snowdec.c
Go to the documentation of this file.
1 /*
2  * Copyright (C) 2004 Michael Niedermayer <michaelni@gmx.at>
3  *
4  * This file is part of Libav.
5  *
6  * Libav is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * Libav is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with Libav; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 #include "libavutil/intmath.h"
22 #include "libavutil/log.h"
23 #include "libavutil/opt.h"
24 #include "avcodec.h"
25 #include "dsputil.h"
26 #include "dwt.h"
27 #include "snow.h"
28 
29 #include "rangecoder.h"
30 #include "mathops.h"
31 
32 #include "mpegvideo.h"
33 #include "h263.h"
34 
35 #undef NDEBUG
36 #include <assert.h>
37 
38 static av_always_inline void predict_slice_buffered(SnowContext *s, slice_buffer * sb, IDWTELEM * old_buffer, int plane_index, int add, int mb_y){
39  Plane *p= &s->plane[plane_index];
40  const int mb_w= s->b_width << s->block_max_depth;
41  const int mb_h= s->b_height << s->block_max_depth;
42  int x, y, mb_x;
43  int block_size = MB_SIZE >> s->block_max_depth;
44  int block_w = plane_index ? block_size/2 : block_size;
45  const uint8_t *obmc = plane_index ? obmc_tab[s->block_max_depth+1] : obmc_tab[s->block_max_depth];
46  int obmc_stride= plane_index ? block_size : 2*block_size;
47  int ref_stride= s->current_picture.linesize[plane_index];
48  uint8_t *dst8= s->current_picture.data[plane_index];
49  int w= p->width;
50  int h= p->height;
51 
52  if(s->keyframe || (s->avctx->debug&512)){
53  if(mb_y==mb_h)
54  return;
55 
56  if(add){
57  for(y=block_w*mb_y; y<FFMIN(h,block_w*(mb_y+1)); y++){
58 // DWTELEM * line = slice_buffer_get_line(sb, y);
59  IDWTELEM * line = sb->line[y];
60  for(x=0; x<w; x++){
61 // int v= buf[x + y*w] + (128<<FRAC_BITS) + (1<<(FRAC_BITS-1));
62  int v= line[x] + (128<<FRAC_BITS) + (1<<(FRAC_BITS-1));
63  v >>= FRAC_BITS;
64  if(v&(~255)) v= ~(v>>31);
65  dst8[x + y*ref_stride]= v;
66  }
67  }
68  }else{
69  for(y=block_w*mb_y; y<FFMIN(h,block_w*(mb_y+1)); y++){
70 // DWTELEM * line = slice_buffer_get_line(sb, y);
71  IDWTELEM * line = sb->line[y];
72  for(x=0; x<w; x++){
73  line[x] -= 128 << FRAC_BITS;
74 // buf[x + y*w]-= 128<<FRAC_BITS;
75  }
76  }
77  }
78 
79  return;
80  }
81 
82  for(mb_x=0; mb_x<=mb_w; mb_x++){
83  add_yblock(s, 1, sb, old_buffer, dst8, obmc,
84  block_w*mb_x - block_w/2,
85  block_w*mb_y - block_w/2,
86  block_w, block_w,
87  w, h,
88  w, ref_stride, obmc_stride,
89  mb_x - 1, mb_y - 1,
90  add, 0, plane_index);
91  }
92 }
93 
94 static inline void decode_subband_slice_buffered(SnowContext *s, SubBand *b, slice_buffer * sb, int start_y, int h, int save_state[1]){
95  const int w= b->width;
96  int y;
97  const int qlog= av_clip(s->qlog + b->qlog, 0, QROOT*16);
98  int qmul= qexp[qlog&(QROOT-1)]<<(qlog>>QSHIFT);
99  int qadd= (s->qbias*qmul)>>QBIAS_SHIFT;
100  int new_index = 0;
101 
102  if(b->ibuf == s->spatial_idwt_buffer || s->qlog == LOSSLESS_QLOG){
103  qadd= 0;
104  qmul= 1<<QEXPSHIFT;
105  }
106 
107  /* If we are on the second or later slice, restore our index. */
108  if (start_y != 0)
109  new_index = save_state[0];
110 
111 
112  for(y=start_y; y<h; y++){
113  int x = 0;
114  int v;
116  memset(line, 0, b->width*sizeof(IDWTELEM));
117  v = b->x_coeff[new_index].coeff;
118  x = b->x_coeff[new_index++].x;
119  while(x < w){
120  register int t= ( (v>>1)*qmul + qadd)>>QEXPSHIFT;
121  register int u= -(v&1);
122  line[x] = (t^u) - u;
123 
124  v = b->x_coeff[new_index].coeff;
125  x = b->x_coeff[new_index++].x;
126  }
127  }
128 
129  /* Save our variables for the next slice. */
130  save_state[0] = new_index;
131 
132  return;
133 }
134 
135 static int decode_q_branch(SnowContext *s, int level, int x, int y){
136  const int w= s->b_width << s->block_max_depth;
137  const int rem_depth= s->block_max_depth - level;
138  const int index= (x + y*w) << rem_depth;
139  int trx= (x+1)<<rem_depth;
140  const BlockNode *left = x ? &s->block[index-1] : &null_block;
141  const BlockNode *top = y ? &s->block[index-w] : &null_block;
142  const BlockNode *tl = y && x ? &s->block[index-w-1] : left;
143  const BlockNode *tr = y && trx<w && ((x&1)==0 || level==0) ? &s->block[index-w+(1<<rem_depth)] : tl; //FIXME use lt
144  int s_context= 2*left->level + 2*top->level + tl->level + tr->level;
145  int res;
146 
147  if(s->keyframe){
149  return 0;
150  }
151 
152  if(level==s->block_max_depth || get_rac(&s->c, &s->block_state[4 + s_context])){
153  int type, mx, my;
154  int l = left->color[0];
155  int cb= left->color[1];
156  int cr= left->color[2];
157  int ref = 0;
158  int ref_context= av_log2(2*left->ref) + av_log2(2*top->ref);
159  int mx_context= av_log2(2*FFABS(left->mx - top->mx)) + 0*av_log2(2*FFABS(tr->mx - top->mx));
160  int my_context= av_log2(2*FFABS(left->my - top->my)) + 0*av_log2(2*FFABS(tr->my - top->my));
161 
162  type= get_rac(&s->c, &s->block_state[1 + left->type + top->type]) ? BLOCK_INTRA : 0;
163 
164  if(type){
165  pred_mv(s, &mx, &my, 0, left, top, tr);
166  l += get_symbol(&s->c, &s->block_state[32], 1);
167  cb+= get_symbol(&s->c, &s->block_state[64], 1);
168  cr+= get_symbol(&s->c, &s->block_state[96], 1);
169  }else{
170  if(s->ref_frames > 1)
171  ref= get_symbol(&s->c, &s->block_state[128 + 1024 + 32*ref_context], 0);
172  if (ref >= s->ref_frames) {
173  av_log(s->avctx, AV_LOG_ERROR, "Invalid ref\n");
174  return AVERROR_INVALIDDATA;
175  }
176  pred_mv(s, &mx, &my, ref, left, top, tr);
177  mx+= get_symbol(&s->c, &s->block_state[128 + 32*(mx_context + 16*!!ref)], 1);
178  my+= get_symbol(&s->c, &s->block_state[128 + 32*(my_context + 16*!!ref)], 1);
179  }
180  set_blocks(s, level, x, y, l, cb, cr, mx, my, ref, type);
181  }else{
182  if ((res = decode_q_branch(s, level+1, 2*x+0, 2*y+0)) < 0 ||
183  (res = decode_q_branch(s, level+1, 2*x+1, 2*y+0)) < 0 ||
184  (res = decode_q_branch(s, level+1, 2*x+0, 2*y+1)) < 0 ||
185  (res = decode_q_branch(s, level+1, 2*x+1, 2*y+1)) < 0)
186  return res;
187  }
188  return 0;
189 }
190 
191 static void dequantize_slice_buffered(SnowContext *s, slice_buffer * sb, SubBand *b, IDWTELEM *src, int stride, int start_y, int end_y){
192  const int w= b->width;
193  const int qlog= av_clip(s->qlog + b->qlog, 0, QROOT*16);
194  const int qmul= qexp[qlog&(QROOT-1)]<<(qlog>>QSHIFT);
195  const int qadd= (s->qbias*qmul)>>QBIAS_SHIFT;
196  int x,y;
197 
198  if(s->qlog == LOSSLESS_QLOG) return;
199 
200  for(y=start_y; y<end_y; y++){
201 // DWTELEM * line = slice_buffer_get_line_from_address(sb, src + (y * stride));
203  for(x=0; x<w; x++){
204  int i= line[x];
205  if(i<0){
206  line[x]= -((-i*qmul + qadd)>>(QEXPSHIFT)); //FIXME try different bias
207  }else if(i>0){
208  line[x]= (( i*qmul + qadd)>>(QEXPSHIFT));
209  }
210  }
211  }
212 }
213 
214 static void correlate_slice_buffered(SnowContext *s, slice_buffer * sb, SubBand *b, IDWTELEM *src, int stride, int inverse, int use_median, int start_y, int end_y){
215  const int w= b->width;
216  int x,y;
217 
218  IDWTELEM * line=0; // silence silly "could be used without having been initialized" warning
219  IDWTELEM * prev;
220 
221  if (start_y != 0)
222  line = slice_buffer_get_line(sb, ((start_y - 1) * b->stride_line) + b->buf_y_offset) + b->buf_x_offset;
223 
224  for(y=start_y; y<end_y; y++){
225  prev = line;
226 // line = slice_buffer_get_line_from_address(sb, src + (y * stride));
227  line = slice_buffer_get_line(sb, (y * b->stride_line) + b->buf_y_offset) + b->buf_x_offset;
228  for(x=0; x<w; x++){
229  if(x){
230  if(use_median){
231  if(y && x+1<w) line[x] += mid_pred(line[x - 1], prev[x], prev[x + 1]);
232  else line[x] += line[x - 1];
233  }else{
234  if(y) line[x] += mid_pred(line[x - 1], prev[x], line[x - 1] + prev[x] - prev[x - 1]);
235  else line[x] += line[x - 1];
236  }
237  }else{
238  if(y) line[x] += prev[x];
239  }
240  }
241  }
242 }
243 
244 static void decode_qlogs(SnowContext *s){
245  int plane_index, level, orientation;
246 
247  for(plane_index=0; plane_index<3; plane_index++){
248  for(level=0; level<s->spatial_decomposition_count; level++){
249  for(orientation=level ? 1:0; orientation<4; orientation++){
250  int q;
251  if (plane_index==2) q= s->plane[1].band[level][orientation].qlog;
252  else if(orientation==2) q= s->plane[plane_index].band[level][1].qlog;
253  else q= get_symbol(&s->c, s->header_state, 1);
254  s->plane[plane_index].band[level][orientation].qlog= q;
255  }
256  }
257  }
258 }
259 
260 #define GET_S(dst, check) \
261  tmp= get_symbol(&s->c, s->header_state, 0);\
262  if(!(check)){\
263  av_log(s->avctx, AV_LOG_ERROR, "Error " #dst " is %d\n", tmp);\
264  return -1;\
265  }\
266  dst= tmp;
267 
268 static int decode_header(SnowContext *s){
269  int plane_index, tmp;
270  uint8_t kstate[32];
271 
272  memset(kstate, MID_STATE, sizeof(kstate));
273 
274  s->keyframe= get_rac(&s->c, kstate);
275  if(s->keyframe || s->always_reset){
278  s->qlog=
279  s->qbias=
280  s->mv_scale=
281  s->block_max_depth= 0;
282  }
283  if(s->keyframe){
284  GET_S(s->version, tmp <= 0U)
285  s->always_reset= get_rac(&s->c, s->header_state);
289  s->colorspace_type= get_symbol(&s->c, s->header_state, 0);
290  s->chroma_h_shift= get_symbol(&s->c, s->header_state, 0);
291  s->chroma_v_shift= get_symbol(&s->c, s->header_state, 0);
293 // s->rate_scalability= get_rac(&s->c, s->header_state);
294  GET_S(s->max_ref_frames, tmp < (unsigned)MAX_REF_FRAMES)
295  s->max_ref_frames++;
296 
297  decode_qlogs(s);
298  }
299 
300  if(!s->keyframe){
301  if(get_rac(&s->c, s->header_state)){
302  for(plane_index=0; plane_index<2; plane_index++){
303  int htaps, i, sum=0;
304  Plane *p= &s->plane[plane_index];
305  p->diag_mc= get_rac(&s->c, s->header_state);
306  htaps= get_symbol(&s->c, s->header_state, 0)*2 + 2;
307  if((unsigned)htaps > HTAPS_MAX || htaps==0)
308  return -1;
309  p->htaps= htaps;
310  for(i= htaps/2; i; i--){
311  p->hcoeff[i]= get_symbol(&s->c, s->header_state, 0) * (1-2*(i&1));
312  sum += p->hcoeff[i];
313  }
314  p->hcoeff[0]= 32-sum;
315  }
316  s->plane[2].diag_mc= s->plane[1].diag_mc;
317  s->plane[2].htaps = s->plane[1].htaps;
318  memcpy(s->plane[2].hcoeff, s->plane[1].hcoeff, sizeof(s->plane[1].hcoeff));
319  }
320  if(get_rac(&s->c, s->header_state)){
322  decode_qlogs(s);
323  }
324  }
325 
327  if(s->spatial_decomposition_type > 1U){
328  av_log(s->avctx, AV_LOG_ERROR, "spatial_decomposition_type %d not supported", s->spatial_decomposition_type);
329  return -1;
330  }
331  if(FFMIN(s->avctx-> width>>s->chroma_h_shift,
332  s->avctx->height>>s->chroma_v_shift) >> (s->spatial_decomposition_count-1) <= 0){
333  av_log(s->avctx, AV_LOG_ERROR, "spatial_decomposition_count %d too large for size", s->spatial_decomposition_count);
334  return -1;
335  }
336 
337  if (s->chroma_h_shift != 1 || s->chroma_v_shift != 1) {
338  av_log(s->avctx, AV_LOG_ERROR, "Invalid chroma shift\n");
339  return AVERROR_PATCHWELCOME;
340  }
341 
342  s->qlog += get_symbol(&s->c, s->header_state, 1);
343  s->mv_scale += get_symbol(&s->c, s->header_state, 1);
344  s->qbias += get_symbol(&s->c, s->header_state, 1);
345  s->block_max_depth+= get_symbol(&s->c, s->header_state, 1);
346  if(s->block_max_depth > 1 || s->block_max_depth < 0){
347  av_log(s->avctx, AV_LOG_ERROR, "block_max_depth= %d is too large", s->block_max_depth);
348  s->block_max_depth= 0;
349  return -1;
350  }
351 
352  return 0;
353 }
354 
356 {
357  int ret;
358 
359  avctx->pix_fmt= PIX_FMT_YUV420P;
360 
361  if ((ret = ff_snow_common_init(avctx)) < 0) {
363  return ret;
364  }
365 
366  return 0;
367 }
368 
369 static int decode_blocks(SnowContext *s){
370  int x, y;
371  int w= s->b_width;
372  int h= s->b_height;
373  int res;
374 
375  for(y=0; y<h; y++){
376  for(x=0; x<w; x++){
377  if ((res = decode_q_branch(s, 0, x, y)) < 0)
378  return res;
379  }
380  }
381  return 0;
382 }
383 
384 static int decode_frame(AVCodecContext *avctx, void *data, int *data_size, AVPacket *avpkt){
385  const uint8_t *buf = avpkt->data;
386  int buf_size = avpkt->size;
387  SnowContext *s = avctx->priv_data;
388  RangeCoder * const c= &s->c;
389  int bytes_read;
390  AVFrame *picture = data;
391  int level, orientation, plane_index;
392  int res;
393 
394  ff_init_range_decoder(c, buf, buf_size);
395  ff_build_rac_states(c, 0.05*(1LL<<32), 256-8);
396 
397  s->current_picture.pict_type= AV_PICTURE_TYPE_I; //FIXME I vs. P
398  if(decode_header(s)<0)
399  return -1;
401 
402  // realloc slice buffer for the case that spatial_decomposition_count changed
405 
406  for(plane_index=0; plane_index<3; plane_index++){
407  Plane *p= &s->plane[plane_index];
408  p->fast_mc= p->diag_mc && p->htaps==6 && p->hcoeff[0]==40
409  && p->hcoeff[1]==-10
410  && p->hcoeff[2]==2;
411  }
412 
414 
415  if(ff_snow_frame_start(s) < 0)
416  return -1;
417  //keyframe flag duplication mess FIXME
418  if(avctx->debug&FF_DEBUG_PICT_INFO)
419  av_log(avctx, AV_LOG_ERROR, "keyframe:%d qlog:%d\n", s->keyframe, s->qlog);
420 
421  if ((res = decode_blocks(s)) < 0)
422  return res;
423 
424  for(plane_index=0; plane_index<3; plane_index++){
425  Plane *p= &s->plane[plane_index];
426  int w= p->width;
427  int h= p->height;
428  int x, y;
429  int decode_state[MAX_DECOMPOSITIONS][4][1]; /* Stored state info for unpack_coeffs. 1 variable per instance. */
430 
431  if(s->avctx->debug&2048){
432  memset(s->spatial_dwt_buffer, 0, sizeof(DWTELEM)*w*h);
433  predict_plane(s, s->spatial_idwt_buffer, plane_index, 1);
434 
435  for(y=0; y<h; y++){
436  for(x=0; x<w; x++){
437  int v= s->current_picture.data[plane_index][y*s->current_picture.linesize[plane_index] + x];
438  s->mconly_picture.data[plane_index][y*s->mconly_picture.linesize[plane_index] + x]= v;
439  }
440  }
441  }
442 
443  {
444  for(level=0; level<s->spatial_decomposition_count; level++){
445  for(orientation=level ? 1 : 0; orientation<4; orientation++){
446  SubBand *b= &p->band[level][orientation];
447  unpack_coeffs(s, b, b->parent, orientation);
448  }
449  }
450  }
451 
452  {
453  const int mb_h= s->b_height << s->block_max_depth;
454  const int block_size = MB_SIZE >> s->block_max_depth;
455  const int block_w = plane_index ? block_size/2 : block_size;
456  int mb_y;
458  int yd=0, yq=0;
459  int y;
460  int end_y;
461 
463  for(mb_y=0; mb_y<=mb_h; mb_y++){
464 
465  int slice_starty = block_w*mb_y;
466  int slice_h = block_w*(mb_y+1);
467  if (!(s->keyframe || s->avctx->debug&512)){
468  slice_starty = FFMAX(0, slice_starty - (block_w >> 1));
469  slice_h -= (block_w >> 1);
470  }
471 
472  for(level=0; level<s->spatial_decomposition_count; level++){
473  for(orientation=level ? 1 : 0; orientation<4; orientation++){
474  SubBand *b= &p->band[level][orientation];
475  int start_y;
476  int end_y;
477  int our_mb_start = mb_y;
478  int our_mb_end = (mb_y + 1);
479  const int extra= 3;
480  start_y = (mb_y ? ((block_w * our_mb_start) >> (s->spatial_decomposition_count - level)) + s->spatial_decomposition_count - level + extra: 0);
481  end_y = (((block_w * our_mb_end) >> (s->spatial_decomposition_count - level)) + s->spatial_decomposition_count - level + extra);
482  if (!(s->keyframe || s->avctx->debug&512)){
483  start_y = FFMAX(0, start_y - (block_w >> (1+s->spatial_decomposition_count - level)));
484  end_y = FFMAX(0, end_y - (block_w >> (1+s->spatial_decomposition_count - level)));
485  }
486  start_y = FFMIN(b->height, start_y);
487  end_y = FFMIN(b->height, end_y);
488 
489  if (start_y != end_y){
490  if (orientation == 0){
491  SubBand * correlate_band = &p->band[0][0];
492  int correlate_end_y = FFMIN(b->height, end_y + 1);
493  int correlate_start_y = FFMIN(b->height, (start_y ? start_y + 1 : 0));
494  decode_subband_slice_buffered(s, correlate_band, &s->sb, correlate_start_y, correlate_end_y, decode_state[0][0]);
495  correlate_slice_buffered(s, &s->sb, correlate_band, correlate_band->ibuf, correlate_band->stride, 1, 0, correlate_start_y, correlate_end_y);
496  dequantize_slice_buffered(s, &s->sb, correlate_band, correlate_band->ibuf, correlate_band->stride, start_y, end_y);
497  }
498  else
499  decode_subband_slice_buffered(s, b, &s->sb, start_y, end_y, decode_state[level][orientation]);
500  }
501  }
502  }
503 
504  for(; yd<slice_h; yd+=4){
506  }
507 
508  if(s->qlog == LOSSLESS_QLOG){
509  for(; yq<slice_h && yq<h; yq++){
510  IDWTELEM * line = slice_buffer_get_line(&s->sb, yq);
511  for(x=0; x<w; x++){
512  line[x] <<= FRAC_BITS;
513  }
514  }
515  }
516 
517  predict_slice_buffered(s, &s->sb, s->spatial_idwt_buffer, plane_index, 1, mb_y);
518 
519  y = FFMIN(p->height, slice_starty);
520  end_y = FFMIN(p->height, slice_h);
521  while(y < end_y)
522  ff_slice_buffer_release(&s->sb, y++);
523  }
524 
526  }
527 
528  }
529 
530  emms_c();
531 
532  ff_snow_release_buffer(avctx);
533 
534  if(!(s->avctx->debug&2048))
535  *picture= s->current_picture;
536  else
537  *picture= s->mconly_picture;
538 
539  *data_size = sizeof(AVFrame);
540 
541  bytes_read= c->bytestream - c->bytestream_start;
542  if(bytes_read ==0) av_log(s->avctx, AV_LOG_ERROR, "error at end of frame\n"); //FIXME
543 
544  return bytes_read;
545 }
546 
548 {
549  SnowContext *s = avctx->priv_data;
550 
552 
554 
555  return 0;
556 }
557 
559  .name = "snow",
560  .type = AVMEDIA_TYPE_VIDEO,
561  .id = CODEC_ID_SNOW,
562  .priv_data_size = sizeof(SnowContext),
563  .init = decode_init,
564  .close = decode_end,
565  .decode = decode_frame,
566  .capabilities = CODEC_CAP_DR1 /*| CODEC_CAP_DRAW_HORIZ_BAND*/,
567  .long_name = NULL_IF_CONFIG_SMALL("Snow"),
568 };