FFmpeg
vc2enc.c
Go to the documentation of this file.
1 /*
2  * Copyright (C) 2016 Open Broadcast Systems Ltd.
3  * Author 2016 Rostislav Pehlivanov <atomnuker@gmail.com>
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * FFmpeg is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with FFmpeg; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
22 #include "libavutil/pixdesc.h"
23 #include "libavutil/opt.h"
24 #include "libavutil/version.h"
25 #include "codec_internal.h"
26 #include "dirac.h"
27 #include "encode.h"
28 #include "put_bits.h"
29 #include "version.h"
30 
31 #include "vc2enc_dwt.h"
32 #include "diractab.h"
33 
34 /* The limited size resolution of each slice forces us to do this */
35 #define SSIZE_ROUND(b) (FFALIGN((b), s->size_scaler) + 4 + s->prefix_bytes)
36 
37 /* Decides the cutoff point in # of slices to distribute the leftover bytes */
38 #define SLICE_REDIST_TOTAL 150
39 
40 typedef struct VC2BaseVideoFormat {
44  const char *name;
46 
48  { 0 }, /* Custom format, here just to make indexing equal to base_vf */
49  { AV_PIX_FMT_YUV420P, { 1001, 15000 }, 176, 120, 0, 1, "QSIF525" },
50  { AV_PIX_FMT_YUV420P, { 2, 25 }, 176, 144, 0, 1, "QCIF" },
51  { AV_PIX_FMT_YUV420P, { 1001, 15000 }, 352, 240, 0, 1, "SIF525" },
52  { AV_PIX_FMT_YUV420P, { 2, 25 }, 352, 288, 0, 1, "CIF" },
53  { AV_PIX_FMT_YUV420P, { 1001, 15000 }, 704, 480, 0, 1, "4SIF525" },
54  { AV_PIX_FMT_YUV420P, { 2, 25 }, 704, 576, 0, 1, "4CIF" },
55 
56  { AV_PIX_FMT_YUV422P10, { 1001, 30000 }, 720, 480, 1, 2, "SD480I-60" },
57  { AV_PIX_FMT_YUV422P10, { 1, 25 }, 720, 576, 1, 2, "SD576I-50" },
58 
59  { AV_PIX_FMT_YUV422P10, { 1001, 60000 }, 1280, 720, 0, 3, "HD720P-60" },
60  { AV_PIX_FMT_YUV422P10, { 1, 50 }, 1280, 720, 0, 3, "HD720P-50" },
61  { AV_PIX_FMT_YUV422P10, { 1001, 30000 }, 1920, 1080, 1, 3, "HD1080I-60" },
62  { AV_PIX_FMT_YUV422P10, { 1, 25 }, 1920, 1080, 1, 3, "HD1080I-50" },
63  { AV_PIX_FMT_YUV422P10, { 1001, 60000 }, 1920, 1080, 0, 3, "HD1080P-60" },
64  { AV_PIX_FMT_YUV422P10, { 1, 50 }, 1920, 1080, 0, 3, "HD1080P-50" },
65 
66  { AV_PIX_FMT_YUV444P12, { 1, 24 }, 2048, 1080, 0, 4, "DC2K" },
67  { AV_PIX_FMT_YUV444P12, { 1, 24 }, 4096, 2160, 0, 5, "DC4K" },
68 
69  { AV_PIX_FMT_YUV422P10, { 1001, 60000 }, 3840, 2160, 0, 6, "UHDTV 4K-60" },
70  { AV_PIX_FMT_YUV422P10, { 1, 50 }, 3840, 2160, 0, 6, "UHDTV 4K-50" },
71 
72  { AV_PIX_FMT_YUV422P10, { 1001, 60000 }, 7680, 4320, 0, 7, "UHDTV 8K-60" },
73  { AV_PIX_FMT_YUV422P10, { 1, 50 }, 7680, 4320, 0, 7, "UHDTV 8K-50" },
74 
75  { AV_PIX_FMT_YUV422P10, { 1001, 24000 }, 1920, 1080, 0, 3, "HD1080P-24" },
76  { AV_PIX_FMT_YUV422P10, { 1001, 30000 }, 720, 486, 1, 2, "SD Pro486" },
77 };
79 
80 enum VC2_QM {
84 
86 };
87 
88 typedef struct SubBand {
90  ptrdiff_t stride;
91  int width;
92  int height;
93 } SubBand;
94 
95 typedef struct Plane {
98  int width;
99  int height;
102  ptrdiff_t coef_stride;
103 } Plane;
104 
105 typedef struct SliceArgs {
108  void *ctx;
109  int x;
110  int y;
114  int bytes;
115 } SliceArgs;
116 
117 typedef struct TransformArgs {
118  void *ctx;
120  const void *idata;
121  ptrdiff_t istride;
122  int field;
124 } TransformArgs;
125 
126 typedef struct VC2EncContext {
132 
135 
136  /* For conversion from unsigned pixel values to signed */
138  int bpp;
139  int bpp_idx;
140 
141  /* Picture number */
142  uint32_t picture_number;
143 
144  /* Base video format */
145  int base_vf;
146  int level;
147  int profile;
148 
149  /* Quantization matrix */
150  uint8_t quant[MAX_DWT_LEVELS][4];
152 
153  /* Division LUT */
154  uint32_t qmagic_lut[116][2];
155 
156  int num_x; /* #slices horizontally */
157  int num_y; /* #slices vertically */
162 
163  /* Rate control stuff */
167  int q_ceil;
168  int q_avg;
169 
170  /* Options */
171  double tolerance;
179 
180  /* Parse code state */
183 } VC2EncContext;
184 
186 {
187  int i;
188  int pbits = 0, bits = 0, topbit = 1, maxval = 1;
189 
190  if (!val++) {
191  put_bits(pb, 1, 1);
192  return;
193  }
194 
195  while (val > maxval) {
196  topbit <<= 1;
197  maxval <<= 1;
198  maxval |= 1;
199  }
200 
201  bits = ff_log2(topbit);
202 
203  for (i = 0; i < bits; i++) {
204  topbit >>= 1;
205  pbits <<= 2;
206  if (val & topbit)
207  pbits |= 0x1;
208  }
209 
210  put_bits(pb, bits*2 + 1, (pbits << 1) | 1);
211 }
212 
214 {
215  int topbit = 1, maxval = 1;
216 
217  if (!val++)
218  return 1;
219 
220  while (val > maxval) {
221  topbit <<= 1;
222  maxval <<= 1;
223  maxval |= 1;
224  }
225 
226  return ff_log2(topbit)*2 + 1;
227 }
228 
229 /* VC-2 10.4 - parse_info() */
231 {
232  uint32_t cur_pos, dist;
233 
234  align_put_bits(&s->pb);
235 
236  cur_pos = put_bits_count(&s->pb) >> 3;
237 
238  /* Magic string */
239  ff_put_string(&s->pb, "BBCD", 0);
240 
241  /* Parse code */
242  put_bits(&s->pb, 8, pcode);
243 
244  /* Next parse offset */
245  dist = cur_pos - s->next_parse_offset;
246  AV_WB32(s->pb.buf + s->next_parse_offset + 5, dist);
247  s->next_parse_offset = cur_pos;
248  put_bits32(&s->pb, pcode == DIRAC_PCODE_END_SEQ ? 13 : 0);
249 
250  /* Last parse offset */
251  put_bits32(&s->pb, s->last_parse_code == DIRAC_PCODE_END_SEQ ? 13 : dist);
252 
253  s->last_parse_code = pcode;
254 }
255 
256 /* VC-2 11.1 - parse_parameters()
257  * The level dictates what the decoder should expect in terms of resolution
258  * and allows it to quickly reject whatever it can't support. Remember,
259  * this codec kinda targets cheapo FPGAs without much memory. Unfortunately
260  * it also limits us greatly in our choice of formats, hence the flag to disable
261  * strict_compliance */
263 {
264  put_vc2_ue_uint(&s->pb, s->ver.major); /* VC-2 demands this to be 2 */
265  put_vc2_ue_uint(&s->pb, s->ver.minor); /* ^^ and this to be 0 */
266  put_vc2_ue_uint(&s->pb, s->profile); /* 3 to signal HQ profile */
267  put_vc2_ue_uint(&s->pb, s->level); /* 3 - 1080/720, 6 - 4K */
268 }
269 
270 /* VC-2 11.3 - frame_size() */
272 {
273  put_bits(&s->pb, 1, !s->strict_compliance);
274  if (!s->strict_compliance) {
275  AVCodecContext *avctx = s->avctx;
276  put_vc2_ue_uint(&s->pb, avctx->width);
277  put_vc2_ue_uint(&s->pb, avctx->height);
278  }
279 }
280 
281 /* VC-2 11.3.3 - color_diff_sampling_format() */
283 {
284  put_bits(&s->pb, 1, !s->strict_compliance);
285  if (!s->strict_compliance) {
286  int idx;
287  if (s->chroma_x_shift == 1 && s->chroma_y_shift == 0)
288  idx = 1; /* 422 */
289  else if (s->chroma_x_shift == 1 && s->chroma_y_shift == 1)
290  idx = 2; /* 420 */
291  else
292  idx = 0; /* 444 */
293  put_vc2_ue_uint(&s->pb, idx);
294  }
295 }
296 
297 /* VC-2 11.3.4 - scan_format() */
299 {
300  put_bits(&s->pb, 1, !s->strict_compliance);
301  if (!s->strict_compliance)
302  put_vc2_ue_uint(&s->pb, s->interlaced);
303 }
304 
305 /* VC-2 11.3.5 - frame_rate() */
307 {
308  put_bits(&s->pb, 1, !s->strict_compliance);
309  if (!s->strict_compliance) {
310  AVCodecContext *avctx = s->avctx;
311  put_vc2_ue_uint(&s->pb, 0);
312  put_vc2_ue_uint(&s->pb, avctx->time_base.den);
313  put_vc2_ue_uint(&s->pb, avctx->time_base.num);
314  }
315 }
316 
317 /* VC-2 11.3.6 - aspect_ratio() */
319 {
320  put_bits(&s->pb, 1, !s->strict_compliance);
321  if (!s->strict_compliance) {
322  AVCodecContext *avctx = s->avctx;
323  put_vc2_ue_uint(&s->pb, 0);
326  }
327 }
328 
329 /* VC-2 11.3.7 - clean_area() */
331 {
332  put_bits(&s->pb, 1, 0);
333 }
334 
335 /* VC-2 11.3.8 - signal_range() */
337 {
338  put_bits(&s->pb, 1, !s->strict_compliance);
339  if (!s->strict_compliance)
340  put_vc2_ue_uint(&s->pb, s->bpp_idx);
341 }
342 
343 /* VC-2 11.3.9 - color_spec() */
345 {
346  AVCodecContext *avctx = s->avctx;
347  put_bits(&s->pb, 1, !s->strict_compliance);
348  if (!s->strict_compliance) {
349  int val;
350  put_vc2_ue_uint(&s->pb, 0);
351 
352  /* primaries */
353  put_bits(&s->pb, 1, 1);
354  if (avctx->color_primaries == AVCOL_PRI_BT470BG)
355  val = 2;
356  else if (avctx->color_primaries == AVCOL_PRI_SMPTE170M)
357  val = 1;
358  else if (avctx->color_primaries == AVCOL_PRI_SMPTE240M)
359  val = 1;
360  else
361  val = 0;
362  put_vc2_ue_uint(&s->pb, val);
363 
364  /* color matrix */
365  put_bits(&s->pb, 1, 1);
366  if (avctx->colorspace == AVCOL_SPC_RGB)
367  val = 3;
368  else if (avctx->colorspace == AVCOL_SPC_YCOCG)
369  val = 2;
370  else if (avctx->colorspace == AVCOL_SPC_BT470BG)
371  val = 1;
372  else
373  val = 0;
374  put_vc2_ue_uint(&s->pb, val);
375 
376  /* transfer function */
377  put_bits(&s->pb, 1, 1);
378  if (avctx->color_trc == AVCOL_TRC_LINEAR)
379  val = 2;
380  else if (avctx->color_trc == AVCOL_TRC_BT1361_ECG)
381  val = 1;
382  else
383  val = 0;
384  put_vc2_ue_uint(&s->pb, val);
385  }
386 }
387 
388 /* VC-2 11.3 - source_parameters() */
390 {
399 }
400 
401 /* VC-2 11 - sequence_header() */
403 {
404  align_put_bits(&s->pb);
406  put_vc2_ue_uint(&s->pb, s->base_vf);
408  put_vc2_ue_uint(&s->pb, s->interlaced); /* Frames or fields coding */
409 }
410 
411 /* VC-2 12.1 - picture_header() */
413 {
414  align_put_bits(&s->pb);
415  put_bits32(&s->pb, s->picture_number++);
416 }
417 
418 /* VC-2 12.3.4.1 - slice_parameters() */
420 {
421  put_vc2_ue_uint(&s->pb, s->num_x);
422  put_vc2_ue_uint(&s->pb, s->num_y);
423  put_vc2_ue_uint(&s->pb, s->prefix_bytes);
424  put_vc2_ue_uint(&s->pb, s->size_scaler);
425 }
426 
427 /* 1st idx = LL, second - vertical, third - horizontal, fourth - total */
428 static const uint8_t vc2_qm_col_tab[][4] = {
429  {20, 9, 15, 4},
430  { 0, 6, 6, 4},
431  { 0, 3, 3, 5},
432  { 0, 3, 5, 1},
433  { 0, 11, 10, 11}
434 };
435 
436 static const uint8_t vc2_qm_flat_tab[][4] = {
437  { 0, 0, 0, 0},
438  { 0, 0, 0, 0},
439  { 0, 0, 0, 0},
440  { 0, 0, 0, 0},
441  { 0, 0, 0, 0}
442 };
443 
445 {
446  int level, orientation;
447 
448  if (s->wavelet_depth <= 4 && s->quant_matrix == VC2_QM_DEF) {
449  s->custom_quant_matrix = 0;
450  for (level = 0; level < s->wavelet_depth; level++) {
451  s->quant[level][0] = ff_dirac_default_qmat[s->wavelet_idx][level][0];
452  s->quant[level][1] = ff_dirac_default_qmat[s->wavelet_idx][level][1];
453  s->quant[level][2] = ff_dirac_default_qmat[s->wavelet_idx][level][2];
454  s->quant[level][3] = ff_dirac_default_qmat[s->wavelet_idx][level][3];
455  }
456  return;
457  }
458 
459  s->custom_quant_matrix = 1;
460 
461  if (s->quant_matrix == VC2_QM_DEF) {
462  for (level = 0; level < s->wavelet_depth; level++) {
463  for (orientation = 0; orientation < 4; orientation++) {
464  if (level <= 3)
465  s->quant[level][orientation] = ff_dirac_default_qmat[s->wavelet_idx][level][orientation];
466  else
467  s->quant[level][orientation] = vc2_qm_col_tab[level][orientation];
468  }
469  }
470  } else if (s->quant_matrix == VC2_QM_COL) {
471  for (level = 0; level < s->wavelet_depth; level++) {
472  for (orientation = 0; orientation < 4; orientation++) {
473  s->quant[level][orientation] = vc2_qm_col_tab[level][orientation];
474  }
475  }
476  } else {
477  for (level = 0; level < s->wavelet_depth; level++) {
478  for (orientation = 0; orientation < 4; orientation++) {
479  s->quant[level][orientation] = vc2_qm_flat_tab[level][orientation];
480  }
481  }
482  }
483 }
484 
485 /* VC-2 12.3.4.2 - quant_matrix() */
487 {
488  int level;
489  put_bits(&s->pb, 1, s->custom_quant_matrix);
490  if (s->custom_quant_matrix) {
491  put_vc2_ue_uint(&s->pb, s->quant[0][0]);
492  for (level = 0; level < s->wavelet_depth; level++) {
493  put_vc2_ue_uint(&s->pb, s->quant[level][1]);
494  put_vc2_ue_uint(&s->pb, s->quant[level][2]);
495  put_vc2_ue_uint(&s->pb, s->quant[level][3]);
496  }
497  }
498 }
499 
500 /* VC-2 12.3 - transform_parameters() */
502 {
503  put_vc2_ue_uint(&s->pb, s->wavelet_idx);
504  put_vc2_ue_uint(&s->pb, s->wavelet_depth);
505 
508 }
509 
510 /* VC-2 12.2 - wavelet_transform() */
512 {
514  align_put_bits(&s->pb);
515 }
516 
517 /* VC-2 12 - picture_parse() */
519 {
520  align_put_bits(&s->pb);
522  align_put_bits(&s->pb);
524 }
525 
526 #define QUANT(c, mul, add, shift) (((mul) * (c) + (add)) >> (shift))
527 
528 /* VC-2 13.5.5.2 - slice_band() */
529 static void encode_subband(VC2EncContext *s, PutBitContext *pb, int sx, int sy,
530  SubBand *b, int quant)
531 {
532  int x, y;
533 
534  const int left = b->width * (sx+0) / s->num_x;
535  const int right = b->width * (sx+1) / s->num_x;
536  const int top = b->height * (sy+0) / s->num_y;
537  const int bottom = b->height * (sy+1) / s->num_y;
538 
539  dwtcoef *coeff = b->buf + top * b->stride;
540  const uint64_t q_m = ((uint64_t)(s->qmagic_lut[quant][0])) << 2;
541  const uint64_t q_a = s->qmagic_lut[quant][1];
542  const int q_s = av_log2(ff_dirac_qscale_tab[quant]) + 32;
543 
544  for (y = top; y < bottom; y++) {
545  for (x = left; x < right; x++) {
546  uint32_t c_abs = QUANT(FFABS(coeff[x]), q_m, q_a, q_s);
547  put_vc2_ue_uint(pb, c_abs);
548  if (c_abs)
549  put_bits(pb, 1, coeff[x] < 0);
550  }
551  coeff += b->stride;
552  }
553 }
554 
555 static int count_hq_slice(SliceArgs *slice, int quant_idx)
556 {
557  int x, y;
558  uint8_t quants[MAX_DWT_LEVELS][4];
559  int bits = 0, p, level, orientation;
560  VC2EncContext *s = slice->ctx;
561 
562  if (slice->cache[quant_idx])
563  return slice->cache[quant_idx];
564 
565  bits += 8*s->prefix_bytes;
566  bits += 8; /* quant_idx */
567 
568  for (level = 0; level < s->wavelet_depth; level++)
569  for (orientation = !!level; orientation < 4; orientation++)
570  quants[level][orientation] = FFMAX(quant_idx - s->quant[level][orientation], 0);
571 
572  for (p = 0; p < 3; p++) {
573  int bytes_start, bytes_len, pad_s, pad_c;
574  bytes_start = bits >> 3;
575  bits += 8;
576  for (level = 0; level < s->wavelet_depth; level++) {
577  for (orientation = !!level; orientation < 4; orientation++) {
578  SubBand *b = &s->plane[p].band[level][orientation];
579 
580  const int q_idx = quants[level][orientation];
581  const uint64_t q_m = ((uint64_t)s->qmagic_lut[q_idx][0]) << 2;
582  const uint64_t q_a = s->qmagic_lut[q_idx][1];
583  const int q_s = av_log2(ff_dirac_qscale_tab[q_idx]) + 32;
584 
585  const int left = b->width * slice->x / s->num_x;
586  const int right = b->width *(slice->x+1) / s->num_x;
587  const int top = b->height * slice->y / s->num_y;
588  const int bottom = b->height *(slice->y+1) / s->num_y;
589 
590  dwtcoef *buf = b->buf + top * b->stride;
591 
592  for (y = top; y < bottom; y++) {
593  for (x = left; x < right; x++) {
594  uint32_t c_abs = QUANT(FFABS(buf[x]), q_m, q_a, q_s);
595  bits += count_vc2_ue_uint(c_abs);
596  bits += !!c_abs;
597  }
598  buf += b->stride;
599  }
600  }
601  }
602  bits += FFALIGN(bits, 8) - bits;
603  bytes_len = (bits >> 3) - bytes_start - 1;
604  pad_s = FFALIGN(bytes_len, s->size_scaler)/s->size_scaler;
605  pad_c = (pad_s*s->size_scaler) - bytes_len;
606  bits += pad_c*8;
607  }
608 
609  slice->cache[quant_idx] = bits;
610 
611  return bits;
612 }
613 
614 /* Approaches the best possible quantizer asymptotically, its kinda exaustive
615  * but we have a LUT to get the coefficient size in bits. Guaranteed to never
616  * overshoot, which is apparently very important when streaming */
617 static int rate_control(AVCodecContext *avctx, void *arg)
618 {
619  SliceArgs *slice_dat = arg;
620  VC2EncContext *s = slice_dat->ctx;
621  const int top = slice_dat->bits_ceil;
622  const int bottom = slice_dat->bits_floor;
623  int quant_buf[2] = {-1, -1};
624  int quant = slice_dat->quant_idx, step = 1;
625  int bits_last, bits = count_hq_slice(slice_dat, quant);
626  while ((bits > top) || (bits < bottom)) {
627  const int signed_step = bits > top ? +step : -step;
628  quant = av_clip(quant + signed_step, 0, s->q_ceil-1);
629  bits = count_hq_slice(slice_dat, quant);
630  if (quant_buf[1] == quant) {
631  quant = FFMAX(quant_buf[0], quant);
632  bits = quant == quant_buf[0] ? bits_last : bits;
633  break;
634  }
635  step = av_clip(step/2, 1, (s->q_ceil-1)/2);
636  quant_buf[1] = quant_buf[0];
637  quant_buf[0] = quant;
638  bits_last = bits;
639  }
640  slice_dat->quant_idx = av_clip(quant, 0, s->q_ceil-1);
641  slice_dat->bytes = SSIZE_ROUND(bits >> 3);
642  return 0;
643 }
644 
646 {
647  int i, j, slice_x, slice_y, bytes_left = 0;
648  int bytes_top[SLICE_REDIST_TOTAL] = {0};
649  int64_t total_bytes_needed = 0;
650  int slice_redist_range = FFMIN(SLICE_REDIST_TOTAL, s->num_x*s->num_y);
651  SliceArgs *enc_args = s->slice_args;
652  SliceArgs *top_loc[SLICE_REDIST_TOTAL] = {NULL};
653 
655 
656  for (slice_y = 0; slice_y < s->num_y; slice_y++) {
657  for (slice_x = 0; slice_x < s->num_x; slice_x++) {
658  SliceArgs *args = &enc_args[s->num_x*slice_y + slice_x];
659  args->ctx = s;
660  args->x = slice_x;
661  args->y = slice_y;
662  args->bits_ceil = s->slice_max_bytes << 3;
663  args->bits_floor = s->slice_min_bytes << 3;
664  memset(args->cache, 0, s->q_ceil*sizeof(*args->cache));
665  }
666  }
667 
668  /* First pass - determine baseline slice sizes w.r.t. max_slice_size */
669  s->avctx->execute(s->avctx, rate_control, enc_args, NULL, s->num_x*s->num_y,
670  sizeof(SliceArgs));
671 
672  for (i = 0; i < s->num_x*s->num_y; i++) {
673  SliceArgs *args = &enc_args[i];
674  bytes_left += args->bytes;
675  for (j = 0; j < slice_redist_range; j++) {
676  if (args->bytes > bytes_top[j]) {
677  bytes_top[j] = args->bytes;
678  top_loc[j] = args;
679  break;
680  }
681  }
682  }
683 
684  bytes_left = s->frame_max_bytes - bytes_left;
685 
686  /* Second pass - distribute leftover bytes */
687  while (bytes_left > 0) {
688  int distributed = 0;
689  for (i = 0; i < slice_redist_range; i++) {
690  SliceArgs *args;
691  int bits, bytes, diff, prev_bytes, new_idx;
692  if (bytes_left <= 0)
693  break;
694  if (!top_loc[i] || !top_loc[i]->quant_idx)
695  break;
696  args = top_loc[i];
697  prev_bytes = args->bytes;
698  new_idx = FFMAX(args->quant_idx - 1, 0);
699  bits = count_hq_slice(args, new_idx);
700  bytes = SSIZE_ROUND(bits >> 3);
701  diff = bytes - prev_bytes;
702  if ((bytes_left - diff) > 0) {
703  args->quant_idx = new_idx;
704  args->bytes = bytes;
705  bytes_left -= diff;
706  distributed++;
707  }
708  }
709  if (!distributed)
710  break;
711  }
712 
713  for (i = 0; i < s->num_x*s->num_y; i++) {
714  SliceArgs *args = &enc_args[i];
715  total_bytes_needed += args->bytes;
716  s->q_avg = (s->q_avg + args->quant_idx)/2;
717  }
718 
719  return total_bytes_needed;
720 }
721 
722 /* VC-2 13.5.3 - hq_slice */
723 static int encode_hq_slice(AVCodecContext *avctx, void *arg)
724 {
725  SliceArgs *slice_dat = arg;
726  VC2EncContext *s = slice_dat->ctx;
727  PutBitContext *pb = &slice_dat->pb;
728  const int slice_x = slice_dat->x;
729  const int slice_y = slice_dat->y;
730  const int quant_idx = slice_dat->quant_idx;
731  const int slice_bytes_max = slice_dat->bytes;
732  uint8_t quants[MAX_DWT_LEVELS][4];
733  int p, level, orientation;
734 
735  /* The reference decoder ignores it, and its typical length is 0 */
736  memset(put_bits_ptr(pb), 0, s->prefix_bytes);
737  skip_put_bytes(pb, s->prefix_bytes);
738 
739  put_bits(pb, 8, quant_idx);
740 
741  /* Slice quantization (slice_quantizers() in the specs) */
742  for (level = 0; level < s->wavelet_depth; level++)
743  for (orientation = !!level; orientation < 4; orientation++)
744  quants[level][orientation] = FFMAX(quant_idx - s->quant[level][orientation], 0);
745 
746  /* Luma + 2 Chroma planes */
747  for (p = 0; p < 3; p++) {
748  int bytes_start, bytes_len, pad_s, pad_c;
749  bytes_start = put_bits_count(pb) >> 3;
750  put_bits(pb, 8, 0);
751  for (level = 0; level < s->wavelet_depth; level++) {
752  for (orientation = !!level; orientation < 4; orientation++) {
753  encode_subband(s, pb, slice_x, slice_y,
754  &s->plane[p].band[level][orientation],
755  quants[level][orientation]);
756  }
757  }
758  align_put_bits(pb);
759  bytes_len = (put_bits_count(pb) >> 3) - bytes_start - 1;
760  if (p == 2) {
761  int len_diff = slice_bytes_max - (put_bits_count(pb) >> 3);
762  pad_s = FFALIGN((bytes_len + len_diff), s->size_scaler)/s->size_scaler;
763  pad_c = (pad_s*s->size_scaler) - bytes_len;
764  } else {
765  pad_s = FFALIGN(bytes_len, s->size_scaler)/s->size_scaler;
766  pad_c = (pad_s*s->size_scaler) - bytes_len;
767  }
768  pb->buf[bytes_start] = pad_s;
769  flush_put_bits(pb);
770  /* vc2-reference uses that padding that decodes to '0' coeffs */
771  memset(put_bits_ptr(pb), 0xFF, pad_c);
772  skip_put_bytes(pb, pad_c);
773  }
774 
775  return 0;
776 }
777 
778 /* VC-2 13.5.1 - low_delay_transform_data() */
780 {
781  uint8_t *buf;
782  int slice_x, slice_y, skip = 0;
783  SliceArgs *enc_args = s->slice_args;
784 
785  flush_put_bits(&s->pb);
786  buf = put_bits_ptr(&s->pb);
787 
788  for (slice_y = 0; slice_y < s->num_y; slice_y++) {
789  for (slice_x = 0; slice_x < s->num_x; slice_x++) {
790  SliceArgs *args = &enc_args[s->num_x*slice_y + slice_x];
791  init_put_bits(&args->pb, buf + skip, args->bytes+s->prefix_bytes);
792  skip += args->bytes;
793  }
794  }
795 
796  s->avctx->execute(s->avctx, encode_hq_slice, enc_args, NULL, s->num_x*s->num_y,
797  sizeof(SliceArgs));
798 
799  skip_put_bytes(&s->pb, skip);
800 
801  return 0;
802 }
803 
804 /*
805  * Transform basics for a 3 level transform
806  * |---------------------------------------------------------------------|
807  * | LL-0 | HL-0 | | |
808  * |--------|-------| HL-1 | |
809  * | LH-0 | HH-0 | | |
810  * |----------------|-----------------| HL-2 |
811  * | | | |
812  * | LH-1 | HH-1 | |
813  * | | | |
814  * |----------------------------------|----------------------------------|
815  * | | |
816  * | | |
817  * | | |
818  * | LH-2 | HH-2 |
819  * | | |
820  * | | |
821  * | | |
822  * |---------------------------------------------------------------------|
823  *
824  * DWT transforms are generally applied by splitting the image in two vertically
825  * and applying a low pass transform on the left part and a corresponding high
826  * pass transform on the right hand side. This is known as the horizontal filter
827  * stage.
828  * After that, the same operation is performed except the image is divided
829  * horizontally, with the high pass on the lower and the low pass on the higher
830  * side.
831  * Therefore, you're left with 4 subdivisions - known as low-low, low-high,
832  * high-low and high-high. They're referred to as orientations in the decoder
833  * and encoder.
834  *
835  * The LL (low-low) area contains the original image downsampled by the amount
836  * of levels. The rest of the areas can be thought as the details needed
837  * to restore the image perfectly to its original size.
838  */
839 static int dwt_plane(AVCodecContext *avctx, void *arg)
840 {
841  TransformArgs *transform_dat = arg;
842  VC2EncContext *s = transform_dat->ctx;
843  const void *frame_data = transform_dat->idata;
844  const ptrdiff_t linesize = transform_dat->istride;
845  const int field = transform_dat->field;
846  const Plane *p = transform_dat->plane;
847  VC2TransformContext *t = &transform_dat->t;
848  dwtcoef *buf = p->coef_buf;
849  const int idx = s->wavelet_idx;
850  const int skip = 1 + s->interlaced;
851 
852  int x, y, level, offset;
853  ptrdiff_t pix_stride = linesize >> (s->bpp - 1);
854 
855  if (field == 1) {
856  offset = 0;
857  pix_stride <<= 1;
858  } else if (field == 2) {
859  offset = pix_stride;
860  pix_stride <<= 1;
861  } else {
862  offset = 0;
863  }
864 
865  if (s->bpp == 1) {
866  const uint8_t *pix = (const uint8_t *)frame_data + offset;
867  for (y = 0; y < p->height*skip; y+=skip) {
868  for (x = 0; x < p->width; x++) {
869  buf[x] = pix[x] - s->diff_offset;
870  }
871  memset(&buf[x], 0, (p->coef_stride - p->width)*sizeof(dwtcoef));
872  buf += p->coef_stride;
873  pix += pix_stride;
874  }
875  } else {
876  const uint16_t *pix = (const uint16_t *)frame_data + offset;
877  for (y = 0; y < p->height*skip; y+=skip) {
878  for (x = 0; x < p->width; x++) {
879  buf[x] = pix[x] - s->diff_offset;
880  }
881  memset(&buf[x], 0, (p->coef_stride - p->width)*sizeof(dwtcoef));
882  buf += p->coef_stride;
883  pix += pix_stride;
884  }
885  }
886 
887  memset(buf, 0, p->coef_stride * (p->dwt_height - p->height) * sizeof(dwtcoef));
888 
889  for (level = s->wavelet_depth-1; level >= 0; level--) {
890  const SubBand *b = &p->band[level][0];
891  t->vc2_subband_dwt[idx](t, p->coef_buf, p->coef_stride,
892  b->width, b->height);
893  }
894 
895  return 0;
896 }
897 
898 static int encode_frame(VC2EncContext *s, AVPacket *avpkt, const AVFrame *frame,
899  const char *aux_data, const int header_size, int field)
900 {
901  int i, ret;
902  int64_t max_frame_bytes;
903 
904  /* Threaded DWT transform */
905  for (i = 0; i < 3; i++) {
906  s->transform_args[i].ctx = s;
907  s->transform_args[i].field = field;
908  s->transform_args[i].plane = &s->plane[i];
909  s->transform_args[i].idata = frame->data[i];
910  s->transform_args[i].istride = frame->linesize[i];
911  }
912  s->avctx->execute(s->avctx, dwt_plane, s->transform_args, NULL, 3,
913  sizeof(TransformArgs));
914 
915  /* Calculate per-slice quantizers and sizes */
916  max_frame_bytes = header_size + calc_slice_sizes(s);
917 
918  if (field < 2) {
919  ret = ff_get_encode_buffer(s->avctx, avpkt,
920  max_frame_bytes << s->interlaced, 0);
921  if (ret) {
922  av_log(s->avctx, AV_LOG_ERROR, "Error getting output packet.\n");
923  return ret;
924  }
925  init_put_bits(&s->pb, avpkt->data, avpkt->size);
926  }
927 
928  /* Sequence header */
931 
932  /* Encoder version */
933  if (aux_data) {
935  ff_put_string(&s->pb, aux_data, 1);
936  }
937 
938  /* Picture header */
941 
942  /* Encode slices */
943  encode_slices(s);
944 
945  /* End sequence */
947 
948  return 0;
949 }
950 
952  const AVFrame *frame, int *got_packet)
953 {
954  int ret = 0;
955  int slice_ceil, sig_size = 256;
956  VC2EncContext *s = avctx->priv_data;
957  const int bitexact = avctx->flags & AV_CODEC_FLAG_BITEXACT;
958  const char *aux_data = bitexact ? "Lavc" : LIBAVCODEC_IDENT;
959  const int aux_data_size = bitexact ? sizeof("Lavc") : sizeof(LIBAVCODEC_IDENT);
960  const int header_size = 100 + aux_data_size;
961  int64_t r_bitrate = avctx->bit_rate >> (s->interlaced);
962 
963  s->avctx = avctx;
964  s->size_scaler = 2;
965  s->prefix_bytes = 0;
966  s->last_parse_code = 0;
967  s->next_parse_offset = 0;
968 
969  /* Rate control */
970  s->frame_max_bytes = (av_rescale(r_bitrate, s->avctx->time_base.num,
971  s->avctx->time_base.den) >> 3) - header_size;
972  s->slice_max_bytes = slice_ceil = av_rescale(s->frame_max_bytes, 1, s->num_x*s->num_y);
973 
974  /* Find an appropriate size scaler */
975  while (sig_size > 255) {
976  int r_size = SSIZE_ROUND(s->slice_max_bytes);
977  if (r_size > slice_ceil) {
978  s->slice_max_bytes -= r_size - slice_ceil;
979  r_size = SSIZE_ROUND(s->slice_max_bytes);
980  }
981  sig_size = r_size/s->size_scaler; /* Signalled slize size */
982  s->size_scaler <<= 1;
983  }
984 
985  s->slice_min_bytes = s->slice_max_bytes - s->slice_max_bytes*(s->tolerance/100.0f);
986  if (s->slice_min_bytes < 0)
987  return AVERROR(EINVAL);
988 
989  ret = encode_frame(s, avpkt, frame, aux_data, header_size, s->interlaced);
990  if (ret)
991  return ret;
992  if (s->interlaced) {
993  ret = encode_frame(s, avpkt, frame, aux_data, header_size, 2);
994  if (ret)
995  return ret;
996  }
997 
998  flush_put_bits(&s->pb);
999  av_shrink_packet(avpkt, put_bytes_output(&s->pb));
1000 
1001  *got_packet = 1;
1002 
1003  return 0;
1004 }
1005 
1007 {
1008  int i;
1009  VC2EncContext *s = avctx->priv_data;
1010 
1011  av_log(avctx, AV_LOG_INFO, "Qavg: %i\n", s->q_avg);
1012 
1013  for (i = 0; i < 3; i++) {
1014  ff_vc2enc_free_transforms(&s->transform_args[i].t);
1015  av_freep(&s->plane[i].coef_buf);
1016  }
1017 
1018  av_freep(&s->slice_args);
1019 
1020  return 0;
1021 }
1022 
1024 {
1025  Plane *p;
1026  SubBand *b;
1027  int i, level, o, shift, ret;
1028  const AVPixFmtDescriptor *fmt = av_pix_fmt_desc_get(avctx->pix_fmt);
1029  const int depth = fmt->comp[0].depth;
1030  VC2EncContext *s = avctx->priv_data;
1031 
1032  s->picture_number = 0;
1033 
1034  /* Total allowed quantization range */
1035  s->q_ceil = DIRAC_MAX_QUANT_INDEX;
1036 
1037  s->ver.major = 2;
1038  s->ver.minor = 0;
1039  s->profile = 3;
1040  s->level = 3;
1041 
1042  s->base_vf = -1;
1043  s->strict_compliance = 1;
1044 
1045  s->q_avg = 0;
1046  s->slice_max_bytes = 0;
1047  s->slice_min_bytes = 0;
1048 
1049  /* Mark unknown as progressive */
1050  s->interlaced = !((avctx->field_order == AV_FIELD_UNKNOWN) ||
1051  (avctx->field_order == AV_FIELD_PROGRESSIVE));
1052 
1053  for (i = 0; i < base_video_fmts_len; i++) {
1054  const VC2BaseVideoFormat *fmt = &base_video_fmts[i];
1055  if (avctx->pix_fmt != fmt->pix_fmt)
1056  continue;
1057  if (avctx->time_base.num != fmt->time_base.num)
1058  continue;
1059  if (avctx->time_base.den != fmt->time_base.den)
1060  continue;
1061  if (avctx->width != fmt->width)
1062  continue;
1063  if (avctx->height != fmt->height)
1064  continue;
1065  if (s->interlaced != fmt->interlaced)
1066  continue;
1067  s->base_vf = i;
1068  s->level = base_video_fmts[i].level;
1069  break;
1070  }
1071 
1072  if (s->interlaced)
1073  av_log(avctx, AV_LOG_WARNING, "Interlacing enabled!\n");
1074 
1075  if ((s->slice_width & (s->slice_width - 1)) ||
1076  (s->slice_height & (s->slice_height - 1))) {
1077  av_log(avctx, AV_LOG_ERROR, "Slice size is not a power of two!\n");
1078  return AVERROR_UNKNOWN;
1079  }
1080 
1081  if ((s->slice_width > avctx->width) ||
1082  (s->slice_height > avctx->height)) {
1083  av_log(avctx, AV_LOG_ERROR, "Slice size is bigger than the image!\n");
1084  return AVERROR_UNKNOWN;
1085  }
1086 
1087  if (s->base_vf <= 0) {
1089  s->strict_compliance = s->base_vf = 0;
1090  av_log(avctx, AV_LOG_WARNING, "Format does not strictly comply with VC2 specs\n");
1091  } else {
1092  av_log(avctx, AV_LOG_WARNING, "Given format does not strictly comply with "
1093  "the specifications, decrease strictness to use it.\n");
1094  return AVERROR_UNKNOWN;
1095  }
1096  } else {
1097  av_log(avctx, AV_LOG_INFO, "Selected base video format = %i (%s)\n",
1098  s->base_vf, base_video_fmts[s->base_vf].name);
1099  }
1100 
1101  /* Chroma subsampling */
1102  ret = av_pix_fmt_get_chroma_sub_sample(avctx->pix_fmt, &s->chroma_x_shift, &s->chroma_y_shift);
1103  if (ret)
1104  return ret;
1105 
1106  /* Bit depth and color range index */
1107  if (depth == 8 && avctx->color_range == AVCOL_RANGE_JPEG) {
1108  s->bpp = 1;
1109  s->bpp_idx = 1;
1110  s->diff_offset = 128;
1111  } else if (depth == 8 && (avctx->color_range == AVCOL_RANGE_MPEG ||
1112  avctx->color_range == AVCOL_RANGE_UNSPECIFIED)) {
1113  s->bpp = 1;
1114  s->bpp_idx = 2;
1115  s->diff_offset = 128;
1116  } else if (depth == 10) {
1117  s->bpp = 2;
1118  s->bpp_idx = 3;
1119  s->diff_offset = 512;
1120  } else {
1121  s->bpp = 2;
1122  s->bpp_idx = 4;
1123  s->diff_offset = 2048;
1124  }
1125 
1126  /* Planes initialization */
1127  for (i = 0; i < 3; i++) {
1128  int w, h;
1129  p = &s->plane[i];
1130  p->width = avctx->width >> (i ? s->chroma_x_shift : 0);
1131  p->height = avctx->height >> (i ? s->chroma_y_shift : 0);
1132  if (s->interlaced)
1133  p->height >>= 1;
1134  p->dwt_width = w = FFALIGN(p->width, (1 << s->wavelet_depth));
1135  p->dwt_height = h = FFALIGN(p->height, (1 << s->wavelet_depth));
1136  p->coef_stride = FFALIGN(p->dwt_width, 32);
1137  p->coef_buf = av_mallocz(p->coef_stride*p->dwt_height*sizeof(dwtcoef));
1138  if (!p->coef_buf)
1139  return AVERROR(ENOMEM);
1140  for (level = s->wavelet_depth-1; level >= 0; level--) {
1141  w = w >> 1;
1142  h = h >> 1;
1143  for (o = 0; o < 4; o++) {
1144  b = &p->band[level][o];
1145  b->width = w;
1146  b->height = h;
1147  b->stride = p->coef_stride;
1148  shift = (o > 1)*b->height*b->stride + (o & 1)*b->width;
1149  b->buf = p->coef_buf + shift;
1150  }
1151  }
1152 
1153  /* DWT init */
1154  if (ff_vc2enc_init_transforms(&s->transform_args[i].t,
1155  s->plane[i].coef_stride,
1156  s->plane[i].dwt_height,
1157  s->slice_width, s->slice_height))
1158  return AVERROR(ENOMEM);
1159  }
1160 
1161  /* Slices */
1162  s->num_x = s->plane[0].dwt_width/s->slice_width;
1163  s->num_y = s->plane[0].dwt_height/s->slice_height;
1164 
1165  s->slice_args = av_calloc(s->num_x*s->num_y, sizeof(SliceArgs));
1166  if (!s->slice_args)
1167  return AVERROR(ENOMEM);
1168 
1169  for (i = 0; i < 116; i++) {
1170  const uint64_t qf = ff_dirac_qscale_tab[i];
1171  const uint32_t m = av_log2(qf);
1172  const uint32_t t = (1ULL << (m + 32)) / qf;
1173  const uint32_t r = (t*qf + qf) & UINT32_MAX;
1174  if (!(qf & (qf - 1))) {
1175  s->qmagic_lut[i][0] = 0xFFFFFFFF;
1176  s->qmagic_lut[i][1] = 0xFFFFFFFF;
1177  } else if (r <= 1 << m) {
1178  s->qmagic_lut[i][0] = t + 1;
1179  s->qmagic_lut[i][1] = 0;
1180  } else {
1181  s->qmagic_lut[i][0] = t;
1182  s->qmagic_lut[i][1] = t;
1183  }
1184  }
1185 
1186  return 0;
1187 }
1188 
1189 #define VC2ENC_FLAGS (AV_OPT_FLAG_ENCODING_PARAM | AV_OPT_FLAG_VIDEO_PARAM)
1190 static const AVOption vc2enc_options[] = {
1191  {"tolerance", "Max undershoot in percent", offsetof(VC2EncContext, tolerance), AV_OPT_TYPE_DOUBLE, {.dbl = 5.0f}, 0.0f, 45.0f, VC2ENC_FLAGS, "tolerance"},
1192  {"slice_width", "Slice width", offsetof(VC2EncContext, slice_width), AV_OPT_TYPE_INT, {.i64 = 32}, 32, 1024, VC2ENC_FLAGS, "slice_width"},
1193  {"slice_height", "Slice height", offsetof(VC2EncContext, slice_height), AV_OPT_TYPE_INT, {.i64 = 16}, 8, 1024, VC2ENC_FLAGS, "slice_height"},
1194  {"wavelet_depth", "Transform depth", offsetof(VC2EncContext, wavelet_depth), AV_OPT_TYPE_INT, {.i64 = 4}, 1, 5, VC2ENC_FLAGS, "wavelet_depth"},
1195  {"wavelet_type", "Transform type", offsetof(VC2EncContext, wavelet_idx), AV_OPT_TYPE_INT, {.i64 = VC2_TRANSFORM_9_7}, 0, VC2_TRANSFORMS_NB, VC2ENC_FLAGS, "wavelet_idx"},
1196  {"9_7", "Deslauriers-Dubuc (9,7)", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_TRANSFORM_9_7}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "wavelet_idx"},
1197  {"5_3", "LeGall (5,3)", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_TRANSFORM_5_3}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "wavelet_idx"},
1198  {"haar", "Haar (with shift)", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_TRANSFORM_HAAR_S}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "wavelet_idx"},
1199  {"haar_noshift", "Haar (without shift)", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_TRANSFORM_HAAR}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "wavelet_idx"},
1200  {"qm", "Custom quantization matrix", offsetof(VC2EncContext, quant_matrix), AV_OPT_TYPE_INT, {.i64 = VC2_QM_DEF}, 0, VC2_QM_NB, VC2ENC_FLAGS, "quant_matrix"},
1201  {"default", "Default from the specifications", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_QM_DEF}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "quant_matrix"},
1202  {"color", "Prevents low bitrate discoloration", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_QM_COL}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "quant_matrix"},
1203  {"flat", "Optimize for PSNR", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_QM_FLAT}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "quant_matrix"},
1204  {NULL}
1205 };
1206 
1207 static const AVClass vc2enc_class = {
1208  .class_name = "SMPTE VC-2 encoder",
1209  .category = AV_CLASS_CATEGORY_ENCODER,
1210  .option = vc2enc_options,
1211  .item_name = av_default_item_name,
1212  .version = LIBAVUTIL_VERSION_INT
1213 };
1214 
1216  { "b", "600000000" },
1217  { NULL },
1218 };
1219 
1220 static const enum AVPixelFormat allowed_pix_fmts[] = {
1225 };
1226 
1228  .p.name = "vc2",
1229  .p.long_name = NULL_IF_CONFIG_SMALL("SMPTE VC-2"),
1230  .p.type = AVMEDIA_TYPE_VIDEO,
1231  .p.id = AV_CODEC_ID_DIRAC,
1232  .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_SLICE_THREADS,
1233  .caps_internal = FF_CODEC_CAP_INIT_CLEANUP,
1234  .priv_data_size = sizeof(VC2EncContext),
1235  .init = vc2_encode_init,
1236  .close = vc2_encode_end,
1238  .p.priv_class = &vc2enc_class,
1239  .defaults = vc2enc_defaults,
1240  .p.pix_fmts = allowed_pix_fmts
1241 };
init_quant_matrix
static void init_quant_matrix(VC2EncContext *s)
Definition: vc2enc.c:444
vc2_qm_col_tab
static const uint8_t vc2_qm_col_tab[][4]
Definition: vc2enc.c:428
SliceArgs::bits_floor
int bits_floor
Definition: vc2enc.c:113
AV_LOG_WARNING
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:186
VC2EncContext::slice_args
SliceArgs * slice_args
Definition: vc2enc.c:133
AVPixelFormat
AVPixelFormat
Pixel format.
Definition: pixfmt.h:64
level
uint8_t level
Definition: svq3.c:206
av_clip
#define av_clip
Definition: common.h:95
FF_CODEC_CAP_INIT_CLEANUP
#define FF_CODEC_CAP_INIT_CLEANUP
The codec allows calling the close function for deallocation even if the init function returned a fai...
Definition: codec_internal.h:41
r
const char * r
Definition: vf_curves.c:116
AVERROR
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later. That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another. Frame references ownership and permissions
opt.h
vc2_qm_flat_tab
static const uint8_t vc2_qm_flat_tab[][4]
Definition: vc2enc.c:436
DiracVersionInfo
Definition: dirac.h:76
put_bits32
static void av_unused put_bits32(PutBitContext *s, uint32_t value)
Write exactly 32 bits into a bitstream.
Definition: put_bits.h:291
LIBAVCODEC_IDENT
#define LIBAVCODEC_IDENT
Definition: version.h:43
VC2_TRANSFORM_9_7
@ VC2_TRANSFORM_9_7
Definition: vc2enc_dwt.h:31
AVCodecContext::colorspace
enum AVColorSpace colorspace
YUV colorspace type.
Definition: avcodec.h:975
Plane::coef_buf
dwtcoef * coef_buf
Definition: vc2enc.c:97
align_put_bits
static void align_put_bits(PutBitContext *s)
Pad the bitstream with zeros up to the next byte boundary.
Definition: put_bits.h:413
put_bytes_output
static int put_bytes_output(const PutBitContext *s)
Definition: put_bits.h:89
av_pix_fmt_desc_get
const AVPixFmtDescriptor * av_pix_fmt_desc_get(enum AVPixelFormat pix_fmt)
Definition: pixdesc.c:2675
AVCOL_TRC_LINEAR
@ AVCOL_TRC_LINEAR
"Linear transfer characteristics"
Definition: pixfmt.h:507
MAX_DWT_LEVELS
#define MAX_DWT_LEVELS
The spec limits the number of wavelet decompositions to 4 for both level 1 (VC-2) and 128 (long-gop d...
Definition: dirac.h:45
dwtcoef
int32_t dwtcoef
Definition: vc2enc_dwt.h:28
AV_CODEC_ID_DIRAC
@ AV_CODEC_ID_DIRAC
Definition: codec_id.h:166
VC2EncContext::chroma_x_shift
int chroma_x_shift
Definition: vc2enc.c:160
VC2BaseVideoFormat::interlaced
int interlaced
Definition: vc2enc.c:43
init_put_bits
static void init_put_bits(PutBitContext *s, uint8_t *buffer, int buffer_size)
Initialize the PutBitContext s.
Definition: put_bits.h:62
TransformArgs::plane
Plane * plane
Definition: vc2enc.c:119
AVFrame
This structure describes decoded (raw) audio or video data.
Definition: frame.h:325
put_bits
static void put_bits(Jpeg2000EncoderContext *s, int val, int n)
put n times val bit
Definition: j2kenc.c:221
pixdesc.h
step
trying all byte sequences megabyte in length and selecting the best looking sequence will yield cases to try But a word about which is also called distortion Distortion can be quantified by almost any quality measurement one chooses the sum of squared differences is used but more complex methods that consider psychovisual effects can be used as well It makes no difference in this discussion First step
Definition: rate_distortion.txt:58
AVCodecContext::color_trc
enum AVColorTransferCharacteristic color_trc
Color Transfer Characteristic.
Definition: avcodec.h:968
VC2EncContext::q_avg
int q_avg
Definition: vc2enc.c:168
w
uint8_t w
Definition: llviddspenc.c:38
AVCOL_RANGE_JPEG
@ AVCOL_RANGE_JPEG
Full range content.
Definition: pixfmt.h:601
AVCOL_SPC_YCOCG
@ AVCOL_SPC_YCOCG
Definition: pixfmt.h:537
AVPacket::data
uint8_t * data
Definition: packet.h:374
VC2EncContext::diff_offset
int diff_offset
Definition: vc2enc.c:137
SubBand::width
int width
Definition: cfhd.h:112
AVComponentDescriptor::depth
int depth
Number of bits in the component.
Definition: pixdesc.h:57
SliceArgs::y
int y
Definition: vc2enc.c:110
AVCodecContext::field_order
enum AVFieldOrder field_order
Field order.
Definition: avcodec.h:1004
AVOption
AVOption.
Definition: opt.h:251
encode.h
b
#define b
Definition: input.c:34
SliceArgs::bytes
int bytes
Definition: vc2enc.c:114
encode_scan_format
static void encode_scan_format(VC2EncContext *s)
Definition: vc2enc.c:298
encode_slices
static int encode_slices(VC2EncContext *s)
Definition: vc2enc.c:779
rate_control
static int rate_control(AVCodecContext *avctx, void *arg)
Definition: vc2enc.c:617
AV_PIX_FMT_YUV420P10
#define AV_PIX_FMT_YUV420P10
Definition: pixfmt.h:408
VC2EncContext::pb
PutBitContext pb
Definition: vc2enc.c:128
ff_vc2enc_free_transforms
av_cold void ff_vc2enc_free_transforms(VC2TransformContext *s)
Definition: vc2enc_dwt.c:277
FFCodec
Definition: codec_internal.h:118
version.h
AVCOL_SPC_RGB
@ AVCOL_SPC_RGB
order of coefficients is actually GBR, also IEC 61966-2-1 (sRGB), YZX and ST 428-1
Definition: pixfmt.h:528
base_video_fmts
static const VC2BaseVideoFormat base_video_fmts[]
Definition: vc2enc.c:47
SliceArgs::pb
PutBitContext pb
Definition: vc2enc.c:106
encode_picture_header
static void encode_picture_header(VC2EncContext *s)
Definition: vc2enc.c:412
encode_slice_params
static void encode_slice_params(VC2EncContext *s)
Definition: vc2enc.c:419
SliceArgs::ctx
void * ctx
Definition: vc2enc.c:108
FFMAX
#define FFMAX(a, b)
Definition: macros.h:47
vc2_encode_frame
static av_cold int vc2_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, const AVFrame *frame, int *got_packet)
Definition: vc2enc.c:951
VC2TransformContext::vc2_subband_dwt
void(* vc2_subband_dwt[VC2_TRANSFORMS_NB])(struct VC2TransformContext *t, dwtcoef *data, ptrdiff_t stride, int width, int height)
Definition: vc2enc_dwt.h:45
VC2EncContext::qmagic_lut
uint32_t qmagic_lut[116][2]
Definition: vc2enc.c:154
AVERROR_UNKNOWN
#define AVERROR_UNKNOWN
Unknown error, typically from an external library.
Definition: error.h:73
ff_dirac_qscale_tab
const int32_t ff_dirac_qscale_tab[116]
Definition: diractab.c:34
VC2_QM_NB
@ VC2_QM_NB
Definition: vc2enc.c:85
VC2EncContext::slice_min_bytes
int slice_min_bytes
Definition: vc2enc.c:166
encode_clean_area
static void encode_clean_area(VC2EncContext *s)
Definition: vc2enc.c:330
encode_frame_size
static void encode_frame_size(VC2EncContext *s)
Definition: vc2enc.c:271
encode_quant_matrix
static void encode_quant_matrix(VC2EncContext *s)
Definition: vc2enc.c:486
init
static int init
Definition: av_tx.c:47
diractab.h
ff_dirac_default_qmat
const uint8_t ff_dirac_default_qmat[7][4][4]
Definition: diractab.c:24
VC2EncContext::prefix_bytes
int prefix_bytes
Definition: vc2enc.c:158
AVCOL_SPC_BT470BG
@ AVCOL_SPC_BT470BG
also ITU-R BT601-6 625 / ITU-R BT1358 625 / ITU-R BT1700 625 PAL & SECAM / IEC 61966-2-4 xvYCC601
Definition: pixfmt.h:533
VC2EncContext::custom_quant_matrix
int custom_quant_matrix
Definition: vc2enc.c:151
encode_color_spec
static void encode_color_spec(VC2EncContext *s)
Definition: vc2enc.c:344
count_hq_slice
static int count_hq_slice(SliceArgs *slice, int quant_idx)
Definition: vc2enc.c:555
VC2BaseVideoFormat
Definition: vc2enc.c:40
FFCodecDefault
Definition: codec_internal.h:88
FFCodec::p
AVCodec p
The public AVCodec.
Definition: codec_internal.h:122
VC2_TRANSFORMS_NB
@ VC2_TRANSFORMS_NB
Definition: vc2enc_dwt.h:39
av_shrink_packet
void av_shrink_packet(AVPacket *pkt, int size)
Reduce packet size, correctly zeroing padding.
Definition: avpacket.c:112
TransformArgs
Definition: vc2enc.c:117
VC2_QM_FLAT
@ VC2_QM_FLAT
Definition: vc2enc.c:83
AVCodecContext::flags
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:478
val
static double val(void *priv, double ch)
Definition: aeval.c:77
av_pix_fmt_get_chroma_sub_sample
int av_pix_fmt_get_chroma_sub_sample(enum AVPixelFormat pix_fmt, int *h_shift, int *v_shift)
Utility function to access log2_chroma_w log2_chroma_h from the pixel format AVPixFmtDescriptor.
Definition: pixdesc.c:2703
FF_CODEC_ENCODE_CB
#define FF_CODEC_ENCODE_CB(func)
Definition: codec_internal.h:269
AVRational::num
int num
Numerator.
Definition: rational.h:59
quant
static int quant(float coef, const float Q, const float rounding)
Quantize one coefficient.
Definition: aacenc_utils.h:59
encode_subband
static void encode_subband(VC2EncContext *s, PutBitContext *pb, int sx, int sy, SubBand *b, int quant)
Definition: vc2enc.c:529
SliceArgs::x
int x
Definition: vc2enc.c:109
dirac.h
AV_PIX_FMT_YUV444P10
#define AV_PIX_FMT_YUV444P10
Definition: pixfmt.h:411
AVCodecContext::color_primaries
enum AVColorPrimaries color_primaries
Chromaticity coordinates of the source primaries.
Definition: avcodec.h:961
SSIZE_ROUND
#define SSIZE_ROUND(b)
Definition: vc2enc.c:35
AV_LOG_ERROR
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:180
VC2EncContext::next_parse_offset
uint32_t next_parse_offset
Definition: vc2enc.c:181
FF_ARRAY_ELEMS
#define FF_ARRAY_ELEMS(a)
Definition: sinewin_tablegen.c:29
VC2EncContext::tolerance
double tolerance
Definition: vc2enc.c:171
av_cold
#define av_cold
Definition: attributes.h:90
ff_vc2enc_init_transforms
av_cold int ff_vc2enc_init_transforms(VC2TransformContext *s, int p_stride, int p_height, int slice_w, int slice_h)
Definition: vc2enc_dwt.c:258
DiracParseCodes
DiracParseCodes
Parse code values:
Definition: dirac.h:57
VC2EncContext::slice_width
int slice_width
Definition: vc2enc.c:176
vc2enc_class
static const AVClass vc2enc_class
Definition: vc2enc.c:1207
s
#define s(width, name)
Definition: cbs_vp9.c:256
AVCOL_TRC_BT1361_ECG
@ AVCOL_TRC_BT1361_ECG
ITU-R BT1361 Extended Colour Gamut.
Definition: pixfmt.h:511
TransformArgs::t
VC2TransformContext t
Definition: vc2enc.c:123
VC2EncContext::quant_matrix
enum VC2_QM quant_matrix
Definition: vc2enc.c:178
AV_OPT_TYPE_DOUBLE
@ AV_OPT_TYPE_DOUBLE
Definition: opt.h:227
bits
uint8_t bits
Definition: vp3data.h:141
encode_frame
static int encode_frame(VC2EncContext *s, AVPacket *avpkt, const AVFrame *frame, const char *aux_data, const int header_size, int field)
Definition: vc2enc.c:898
encode_transform_params
static void encode_transform_params(VC2EncContext *s)
Definition: vc2enc.c:501
VC2EncContext::chroma_y_shift
int chroma_y_shift
Definition: vc2enc.c:161
AV_FIELD_UNKNOWN
@ AV_FIELD_UNKNOWN
Definition: codec_par.h:38
dwt_plane
static int dwt_plane(AVCodecContext *avctx, void *arg)
Definition: vc2enc.c:839
vc2enc_dwt.h
VC2EncContext::wavelet_idx
int wavelet_idx
Definition: vc2enc.c:172
ff_put_string
void ff_put_string(PutBitContext *pb, const char *string, int terminate_string)
Put the string string in the bitstream.
Definition: bitstream.c:39
field
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this field
Definition: writing_filters.txt:78
VC2EncContext::slice_max_bytes
int slice_max_bytes
Definition: vc2enc.c:165
AV_PIX_FMT_YUV420P
@ AV_PIX_FMT_YUV420P
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Definition: pixfmt.h:66
AVCOL_PRI_SMPTE240M
@ AVCOL_PRI_SMPTE240M
identical to above, also called "SMPTE C" even though it uses D65
Definition: pixfmt.h:482
PutBitContext
Definition: put_bits.h:50
VC2EncContext::avctx
AVCodecContext * avctx
Definition: vc2enc.c:130
AVCOL_PRI_BT470BG
@ AVCOL_PRI_BT470BG
also ITU-R BT601-6 625 / ITU-R BT1358 625 / ITU-R BT1700 625 PAL & SECAM
Definition: pixfmt.h:480
arg
const char * arg
Definition: jacosubdec.c:67
AVCOL_PRI_SMPTE170M
@ AVCOL_PRI_SMPTE170M
also ITU-R BT601-6 525 / ITU-R BT1358 525 / ITU-R BT1700 NTSC
Definition: pixfmt.h:481
FFABS
#define FFABS(a)
Absolute value, Note, INT_MIN / INT64_MIN result in undefined behavior as they are not representable ...
Definition: common.h:64
VC2_QM
VC2_QM
Definition: vc2enc.c:80
LIBAVUTIL_VERSION_INT
#define LIBAVUTIL_VERSION_INT
Definition: version.h:85
AVClass
Describe the class of an AVClass context structure.
Definition: log.h:66
encode_frame_rate
static void encode_frame_rate(VC2EncContext *s)
Definition: vc2enc.c:306
PutBitContext::buf
uint8_t * buf
Definition: put_bits.h:53
NULL
#define NULL
Definition: coverity.c:32
VC2BaseVideoFormat::pix_fmt
enum AVPixelFormat pix_fmt
Definition: vc2enc.c:41
AVCodecContext::color_range
enum AVColorRange color_range
MPEG vs JPEG YUV range.
Definition: avcodec.h:982
VC2EncContext::quant
uint8_t quant[MAX_DWT_LEVELS][4]
Definition: vc2enc.c:150
Plane::dwt_width
int dwt_width
Definition: vc2enc.c:100
AVRational
Rational number (pair of numerator and denominator).
Definition: rational.h:58
VC2EncContext::size_scaler
int size_scaler
Definition: vc2enc.c:159
encode_wavelet_transform
static void encode_wavelet_transform(VC2EncContext *s)
Definition: vc2enc.c:511
AVCodecContext::bit_rate
int64_t bit_rate
the average bitrate
Definition: avcodec.h:448
vc2enc_options
static const AVOption vc2enc_options[]
Definition: vc2enc.c:1190
base_video_fmts_len
static const int base_video_fmts_len
Definition: vc2enc.c:78
av_default_item_name
const char * av_default_item_name(void *ptr)
Return the context name.
Definition: log.c:237
encode_aspect_ratio
static void encode_aspect_ratio(VC2EncContext *s)
Definition: vc2enc.c:318
SliceArgs::quant_idx
int quant_idx
Definition: vc2enc.c:111
VC2BaseVideoFormat::height
int height
Definition: vc2enc.c:43
ff_log2
#define ff_log2
Definition: intmath.h:50
VC2_QM_COL
@ VC2_QM_COL
Definition: vc2enc.c:82
VC2EncContext::av_class
AVClass * av_class
Definition: vc2enc.c:127
AV_PIX_FMT_YUV422P10
#define AV_PIX_FMT_YUV422P10
Definition: pixfmt.h:409
DIRAC_MAX_QUANT_INDEX
#define DIRAC_MAX_QUANT_INDEX
Definition: diractab.h:41
DIRAC_PCODE_AUX
@ DIRAC_PCODE_AUX
Definition: dirac.h:60
SliceArgs::bits_ceil
int bits_ceil
Definition: vc2enc.c:112
AVCOL_RANGE_UNSPECIFIED
@ AVCOL_RANGE_UNSPECIFIED
Definition: pixfmt.h:567
allowed_pix_fmts
static enum AVPixelFormat allowed_pix_fmts[]
Definition: vc2enc.c:1220
encode_hq_slice
static int encode_hq_slice(AVCodecContext *avctx, void *arg)
Definition: vc2enc.c:723
VC2BaseVideoFormat::level
int level
Definition: vc2enc.c:43
VC2EncContext::interlaced
int interlaced
Definition: vc2enc.c:177
AV_WB32
#define AV_WB32(p, v)
Definition: intreadwrite.h:419
VC2EncContext::picture_number
uint32_t picture_number
Definition: vc2enc.c:142
VC2EncContext::last_parse_code
enum DiracParseCodes last_parse_code
Definition: vc2enc.c:182
AVCodecContext::time_base
AVRational time_base
This is the fundamental unit of time (in seconds) in terms of which frame timestamps are represented.
Definition: avcodec.h:521
AV_CODEC_CAP_DR1
#define AV_CODEC_CAP_DR1
Codec uses get_buffer() or get_encode_buffer() for allocating buffers and supports custom allocators.
Definition: codec.h:52
AVPacket::size
int size
Definition: packet.h:375
NULL_IF_CONFIG_SMALL
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
Definition: internal.h:117
VC2TransformContext
Definition: vc2enc_dwt.h:42
VC2BaseVideoFormat::time_base
AVRational time_base
Definition: vc2enc.c:42
SubBand::stride
ptrdiff_t stride
Definition: cfhd.h:110
codec_internal.h
Plane::height
int height
Definition: cfhd.h:120
FF_COMPLIANCE_STRICT
#define FF_COMPLIANCE_STRICT
Strictly conform to all the things in the spec no matter what consequences.
Definition: avcodec.h:1311
AV_PIX_FMT_YUV422P12
#define AV_PIX_FMT_YUV422P12
Definition: pixfmt.h:413
vc2_encode_init
static av_cold int vc2_encode_init(AVCodecContext *avctx)
Definition: vc2enc.c:1023
VC2EncContext::level
int level
Definition: vc2enc.c:146
AV_PIX_FMT_YUV444P12
#define AV_PIX_FMT_YUV444P12
Definition: pixfmt.h:415
SubBand
Definition: cfhd.h:109
VC2EncContext::bpp
int bpp
Definition: vc2enc.c:138
encode_source_params
static void encode_source_params(VC2EncContext *s)
Definition: vc2enc.c:389
Plane::width
int width
Definition: cfhd.h:119
VC2EncContext::ver
DiracVersionInfo ver
Definition: vc2enc.c:131
AV_CODEC_CAP_SLICE_THREADS
#define AV_CODEC_CAP_SLICE_THREADS
Codec supports slice-based (or partition-based) multithreading.
Definition: codec.h:117
offset
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf offset
Definition: writing_filters.txt:86
encode_picture_start
static void encode_picture_start(VC2EncContext *s)
Definition: vc2enc.c:518
DIRAC_PCODE_SEQ_HEADER
@ DIRAC_PCODE_SEQ_HEADER
Definition: dirac.h:58
Plane::coef_stride
ptrdiff_t coef_stride
Definition: vc2enc.c:102
VC2EncContext::num_y
int num_y
Definition: vc2enc.c:157
AV_LOG_INFO
#define AV_LOG_INFO
Standard information.
Definition: log.h:191
encode_parse_params
static void encode_parse_params(VC2EncContext *s)
Definition: vc2enc.c:262
SliceArgs
Definition: vc2enc.c:105
TransformArgs::ctx
void * ctx
Definition: vc2enc.c:118
i
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:269
put_bits_count
static int put_bits_count(PutBitContext *s)
Definition: put_bits.h:80
VC2EncContext::wavelet_depth
int wavelet_depth
Definition: vc2enc.c:173
av_always_inline
#define av_always_inline
Definition: attributes.h:49
FFMIN
#define FFMIN(a, b)
Definition: macros.h:49
av_mallocz
void * av_mallocz(size_t size)
Allocate a memory block with alignment suitable for all memory accesses (including vectors if availab...
Definition: mem.c:264
AVCodec::name
const char * name
Name of the codec implementation.
Definition: codec.h:211
av_rescale
int64_t av_rescale(int64_t a, int64_t b, int64_t c)
Rescale a 64-bit integer with rounding to nearest.
Definition: mathematics.c:129
AVCodecContext::height
int height
Definition: avcodec.h:571
AVCodecContext::pix_fmt
enum AVPixelFormat pix_fmt
Pixel format, see AV_PIX_FMT_xxx.
Definition: avcodec.h:608
AVCOL_RANGE_MPEG
@ AVCOL_RANGE_MPEG
Narrow or limited range content.
Definition: pixfmt.h:584
av_calloc
void * av_calloc(size_t nmemb, size_t size)
Definition: mem.c:272
encode_parse_info
static void encode_parse_info(VC2EncContext *s, enum DiracParseCodes pcode)
Definition: vc2enc.c:230
ff_vc2_encoder
const FFCodec ff_vc2_encoder
Definition: vc2enc.c:1227
vc2_encode_end
static av_cold int vc2_encode_end(AVCodecContext *avctx)
Definition: vc2enc.c:1006
version.h
SubBand::buf
dwtcoef * buf
Definition: vc2enc.c:89
vc2enc_defaults
static const FFCodecDefault vc2enc_defaults[]
Definition: vc2enc.c:1215
ret
ret
Definition: filter_design.txt:187
VC2BaseVideoFormat::name
const char * name
Definition: vc2enc.c:44
VC2EncContext::transform_args
TransformArgs transform_args[3]
Definition: vc2enc.c:134
AVClass::class_name
const char * class_name
The name of the class; usually it is the same name as the context structure type to which the AVClass...
Definition: log.h:71
frame
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several the filter must be ready for frames arriving randomly on any input any filter with several inputs will most likely require some kind of queuing mechanism It is perfectly acceptable to have a limited queue and to drop frames when the inputs are too unbalanced request_frame For filters that do not use the this method is called when a frame is wanted on an output For a it should directly call filter_frame on the corresponding output For a if there are queued frames already one of these frames should be pushed If the filter should request a frame on one of its repeatedly until at least one frame has been pushed Return or at least make progress towards producing a frame
Definition: filter_design.txt:264
VC2EncContext
Definition: vc2enc.c:126
VC2EncContext::num_x
int num_x
Definition: vc2enc.c:156
AVCodecContext::strict_std_compliance
int strict_std_compliance
strictly follow the standard (MPEG-4, ...).
Definition: avcodec.h:1309
VC2EncContext::q_ceil
int q_ceil
Definition: vc2enc.c:167
AV_PIX_FMT_YUV420P12
#define AV_PIX_FMT_YUV420P12
Definition: pixfmt.h:412
VC2EncContext::profile
int profile
Definition: vc2enc.c:147
left
Tag MUST be and< 10hcoeff half pel interpolation filter coefficients, hcoeff[0] are the 2 middle coefficients[1] are the next outer ones and so on, resulting in a filter like:...eff[2], hcoeff[1], hcoeff[0], hcoeff[0], hcoeff[1], hcoeff[2] ... the sign of the coefficients is not explicitly stored but alternates after each coeff and coeff[0] is positive, so ...,+,-,+,-,+,+,-,+,-,+,... hcoeff[0] is not explicitly stored but found by subtracting the sum of all stored coefficients with signs from 32 hcoeff[0]=32 - hcoeff[1] - hcoeff[2] - ... a good choice for hcoeff and htaps is htaps=6 hcoeff={40,-10, 2} an alternative which requires more computations at both encoder and decoder side and may or may not be better is htaps=8 hcoeff={42,-14, 6,-2}ref_frames minimum of the number of available reference frames and max_ref_frames for example the first frame after a key frame always has ref_frames=1spatial_decomposition_type wavelet type 0 is a 9/7 symmetric compact integer wavelet 1 is a 5/3 symmetric compact integer wavelet others are reserved stored as delta from last, last is reset to 0 if always_reset||keyframeqlog quality(logarithmic quantizer scale) stored as delta from last, last is reset to 0 if always_reset||keyframemv_scale stored as delta from last, last is reset to 0 if always_reset||keyframe FIXME check that everything works fine if this changes between framesqbias dequantization bias stored as delta from last, last is reset to 0 if always_reset||keyframeblock_max_depth maximum depth of the block tree stored as delta from last, last is reset to 0 if always_reset||keyframequant_table quantization tableHighlevel bitstream structure:==============================--------------------------------------------|Header|--------------------------------------------|------------------------------------|||Block0||||split?||||yes no||||......... intra?||||:Block01 :yes no||||:Block02 :....... ..........||||:Block03 ::y DC ::ref index:||||:Block04 ::cb DC ::motion x :||||......... :cr DC ::motion y :||||....... ..........|||------------------------------------||------------------------------------|||Block1|||...|--------------------------------------------|------------ ------------ ------------|||Y subbands||Cb subbands||Cr subbands||||--- ---||--- ---||--- ---|||||LL0||HL0||||LL0||HL0||||LL0||HL0|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||LH0||HH0||||LH0||HH0||||LH0||HH0|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||HL1||LH1||||HL1||LH1||||HL1||LH1|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||HH1||HL2||||HH1||HL2||||HH1||HL2|||||...||...||...|||------------ ------------ ------------|--------------------------------------------Decoding process:=================------------|||Subbands|------------||||------------|Intra DC||||LL0 subband prediction ------------|\ Dequantization ------------------- \||Reference frames|\ IDWT|------- -------|Motion \|||Frame 0||Frame 1||Compensation . OBMC v -------|------- -------|--------------. \------> Frame n output Frame Frame<----------------------------------/|...|------------------- Range Coder:============Binary Range Coder:------------------- The implemented range coder is an adapted version based upon "Range encoding: an algorithm for removing redundancy from a digitised message." by G. N. N. Martin. The symbols encoded by the Snow range coder are bits(0|1). The associated probabilities are not fix but change depending on the symbol mix seen so far. bit seen|new state ---------+----------------------------------------------- 0|256 - state_transition_table[256 - old_state];1|state_transition_table[old_state];state_transition_table={ 0, 0, 0, 0, 0, 0, 0, 0, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 190, 191, 192, 194, 194, 195, 196, 197, 198, 199, 200, 201, 202, 202, 204, 205, 206, 207, 208, 209, 209, 210, 211, 212, 213, 215, 215, 216, 217, 218, 219, 220, 220, 222, 223, 224, 225, 226, 227, 227, 229, 229, 230, 231, 232, 234, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 248, 0, 0, 0, 0, 0, 0, 0};FIXME Range Coding of integers:------------------------- FIXME Neighboring Blocks:===================left and top are set to the respective blocks unless they are outside of the image in which case they are set to the Null block top-left is set to the top left block unless it is outside of the image in which case it is set to the left block if this block has no larger parent block or it is at the left side of its parent block and the top right block is not outside of the image then the top right block is used for top-right else the top-left block is used Null block y, cb, cr are 128 level, ref, mx and my are 0 Motion Vector Prediction:=========================1. the motion vectors of all the neighboring blocks are scaled to compensate for the difference of reference frames scaled_mv=(mv *(256 *(current_reference+1)/(mv.reference+1))+128)> the median of the scaled left
Definition: snow.txt:386
DIRAC_PCODE_END_SEQ
@ DIRAC_PCODE_END_SEQ
Definition: dirac.h:59
AVCodecContext
main external API structure.
Definition: avcodec.h:398
put_bits_ptr
static uint8_t * put_bits_ptr(PutBitContext *s)
Return the pointer to the byte where the bitstream writer will put the next bit.
Definition: put_bits.h:370
SliceArgs::cache
int cache[DIRAC_MAX_QUANT_INDEX]
Definition: vc2enc.c:107
ff_get_encode_buffer
int ff_get_encode_buffer(AVCodecContext *avctx, AVPacket *avpkt, int64_t size, int flags)
Get a buffer for a packet.
Definition: encode.c:79
AVRational::den
int den
Denominator.
Definition: rational.h:60
AV_PIX_FMT_NONE
@ AV_PIX_FMT_NONE
Definition: pixfmt.h:65
AV_OPT_TYPE_INT
@ AV_OPT_TYPE_INT
Definition: opt.h:225
encode_seq_header
static void encode_seq_header(VC2EncContext *s)
Definition: vc2enc.c:402
skip_put_bytes
static void skip_put_bytes(PutBitContext *s, int n)
Skip the given number of bytes.
Definition: put_bits.h:379
AVPixFmtDescriptor::comp
AVComponentDescriptor comp[4]
Parameters that describe how pixels are packed.
Definition: pixdesc.h:105
VC2_TRANSFORM_HAAR
@ VC2_TRANSFORM_HAAR
Definition: vc2enc_dwt.h:34
VC2ENC_FLAGS
#define VC2ENC_FLAGS
Definition: vc2enc.c:1189
Plane
Definition: cfhd.h:118
VC2EncContext::strict_compliance
int strict_compliance
Definition: vc2enc.c:174
put_vc2_ue_uint
static av_always_inline void put_vc2_ue_uint(PutBitContext *pb, uint32_t val)
Definition: vc2enc.c:185
AV_PIX_FMT_YUV444P
@ AV_PIX_FMT_YUV444P
planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples)
Definition: pixfmt.h:71
shift
static int shift(int a, int b)
Definition: sonic.c:88
QUANT
#define QUANT(c, mul, add, shift)
Definition: vc2enc.c:526
Plane::dwt_height
int dwt_height
Definition: vc2enc.c:101
AVMEDIA_TYPE_VIDEO
@ AVMEDIA_TYPE_VIDEO
Definition: avutil.h:201
count_vc2_ue_uint
static av_always_inline int count_vc2_ue_uint(uint32_t val)
Definition: vc2enc.c:213
AV_PIX_FMT_YUV422P
@ AV_PIX_FMT_YUV422P
planar YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples)
Definition: pixfmt.h:70
Plane::band
SubBand band[DWT_LEVELS_3D][4]
Definition: cfhd.h:131
AV_CODEC_FLAG_BITEXACT
#define AV_CODEC_FLAG_BITEXACT
Use only bitexact stuff (except (I)DCT).
Definition: avcodec.h:288
AV_FIELD_PROGRESSIVE
@ AV_FIELD_PROGRESSIVE
Definition: codec_par.h:39
flush_put_bits
static void flush_put_bits(PutBitContext *s)
Pad the end of the output stream with zeros.
Definition: put_bits.h:143
AVPixFmtDescriptor
Descriptor that unambiguously describes how the bits of a pixel are stored in the up to 4 data planes...
Definition: pixdesc.h:69
diff
static av_always_inline int diff(const uint32_t a, const uint32_t b)
Definition: vf_palettegen.c:139
TransformArgs::field
int field
Definition: vc2enc.c:122
FFALIGN
#define FFALIGN(x, a)
Definition: macros.h:78
calc_slice_sizes
static int calc_slice_sizes(VC2EncContext *s)
Definition: vc2enc.c:645
VC2EncContext::base_vf
int base_vf
Definition: vc2enc.c:145
VC2EncContext::slice_height
int slice_height
Definition: vc2enc.c:175
AVCodecContext::priv_data
void * priv_data
Definition: avcodec.h:425
AVPacket
This structure stores compressed data.
Definition: packet.h:351
av_freep
#define av_freep(p)
Definition: tableprint_vlc.h:34
AVCodecContext::width
int width
picture width / height.
Definition: avcodec.h:571
VC2EncContext::plane
Plane plane[3]
Definition: vc2enc.c:129
coeff
static const double coeff[2][5]
Definition: vf_owdenoise.c:78
av_log
#define av_log(a,...)
Definition: tableprint_vlc.h:27
VC2_QM_DEF
@ VC2_QM_DEF
Definition: vc2enc.c:81
AV_CLASS_CATEGORY_ENCODER
@ AV_CLASS_CATEGORY_ENCODER
Definition: log.h:34
h
h
Definition: vp9dsp_template.c:2038
DIRAC_PCODE_PICTURE_HQ
@ DIRAC_PCODE_PICTURE_HQ
Definition: dirac.h:65
TransformArgs::istride
ptrdiff_t istride
Definition: vc2enc.c:121
put_bits.h
AV_OPT_TYPE_CONST
@ AV_OPT_TYPE_CONST
Definition: opt.h:234
SLICE_REDIST_TOTAL
#define SLICE_REDIST_TOTAL
Definition: vc2enc.c:38
av_log2
int av_log2(unsigned v)
Definition: intmath.c:26
VC2_TRANSFORM_HAAR_S
@ VC2_TRANSFORM_HAAR_S
Definition: vc2enc_dwt.h:35
encode_sample_fmt
static void encode_sample_fmt(VC2EncContext *s)
Definition: vc2enc.c:282
VC2EncContext::frame_max_bytes
int frame_max_bytes
Definition: vc2enc.c:164
VC2BaseVideoFormat::width
int width
Definition: vc2enc.c:43
AVCodecContext::sample_aspect_ratio
AVRational sample_aspect_ratio
sample aspect ratio (0 if unknown) That is the width of a pixel divided by the height of the pixel.
Definition: avcodec.h:768
VC2_TRANSFORM_5_3
@ VC2_TRANSFORM_5_3
Definition: vc2enc_dwt.h:32
encode_signal_range
static void encode_signal_range(VC2EncContext *s)
Definition: vc2enc.c:336
TransformArgs::idata
const void * idata
Definition: vc2enc.c:120
SubBand::height
int height
Definition: cfhd.h:114
VC2EncContext::bpp_idx
int bpp_idx
Definition: vc2enc.c:139