 0, 32 * sizeof(int16_t));
3869 memset(s->ac_val[0][xy + wrap], 0, 32 * sizeof(int16_t));
3870 if (s->msmpeg4_version>=3) {
3871 s->coded_block[xy ] =
3872 s->coded_block[xy + 1 ] =
3873 s->coded_block[xy + wrap] =
3874 s->coded_block[xy + 1 + wrap] = 0;
3877 wrap = s->mb_stride;
3878 xy = s->mb_x + s->mb_y * wrap;
3880 s->dc_val[2][xy] = 1024;
3882 memset(s->ac_val[1][xy], 0, 16 * sizeof(int16_t));
3883 memset(s->ac_val[2][xy], 0, 16 * sizeof(int16_t));
3885 s->mbintra_table[xy]= 0;
3888 /* generic function called after a macroblock has been parsed by the
3889 decoder or after it has been encoded by the encoder.
3891 Important variables used:
3892 s->mb_intra : true if intra macroblock
3893 s->mv_dir : motion vector direction
3894 s->mv_type : motion vector type
3895 s->mv : motion vector
3896 s->interlaced_dct : true if interlaced dct used (mpeg2)
3898 static always_inline void MPV_decode_mb_internal(MpegEncContext *s, DCTELEM block[12][64], int lowres_flag)
3901 const int mb_xy = s->mb_y * s->mb_stride + s->mb_x;
3903 if(s->avctx->xvmc_acceleration){
3904 XVMC_decode_mb(s);//xvmc uses pblocks
3912 if(s->avctx->debug&FF_DEBUG_DCT_COEFF) {
3913 /* save DCT coefficients */
3915 DCTELEM *dct = &s->current_picture.dct_coeff[mb_xy*64*6];
3918 *dct++ = block[i][s->dsp.idct_permutation[j]];
3921 s->current_picture.qscale_table[mb_xy]= s->qscale;
3923 /* update DC predictors for P macroblocks */
3925 if (s->h263_pred || s->h263_aic) {
3926 if(s->mbintra_table[mb_xy])
3927 ff_clean_intra_table_entries(s);
3931 s->last_dc[2] = 128 << s->intra_dc_precision;
3934 else if (s->h263_pred || s->h263_aic)
3935 s->mbintra_table[mb_xy]=1;
3937 if ((s->flags&CODEC_FLAG_PSNR) || !(s->encoding && (s->intra_only || s->pict_type==B_TYPE) && s->avctx->mb_decision != FF_MB_DECISION_RD)) { //FIXME precalc
3938 uint8_t *dest_y, *dest_cb, *dest_cr;
3939 int dct_linesize, dct_offset;
3940 op_pixels_func (*op_pix)[4];
3941 qpel_mc_func (*op_qpix)[16];
3942 const int linesize= s->current_picture.linesize[0]; //not s->linesize as this would be wrong for field pics
3943 const int uvlinesize= s->current_picture.linesize[1];
3944 const int readable= s->pict_type != B_TYPE || s->encoding || s->avctx->draw_horiz_band || lowres_flag;
3945 const int block_size= lowres_flag ? 8>>s->avctx->lowres : 8;
3947 /* avoid copy if macroblock skipped in last frame too */
3948 /* skip only during decoding as we might trash the buffers during encoding a bit */
3950 uint8_t *mbskip_ptr = &s->mbskip_table[mb_xy];
3951 const int age= s->current_picture.age;
3955 if (s->mb_skipped) {
3957 assert(s->pict_type!=I_TYPE);
3959 (*mbskip_ptr) ++; /* indicate that this time we skipped it */
3960 if(*mbskip_ptr >99) *mbskip_ptr= 99;
3962 /* if previous was skipped too, then nothing to do ! */
3963 if (*mbskip_ptr >= age && s->current_picture.reference){
3966 } else if(!s->current_picture.reference){
3967 (*mbskip_ptr) ++; /* increase counter so the age can be compared cleanly */
3968 if(*mbskip_ptr >99) *mbskip_ptr= 99;
3970 *mbskip_ptr = 0; /* not skipped */
3974 dct_linesize = linesize << s->interlaced_dct;
3975 dct_offset =(s->interlaced_dct)? linesize : linesize*block_size;
3979 dest_cb= s->dest[1];
3980 dest_cr= s->dest[2];
3982 dest_y = s->b_scratchpad;
3983 dest_cb= s->b_scratchpad+16*linesize;
3984 dest_cr= s->b_scratchpad+32*linesize;
3988 /* motion handling */
3989 /* decoding or more than one mb_type (MC was already done otherwise) */
3992 h264_chroma_mc_func *op_pix = s->dsp.put_h264_chroma_pixels_tab;
3994 if (s->mv_dir & MV_DIR_FORWARD) {
3995 MPV_motion_lowres(s, dest_y, dest_cb, dest_cr, 0, s->last_picture.data, op_pix);
3996 op_pix = s->dsp.avg_h264_chroma_pixels_tab;
3998 if (s->mv_dir & MV_DIR_BACKWARD) {
3999 MPV_motion_lowres(s, dest_y, dest_cb, dest_cr, 1, s->next_picture.data, op_pix);
4002 op_qpix= s->me.qpel_put;
4003 if ((!s->no_rounding) || s->pict_type==B_TYPE){
4004 op_pix = s->dsp.put_pixels_tab;
4006 op_pix = s->dsp.put_no_rnd_pixels_tab;
4008 if (s->mv_dir & MV_DIR_FORWARD) {
4009 MPV_motion(s, dest_y, dest_cb, dest_cr, 0, s->last_picture.data, op_pix, op_qpix);
4010 op_pix = s->dsp.avg_pixels_tab;
4011 op_qpix= s->me.qpel_avg;
4013 if (s->mv_dir & MV_DIR_BACKWARD) {
4014 MPV_motion(s, dest_y, dest_cb, dest_cr, 1, s->next_picture.data, op_pix, op_qpix);
4019 /* skip dequant / idct if we are really late ;) */
4020 if(s->hurry_up>1) goto skip_idct;
4021 if(s->avctx->skip_idct){
4022 if( (s->avctx->skip_idct >= AVDISCARD_NONREF && s->pict_type == B_TYPE)
4023 ||(s->avctx->skip_idct >= AVDISCARD_NONKEY && s->pict_type != I_TYPE)
4024 || s->avctx->skip_idct >= AVDISCARD_ALL)
4028 /* add dct residue */
4029 if(s->encoding || !( s->h263_msmpeg4 || s->codec_id==CODEC_ID_MPEG1VIDEO || s->codec_id==CODEC_ID_MPEG2VIDEO
4030 || (s->codec_id==CODEC_ID_MPEG4 && !s->mpeg_quant))){
4031 add_dequant_dct(s, block[0], 0, dest_y , dct_linesize, s->qscale);
4032 add_dequant_dct(s, block[1], 1, dest_y + block_size, dct_linesize, s->qscale);
4033 add_dequant_dct(s, block[2], 2, dest_y + dct_offset , dct_linesize, s->qscale);
4034 add_dequant_dct(s, block[3], 3, dest_y + dct_offset + block_size, dct_linesize, s->qscale);
4036 if(!(s->flags&CODEC_FLAG_GRAY)){
4037 if (s->chroma_y_shift){
4038 add_dequant_dct(s, block[4], 4, dest_cb, uvlinesize, s->chroma_qscale);
4039 add_dequant_dct(s, block[5], 5, dest_cr, uvlinesize, s->chroma_qscale);
4043 add_dequant_dct(s, block[4], 4, dest_cb, dct_linesize, s->chroma_qscale);
4044 add_dequant_dct(s, block[5], 5, dest_cr, dct_linesize, s->chroma_qscale);
4045 add_dequant_dct(s, block[6], 6, dest_cb + dct_offset, dct_linesize, s->chroma_qscale);
4046 add_dequant_dct(s, block[7], 7, dest_cr + dct_offset, dct_linesize, s->chroma_qscale);
4049 } else if(s->codec_id != CODEC_ID_WMV2){
4050 add_dct(s, block[0], 0, dest_y , dct_linesize);
4051 add_dct(s, block[1], 1, dest_y + block_size, dct_linesize);
4052 add_dct(s, block[2], 2, dest_y + dct_offset , dct_linesize);
4053 add_dct(s, block[3], 3, dest_y + dct_offset + block_size, dct_linesize);
4055 if(!(s->flags&CODEC_FLAG_GRAY)){
4056 if(s->chroma_y_shift){//Chroma420
4057 add_dct(s, block[4], 4, dest_cb, uvlinesize);
4058 add_dct(s, block[5], 5, dest_cr, uvlinesize);
4061 dct_linesize = uvlinesize << s->interlaced_dct;
4062 dct_offset =(s->interlaced_dct)? uvlinesize : uvlinesize*8;
4064 add_dct(s, block[4], 4, dest_cb, dct_linesize);
4065 add_dct(s, block[5], 5, dest_cr, dct_linesize);
4066 add_dct(s, block[6], 6, dest_cb+dct_offset, dct_linesize);
4067 add_dct(s, block[7], 7, dest_cr+dct_offset, dct_linesize);
4068 if(!s->chroma_x_shift){//Chroma444
4069 add_dct(s, block[8], 8, dest_cb+8, dct_linesize);
4070 add_dct(s, block[9], 9, dest_cr+8, dct_linesize);
4071 add_dct(s, block[10], 10, dest_cb+8+dct_offset, dct_linesize);
4072 add_dct(s, block[11], 11, dest_cr+8+dct_offset, dct_linesize);
4078 ff_wmv2_add_mb(s, block, dest_y, dest_cb, dest_cr);
4081 /* dct only in intra block */
4082 if(s->encoding || !(s->codec_id==CODEC_ID_MPEG1VIDEO || s->codec_id==CODEC_ID_MPEG2VIDEO)){
4083 put_dct(s, block[0], 0, dest_y , dct_linesize, s->qscale);
4084 put_dct(s, block[1], 1, dest_y + block_size, dct_linesize, s->qscale);
4085 put_dct(s, block[2], 2, dest_y + dct_offset , dct_linesize, s->qscale);
4086 put_dct(s, block[3], 3, dest_y + dct_offset + block_size, dct_linesize, s->qscale);
4088 if(!(s->flags&CODEC_FLAG_GRAY)){
4089 if(s->chroma_y_shift){
4090 put_dct(s, block[4], 4, dest_cb, uvlinesize, s->chroma_qscale);
4091 put_dct(s, block[5], 5, dest_cr, uvlinesize, s->chroma_qscale);
4095 put_dct(s, block[4], 4, dest_cb, dct_linesize, s->chroma_qscale);
4096 put_dct(s, block[5], 5, dest_cr, dct_linesize, s->chroma_qscale);
4097 put_dct(s, block[6], 6, dest_cb + dct_offset, dct_linesize, s->chroma_qscale);
4098 put_dct(s, block[7], 7, dest_cr + dct_offset, dct_linesize, s->chroma_qscale);
4102 s->dsp.idct_put(dest_y , dct_linesize, block[0]);
4103 s->dsp.idct_put(dest_y + block_size, dct_linesize, block[1]);
4104 s->dsp.idct_put(dest_y + dct_offset , dct_linesize, block[2]);
4105 s->dsp.idct_put(dest_y + dct_offset + block_size, dct_linesize, block[3]);
4107 if(!(s->flags&CODEC_FLAG_GRAY)){
4108 if(s->chroma_y_shift){
4109 s->dsp.idct_put(dest_cb, uvlinesize, block[4]);
4110 s->dsp.idct_put(dest_cr, uvlinesize, block[5]);
4113 dct_linesize = uvlinesize << s->interlaced_dct;
4114 dct_offset =(s->interlaced_dct)? uvlinesize : uvlinesize*8;
4116 s->dsp.idct_put(dest_cb, dct_linesize, block[4]);
4117 s->dsp.idct_put(dest_cr, dct_linesize, block[5]);
4118 s->dsp.idct_put(dest_cb + dct_offset, dct_linesize, block[6]);
4119 s->dsp.idct_put(dest_cr + dct_offset, dct_linesize, block[7]);
4120 if(!s->chroma_x_shift){//Chroma444
4121 s->dsp.idct_put(dest_cb + 8, dct_linesize, block[8]);
4122 s->dsp.idct_put(dest_cr + 8, dct_linesize, block[9]);
4123 s->dsp.idct_put(dest_cb + 8 + dct_offset, dct_linesize, block[10]);
4124 s->dsp.idct_put(dest_cr + 8 + dct_offset, dct_linesize, block[11]);
4132 s->dsp.put_pixels_tab[0][0](s->dest[0], dest_y , linesize,16);
4133 s->dsp.put_pixels_tab[s->chroma_x_shift][0](s->dest[1], dest_cb, uvlinesize,16 >> s->chroma_y_shift);
4134 s->dsp.put_pixels_tab[s->chroma_x_shift][0](s->dest[2], dest_cr, uvlinesize,16 >> s->chroma_y_shift);
4139 void MPV_decode_mb(MpegEncContext *s, DCTELEM block[12][64]){
4140 if(s->avctx->lowres) MPV_decode_mb_internal(s, block, 1);
4141 else MPV_decode_mb_internal(s, block, 0);
4144 #ifdef CONFIG_ENCODERS
4146 static inline void dct_single_coeff_elimination(MpegEncContext *s, int n, int threshold)
4148 static const char tab[64]=
4160 DCTELEM *block= s->block[n];
4161 const int last_index= s->block_last_index[n];
4166 threshold= -threshold;
4170 /* are all which we could set to zero are allready zero? */
4171 if(last_index<=skip_dc - 1) return;
4173 for(i=0; i<=last_index; i++){
4174 const int j = s->intra_scantable.permutated[i];
4175 const int level = FFABS(block[j]);
4177 if(skip_dc && i==0) continue;
4186 if(score >= threshold) return;
4187 for(i=skip_dc; i<=last_index; i++){
4188 const int j = s->intra_scantable.permutated[i];
4191 if(block[0]) s->block_last_index[n]= 0;
4192 else s->block_last_index[n]= -1;
4195 static inline void clip_coeffs(MpegEncContext *s, DCTELEM *block, int last_index)
4198 const int maxlevel= s->max_qcoeff;
4199 const int minlevel= s->min_qcoeff;
4203 i=1; //skip clipping of intra dc
4207 for(;i<=last_index; i++){
4208 const int j= s->intra_scantable.permutated[i];
4209 int level = block[j];
4211 if (level>maxlevel){
4214 }else if(level<minlevel){
4222 if(overflow && s->avctx->mb_decision == FF_MB_DECISION_SIMPLE)
4223 av_log(s->avctx, AV_LOG_INFO, "warning, clipping %d dct coefficients to %d..%d\n", overflow, minlevel, maxlevel);
4226 #endif //CONFIG_ENCODERS
4230 * @param h is the normal height, this will be reduced automatically if needed for the last row
4232 void ff_draw_horiz_band(MpegEncContext *s, int y, int h){
4233 if (s->avctx->draw_horiz_band) {
4237 if(s->picture_structure != PICT_FRAME){
4240 if(s->first_field && !(s->avctx->slice_flags&SLICE_FLAG_ALLOW_FIELD)) return;
4243 h= FFMIN(h, s->avctx->height - y);
4245 if(s->pict_type==B_TYPE || s->low_delay || (s->avctx->slice_flags&SLICE_FLAG_CODED_ORDER))
4246 src= (AVFrame*)s->current_picture_ptr;
4247 else if(s->last_picture_ptr)
4248 src= (AVFrame*)s->last_picture_ptr;
4252 if(s->pict_type==B_TYPE && s->picture_structure == PICT_FRAME && s->out_format != FMT_H264){
4258 offset[0]= y * s->linesize;;
4260 offset[2]= (y >> s->chroma_y_shift) * s->uvlinesize;
4266 s->avctx->draw_horiz_band(s->avctx, src, offset,
4267 y, s->picture_structure, h);
4271 void ff_init_block_index(MpegEncContext *s){ //FIXME maybe rename
4272 const int linesize= s->current_picture.linesize[0]; //not s->linesize as this would be wrong for field pics
4273 const int uvlinesize= s->current_picture.linesize[1];
4274 const int mb_size= 4 - s->avctx->lowres;
4276 s->block_index[0]= s->b8_stride*(s->mb_y*2 ) - 2 + s->mb_x*2;
4277 s->block_index[1]= s->b8_stride*(s->mb_y*2 ) - 1 + s->mb_x*2;
4278 s->block_index[2]= s->b8_stride*(s->mb_y*2 + 1) - 2 + s->mb_x*2;
4279 s->block_index[3]= s->b8_stride*(s->mb_y*2 + 1) - 1 + s->mb_x*2;
4280 s->block_index[4]= s->mb_stride*(s->mb_y + 1) + s->b8_stride*s->mb_height*2 + s->mb_x - 1;
4281 s->block_index[5]= s->mb_stride*(s->mb_y + s->mb_height + 2) + s->b8_stride*s->mb_height*2 + s->mb_x - 1;
4282 //block_index is not used by mpeg2, so it is not affected by chroma_format
4284 s->dest[0] = s->current_picture.data[0] + ((s->mb_x - 1) << mb_size);
4285 s->dest[1] = s->current_picture.data[1] + ((s->mb_x - 1) << (mb_size - s->chroma_x_shift));
4286 s->dest[2] = s->current_picture.data[2] + ((s->mb_x - 1) << (mb_size - s->chroma_x_shift));
4288 if(!(s->pict_type==B_TYPE && s->avctx->draw_horiz_band && s->picture_structure==PICT_FRAME))
4290 s->dest[0] += s->mb_y * linesize << mb_size;
4291 s->dest[1] += s->mb_y * uvlinesize << (mb_size - s->chroma_y_shift);
4292 s->dest[2] += s->mb_y * uvlinesize << (mb_size - s->chroma_y_shift);
4296 #ifdef CONFIG_ENCODERS
4298 static void get_vissual_weight(int16_t *weight, uint8_t *ptr, int stride){
4308 for(y2= FFMAX(y-1, 0); y2 < FFMIN(8, y+2); y2++){
4309 for(x2= FFMAX(x-1, 0); x2 < FFMIN(8, x+2); x2++){
4310 int v= ptr[x2 + y2*stride];
4316 weight[x + 8*y]= (36*ff_sqrt(count*sqr - sum*sum)) / count;
4321 static always_inline void encode_mb_internal(MpegEncContext *s, int motion_x, int motion_y, int mb_block_height, int mb_block_count)
4323 int16_t weight[8][64];
4324 DCTELEM orig[8][64];
4325 const int mb_x= s->mb_x;
4326 const int mb_y= s->mb_y;
4329 int dct_offset = s->linesize*8; //default for progressive frames
4330 uint8_t *ptr_y, *ptr_cb, *ptr_cr;
4333 for(i=0; i<mb_block_count; i++) skip_dct[i]=0;
4335 if(s->adaptive_quant){
4336 const int last_qp= s->qscale;
4337 const int mb_xy= mb_x + mb_y*s->mb_stride;
4339 s->lambda= s->lambda_table[mb_xy];
4342 if(!(s->flags&CODEC_FLAG_QP_RD)){
4343 s->dquant= s->qscale - last_qp;
4345 if(s->out_format==FMT_H263){
4346 s->dquant= clip(s->dquant, -2, 2); //FIXME RD
4348 if(s->codec_id==CODEC_ID_MPEG4){
4350 if(s->pict_type == B_TYPE){
4352 s->dquant= (s->dquant/2)*2;
4353 if(s->mv_dir&MV_DIRECT)
4356 if(s->mv_type==MV_TYPE_8X8)
4362 ff_set_qscale(s, last_qp + s->dquant);
4363 }else if(s->flags&CODEC_FLAG_QP_RD)
4364 ff_set_qscale(s, s->qscale + s->dquant);
4366 wrap_y = s->linesize;
4367 wrap_c = s->uvlinesize;
4368 ptr_y = s->new_picture.data[0] + (mb_y * 16 * wrap_y) + mb_x * 16;
4369 ptr_cb = s->new_picture.data[1] + (mb_y * mb_block_height * wrap_c) + mb_x * 8;
4370 ptr_cr = s->new_picture.data[2] + (mb_y * mb_block_height * wrap_c) + mb_x * 8;
4372 if(mb_x*16+16 > s->width || mb_y*16+16 > s->height){
4373 uint8_t *ebuf= s->edge_emu_buffer + 32;
4374 ff_emulated_edge_mc(ebuf , ptr_y , wrap_y,16,16,mb_x*16,mb_y*16, s->width , s->height);
4376 ff_emulated_edge_mc(ebuf+18*wrap_y , ptr_cb, wrap_c, 8, mb_block_height, mb_x*8, mb_y*8, s->width>>1, s->height>>1);
4377 ptr_cb= ebuf+18*wrap_y;
4378 ff_emulated_edge_mc(ebuf+18*wrap_y+8, ptr_cr, wrap_c, 8, mb_block_height, mb_x*8, mb_y*8, s->width>>1, s->height>>1);
4379 ptr_cr= ebuf+18*wrap_y+8;
4383 if(s->flags&CODEC_FLAG_INTERLACED_DCT){
4384 int progressive_score, interlaced_score;
4386 s->interlaced_dct=0;
4387 progressive_score= s->dsp.ildct_cmp[4](s, ptr_y , NULL, wrap_y, 8)
4388 +s->dsp.ildct_cmp[4](s, ptr_y + wrap_y*8, NULL, wrap_y, 8) - 400;
4390 if(progressive_score > 0){
4391 interlaced_score = s->dsp.ildct_cmp[4](s, ptr_y , NULL, wrap_y*2, 8)
4392 +s->dsp.ildct_cmp[4](s, ptr_y + wrap_y , NULL, wrap_y*2, 8);
4393 if(progressive_score > interlaced_score){
4394 s->interlaced_dct=1;
4398 if (s->chroma_format == CHROMA_422)
4404 s->dsp.get_pixels(s->block[0], ptr_y , wrap_y);
4405 s->dsp.get_pixels(s->block[1], ptr_y + 8, wrap_y);
4406 s->dsp.get_pixels(s->block[2], ptr_y + dct_offset , wrap_y);
4407 s->dsp.get_pixels(s->block[3], ptr_y + dct_offset + 8, wrap_y);
4409 if(s->flags&CODEC_FLAG_GRAY){
4413 s->dsp.get_pixels(s->block[4], ptr_cb, wrap_c);
4414 s->dsp.get_pixels(s->block[5], ptr_cr, wrap_c);
4415 if(!s->chroma_y_shift){ /* 422 */
4416 s->dsp.get_pixels(s->block[6], ptr_cb + (dct_offset>>1), wrap_c);
4417 s->dsp.get_pixels(s->block[7], ptr_cr + (dct_offset>>1), wrap_c);
4421 op_pixels_func (*op_pix)[4];
4422 qpel_mc_func (*op_qpix)[16];
4423 uint8_t *dest_y, *dest_cb, *dest_cr;
4425 dest_y = s->dest[0];
4426 dest_cb = s->dest[1];
4427 dest_cr = s->dest[2];
4429 if ((!s->no_rounding) || s->pict_type==B_TYPE){
4430 op_pix = s->dsp.put_pixels_tab;
4431 op_qpix= s->dsp.put_qpel_pixels_tab;
4433 op_pix = s->dsp.put_no_rnd_pixels_tab;
4434 op_qpix= s->dsp.put_no_rnd_qpel_pixels_tab;
4437 if (s->mv_dir & MV_DIR_FORWARD) {
4438 MPV_motion(s, dest_y, dest_cb, dest_cr, 0, s->last_picture.data, op_pix, op_qpix);
4439 op_pix = s->dsp.avg_pixels_tab;
4440 op_qpix= s->dsp.avg_qpel_pixels_tab;
4442 if (s->mv_dir & MV_DIR_BACKWARD) {
4443 MPV_motion(s, dest_y, dest_cb, dest_cr, 1, s->next_picture.data, op_pix, op_qpix);
4446 if(s->flags&CODEC_FLAG_INTERLACED_DCT){
4447 int progressive_score, interlaced_score;
4449 s->interlaced_dct=0;
4450 progressive_score= s->dsp.ildct_cmp[0](s, dest_y , ptr_y , wrap_y, 8)
4451 +s->dsp.ildct_cmp[0](s, dest_y + wrap_y*8, ptr_y + wrap_y*8, wrap_y, 8) - 400;
4453 if(s->avctx->ildct_cmp == FF_CMP_VSSE) progressive_score -= 400;
4455 if(progressive_score>0){
4456 interlaced_score = s->dsp.ildct_cmp[0](s, dest_y , ptr_y , wrap_y*2, 8)
4457 +s->dsp.ildct_cmp[0](s, dest_y + wrap_y , ptr_y + wrap_y , wrap_y*2, 8);
4459 if(progressive_score > interlaced_score){
4460 s->interlaced_dct=1;
4464 if (s->chroma_format == CHROMA_422)
4470 s->dsp.diff_pixels(s->block[0], ptr_y , dest_y , wrap_y);
4471 s->dsp.diff_pixels(s->block[1], ptr_y + 8, dest_y + 8, wrap_y);
4472 s->dsp.diff_pixels(s->block[2], ptr_y + dct_offset , dest_y + dct_offset , wrap_y);
4473 s->dsp.diff_pixels(s->block[3], ptr_y + dct_offset + 8, dest_y + dct_offset + 8, wrap_y);
4475 if(s->flags&CODEC_FLAG_GRAY){
4479 s->dsp.diff_pixels(s->block[4], ptr_cb, dest_cb, wrap_c);
4480 s->dsp.diff_pixels(s->block[5], ptr_cr, dest_cr, wrap_c);
4481 if(!s->chroma_y_shift){ /* 422 */
4482 s->dsp.diff_pixels(s->block[6], ptr_cb + (dct_offset>>1), dest_cb + (dct_offset>>1), wrap_c);
4483 s->dsp.diff_pixels(s->block[7], ptr_cr + (dct_offset>>1), dest_cr + (dct_offset>>1), wrap_c);
4486 /* pre quantization */
4487 if(s->current_picture.mc_mb_var[s->mb_stride*mb_y+ mb_x]<2*s->qscale*s->qscale){
4489 if(s->dsp.sad[1](NULL, ptr_y , dest_y , wrap_y, 8) < 20*s->qscale) skip_dct[0]= 1;
4490 if(s->dsp.sad[1](NULL, ptr_y + 8, dest_y + 8, wrap_y, 8) < 20*s->qscale) skip_dct[1]= 1;
4491 if(s->dsp.sad[1](NULL, ptr_y +dct_offset , dest_y +dct_offset , wrap_y, 8) < 20*s->qscale) skip_dct[2]= 1;
4492 if(s->dsp.sad[1](NULL, ptr_y +dct_offset+ 8, dest_y +dct_offset+ 8, wrap_y, 8) < 20*s->qscale) skip_dct[3]= 1;
4493 if(s->dsp.sad[1](NULL, ptr_cb , dest_cb , wrap_c, 8) < 20*s->qscale) skip_dct[4]= 1;
4494 if(s->dsp.sad[1](NULL, ptr_cr , dest_cr , wrap_c, 8) < 20*s->qscale) skip_dct[5]= 1;
4495 if(!s->chroma_y_shift){ /* 422 */
4496 if(s->dsp.sad[1](NULL, ptr_cb +(dct_offset>>1), dest_cb +(dct_offset>>1), wrap_c, 8) < 20*s->qscale) skip_dct[6]= 1;
4497 if(s->dsp.sad[1](NULL, ptr_cr +(dct_offset>>1), dest_cr +(dct_offset>>1), wrap_c, 8) < 20*s->qscale) skip_dct[7]= 1;
4502 if(s->avctx->quantizer_noise_shaping){
4503 if(!skip_dct[0]) get_vissual_weight(weight[0], ptr_y , wrap_y);
4504 if(!skip_dct[1]) get_vissual_weight(weight[1], ptr_y + 8, wrap_y);
4505 if(!skip_dct[2]) get_vissual_weight(weight[2], ptr_y + dct_offset , wrap_y);
4506 if(!skip_dct[3]) get_vissual_weight(weight[3], ptr_y + dct_offset + 8, wrap_y);
4507 if(!skip_dct[4]) get_vissual_weight(weight[4], ptr_cb , wrap_c);
4508 if(!skip_dct[5]) get_vissual_weight(weight[5], ptr_cr , wrap_c);
4509 if(!s->chroma_y_shift){ /* 422 */
4510 if(!skip_dct[6]) get_vissual_weight(weight[6], ptr_cb + (dct_offset>>1), wrap_c);
4511 if(!skip_dct[7]) get_vissual_weight(weight[7], ptr_cr + (dct_offset>>1), wrap_c);
4513 memcpy(orig[0], s->block[0], sizeof(DCTELEM)*64*mb_block_count);
4516 /* DCT & quantize */
4517 assert(s->out_format!=FMT_MJPEG || s->qscale==8);
4519 for(i=0;i<mb_block_count;i++) {
4522 s->block_last_index[i] = s->dct_quantize(s, s->block[i], i, s->qscale, &overflow);
4523 // FIXME we could decide to change to quantizer instead of clipping
4524 // JS: I don't think that would be a good idea it could lower quality instead
4525 // of improve it. Just INTRADC clipping deserves changes in quantizer
4526 if (overflow) clip_coeffs(s, s->block[i], s->block_last_index[i]);
4528 s->block_last_index[i]= -1;
4530 if(s->avctx->quantizer_noise_shaping){
4531 for(i=0;i<mb_block_count;i++) {
4533 s->block_last_index[i] = dct_quantize_refine(s, s->block[i], weight[i], orig[i], i, s->qscale);
4538 if(s->luma_elim_threshold && !s->mb_intra)
4540 dct_single_coeff_elimination(s, i, s->luma_elim_threshold);
4541 if(s->chroma_elim_threshold && !s->mb_intra)
4542 for(i=4; i<mb_block_count; i++)
4543 dct_single_coeff_elimination(s, i, s->chroma_elim_threshold);
4545 if(s->flags & CODEC_FLAG_CBP_RD){
4546 for(i=0;i<mb_block_count;i++) {
4547 if(s->block_last_index[i] == -1)
4548 s->coded_score[i]= INT_MAX/256;
4553 if((s->flags&CODEC_FLAG_GRAY) && s->mb_intra){
4554 s->block_last_index[4]=
4555 s->block_last_index[5]= 0;
4557 s->block[5][0]= (1024 + s->c_dc_scale/2)/ s->c_dc_scale;
4560 //non c quantize code returns incorrect block_last_index FIXME
4561 if(s->alternate_scan && s->dct_quantize != dct_quantize_c){
4562 for(i=0; i<mb_block_count; i++){
4564 if(s->block_last_index[i]>0){
4565 for(j=63; j>0; j--){
4566 if(s->block[i][ s->intra_scantable.permutated[j] ]) break;
4568 s->block_last_index[i]= j;
4573 /* huffman encode */
4574 switch(s->codec_id){ //FIXME funct ptr could be slightly faster
4575 case CODEC_ID_MPEG1VIDEO:
4576 case CODEC_ID_MPEG2VIDEO:
4577 mpeg1_encode_mb(s, s->block, motion_x, motion_y); break;
4578 case CODEC_ID_MPEG4:
4579 mpeg4_encode_mb(s, s->block, motion_x, motion_y); break;
4580 case CODEC_ID_MSMPEG4V2:
4581 case CODEC_ID_MSMPEG4V3:
4583 msmpeg4_encode_mb(s, s->block, motion_x, motion_y); break;
4585 ff_wmv2_encode_mb(s, s->block, motion_x, motion_y); break;
4586 #ifdef CONFIG_H261_ENCODER
4588 ff_h261_encode_mb(s, s->block, motion_x, motion_y); break;
4591 case CODEC_ID_H263P:
4595 h263_encode_mb(s, s->block, motion_x, motion_y); break;
4596 case CODEC_ID_MJPEG:
4597 mjpeg_encode_mb(s, s->block); break;
4603 static always_inline void encode_mb(MpegEncContext *s, int motion_x, int motion_y)
4605 if (s->chroma_format == CHROMA_420) encode_mb_internal(s, motion_x, motion_y, 8, 6);
4606 else encode_mb_internal(s, motion_x, motion_y, 16, 8);
4609 #endif //CONFIG_ENCODERS
4611 void ff_mpeg_flush(AVCodecContext *avctx){
4613 MpegEncContext *s = avctx->priv_data;
4615 if(s==NULL || s->picture==NULL)
4618 for(i=0; i<MAX_PICTURE_COUNT; i++){
4619 if(s->picture[i].data[0] && ( s->picture[i].type == FF_BUFFER_TYPE_INTERNAL
4620 || s->picture[i].type == FF_BUFFER_TYPE_USER))
4621 avctx->release_buffer(avctx, (AVFrame*)&s->picture[i]);
4623 s->current_picture_ptr = s->last_picture_ptr = s->next_picture_ptr = NULL;
4625 s->mb_x= s->mb_y= 0;
4627 s->parse_context.state= -1;
4628 s->parse_context.frame_start_found= 0;
4629 s->parse_context.overread= 0;
4630 s->parse_context.overread_index= 0;
4631 s->parse_context.index= 0;
4632 s->parse_context.last_index= 0;
4633 s->bitstream_buffer_size=0;
4636 #ifdef CONFIG_ENCODERS
4637 void ff_copy_bits(PutBitContext *pb, uint8_t *src, int length)
4639 const uint16_t *srcw= (uint16_t*)src;
4640 int words= length>>4;
4641 int bits= length&15;
4644 if(length==0) return;
4647 for(i=0; i<words; i++) put_bits(pb, 16, be2me_16(srcw[i]));
4648 }else if(put_bits_count(pb)&7){
4649 for(i=0; i<words; i++) put_bits(pb, 16, be2me_16(srcw[i]));
4651 for(i=0; put_bits_count(pb)&31; i++)
4652 put_bits(pb, 8, src[i]);
4654 memcpy(pbBufPtr(pb), src+i, 2*words-i);
4655 skip_put_bytes(pb, 2*words-i);
4658 put_bits(pb, bits, be2me_16(srcw[words])>>(16-bits));
4661 static inline void copy_context_before_encode(MpegEncContext *d, MpegEncContext *s, int type){
4664 memcpy(d->last_mv, s->last_mv, 2*2*2*sizeof(int)); //FIXME is memcpy faster then a loop?
4667 d->mb_skip_run= s->mb_skip_run;
4669 d->last_dc[i]= s->last_dc[i];
4672 d->mv_bits= s->mv_bits;
4673 d->i_tex_bits= s->i_tex_bits;
4674 d->p_tex_bits= s->p_tex_bits;
4675 d->i_count= s->i_count;
4676 d->f_count= s->f_count;
4677 d->b_count= s->b_count;
4678 d->skip_count= s->skip_count;
4679 d->misc_bits= s->misc_bits;
4683 d->qscale= s->qscale;
4684 d->dquant= s->dquant;
4687 static inline void copy_context_after_encode(MpegEncContext *d, MpegEncContext *s, int type){
4690 memcpy(d->mv, s->mv, 2*4*2*sizeof(int));
4691 memcpy(d->last_mv, s->last_mv, 2*2*2*sizeof(int)); //FIXME is memcpy faster then a loop?
4694 d->mb_skip_run= s->mb_skip_run;
4696 d->last_dc[i]= s->last_dc[i];
4699 d->mv_bits= s->mv_bits;
4700 d->i_tex_bits= s->i_tex_bits;
4701 d->p_tex_bits= s->p_tex_bits;
4702 d->i_count= s->i_count;
4703 d->f_count= s->f_count;
4704 d->b_count= s->b_count;
4705 d->skip_count= s->skip_count;
4706 d->misc_bits= s->misc_bits;
4708 d->mb_intra= s->mb_intra;
4709 d->mb_skipped= s->mb_skipped;
4710 d->mv_type= s->mv_type;
4711 d->mv_dir= s->mv_dir;
4713 if(s->data_partitioning){
4715 d->tex_pb= s->tex_pb;
4719 d->block_last_index[i]= s->block_last_index[i];
4720 d->interlaced_dct= s->interlaced_dct;
4721 d->qscale= s->qscale;
4724 static inline void encode_mb_hq(MpegEncContext *s, MpegEncContext *backup, MpegEncContext *best, int type,
4725 PutBitContext pb[2], PutBitContext pb2[2], PutBitContext tex_pb[2],
4726 int *dmin, int *next_block, int motion_x, int motion_y)
4729 uint8_t *dest_backup[3];
4731 copy_context_before_encode(s, backup, type);
4733 s->block= s->blocks[*next_block];
4734 s->pb= pb[*next_block];
4735 if(s->data_partitioning){
4736 s->pb2 = pb2 [*next_block];
4737 s->tex_pb= tex_pb[*next_block];
4741 memcpy(dest_backup, s->dest, sizeof(s->dest));
4742 s->dest[0] = s->rd_scratchpad;
4743 s->dest[1] = s->rd_scratchpad + 16*s->linesize;
4744 s->dest[2] = s->rd_scratchpad + 16*s->linesize + 8;
4745 assert(s->linesize >= 32); //FIXME
4748 encode_mb(s, motion_x, motion_y);
4750 score= put_bits_count(&s->pb);
4751 if(s->data_partitioning){
4752 score+= put_bits_count(&s->pb2);
4753 score+= put_bits_count(&s->tex_pb);
4756 if(s->avctx->mb_decision == FF_MB_DECISION_RD){
4757 MPV_decode_mb(s, s->block);
4759 score *= s->lambda2;
4760 score += sse_mb(s) << FF_LAMBDA_SHIFT;
4764 memcpy(s->dest, dest_backup, sizeof(s->dest));
4771 copy_context_after_encode(best, s, type);
4775 static int sse(MpegEncContext *s, uint8_t *src1, uint8_t *src2, int w, int h, int stride){
4776 uint32_t *sq = ff_squareTbl + 256;
4781 return s->dsp.sse[0](NULL, src1, src2, stride, 16);
4782 else if(w==8 && h==8)
4783 return s->dsp.sse[1](NULL, src1, src2, stride, 8);
4787 acc+= sq[src1[x + y*stride] - src2[x + y*stride]];
4796 static int sse_mb(MpegEncContext *s){
4800 if(s->mb_x*16 + 16 > s->width ) w= s->width - s->mb_x*16;
4801 if(s->mb_y*16 + 16 > s->height) h= s->height- s->mb_y*16;
4804 if(s->avctx->mb_cmp == FF_CMP_NSSE){
4805 return s->dsp.nsse[0](s, s->new_picture.data[0] + s->mb_x*16 + s->mb_y*s->linesize*16, s->dest[0], s->linesize, 16)
4806 +s->dsp.nsse[1](s, s->new_picture.data[1] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,s->dest[1], s->uvlinesize, 8)
4807 +s->dsp.nsse[1](s, s->new_picture.data[2] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,s->dest[2], s->uvlinesize, 8);
4809 return s->dsp.sse[0](NULL, s->new_picture.data[0] + s->mb_x*16 + s->mb_y*s->linesize*16, s->dest[0], s->linesize, 16)
4810 +s->dsp.sse[1](NULL, s->new_picture.data[1] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,s->dest[1], s->uvlinesize, 8)
4811 +s->dsp.sse[1](NULL, s->new_picture.data[2] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,s->dest[2], s->uvlinesize, 8);
4814 return sse(s, s->new_picture.data[0] + s->mb_x*16 + s->mb_y*s->linesize*16, s->dest[0], w, h, s->linesize)
4815 +sse(s, s->new_picture.data[1] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,s->dest[1], w>>1, h>>1, s->uvlinesize)
4816 +sse(s, s->new_picture.data[2] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,s->dest[2], w>>1, h>>1, s->uvlinesize);
4819 static int pre_estimate_motion_thread(AVCodecContext *c, void *arg){
4820 MpegEncContext *s= arg;
4824 s->me.dia_size= s->avctx->pre_dia_size;
4825 s->first_slice_line=1;
4826 for(s->mb_y= s->end_mb_y-1; s->mb_y >= s->start_mb_y; s->mb_y--) {
4827 for(s->mb_x=s->mb_width-1; s->mb_x >=0 ;s->mb_x--) {
4828 ff_pre_estimate_p_frame_motion(s, s->mb_x, s->mb_y);
4830 s->first_slice_line=0;
4838 static int estimate_motion_thread(AVCodecContext *c, void *arg){
4839 MpegEncContext *s= arg;
4841 ff_check_alignment();
4843 s->me.dia_size= s->avctx->dia_size;
4844 s->first_slice_line=1;
4845 for(s->mb_y= s->start_mb_y; s->mb_y < s->end_mb_y; s->mb_y++) {
4846 s->mb_x=0; //for block init below
4847 ff_init_block_index(s);
4848 for(s->mb_x=0; s->mb_x < s->mb_width; s->mb_x++) {
4849 s->block_index[0]+=2;
4850 s->block_index[1]+=2;
4851 s->block_index[2]+=2;
4852 s->block_index[3]+=2;
4854 /* compute motion vector & mb_type and store in context */
4855 if(s->pict_type==B_TYPE)
4856 ff_estimate_b_frame_motion(s, s->mb_x, s->mb_y);
4858 ff_estimate_p_frame_motion(s, s->mb_x, s->mb_y);
4860 s->first_slice_line=0;
4865 static int mb_var_thread(AVCodecContext *c, void *arg){
4866 MpegEncContext *s= arg;
4869 ff_check_alignment();
4871 for(mb_y=s->start_mb_y; mb_y < s->end_mb_y; mb_y++) {
4872 for(mb_x=0; mb_x < s->mb_width; mb_x++) {
4875 uint8_t *pix = s->new_picture.data[0] + (yy * s->linesize) + xx;
4877 int sum = s->dsp.pix_sum(pix, s->linesize);
4879 varc = (s->dsp.pix_norm1(pix, s->linesize) - (((unsigned)(sum*sum))>>8) + 500 + 128)>>8;
4881 s->current_picture.mb_var [s->mb_stride * mb_y + mb_x] = varc;
4882 s->current_picture.mb_mean[s->mb_stride * mb_y + mb_x] = (sum+128)>>8;
4883 s->me.mb_var_sum_temp += varc;
4889 static void write_slice_end(MpegEncContext *s){
4890 if(s->codec_id==CODEC_ID_MPEG4){
4891 if(s->partitioned_frame){
4892 ff_mpeg4_merge_partitions(s);
4895 ff_mpeg4_stuffing(&s->pb);
4896 }else if(s->out_format == FMT_MJPEG){
4897 ff_mjpeg_stuffing(&s->pb);
4900 align_put_bits(&s->pb);
4901 flush_put_bits(&s->pb);
4903 if((s->flags&CODEC_FLAG_PASS1) && !s->partitioned_frame)
4904 s->misc_bits+= get_bits_diff(s);
4907 static int encode_thread(AVCodecContext *c, void *arg){
4908 MpegEncContext *s= arg;
4909 int mb_x, mb_y, pdif = 0;
4911 MpegEncContext best_s, backup_s;
4912 uint8_t bit_buf[2][MAX_MB_BYTES];
4913 uint8_t bit_buf2[2][MAX_MB_BYTES];
4914 uint8_t bit_buf_tex[2][MAX_MB_BYTES];
4915 PutBitContext pb[2], pb2[2], tex_pb[2];
4916 //printf("%d->%d\n", s->resync_mb_y, s->end_mb_y);
4918 ff_check_alignment();
4921 init_put_bits(&pb [i], bit_buf [i], MAX_MB_BYTES);
4922 init_put_bits(&pb2 [i], bit_buf2 [i], MAX_MB_BYTES);
4923 init_put_bits(&tex_pb[i], bit_buf_tex[i], MAX_MB_BYTES);
4926 s->last_bits= put_bits_count(&s->pb);
4937 /* init last dc values */
4938 /* note: quant matrix value (8) is implied here */
4939 s->last_dc[i] = 128 << s->intra_dc_precision;
4941 s->current_picture.error[i] = 0;
4944 memset(s->last_mv, 0, sizeof(s->last_mv));
4948 switch(s->codec_id){
4950 case CODEC_ID_H263P:
4952 s->gob_index = ff_h263_get_gob_height(s);
4954 case CODEC_ID_MPEG4:
4955 if(s->partitioned_frame)
4956 ff_mpeg4_init_partitions(s);
4962 s->first_slice_line = 1;
4963 s->ptr_lastgob = s->pb.buf;
4964 for(mb_y= s->start_mb_y; mb_y < s->end_mb_y; mb_y++) {
4965 // printf("row %d at %X\n", s->mb_y, (int)s);
4969 ff_set_qscale(s, s->qscale);
4970 ff_init_block_index(s);
4972 for(mb_x=0; mb_x < s->mb_width; mb_x++) {
4973 int xy= mb_y*s->mb_stride + mb_x; // removed const, H261 needs to adjust this
4974 int mb_type= s->mb_type[xy];
4979 if(s->pb.buf_end - s->pb.buf - (put_bits_count(&s->pb)>>3) < MAX_MB_BYTES){
4980 av_log(s->avctx, AV_LOG_ERROR, "encoded frame too large\n");
4983 if(s->data_partitioning){
4984 if( s->pb2 .buf_end - s->pb2 .buf - (put_bits_count(&s-> pb2)>>3) < MAX_MB_BYTES
4985 || s->tex_pb.buf_end - s->tex_pb.buf - (put_bits_count(&s->tex_pb )>>3) < MAX_MB_BYTES){
4986 av_log(s->avctx, AV_LOG_ERROR, "encoded frame too large\n");
4992 s->mb_y = mb_y; // moved into loop, can get changed by H.261
4993 ff_update_block_index(s);
4995 #ifdef CONFIG_H261_ENCODER
4996 if(s->codec_id == CODEC_ID_H261){
4997 ff_h261_reorder_mb_index(s);
4998 xy= s->mb_y*s->mb_stride + s->mb_x;
4999 mb_type= s->mb_type[xy];
5003 /* write gob / video packet header */
5005 int current_packet_size, is_gob_start;
5007 current_packet_size= ((put_bits_count(&s->pb)+7)>>3) - (s->ptr_lastgob - s->pb.buf);
5009 is_gob_start= s->avctx->rtp_payload_size && current_packet_size >= s->avctx->rtp_payload_size && mb_y + mb_x>0;
5011 if(s->start_mb_y == mb_y && mb_y > 0 && mb_x==0) is_gob_start=1;
5013 switch(s->codec_id){
5015 case CODEC_ID_H263P:
5016 if(!s->h263_slice_structured)
5017 if(s->mb_x || s->mb_y%s->gob_index) is_gob_start=0;
5019 case CODEC_ID_MPEG2VIDEO:
5020 if(s->mb_x==0 && s->mb_y!=0) is_gob_start=1;
5021 case CODEC_ID_MPEG1VIDEO:
5022 if(s->mb_skip_run) is_gob_start=0;
5027 if(s->start_mb_y != mb_y || mb_x!=0){
5030 if(s->codec_id==CODEC_ID_MPEG4 && s->partitioned_frame){
5031 ff_mpeg4_init_partitions(s);
5035 assert((put_bits_count(&s->pb)&7) == 0);
5036 current_packet_size= pbBufPtr(&s->pb) - s->ptr_lastgob;
5038 if(s->avctx->error_rate && s->resync_mb_x + s->resync_mb_y > 0){
5039 int r= put_bits_count(&s->pb)/8 + s->picture_number + 16 + s->mb_x + s->mb_y;
5040 int d= 100 / s->avctx->error_rate;
5042 current_packet_size=0;
5043 #ifndef ALT_BITSTREAM_WRITER
5044 s->pb.buf_ptr= s->ptr_lastgob;
5046 assert(pbBufPtr(&s->pb) == s->ptr_lastgob);
5050 if (s->avctx->rtp_callback){
5051 int number_mb = (mb_y - s->resync_mb_y)*s->mb_width + mb_x - s->resync_mb_x;
5052 s->avctx->rtp_callback(s->avctx, s->ptr_lastgob, current_packet_size, number_mb);
5055 switch(s->codec_id){
5056 case CODEC_ID_MPEG4:
5057 ff_mpeg4_encode_video_packet_header(s);
5058 ff_mpeg4_clean_buffers(s);
5060 case CODEC_ID_MPEG1VIDEO:
5061 case CODEC_ID_MPEG2VIDEO:
5062 ff_mpeg1_encode_slice_header(s);
5063 ff_mpeg1_clean_buffers(s);
5066 case CODEC_ID_H263P:
5067 h263_encode_gob_header(s, mb_y);
5071 if(s->flags&CODEC_FLAG_PASS1){
5072 int bits= put_bits_count(&s->pb);
5073 s->misc_bits+= bits - s->last_bits;
5077 s->ptr_lastgob += current_packet_size;
5078 s->first_slice_line=1;
5079 s->resync_mb_x=mb_x;
5080 s->resync_mb_y=mb_y;
5084 if( (s->resync_mb_x == s->mb_x)
5085 && s->resync_mb_y+1 == s->mb_y){
5086 s->first_slice_line=0;
5090 s->dquant=0; //only for QP_RD
5092 if(mb_type & (mb_type-1) || (s->flags & CODEC_FLAG_QP_RD)){ // more than 1 MB type possible or CODEC_FLAG_QP_RD
5094 int pb_bits_count, pb2_bits_count, tex_pb_bits_count;
5096 copy_context_before_encode(&backup_s, s, -1);
5098 best_s.data_partitioning= s->data_partitioning;
5099 best_s.partitioned_frame= s->partitioned_frame;
5100 if(s->data_partitioning){
5101 backup_s.pb2= s->pb2;
5102 backup_s.tex_pb= s->tex_pb;
5105 if(mb_type&CANDIDATE_MB_TYPE_INTER){
5106 s->mv_dir = MV_DIR_FORWARD;
5107 s->mv_type = MV_TYPE_16X16;
5109 s->mv[0][0][0] = s->p_mv_table[xy][0];
5110 s->mv[0][0][1] = s->p_mv_table[xy][1];
5111 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_INTER, pb, pb2, tex_pb,
5112 &dmin, &next_block, s->mv[0][0][0], s->mv[0][0][1]);
5114 if(mb_type&CANDIDATE_MB_TYPE_INTER_I){
5115 s->mv_dir = MV_DIR_FORWARD;
5116 s->mv_type = MV_TYPE_FIELD;
5119 j= s->field_select[0][i] = s->p_field_select_table[i][xy];
5120 s->mv[0][i][0] = s->p_field_mv_table[i][j][xy][0];
5121 s->mv[0][i][1] = s->p_field_mv_table[i][j][xy][1];
5123 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_INTER_I, pb, pb2, tex_pb,
5124 &dmin, &next_block, 0, 0);
5126 if(mb_type&CANDIDATE_MB_TYPE_SKIPPED){
5127 s->mv_dir = MV_DIR_FORWARD;
5128 s->mv_type = MV_TYPE_16X16;
5132 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_SKIPPED, pb, pb2, tex_pb,
5133 &dmin, &next_block, s->mv[0][0][0], s->mv[0][0][1]);
5135 if(mb_type&CANDIDATE_MB_TYPE_INTER4V){
5136 s->mv_dir = MV_DIR_FORWARD;
5137 s->mv_type = MV_TYPE_8X8;
5140 s->mv[0][i][0] = s->current_picture.motion_val[0][s->block_index[i]][0];
5141 s->mv[0][i][1] = s->current_picture.motion_val[0][s->block_index[i]][1];
5143 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_INTER4V, pb, pb2, tex_pb,
5144 &dmin, &next_block, 0, 0);
5146 if(mb_type&CANDIDATE_MB_TYPE_FORWARD){
5147 s->mv_dir = MV_DIR_FORWARD;
5148 s->mv_type = MV_TYPE_16X16;
5150 s->mv[0][0][0] = s->b_forw_mv_table[xy][0];
5151 s->mv[0][0][1] = s->b_forw_mv_table[xy][1];
5152 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_FORWARD, pb, pb2, tex_pb,
5153 &dmin, &next_block, s->mv[0][0][0], s->mv[0][0][1]);
5155 if(mb_type&CANDIDATE_MB_TYPE_BACKWARD){
5156 s->mv_dir = MV_DIR_BACKWARD;
5157 s->mv_type = MV_TYPE_16X16;
5159 s->mv[1][0][0] = s->b_back_mv_table[xy][0];
5160 s->mv[1][0][1] = s->b_back_mv_table[xy][1];
5161 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_BACKWARD, pb, pb2, tex_pb,
5162 &dmin, &next_block, s->mv[1][0][0], s->mv[1][0][1]);
5164 if(mb_type&CANDIDATE_MB_TYPE_BIDIR){
5165 s->mv_dir = MV_DIR_FORWARD | MV_DIR_BACKWARD;
5166 s->mv_type = MV_TYPE_16X16;
5168 s->mv[0][0][0] = s->b_bidir_forw_mv_table[xy][0];
5169 s->mv[0][0][1] = s->b_bidir_forw_mv_table[xy][1];
5170 s->mv[1][0][0] = s->b_bidir_back_mv_table[xy][0];
5171 s->mv[1][0][1] = s->b_bidir_back_mv_table[xy][1];
5172 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_BIDIR, pb, pb2, tex_pb,
5173 &dmin, &next_block, 0, 0);
5175 if(mb_type&CANDIDATE_MB_TYPE_DIRECT){
5176 int mx= s->b_direct_mv_table[xy][0];
5177 int my= s->b_direct_mv_table[xy][1];
5179 s->mv_dir = MV_DIR_FORWARD | MV_DIR_BACKWARD | MV_DIRECT;
5181 ff_mpeg4_set_direct_mv(s, mx, my);
5182 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_DIRECT, pb, pb2, tex_pb,
5183 &dmin, &next_block, mx, my);
5185 if(mb_type&CANDIDATE_MB_TYPE_FORWARD_I){
5186 s->mv_dir = MV_DIR_FORWARD;
5187 s->mv_type = MV_TYPE_FIELD;
5190 j= s->field_select[0][i] = s->b_field_select_table[0][i][xy];
5191 s->mv[0][i][0] = s->b_field_mv_table[0][i][j][xy][0];
5192 s->mv[0][i][1] = s->b_field_mv_table[0][i][j][xy][1];
5194 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_FORWARD_I, pb, pb2, tex_pb,
5195 &dmin, &next_block, 0, 0);
5197 if(mb_type&CANDIDATE_MB_TYPE_BACKWARD_I){
5198 s->mv_dir = MV_DIR_BACKWARD;
5199 s->mv_type = MV_TYPE_FIELD;
5202 j= s->field_select[1][i] = s->b_field_select_table[1][i][xy];
5203 s->mv[1][i][0] = s->b_field_mv_table[1][i][j][xy][0];
5204 s->mv[1][i][1] = s->b_field_mv_table[1][i][j][xy][1];
5206 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_BACKWARD_I, pb, pb2, tex_pb,
5207 &dmin, &next_block, 0, 0);
5209 if(mb_type&CANDIDATE_MB_TYPE_BIDIR_I){
5210 s->mv_dir = MV_DIR_FORWARD | MV_DIR_BACKWARD;
5211 s->mv_type = MV_TYPE_FIELD;
5213 for(dir=0; dir<2; dir++){
5215 j= s->field_select[dir][i] = s->b_field_select_table[dir][i][xy];
5216 s->mv[dir][i][0] = s->b_field_mv_table[dir][i][j][xy][0];
5217 s->mv[dir][i][1] = s->b_field_mv_table[dir][i][j][xy][1];
5220 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_BIDIR_I, pb, pb2, tex_pb,
5221 &dmin, &next_block, 0, 0);
5223 if(mb_type&CANDIDATE_MB_TYPE_INTRA){
5225 s->mv_type = MV_TYPE_16X16;
5229 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_INTRA, pb, pb2, tex_pb,
5230 &dmin, &next_block, 0, 0);
5231 if(s->h263_pred || s->h263_aic){
5233 s->mbintra_table[mb_x + mb_y*s->mb_stride]=1;
5235 ff_clean_intra_table_entries(s); //old mode?
5239 if(s->flags & CODEC_FLAG_QP_RD){
5240 if(best_s.mv_type==MV_TYPE_16X16 && !(best_s.mv_dir&MV_DIRECT)){
5241 const int last_qp= backup_s.qscale;
5244 const int mvdir= (best_s.mv_dir&MV_DIR_BACKWARD) ? 1 : 0;
5245 static const int dquant_tab[4]={-1,1,-2,2};
5247 assert(backup_s.dquant == 0);
5250 s->mv_dir= best_s.mv_dir;
5251 s->mv_type = MV_TYPE_16X16;
5252 s->mb_intra= best_s.mb_intra;
5253 s->mv[0][0][0] = best_s.mv[0][0][0];
5254 s->mv[0][0][1] = best_s.mv[0][0][1];
5255 s->mv[1][0][0] = best_s.mv[1][0][0];
5256 s->mv[1][0][1] = best_s.mv[1][0][1];
5258 qpi = s->pict_type == B_TYPE ? 2 : 0;
5259 for(; qpi<4; qpi++){
5260 int dquant= dquant_tab[qpi];
5261 qp= last_qp + dquant;
5262 if(qp < s->avctx->qmin || qp > s->avctx->qmax)
5264 backup_s.dquant= dquant;
5265 if(s->mb_intra && s->dc_val[0]){
5267 dc[i]= s->dc_val[0][ s->block_index[i] ];
5268 memcpy(ac[i], s->ac_val[0][s->block_index[i]], sizeof(DCTELEM)*16);
5272 encode_mb_hq(s, &backup_s, &best_s, CANDIDATE_MB_TYPE_INTER /* wrong but unused */, pb, pb2, tex_pb,
5273 &dmin, &next_block, s->mv[mvdir][0][0], s->mv[mvdir][0][1]);
5274 if(best_s.qscale != qp){
5275 if(s->mb_intra && s->dc_val[0]){
5277 s->dc_val[0][ s->block_index[i] ]= dc[i];
5278 memcpy(s->ac_val[0][s->block_index[i]], ac[i], sizeof(DCTELEM)*16);
5284 s->current_picture.qscale_table[xy]= qp;
5288 copy_context_after_encode(s, &best_s, -1);
5290 pb_bits_count= put_bits_count(&s->pb);
5291 flush_put_bits(&s->pb);
5292 ff_copy_bits(&backup_s.pb, bit_buf[next_block^1], pb_bits_count);
5295 if(s->data_partitioning){
5296 pb2_bits_count= put_bits_count(&s->pb2);
5297 flush_put_bits(&s->pb2);
5298 ff_copy_bits(&backup_s.pb2, bit_buf2[next_block^1], pb2_bits_count);
5299 s->pb2= backup_s.pb2;
5301 tex_pb_bits_count= put_bits_count(&s->tex_pb);
5302 flush_put_bits(&s->tex_pb);
5303 ff_copy_bits(&backup_s.tex_pb, bit_buf_tex[next_block^1], tex_pb_bits_count);
5304 s->tex_pb= backup_s.tex_pb;
5306 s->last_bits= put_bits_count(&s->pb);
5308 if (s->out_format == FMT_H263 && s->pict_type!=B_TYPE)
5309 ff_h263_update_motion_val(s);
5311 if(next_block==0){ //FIXME 16 vs linesize16
5312 s->dsp.put_pixels_tab[0][0](s->dest[0], s->rd_scratchpad , s->linesize ,16);
5313 s->dsp.put_pixels_tab[1][0](s->dest[1], s->rd_scratchpad + 16*s->linesize , s->uvlinesize, 8);
5314 s->dsp.put_pixels_tab[1][0](s->dest[2], s->rd_scratchpad + 16*s->linesize + 8, s->uvlinesize, 8);
5317 if(s->avctx->mb_decision == FF_MB_DECISION_BITS)
5318 MPV_decode_mb(s, s->block);
5320 int motion_x, motion_y;
5321 s->mv_type=MV_TYPE_16X16;
5322 // only one MB-Type possible
5325 case CANDIDATE_MB_TYPE_INTRA:
5328 motion_x= s->mv[0][0][0] = 0;
5329 motion_y= s->mv[0][0][1] = 0;
5331 case CANDIDATE_MB_TYPE_INTER:
5332 s->mv_dir = MV_DIR_FORWARD;
5334 motion_x= s->mv[0][0][0] = s->p_mv_table[xy][0];
5335 motion_y= s->mv[0][0][1] = s->p_mv_table[xy][1];
5337 case CANDIDATE_MB_TYPE_INTER_I:
5338 s->mv_dir = MV_DIR_FORWARD;
5339 s->mv_type = MV_TYPE_FIELD;
5342 j= s->field_select[0][i] = s->p_field_select_table[i][xy];
5343 s->mv[0][i][0] = s->p_field_mv_table[i][j][xy][0];
5344 s->mv[0][i][1] = s->p_field_mv_table[i][j][xy][1];
5346 motion_x = motion_y = 0;
5348 case CANDIDATE_MB_TYPE_INTER4V:
5349 s->mv_dir = MV_DIR_FORWARD;
5350 s->mv_type = MV_TYPE_8X8;
5353 s->mv[0][i][0] = s->current_picture.motion_val[0][s->block_index[i]][0];
5354 s->mv[0][i][1] = s->current_picture.motion_val[0][s->block_index[i]][1];
5356 motion_x= motion_y= 0;
5358 case CANDIDATE_MB_TYPE_DIRECT:
5359 s->mv_dir = MV_DIR_FORWARD | MV_DIR_BACKWARD | MV_DIRECT;
5361 motion_x=s->b_direct_mv_table[xy][0];
5362 motion_y=s->b_direct_mv_table[xy][1];
5363 ff_mpeg4_set_direct_mv(s, motion_x, motion_y);
5365 case CANDIDATE_MB_TYPE_BIDIR:
5366 s->mv_dir = MV_DIR_FORWARD | MV_DIR_BACKWARD;
5370 s->mv[0][0][0] = s->b_bidir_forw_mv_table[xy][0];
5371 s->mv[0][0][1] = s->b_bidir_forw_mv_table[xy][1];
5372 s->mv[1][0][0] = s->b_bidir_back_mv_table[xy][0];
5373 s->mv[1][0][1] = s->b_bidir_back_mv_table[xy][1];
5375 case CANDIDATE_MB_TYPE_BACKWARD:
5376 s->mv_dir = MV_DIR_BACKWARD;
5378 motion_x= s->mv[1][0][0] = s->b_back_mv_table[xy][0];
5379 motion_y= s->mv[1][0][1] = s->b_back_mv_table[xy][1];
5381 case CANDIDATE_MB_TYPE_FORWARD:
5382 s->mv_dir = MV_DIR_FORWARD;
5384 motion_x= s->mv[0][0][0] = s->b_forw_mv_table[xy][0];
5385 motion_y= s->mv[0][0][1] = s->b_forw_mv_table[xy][1];
5386 // printf(" %d %d ", motion_x, motion_y);
5388 case CANDIDATE_MB_TYPE_FORWARD_I:
5389 s->mv_dir = MV_DIR_FORWARD;
5390 s->mv_type = MV_TYPE_FIELD;
5393 j= s->field_select[0][i] = s->b_field_select_table[0][i][xy];
5394 s->mv[0][i][0] = s->b_field_mv_table[0][i][j][xy][0];
5395 s->mv[0][i][1] = s->b_field_mv_table[0][i][j][xy][1];
5397 motion_x=motion_y=0;
5399 case CANDIDATE_MB_TYPE_BACKWARD_I:
5400 s->mv_dir = MV_DIR_BACKWARD;
5401 s->mv_type = MV_TYPE_FIELD;
5404 j= s->field_select[1][i] = s->b_field_select_table[1][i][xy];
5405 s->mv[1][i][0] = s->b_field_mv_table[1][i][j][xy][0];
5406 s->mv[1][i][1] = s->b_field_mv_table[1][i][j][xy][1];
5408 motion_x=motion_y=0;
5410 case CANDIDATE_MB_TYPE_BIDIR_I:
5411 s->mv_dir = MV_DIR_FORWARD | MV_DIR_BACKWARD;
5412 s->mv_type = MV_TYPE_FIELD;
5414 for(dir=0; dir<2; dir++){
5416 j= s->field_select[dir][i] = s->b_field_select_table[dir][i][xy];
5417 s->mv[dir][i][0] = s->b_field_mv_table[dir][i][j][xy][0];
5418 s->mv[dir][i][1] = s->b_field_mv_table[dir][i][j][xy][1];
5421 motion_x=motion_y=0;
5424 motion_x=motion_y=0; //gcc warning fix
5425 av_log(s->avctx, AV_LOG_ERROR, "illegal MB type\n");
5428 encode_mb(s, motion_x, motion_y);
5430 // RAL: Update last macroblock type
5431 s->last_mv_dir = s->mv_dir;
5433 if (s->out_format == FMT_H263 && s->pict_type!=B_TYPE)
5434 ff_h263_update_motion_val(s);
5436 MPV_decode_mb(s, s->block);
5439 /* clean the MV table in IPS frames for direct mode in B frames */
5440 if(s->mb_intra /* && I,P,S_TYPE */){
5441 s->p_mv_table[xy][0]=0;
5442 s->p_mv_table[xy][1]=0;
5445 if(s->flags&CODEC_FLAG_PSNR){
5449 if(s->mb_x*16 + 16 > s->width ) w= s->width - s->mb_x*16;
5450 if(s->mb_y*16 + 16 > s->height) h= s->height- s->mb_y*16;
5452 s->current_picture.error[0] += sse(
5453 s, s->new_picture.data[0] + s->mb_x*16 + s->mb_y*s->linesize*16,
5454 s->dest[0], w, h, s->linesize);
5455 s->current_picture.error[1] += sse(
5456 s, s->new_picture.data[1] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,
5457 s->dest[1], w>>1, h>>1, s->uvlinesize);
5458 s->current_picture.error[2] += sse(
5459 s, s->new_picture .data[2] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,
5460 s->dest[2], w>>1, h>>1, s->uvlinesize);
5463 if(s->out_format == FMT_H263)
5464 ff_h263_loop_filter(s);
5466 //printf("MB %d %d bits\n", s->mb_x+s->mb_y*s->mb_stride, put_bits_count(&s->pb));
5470 //not beautiful here but we must write it before flushing so it has to be here
5471 if (s->msmpeg4_version && s->msmpeg4_version<4 && s->pict_type == I_TYPE)
5472 msmpeg4_encode_ext_header(s);
5476 /* Send the last GOB if RTP */
5477 if (s->avctx->rtp_callback) {
5478 int number_mb = (mb_y - s->resync_mb_y)*s->mb_width - s->resync_mb_x;
5479 pdif = pbBufPtr(&s->pb) - s->ptr_lastgob;
5480 /* Call the RTP callback to send the last GOB */
5482 s->avctx->rtp_callback(s->avctx, s->ptr_lastgob, pdif, number_mb);
5488 #define MERGE(field) dst->field += src->field; src->field=0
5489 static void merge_context_after_me(MpegEncContext *dst, MpegEncContext *src){
5490 MERGE(me.scene_change_score);
5491 MERGE(me.mc_mb_var_sum_temp);
5492 MERGE(me.mb_var_sum_temp);
5495 static void merge_context_after_encode(MpegEncContext *dst, MpegEncContext *src){
5498 MERGE(dct_count[0]); //note, the other dct vars are not part of the context
5499 MERGE(dct_count[1]);
5509 MERGE(padding_bug_score);
5510 MERGE(current_picture.error[0]);
5511 MERGE(current_picture.error[1]);
5512 MERGE(current_picture.error[2]);
5514 if(dst->avctx->noise_reduction){
5515 for(i=0; i<64; i++){
5516 MERGE(dct_error_sum[0][i]);
5517 MERGE(dct_error_sum[1][i]);
5521 assert(put_bits_count(&src->pb) % 8 ==0);
5522 assert(put_bits_count(&dst->pb) % 8 ==0);
5523 ff_copy_bits(&dst->pb, src->pb.buf, put_bits_count(&src->pb));
5524 flush_put_bits(&dst->pb);
5527 static int estimate_qp(MpegEncContext *s, int dry_run){
5528 if (s->next_lambda){
5529 s->current_picture_ptr->quality=
5530 s->current_picture.quality = s->next_lambda;
5531 if(!dry_run) s->next_lambda= 0;
5532 } else if (!s->fixed_qscale) {
5533 s->current_picture_ptr->quality=
5534 s->current_picture.quality = ff_rate_estimate_qscale(s, dry_run);
5535 if (s->current_picture.quality < 0)
5539 if(s->adaptive_quant){
5540 switch(s->codec_id){
5541 case CODEC_ID_MPEG4:
5542 ff_clean_mpeg4_qscales(s);
5545 case CODEC_ID_H263P:
5547 ff_clean_h263_qscales(s);
5551 s->lambda= s->lambda_table[0];
5554 s->lambda= s->current_picture.quality;
5555 //printf("%d %d\n", s->avctx->global_quality, s->current_picture.quality);
5560 static int encode_picture(MpegEncContext *s, int picture_number)
5565 s->picture_number = picture_number;
5567 /* Reset the average MB variance */
5568 s->me.mb_var_sum_temp =
5569 s->me.mc_mb_var_sum_temp = 0;
5571 /* we need to initialize some time vars before we can encode b-frames */
5572 // RAL: Condition added for MPEG1VIDEO
5573 if (s->codec_id == CODEC_ID_MPEG1VIDEO || s->codec_id == CODEC_ID_MPEG2VIDEO || (s->h263_pred && !s->h263_msmpeg4))
5574 ff_set_mpeg4_time(s, s->picture_number); //FIXME rename and use has_b_frames or similar
5576 s->me.scene_change_score=0;
5578 // s->lambda= s->current_picture_ptr->quality; //FIXME qscale / ... stuff for ME ratedistoration
5580 if(s->pict_type==I_TYPE){
5581 if(s->msmpeg4_version >= 3) s->no_rounding=1;
5582 else s->no_rounding=0;
5583 }else if(s->pict_type!=B_TYPE){
5584 if(s->flipflop_rounding || s->codec_id == CODEC_ID_H263P || s->codec_id == CODEC_ID_MPEG4)
5585 s->no_rounding ^= 1;
5588 if(s->flags & CODEC_FLAG_PASS2){
5589 if (estimate_qp(s,1) < 0)
5591 ff_get_2pass_fcode(s);
5592 }else if(!(s->flags & CODEC_FLAG_QSCALE)){
5593 if(s->pict_type==B_TYPE)
5594 s->lambda= s->last_lambda_for[s->pict_type];
5596 s->lambda= s->last_lambda_for[s->last_non_b_pict_type];
5600 s->mb_intra=0; //for the rate distortion & bit compare functions
5601 for(i=1; i<s->avctx->thread_count; i++){
5602 ff_update_duplicate_context(s->thread_context[i], s);
5607 /* Estimate motion for every MB */
5608 if(s->pict_type != I_TYPE){
5609 s->lambda = (s->lambda * s->avctx->me_penalty_compensation + 128)>>8;
5610 s->lambda2= (s->lambda2* (int64_t)s->avctx->me_penalty_compensation + 128)>>8;
5611 if(s->pict_type != B_TYPE && s->avctx->me_threshold==0){
5612 if((s->avctx->pre_me && s->last_non_b_pict_type==I_TYPE) || s->avctx->pre_me==2){
5613 s->avctx->execute(s->avctx, pre_estimate_motion_thread, (void**)&(s->thread_context[0]), NULL, s->avctx->thread_count);
5617 s->avctx->execute(s->avctx, estimate_motion_thread, (void**)&(s->thread_context[0]), NULL, s->avctx->thread_count);
5618 }else /* if(s->pict_type == I_TYPE) */{
5620 for(i=0; i<s->mb_stride*s->mb_height; i++)
5621 s->mb_type[i]= CANDIDATE_MB_TYPE_INTRA;
5623 if(!s->fixed_qscale){
5624 /* finding spatial complexity for I-frame rate control */
5625 s->avctx->execute(s->avctx, mb_var_thread, (void**)&(s->thread_context[0]), NULL, s->avctx->thread_count);
5628 for(i=1; i<s->avctx->thread_count; i++){
5629 merge_context_after_me(s, s->thread_context[i]);
5631 s->current_picture.mc_mb_var_sum= s->current_picture_ptr->mc_mb_var_sum= s->me.mc_mb_var_sum_temp;
5632 s->current_picture. mb_var_sum= s->current_picture_ptr-> mb_var_sum= s->me. mb_var_sum_temp;
5635 if(s->me.scene_change_score > s->avctx->scenechange_threshold && s->pict_type == P_TYPE){
5636 s->pict_type= I_TYPE;
5637 for(i=0; i<s->mb_stride*s->mb_height; i++)
5638 s->mb_type[i]= CANDIDATE_MB_TYPE_INTRA;
5639 //printf("Scene change detected, encoding as I Frame %d %d\n", s->current_picture.mb_var_sum, s->current_picture.mc_mb_var_sum);
5643 if(s->pict_type==P_TYPE || s->pict_type==S_TYPE) {
5644 s->f_code= ff_get_best_fcode(s, s->p_mv_table, CANDIDATE_MB_TYPE_INTER);
5646 if(s->flags & CODEC_FLAG_INTERLACED_ME){
5648 a= ff_get_best_fcode(s, s->p_field_mv_table[0][0], CANDIDATE_MB_TYPE_INTER_I); //FIXME field_select
5649 b= ff_get_best_fcode(s, s->p_field_mv_table[1][1], CANDIDATE_MB_TYPE_INTER_I);
5650 s->f_code= FFMAX(s->f_code, FFMAX(a,b));
5653 ff_fix_long_p_mvs(s);
5654 ff_fix_long_mvs(s, NULL, 0, s->p_mv_table, s->f_code, CANDIDATE_MB_TYPE_INTER, 0);
5655 if(s->flags & CODEC_FLAG_INTERLACED_ME){
5659 ff_fix_long_mvs(s, s->p_field_select_table[i], j,
5660 s->p_field_mv_table[i][j], s->f_code, CANDIDATE_MB_TYPE_INTER_I, 0);
5665 if(s->pict_type==B_TYPE){
5668 a = ff_get_best_fcode(s, s->b_forw_mv_table, CANDIDATE_MB_TYPE_FORWARD);
5669 b = ff_get_best_fcode(s, s->b_bidir_forw_mv_table, CANDIDATE_MB_TYPE_BIDIR);
5670 s->f_code = FFMAX(a, b);
5672 a = ff_get_best_fcode(s, s->b_back_mv_table, CANDIDATE_MB_TYPE_BACKWARD);
5673 b = ff_get_best_fcode(s, s->b_bidir_back_mv_table, CANDIDATE_MB_TYPE_BIDIR);
5674 s->b_code = FFMAX(a, b);
5676 ff_fix_long_mvs(s, NULL, 0, s->b_forw_mv_table, s->f_code, CANDIDATE_MB_TYPE_FORWARD, 1);
5677 ff_fix_long_mvs(s, NULL, 0, s->b_back_mv_table, s->b_code, CANDIDATE_MB_TYPE_BACKWARD, 1);
5678 ff_fix_long_mvs(s, NULL, 0, s->b_bidir_forw_mv_table, s->f_code, CANDIDATE_MB_TYPE_BIDIR, 1);
5679 ff_fix_long_mvs(s, NULL, 0, s->b_bidir_back_mv_table, s->b_code, CANDIDATE_MB_TYPE_BIDIR, 1);
5680 if(s->flags & CODEC_FLAG_INTERLACED_ME){
5682 for(dir=0; dir<2; dir++){
5685 int type= dir ? (CANDIDATE_MB_TYPE_BACKWARD_I|CANDIDATE_MB_TYPE_BIDIR_I)
5686 : (CANDIDATE_MB_TYPE_FORWARD_I |CANDIDATE_MB_TYPE_BIDIR_I);
5687 ff_fix_long_mvs(s, s->b_field_select_table[dir][i], j,
5688 s->b_field_mv_table[dir][i][j], dir ? s->b_code : s->f_code, type, 1);
5696 if (estimate_qp(s, 0) < 0)
5699 if(s->qscale < 3 && s->max_qcoeff<=128 && s->pict_type==I_TYPE && !(s->flags & CODEC_FLAG_QSCALE))
5700 s->qscale= 3; //reduce clipping problems
5702 if (s->out_format == FMT_MJPEG) {
5703 /* for mjpeg, we do include qscale in the matrix */
5704 s->intra_matrix[0] = ff_mpeg1_default_intra_matrix[0];
5706 int j= s->dsp.idct_permutation[i];
5708 s->intra_matrix[j] = clip_uint8((ff_mpeg1_default_intra_matrix[i] * s->qscale) >> 3);
5710 convert_matrix(&s->dsp, s->q_intra_matrix, s->q_intra_matrix16,
5711 s->intra_matrix, s->intra_quant_bias, 8, 8, 1);
5715 //FIXME var duplication
5716 s->current_picture_ptr->key_frame=
5717 s->current_picture.key_frame= s->pict_type == I_TYPE; //FIXME pic_ptr
5718 s->current_picture_ptr->pict_type=
5719 s->current_picture.pict_type= s->pict_type;
5721 if(s->current_picture.key_frame)
5722 s->picture_in_gop_number=0;
5724 s->last_bits= put_bits_count(&s->pb);
5725 switch(s->out_format) {
5727 mjpeg_picture_header(s);
5729 #ifdef CONFIG_H261_ENCODER
5731 ff_h261_encode_picture_header(s, picture_number);
5735 if (s->codec_id == CODEC_ID_WMV2)
5736 ff_wmv2_encode_picture_header(s, picture_number);
5737 else if (s->h263_msmpeg4)
5738 msmpeg4_encode_picture_header(s, picture_number);
5739 else if (s->h263_pred)
5740 mpeg4_encode_picture_header(s, picture_number);
5741 #ifdef CONFIG_RV10_ENCODER
5742 else if (s->codec_id == CODEC_ID_RV10)
5743 rv10_encode_picture_header(s, picture_number);
5745 #ifdef CONFIG_RV20_ENCODER
5746 else if (s->codec_id == CODEC_ID_RV20)
5747 rv20_encode_picture_header(s, picture_number);
5749 else if (s->codec_id == CODEC_ID_FLV1)
5750 ff_flv_encode_picture_header(s, picture_number);
5752 h263_encode_picture_header(s, picture_number);
5755 mpeg1_encode_picture_header(s, picture_number);
5762 bits= put_bits_count(&s->pb);
5763 s->header_bits= bits - s->last_bits;
5765 for(i=1; i<s->avctx->thread_count; i++){
5766 update_duplicate_context_after_me(s->thread_context[i], s);
5768 s->avctx->execute(s->avctx, encode_thread, (void**)&(s->thread_context[0]), NULL, s->avctx->thread_count);
5769 for(i=1; i<s->avctx->thread_count; i++){
5770 merge_context_after_encode(s, s->thread_context[i]);
5776 static void denoise_dct_c(MpegEncContext *s, DCTELEM *block){
5777 const int intra= s->mb_intra;
5780 s->dct_count[intra]++;
5782 for(i=0; i<64; i++){
5783 int level= block[i];
5787 s->dct_error_sum[intra][i] += level;
5788 level -= s->dct_offset[intra][i];
5789 if(level<0) level=0;
5791 s->dct_error_sum[intra][i] -= level;
5792 level += s->dct_offset[intra][i];
5793 if(level>0) level=0;
5800 static int dct_quantize_trellis_c(MpegEncContext *s,
5801 DCTELEM *block, int n,
5802 int qscale, int *overflow){
5804 const uint8_t *scantable= s->intra_scantable.scantable;
5805 const uint8_t *perm_scantable= s->intra_scantable.permutated;
5807 unsigned int threshold1, threshold2;
5819 int coeff_count[64];
5820 int qmul, qadd, start_i, last_non_zero, i, dc;
5821 const int esc_length= s->ac_esc_length;
5823 uint8_t * last_length;
5824 const int lambda= s->lambda2 >> (FF_LAMBDA_SHIFT - 6);
5826 s->dsp.fdct (block);
5828 if(s->dct_error_sum)
5829 s->denoise_dct(s, block);
5831 qadd= ((qscale-1)|1)*8;
5842 /* For AIC we skip quant/dequant of INTRADC */
5847 /* note: block[0] is assumed to be positive */
5848 block[0] = (block[0] + (q >> 1)) / q;
5851 qmat = s->q_intra_matrix[qscale];
5852 if(s->mpeg_quant || s->out_format == FMT_MPEG1)
5853 bias= 1<<(QMAT_SHIFT-1);
5854 length = s->intra_ac_vlc_length;
5855 last_length= s->intra_ac_vlc_last_length;
5859 qmat = s->q_inter_matrix[qscale];
5860 length = s->inter_ac_vlc_length;
5861 last_length= s->inter_ac_vlc_last_length;
5865 threshold1= (1<<QMAT_SHIFT) - bias - 1;
5866 threshold2= (threshold1<<1);
5868 for(i=63; i>=start_i; i--) {
5869 const int j = scantable[i];
5870 int level = block[j] * qmat[j];
5872 if(((unsigned)(level+threshold1))>threshold2){
5878 for(i=start_i; i<=last_non_zero; i++) {
5879 const int j = scantable[i];
5880 int level = block[j] * qmat[j];
5882 // if( bias+level >= (1<<(QMAT_SHIFT - 3))
5883 // || bias-level >= (1<<(QMAT_SHIFT - 3))){
5884 if(((unsigned)(level+threshold1))>threshold2){
5886 level= (bias + level)>>QMAT_SHIFT;
5888 coeff[1][i]= level-1;
5889 // coeff[2][k]= level-2;
5891 level= (bias - level)>>QMAT_SHIFT;
5892 coeff[0][i]= -level;
5893 coeff[1][i]= -level+1;
5894 // coeff[2][k]= -level+2;
5896 coeff_count[i]= FFMIN(level, 2);
5897 assert(coeff_count[i]);
5900 coeff[0][i]= (level>>31)|1;
5905 *overflow= s->max_qcoeff < max; //overflow might have happened
5907 if(last_non_zero < start_i){
5908 memset(block + start_i, 0, (64-start_i)*sizeof(DCTELEM));
5909 return last_non_zero;
5912 score_tab[start_i]= 0;
5913 survivor[0]= start_i;
5916 for(i=start_i; i<=last_non_zero; i++){
5918 const int dct_coeff= FFABS(block[ scantable[i] ]);
5919 const int zero_distoration= dct_coeff*dct_coeff;
5920 int best_score=256*256*256*120;
5921 for(level_index=0; level_index < coeff_count[i]; level_index++){
5923 int level= coeff[level_index][i];
5924 const int alevel= FFABS(level);
5929 if(s->out_format == FMT_H263){
5930 unquant_coeff= alevel*qmul + qadd;
5932 j= s->dsp.idct_permutation[ scantable[i] ]; //FIXME optimize
5934 unquant_coeff = (int)( alevel * qscale * s->intra_matrix[j]) >> 3;
5935 unquant_coeff = (unquant_coeff - 1) | 1;
5937 unquant_coeff = ((( alevel << 1) + 1) * qscale * ((int) s->inter_matrix[j])) >> 4;
5938 unquant_coeff = (unquant_coeff - 1) | 1;
5943 distoration= (unquant_coeff - dct_coeff) * (unquant_coeff - dct_coeff) - zero_distoration;
5945 if((level&(~127)) == 0){
5946 for(j=survivor_count-1; j>=0; j--){
5947 int run= i - survivor[j];
5948 int score= distoration + length[UNI_AC_ENC_INDEX(run, level)]*lambda;
5949 score += score_tab[i-run];
5951 if(score < best_score){
5954 level_tab[i+1]= level-64;
5958 if(s->out_format == FMT_H263){
5959 for(j=survivor_count-1; j>=0; j--){
5960 int run= i - survivor[j];
5961 int score= distoration + last_length[UNI_AC_ENC_INDEX(run, level)]*lambda;
5962 score += score_tab[i-run];
5963 if(score < last_score){
5966 last_level= level-64;
5972 distoration += esc_length*lambda;
5973 for(j=survivor_count-1; j>=0; j--){
5974 int run= i - survivor[j];
5975 int score= distoration + score_tab[i-run];
5977 if(score < best_score){
5980 level_tab[i+1]= level-64;
5984 if(s->out_format == FMT_H263){
5985 for(j=survivor_count-1; j>=0; j--){
5986 int run= i - survivor[j];
5987 int score= distoration + score_tab[i-run];
5988 if(score < last_score){
5991 last_level= level-64;
5999 score_tab[i+1]= best_score;
6001 //Note: there is a vlc code in mpeg4 which is 1 bit shorter then another one with a shorter run and the same level
6002 if(last_non_zero <= 27){
6003 for(; survivor_count; survivor_count--){
6004 if(score_tab[ survivor[survivor_count-1] ] <= best_score)
6008 for(; survivor_count; survivor_count--){
6009 if(score_tab[ survivor[survivor_count-1] ] <= best_score + lambda)
6014 survivor[ survivor_count++ ]= i+1;
6017 if(s->out_format != FMT_H263){
6018 last_score= 256*256*256*120;
6019 for(i= survivor[0]; i<=last_non_zero + 1; i++){
6020 int score= score_tab[i];
6021 if(i) score += lambda*2; //FIXME exacter?
6023 if(score < last_score){
6026 last_level= level_tab[i];
6027 last_run= run_tab[i];
6032 s->coded_score[n] = last_score;
6034 dc= FFABS(block[0]);
6035 last_non_zero= last_i - 1;
6036 memset(block + start_i, 0, (64-start_i)*sizeof(DCTELEM));
6038 if(last_non_zero < start_i)
6039 return last_non_zero;
6041 if(last_non_zero == 0 && start_i == 0){
6043 int best_score= dc * dc;
6045 for(i=0; i<coeff_count[0]; i++){
6046 int level= coeff[i][0];
6047 int alevel= FFABS(level);
6048 int unquant_coeff, score, distortion;
6050 if(s->out_format == FMT_H263){
6051 unquant_coeff= (alevel*qmul + qadd)>>3;
6053 unquant_coeff = ((( alevel << 1) + 1) * qscale * ((int) s->inter_matrix[0])) >> 4;
6054 unquant_coeff = (unquant_coeff - 1) | 1;
6056 unquant_coeff = (unquant_coeff + 4) >> 3;
6057 unquant_coeff<<= 3 + 3;
6059 distortion= (unquant_coeff - dc) * (unquant_coeff - dc);
6061 if((level&(~127)) == 0) score= distortion + last_length[UNI_AC_ENC_INDEX(0, level)]*lambda;
6062 else score= distortion + esc_length*lambda;
6064 if(score < best_score){
6066 best_level= level - 64;
6069 block[0]= best_level;
6070 s->coded_score[n] = best_score - dc*dc;
6071 if(best_level == 0) return -1;
6072 else return last_non_zero;
6078 block[ perm_scantable[last_non_zero] ]= last_level;
6081 for(; i>start_i; i -= run_tab[i] + 1){
6082 block[ perm_scantable[i-1] ]= level_tab[i];
6085 return last_non_zero;
6088 //#define REFINE_STATS 1
6089 static int16_t basis[64][64];
6091 static void build_basis(uint8_t *perm){
6098 double s= 0.25*(1<<BASIS_SHIFT);
6100 int perm_index= perm[index];
6101 if(i==0) s*= sqrt(0.5);
6102 if(j==0) s*= sqrt(0.5);
6103 basis[perm_index][8*x + y]= lrintf(s * cos((M_PI/8.0)*i*(x+0.5)) * cos((M_PI/8.0)*j*(y+0.5)));
6110 static int dct_quantize_refine(MpegEncContext *s, //FIXME breaks denoise?
6111 DCTELEM *block, int16_t *weight, DCTELEM *orig,
6114 DECLARE_ALIGNED_16(DCTELEM, d1[64]);
6116 const uint8_t *scantable= s->intra_scantable.scantable;
6117 const uint8_t *perm_scantable= s->intra_scantable.permutated;
6118 // unsigned int threshold1, threshold2;
6123 int qmul, qadd, start_i, last_non_zero, i, dc;
6125 uint8_t * last_length;
6127 int rle_index, run, q, sum;
6130 static int after_last=0;
6131 static int to_zero=0;
6132 static int from_zero=0;
6135 static int messed_sign=0;
6138 if(basis[0][0] == 0)
6139 build_basis(s->dsp.idct_permutation);
6150 /* For AIC we skip quant/dequant of INTRADC */
6154 q <<= RECON_SHIFT-3;
6155 /* note: block[0] is assumed to be positive */
6157 // block[0] = (block[0] + (q >> 1)) / q;
6159 qmat = s->q_intra_matrix[qscale];
6160 // if(s->mpeg_quant || s->out_format == FMT_MPEG1)
6161 // bias= 1<<(QMAT_SHIFT-1);
6162 length = s->intra_ac_vlc_length;
6163 last_length= s->intra_ac_vlc_last_length;
6167 qmat = s->q_inter_matrix[qscale];
6168 length = s->inter_ac_vlc_length;
6169 last_length= s->inter_ac_vlc_last_length;
6171 last_non_zero = s->block_last_index[n];
6176 dc += (1<<(RECON_SHIFT-1));
6177 for(i=0; i<64; i++){
6178 rem[i]= dc - (orig[i]<<RECON_SHIFT); //FIXME use orig dirrectly instead of copying to rem[]
6181 STOP_TIMER("memset rem[]")}
6184 for(i=0; i<64; i++){
6189 w= FFABS(weight[i]) + qns*one;
6190 w= 15 + (48*qns*one + w/2)/w; // 16 .. 63
6193 // w=weight[i] = (63*qns + (w/2)) / w;
6199 lambda= sum*(uint64_t)s->lambda2 >> (FF_LAMBDA_SHIFT - 6 + 6 + 6 + 6);
6205 for(i=start_i; i<=last_non_zero; i++){
6206 int j= perm_scantable[i];
6207 const int level= block[j];
6211 if(level<0) coeff= qmul*level - qadd;
6212 else coeff= qmul*level + qadd;
6213 run_tab[rle_index++]=run;
6216 s->dsp.add_8x8basis(rem, basis[j], coeff);
6222 if(last_non_zero>0){
6223 STOP_TIMER("init rem[]")
6230 int best_score=s->dsp.try_8x8basis(rem, weight, basis[0], 0);
6233 int run2, best_unquant_change=0, analyze_gradient;
6237 analyze_gradient = last_non_zero > 2 || s->avctx->quantizer_noise_shaping >= 3;
6239 if(analyze_gradient){
6243 for(i=0; i<64; i++){
6246 d1[i] = (rem[i]*w*w + (1<<(RECON_SHIFT+12-1)))>>(RECON_SHIFT+12);
6249 STOP_TIMER("rem*w*w")}
6259 const int level= block[0];
6260 int change, old_coeff;
6262 assert(s->mb_intra);
6266 for(change=-1; change<=1; change+=2){
6267 int new_level= level + change;
6268 int score, new_coeff;
6270 new_coeff= q*new_level;
6271 if(new_coeff >= 2048 || new_coeff < 0)
6274 score= s->dsp.try_8x8basis(rem, weight, basis[0], new_coeff - old_coeff);
6275 if(score<best_score){
6278 best_change= change;
6279 best_unquant_change= new_coeff - old_coeff;
6286 run2= run_tab[rle_index++];
6290 for(i=start_i; i<64; i++){
6291 int j= perm_scantable[i];
6292 const int level= block[j];
6293 int change, old_coeff;
6295 if(s->avctx->quantizer_noise_shaping < 3 && i > last_non_zero + 1)
6299 if(level<0) old_coeff= qmul*level - qadd;
6300 else old_coeff= qmul*level + qadd;
6301 run2= run_tab[rle_index++]; //FIXME ! maybe after last
6305 assert(run2>=0 || i >= last_non_zero );
6308 for(change=-1; change<=1; change+=2){
6309 int new_level= level + change;
6310 int score, new_coeff, unquant_change;
6313 if(s->avctx->quantizer_noise_shaping < 2 && FFABS(new_level) > FFABS(level))
6317 if(new_level<0) new_coeff= qmul*new_level - qadd;
6318 else new_coeff= qmul*new_level + qadd;
6319 if(new_coeff >= 2048 || new_coeff <= -2048)
6321 //FIXME check for overflow
6324 if(level < 63 && level > -63){
6325 if(i < last_non_zero)
6326 score += length[UNI_AC_ENC_INDEX(run, new_level+64)]
6327 - length[UNI_AC_ENC_INDEX(run, level+64)];
6329 score += last_length[UNI_AC_ENC_INDEX(run, new_level+64)]
6330 - last_length[UNI_AC_ENC_INDEX(run, level+64)];
6333 assert(FFABS(new_level)==1);
6335 if(analyze_gradient){
6336 int g= d1[ scantable[i] ];
6337 if(g && (g^new_level) >= 0)
6341 if(i < last_non_zero){
6342 int next_i= i + run2 + 1;
6343 int next_level= block[ perm_scantable[next_i] ] + 64;
6345 if(next_level&(~127))
6348 if(next_i < last_non_zero)
6349 score += length[UNI_AC_ENC_INDEX(run, 65)]
6350 + length[UNI_AC_ENC_INDEX(run2, next_level)]
6351 - length[UNI_AC_ENC_INDEX(run + run2 + 1, next_level)];
6353 score += length[UNI_AC_ENC_INDEX(run, 65)]
6354 + last_length[UNI_AC_ENC_INDEX(run2, next_level)]
6355 - last_length[UNI_AC_ENC_INDEX(run + run2 + 1, next_level)];
6357 score += last_length[UNI_AC_ENC_INDEX(run, 65)];
6359 score += length[UNI_AC_ENC_INDEX(prev_run, prev_level)]
6360 - last_length[UNI_AC_ENC_INDEX(prev_run, prev_level)];
6366 assert(FFABS(level)==1);
6368 if(i < last_non_zero){
6369 int next_i= i + run2 + 1;
6370 int next_level= block[ perm_scantable[next_i] ] + 64;
6372 if(next_level&(~127))
6375 if(next_i < last_non_zero)
6376 score += length[UNI_AC_ENC_INDEX(run + run2 + 1, next_level)]
6377 - length[UNI_AC_ENC_INDEX(run2, next_level)]
6378 - length[UNI_AC_ENC_INDEX(run, 65)];
6380 score += last_length[UNI_AC_ENC_INDEX(run + run2 + 1, next_level)]
6381 - last_length[UNI_AC_ENC_INDEX(run2, next_level)]
6382 - length[UNI_AC_ENC_INDEX(run, 65)];
6384 score += -last_length[UNI_AC_ENC_INDEX(run, 65)];
6386 score += last_length[UNI_AC_ENC_INDEX(prev_run, prev_level)]
6387 - length[UNI_AC_ENC_INDEX(prev_run, prev_level)];
6394 unquant_change= new_coeff - old_coeff;
6395 assert((score < 100*lambda && score > -100*lambda) || lambda==0);
6397 score+= s->dsp.try_8x8basis(rem, weight, basis[j], unquant_change);
6398 if(score<best_score){
6401 best_change= change;
6402 best_unquant_change= unquant_change;
6406 prev_level= level + 64;
6407 if(prev_level&(~127))
6416 STOP_TIMER("iterative step")}
6420 int j= perm_scantable[ best_coeff ];
6422 block[j] += best_change;
6424 if(best_coeff > last_non_zero){
6425 last_non_zero= best_coeff;
6433 if(block[j] - best_change){
6434 if(FFABS(block[j]) > FFABS(block[j] - best_change)){
6446 for(; last_non_zero>=start_i; last_non_zero--){
6447 if(block[perm_scantable[last_non_zero]])
6453 if(256*256*256*64 % count == 0){
6454 printf("after_last:%d to_zero:%d from_zero:%d raise:%d lower:%d sign:%d xyp:%d/%d/%d\n", after_last, to_zero, from_zero, raise, lower, messed_sign, s->mb_x, s->mb_y, s->picture_number);
6459 for(i=start_i; i<=last_non_zero; i++){
6460 int j= perm_scantable[i];
6461 const int level= block[j];
6464 run_tab[rle_index++]=run;
6471 s->dsp.add_8x8basis(rem, basis[j], best_unquant_change);
6477 if(last_non_zero>0){
6478 STOP_TIMER("iterative search")
6483 return last_non_zero;
6486 static int dct_quantize_c(MpegEncContext *s,
6487 DCTELEM *block, int n,
6488 int qscale, int *overflow)
6490 int i, j, level, last_non_zero, q, start_i;
6492 const uint8_t *scantable= s->intra_scantable.scantable;
6495 unsigned int threshold1, threshold2;
6497 s->dsp.fdct (block);
6499 if(s->dct_error_sum)
6500 s->denoise_dct(s, block);
6510 /* For AIC we skip quant/dequant of INTRADC */
6513 /* note: block[0] is assumed to be positive */
6514 block[0] = (block[0] + (q >> 1)) / q;
6517 qmat = s->q_intra_matrix[qscale];
6518 bias= s->intra_quant_bias<<(QMAT_SHIFT - QUANT_BIAS_SHIFT);
6522 qmat = s->q_inter_matrix[qscale];
6523 bias= s->inter_quant_bias<<(QMAT_SHIFT - QUANT_BIAS_SHIFT);
6525 threshold1= (1<<QMAT_SHIFT) - bias - 1;
6526 threshold2= (threshold1<<1);
6527 for(i=63;i>=start_i;i--) {
6529 level = block[j] * qmat[j];
6531 if(((unsigned)(level+threshold1))>threshold2){
6538 for(i=start_i; i<=last_non_zero; i++) {
6540 level = block[j] * qmat[j];
6542 // if( bias+level >= (1<<QMAT_SHIFT)
6543 // || bias-level >= (1<<QMAT_SHIFT)){
6544 if(((unsigned)(level+threshold1))>threshold2){
6546 level= (bias + level)>>QMAT_SHIFT;
6549 level= (bias - level)>>QMAT_SHIFT;
6557 *overflow= s->max_qcoeff < max; //overflow might have happened
6559 /* we need this permutation so that we correct the IDCT, we only permute the !=0 elements */
6560 if (s->dsp.idct_permutation_type != FF_NO_IDCT_PERM)
6561 ff_block_permute(block, s->dsp.idct_permutation, scantable, last_non_zero);
6563 return last_non_zero;
6566 #endif //CONFIG_ENCODERS
6568 static void dct_unquantize_mpeg1_intra_c(MpegEncContext *s,
6569 DCTELEM *block, int n, int qscale)
6571 int i, level, nCoeffs;
6572 const uint16_t *quant_matrix;
6574 nCoeffs= s->block_last_index[n];
6577 block[0] = block[0] * s->y_dc_scale;
6579 block[0] = block[0] * s->c_dc_scale;
6580 /* XXX: only mpeg1 */
6581 quant_matrix = s->intra_matrix;
6582 for(i=1;i<=nCoeffs;i++) {
6583 int j= s->intra_scantable.permutated[i];
6588 level = (int)(level * qscale * quant_matrix[j]) >> 3;
6589 level = (level - 1) | 1;
6592 level = (int)(level * qscale * quant_matrix[j]) >> 3;
6593 level = (level - 1) | 1;
6600 static void dct_unquantize_mpeg1_inter_c(MpegEncContext *s,
6601 DCTELEM *block, int n, int qscale)
6603 int i, level, nCoeffs;
6604 const uint16_t *quant_matrix;
6606 nCoeffs= s->block_last_index[n];
6608 quant_matrix = s->inter_matrix;
6609 for(i=0; i<=nCoeffs; i++) {
6610 int j= s->intra_scantable.permutated[i];
6615 level = (((level << 1) + 1) * qscale *
6616 ((int) (quant_matrix[j]))) >> 4;
6617 level = (level - 1) | 1;
6620 level = (((level << 1) + 1) * qscale *
6621 ((int) (quant_matrix[j]))) >> 4;
6622 level = (level - 1) | 1;
6629 static void dct_unquantize_mpeg2_intra_c(MpegEncContext *s,
6630 DCTELEM *block, int n, int qscale)
6632 int i, level, nCoeffs;
6633 const uint16_t *quant_matrix;
6635 if(s->alternate_scan) nCoeffs= 63;
6636 else nCoeffs= s->block_last_index[n];
6639 block[0] = block[0] * s->y_dc_scale;
6641 block[0] = block[0] * s->c_dc_scale;
6642 quant_matrix = s->intra_matrix;
6643 for(i=1;i<=nCoeffs;i++) {
6644 int j= s->intra_scantable.permutated[i];
6649 level = (int)(level * qscale * quant_matrix[j]) >> 3;
6652 level = (int)(level * qscale * quant_matrix[j]) >> 3;
6659 static void dct_unquantize_mpeg2_intra_bitexact(MpegEncContext *s,
6660 DCTELEM *block, int n, int qscale)
6662 int i, level, nCoeffs;
6663 const uint16_t *quant_matrix;
6666 if(s->alternate_scan) nCoeffs= 63;
6667 else nCoeffs= s->block_last_index[n];
6670 block[0] = block[0] * s->y_dc_scale;
6672 block[0] = block[0] * s->c_dc_scale;
6673 quant_matrix = s->intra_matrix;
6674 for(i=1;i<=nCoeffs;i++) {
6675 int j= s->intra_scantable.permutated[i];
6680 level = (int)(level * qscale * quant_matrix[j]) >> 3;
6683 level = (int)(level * qscale * quant_matrix[j]) >> 3;
6692 static void dct_unquantize_mpeg2_inter_c(MpegEncContext *s,
6693 DCTELEM *block, int n, int qscale)
6695 int i, level, nCoeffs;
6696 const uint16_t *quant_matrix;
6699 if(s->alternate_scan) nCoeffs= 63;
6700 else nCoeffs= s->block_last_index[n];
6702 quant_matrix = s->inter_matrix;
6703 for(i=0; i<=nCoeffs; i++) {
6704 int j= s->intra_scantable.permutated[i];
6709 level = (((level << 1) + 1) * qscale *
6710 ((int) (quant_matrix[j]))) >> 4;
6713 level = (((level << 1) + 1) * qscale *
6714 ((int) (quant_matrix[j]))) >> 4;
6723 static void dct_unquantize_h263_intra_c(MpegEncContext *s,
6724 DCTELEM *block, int n, int qscale)
6726 int i, level, qmul, qadd;
6729 assert(s->block_last_index[n]>=0);
6735 block[0] = block[0] * s->y_dc_scale;
6737 block[0] = block[0] * s->c_dc_scale;
6738 qadd = (qscale - 1) | 1;
6745 nCoeffs= s->inter_scantable.raster_end[ s->block_last_index[n] ];
6747 for(i=1; i<=nCoeffs; i++) {
6751 level = level * qmul - qadd;
6753 level = level * qmul + qadd;
6760 static void dct_unquantize_h263_inter_c(MpegEncContext *s,
6761 DCTELEM *block, int n, int qscale)
6763 int i, level, qmul, qadd;
6766 assert(s->block_last_index[n]>=0);
6768 qadd = (qscale - 1) | 1;
6771 nCoeffs= s->inter_scantable.raster_end[ s->block_last_index[n] ];
6773 for(i=0; i<=nCoeffs; i++) {
6777 level = level * qmul - qadd;
6779 level = level * qmul + qadd;
6786 #ifdef CONFIG_ENCODERS
6787 AVCodec h263_encoder = {
6791 sizeof(MpegEncContext),
6795 .pix_fmts= (enum PixelFormat[]){PIX_FMT_YUV420P, -1},
6798 AVCodec h263p_encoder = {
6802 sizeof(MpegEncContext),
6806 .pix_fmts= (enum PixelFormat[]){PIX_FMT_YUV420P, -1},
6809 AVCodec flv_encoder = {
6813 sizeof(MpegEncContext),
6817 .pix_fmts= (enum PixelFormat[]){PIX_FMT_YUV420P, -1},
6820 AVCodec rv10_encoder = {
6824 sizeof(MpegEncContext),
6828 .pix_fmts= (enum PixelFormat[]){PIX_FMT_YUV420P, -1},
6831 AVCodec rv20_encoder = {
6835 sizeof(MpegEncContext),
6839 .pix_fmts= (enum PixelFormat[]){PIX_FMT_YUV420P, -1},
6842 AVCodec mpeg4_encoder = {
6846 sizeof(MpegEncContext),
6850 .pix_fmts= (enum PixelFormat[]){PIX_FMT_YUV420P, -1},
6851 .capabilities= CODEC_CAP_DELAY,
6854 AVCodec msmpeg4v1_encoder = {
6858 sizeof(MpegEncContext),
6862 .pix_fmts= (enum PixelFormat[]){PIX_FMT_YUV420P, -1},
6865 AVCodec msmpeg4v2_encoder = {
6869 sizeof(MpegEncContext),
6873 .pix_fmts= (enum PixelFormat[]){PIX_FMT_YUV420P, -1},
6876 AVCodec msmpeg4v3_encoder = {
6880 sizeof(MpegEncContext),
6884 .pix_fmts= (enum PixelFormat[]){PIX_FMT_YUV420P, -1},
6887 AVCodec wmv1_encoder = {
6891 sizeof(MpegEncContext),
6895 .pix_fmts= (enum PixelFormat[]){PIX_FMT_YUV420P, -1},
6898 AVCodec mjpeg_encoder = {
6902 sizeof(MpegEncContext),
6906 .pix_fmts= (enum PixelFormat[]){PIX_FMT_YUVJ420P, PIX_FMT_YUVJ422P, -1},
6909 #endif //CONFIG_ENCODERS