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 "dirac.h"
25 #include "put_bits.h"
26 #include "internal.h"
27 #include "version.h"
28 
29 #include "vc2enc_dwt.h"
30 #include "diractab.h"
31 
32 /* The limited size resolution of each slice forces us to do this */
33 #define SSIZE_ROUND(b) (FFALIGN((b), s->size_scaler) + 4 + s->prefix_bytes)
34 
35 /* Decides the cutoff point in # of slices to distribute the leftover bytes */
36 #define SLICE_REDIST_TOTAL 150
37 
38 typedef struct VC2BaseVideoFormat {
42  const char *name;
44 
46  { 0 }, /* Custom format, here just to make indexing equal to base_vf */
47  { AV_PIX_FMT_YUV420P, { 1001, 15000 }, 176, 120, 0, 1, "QSIF525" },
48  { AV_PIX_FMT_YUV420P, { 2, 25 }, 176, 144, 0, 1, "QCIF" },
49  { AV_PIX_FMT_YUV420P, { 1001, 15000 }, 352, 240, 0, 1, "SIF525" },
50  { AV_PIX_FMT_YUV420P, { 2, 25 }, 352, 288, 0, 1, "CIF" },
51  { AV_PIX_FMT_YUV420P, { 1001, 15000 }, 704, 480, 0, 1, "4SIF525" },
52  { AV_PIX_FMT_YUV420P, { 2, 25 }, 704, 576, 0, 1, "4CIF" },
53 
54  { AV_PIX_FMT_YUV422P10, { 1001, 30000 }, 720, 480, 1, 2, "SD480I-60" },
55  { AV_PIX_FMT_YUV422P10, { 1, 25 }, 720, 576, 1, 2, "SD576I-50" },
56 
57  { AV_PIX_FMT_YUV422P10, { 1001, 60000 }, 1280, 720, 0, 3, "HD720P-60" },
58  { AV_PIX_FMT_YUV422P10, { 1, 50 }, 1280, 720, 0, 3, "HD720P-50" },
59  { AV_PIX_FMT_YUV422P10, { 1001, 30000 }, 1920, 1080, 1, 3, "HD1080I-60" },
60  { AV_PIX_FMT_YUV422P10, { 1, 25 }, 1920, 1080, 1, 3, "HD1080I-50" },
61  { AV_PIX_FMT_YUV422P10, { 1001, 60000 }, 1920, 1080, 0, 3, "HD1080P-60" },
62  { AV_PIX_FMT_YUV422P10, { 1, 50 }, 1920, 1080, 0, 3, "HD1080P-50" },
63 
64  { AV_PIX_FMT_YUV444P12, { 1, 24 }, 2048, 1080, 0, 4, "DC2K" },
65  { AV_PIX_FMT_YUV444P12, { 1, 24 }, 4096, 2160, 0, 5, "DC4K" },
66 
67  { AV_PIX_FMT_YUV422P10, { 1001, 60000 }, 3840, 2160, 0, 6, "UHDTV 4K-60" },
68  { AV_PIX_FMT_YUV422P10, { 1, 50 }, 3840, 2160, 0, 6, "UHDTV 4K-50" },
69 
70  { AV_PIX_FMT_YUV422P10, { 1001, 60000 }, 7680, 4320, 0, 7, "UHDTV 8K-60" },
71  { AV_PIX_FMT_YUV422P10, { 1, 50 }, 7680, 4320, 0, 7, "UHDTV 8K-50" },
72 
73  { AV_PIX_FMT_YUV422P10, { 1001, 24000 }, 1920, 1080, 0, 3, "HD1080P-24" },
74  { AV_PIX_FMT_YUV422P10, { 1001, 30000 }, 720, 486, 1, 2, "SD Pro486" },
75 };
76 static const int base_video_fmts_len = FF_ARRAY_ELEMS(base_video_fmts);
77 
78 enum VC2_QM {
82 
84 };
85 
86 typedef struct SubBand {
88  ptrdiff_t stride;
89  int width;
90  int height;
91 } SubBand;
92 
93 typedef struct Plane {
94  SubBand band[MAX_DWT_LEVELS][4];
96  int width;
97  int height;
98  int dwt_width;
100  ptrdiff_t coef_stride;
101 } Plane;
102 
103 typedef struct SliceArgs {
106  void *ctx;
107  int x;
108  int y;
112  int bytes;
113 } SliceArgs;
114 
115 typedef struct TransformArgs {
116  void *ctx;
118  void *idata;
119  ptrdiff_t istride;
120  int field;
122 } TransformArgs;
123 
124 typedef struct VC2EncContext {
127  Plane plane[3];
130 
132  TransformArgs transform_args[3];
133 
134  /* For conversion from unsigned pixel values to signed */
136  int bpp;
137  int bpp_idx;
138 
139  /* Picture number */
140  uint32_t picture_number;
141 
142  /* Base video format */
143  int base_vf;
144  int level;
145  int profile;
146 
147  /* Quantization matrix */
150 
151  /* Division LUT */
152  uint32_t qmagic_lut[116][2];
153 
154  int num_x; /* #slices horizontally */
155  int num_y; /* #slices vertically */
160 
161  /* Rate control stuff */
165  int q_ceil;
166  int q_avg;
167 
168  /* Options */
169  double tolerance;
176  enum VC2_QM quant_matrix;
177 
178  /* Parse code state */
180  enum DiracParseCodes last_parse_code;
181 } VC2EncContext;
182 
184 {
185  int i;
186  int pbits = 0, bits = 0, topbit = 1, maxval = 1;
187 
188  if (!val++) {
189  put_bits(pb, 1, 1);
190  return;
191  }
192 
193  while (val > maxval) {
194  topbit <<= 1;
195  maxval <<= 1;
196  maxval |= 1;
197  }
198 
199  bits = ff_log2(topbit);
200 
201  for (i = 0; i < bits; i++) {
202  topbit >>= 1;
203  pbits <<= 2;
204  if (val & topbit)
205  pbits |= 0x1;
206  }
207 
208  put_bits(pb, bits*2 + 1, (pbits << 1) | 1);
209 }
210 
212 {
213  int topbit = 1, maxval = 1;
214 
215  if (!val++)
216  return 1;
217 
218  while (val > maxval) {
219  topbit <<= 1;
220  maxval <<= 1;
221  maxval |= 1;
222  }
223 
224  return ff_log2(topbit)*2 + 1;
225 }
226 
227 /* VC-2 10.4 - parse_info() */
229 {
230  uint32_t cur_pos, dist;
231 
233 
234  cur_pos = put_bits_count(&s->pb) >> 3;
235 
236  /* Magic string */
237  avpriv_put_string(&s->pb, "BBCD", 0);
238 
239  /* Parse code */
240  put_bits(&s->pb, 8, pcode);
241 
242  /* Next parse offset */
243  dist = cur_pos - s->next_parse_offset;
244  AV_WB32(s->pb.buf + s->next_parse_offset + 5, dist);
245  s->next_parse_offset = cur_pos;
246  put_bits32(&s->pb, pcode == DIRAC_PCODE_END_SEQ ? 13 : 0);
247 
248  /* Last parse offset */
249  put_bits32(&s->pb, s->last_parse_code == DIRAC_PCODE_END_SEQ ? 13 : dist);
250 
251  s->last_parse_code = pcode;
252 }
253 
254 /* VC-2 11.1 - parse_parameters()
255  * The level dictates what the decoder should expect in terms of resolution
256  * and allows it to quickly reject whatever it can't support. Remember,
257  * this codec kinda targets cheapo FPGAs without much memory. Unfortunately
258  * it also limits us greatly in our choice of formats, hence the flag to disable
259  * strict_compliance */
261 {
262  put_vc2_ue_uint(&s->pb, s->ver.major); /* VC-2 demands this to be 2 */
263  put_vc2_ue_uint(&s->pb, s->ver.minor); /* ^^ and this to be 0 */
264  put_vc2_ue_uint(&s->pb, s->profile); /* 3 to signal HQ profile */
265  put_vc2_ue_uint(&s->pb, s->level); /* 3 - 1080/720, 6 - 4K */
266 }
267 
268 /* VC-2 11.3 - frame_size() */
270 {
271  put_bits(&s->pb, 1, !s->strict_compliance);
272  if (!s->strict_compliance) {
273  AVCodecContext *avctx = s->avctx;
274  put_vc2_ue_uint(&s->pb, avctx->width);
275  put_vc2_ue_uint(&s->pb, avctx->height);
276  }
277 }
278 
279 /* VC-2 11.3.3 - color_diff_sampling_format() */
281 {
282  put_bits(&s->pb, 1, !s->strict_compliance);
283  if (!s->strict_compliance) {
284  int idx;
285  if (s->chroma_x_shift == 1 && s->chroma_y_shift == 0)
286  idx = 1; /* 422 */
287  else if (s->chroma_x_shift == 1 && s->chroma_y_shift == 1)
288  idx = 2; /* 420 */
289  else
290  idx = 0; /* 444 */
291  put_vc2_ue_uint(&s->pb, idx);
292  }
293 }
294 
295 /* VC-2 11.3.4 - scan_format() */
297 {
298  put_bits(&s->pb, 1, !s->strict_compliance);
299  if (!s->strict_compliance)
300  put_vc2_ue_uint(&s->pb, s->interlaced);
301 }
302 
303 /* VC-2 11.3.5 - frame_rate() */
305 {
306  put_bits(&s->pb, 1, !s->strict_compliance);
307  if (!s->strict_compliance) {
308  AVCodecContext *avctx = s->avctx;
309  put_vc2_ue_uint(&s->pb, 0);
310  put_vc2_ue_uint(&s->pb, avctx->time_base.den);
311  put_vc2_ue_uint(&s->pb, avctx->time_base.num);
312  }
313 }
314 
315 /* VC-2 11.3.6 - aspect_ratio() */
317 {
318  put_bits(&s->pb, 1, !s->strict_compliance);
319  if (!s->strict_compliance) {
320  AVCodecContext *avctx = s->avctx;
321  put_vc2_ue_uint(&s->pb, 0);
324  }
325 }
326 
327 /* VC-2 11.3.7 - clean_area() */
329 {
330  put_bits(&s->pb, 1, 0);
331 }
332 
333 /* VC-2 11.3.8 - signal_range() */
335 {
336  put_bits(&s->pb, 1, !s->strict_compliance);
337  if (!s->strict_compliance)
338  put_vc2_ue_uint(&s->pb, s->bpp_idx);
339 }
340 
341 /* VC-2 11.3.9 - color_spec() */
343 {
344  AVCodecContext *avctx = s->avctx;
345  put_bits(&s->pb, 1, !s->strict_compliance);
346  if (!s->strict_compliance) {
347  int val;
348  put_vc2_ue_uint(&s->pb, 0);
349 
350  /* primaries */
351  put_bits(&s->pb, 1, 1);
352  if (avctx->color_primaries == AVCOL_PRI_BT470BG)
353  val = 2;
354  else if (avctx->color_primaries == AVCOL_PRI_SMPTE170M)
355  val = 1;
356  else if (avctx->color_primaries == AVCOL_PRI_SMPTE240M)
357  val = 1;
358  else
359  val = 0;
360  put_vc2_ue_uint(&s->pb, val);
361 
362  /* color matrix */
363  put_bits(&s->pb, 1, 1);
364  if (avctx->colorspace == AVCOL_SPC_RGB)
365  val = 3;
366  else if (avctx->colorspace == AVCOL_SPC_YCOCG)
367  val = 2;
368  else if (avctx->colorspace == AVCOL_SPC_BT470BG)
369  val = 1;
370  else
371  val = 0;
372  put_vc2_ue_uint(&s->pb, val);
373 
374  /* transfer function */
375  put_bits(&s->pb, 1, 1);
376  if (avctx->color_trc == AVCOL_TRC_LINEAR)
377  val = 2;
378  else if (avctx->color_trc == AVCOL_TRC_BT1361_ECG)
379  val = 1;
380  else
381  val = 0;
382  put_vc2_ue_uint(&s->pb, val);
383  }
384 }
385 
386 /* VC-2 11.3 - source_parameters() */
388 {
397 }
398 
399 /* VC-2 11 - sequence_header() */
401 {
404  put_vc2_ue_uint(&s->pb, s->base_vf);
406  put_vc2_ue_uint(&s->pb, s->interlaced); /* Frames or fields coding */
407 }
408 
409 /* VC-2 12.1 - picture_header() */
411 {
413  put_bits32(&s->pb, s->picture_number++);
414 }
415 
416 /* VC-2 12.3.4.1 - slice_parameters() */
418 {
419  put_vc2_ue_uint(&s->pb, s->num_x);
420  put_vc2_ue_uint(&s->pb, s->num_y);
422  put_vc2_ue_uint(&s->pb, s->size_scaler);
423 }
424 
425 /* 1st idx = LL, second - vertical, third - horizontal, fourth - total */
426 const uint8_t vc2_qm_col_tab[][4] = {
427  {20, 9, 15, 4},
428  { 0, 6, 6, 4},
429  { 0, 3, 3, 5},
430  { 0, 3, 5, 1},
431  { 0, 11, 10, 11}
432 };
433 
434 const uint8_t vc2_qm_flat_tab[][4] = {
435  { 0, 0, 0, 0},
436  { 0, 0, 0, 0},
437  { 0, 0, 0, 0},
438  { 0, 0, 0, 0},
439  { 0, 0, 0, 0}
440 };
441 
443 {
444  int level, orientation;
445 
446  if (s->wavelet_depth <= 4 && s->quant_matrix == VC2_QM_DEF) {
447  s->custom_quant_matrix = 0;
448  for (level = 0; level < s->wavelet_depth; level++) {
453  }
454  return;
455  }
456 
457  s->custom_quant_matrix = 1;
458 
459  if (s->quant_matrix == VC2_QM_DEF) {
460  for (level = 0; level < s->wavelet_depth; level++) {
461  for (orientation = 0; orientation < 4; orientation++) {
462  if (level <= 3)
463  s->quant[level][orientation] = ff_dirac_default_qmat[s->wavelet_idx][level][orientation];
464  else
465  s->quant[level][orientation] = vc2_qm_col_tab[level][orientation];
466  }
467  }
468  } else if (s->quant_matrix == VC2_QM_COL) {
469  for (level = 0; level < s->wavelet_depth; level++) {
470  for (orientation = 0; orientation < 4; orientation++) {
471  s->quant[level][orientation] = vc2_qm_col_tab[level][orientation];
472  }
473  }
474  } else {
475  for (level = 0; level < s->wavelet_depth; level++) {
476  for (orientation = 0; orientation < 4; orientation++) {
477  s->quant[level][orientation] = vc2_qm_flat_tab[level][orientation];
478  }
479  }
480  }
481 }
482 
483 /* VC-2 12.3.4.2 - quant_matrix() */
485 {
486  int level;
487  put_bits(&s->pb, 1, s->custom_quant_matrix);
488  if (s->custom_quant_matrix) {
489  put_vc2_ue_uint(&s->pb, s->quant[0][0]);
490  for (level = 0; level < s->wavelet_depth; level++) {
491  put_vc2_ue_uint(&s->pb, s->quant[level][1]);
492  put_vc2_ue_uint(&s->pb, s->quant[level][2]);
493  put_vc2_ue_uint(&s->pb, s->quant[level][3]);
494  }
495  }
496 }
497 
498 /* VC-2 12.3 - transform_parameters() */
500 {
501  put_vc2_ue_uint(&s->pb, s->wavelet_idx);
503 
506 }
507 
508 /* VC-2 12.2 - wavelet_transform() */
510 {
513 }
514 
515 /* VC-2 12 - picture_parse() */
517 {
522 }
523 
524 #define QUANT(c, mul, add, shift) (((mul) * (c) + (add)) >> (shift))
525 
526 /* VC-2 13.5.5.2 - slice_band() */
527 static void encode_subband(VC2EncContext *s, PutBitContext *pb, int sx, int sy,
528  SubBand *b, int quant)
529 {
530  int x, y;
531 
532  const int left = b->width * (sx+0) / s->num_x;
533  const int right = b->width * (sx+1) / s->num_x;
534  const int top = b->height * (sy+0) / s->num_y;
535  const int bottom = b->height * (sy+1) / s->num_y;
536 
537  dwtcoef *coeff = b->buf + top * b->stride;
538  const uint64_t q_m = ((uint64_t)(s->qmagic_lut[quant][0])) << 2;
539  const uint64_t q_a = s->qmagic_lut[quant][1];
540  const int q_s = av_log2(ff_dirac_qscale_tab[quant]) + 32;
541 
542  for (y = top; y < bottom; y++) {
543  for (x = left; x < right; x++) {
544  uint32_t c_abs = QUANT(FFABS(coeff[x]), q_m, q_a, q_s);
545  put_vc2_ue_uint(pb, c_abs);
546  if (c_abs)
547  put_bits(pb, 1, coeff[x] < 0);
548  }
549  coeff += b->stride;
550  }
551 }
552 
553 static int count_hq_slice(SliceArgs *slice, int quant_idx)
554 {
555  int x, y;
556  uint8_t quants[MAX_DWT_LEVELS][4];
557  int bits = 0, p, level, orientation;
558  VC2EncContext *s = slice->ctx;
559 
560  if (slice->cache[quant_idx])
561  return slice->cache[quant_idx];
562 
563  bits += 8*s->prefix_bytes;
564  bits += 8; /* quant_idx */
565 
566  for (level = 0; level < s->wavelet_depth; level++)
567  for (orientation = !!level; orientation < 4; orientation++)
568  quants[level][orientation] = FFMAX(quant_idx - s->quant[level][orientation], 0);
569 
570  for (p = 0; p < 3; p++) {
571  int bytes_start, bytes_len, pad_s, pad_c;
572  bytes_start = bits >> 3;
573  bits += 8;
574  for (level = 0; level < s->wavelet_depth; level++) {
575  for (orientation = !!level; orientation < 4; orientation++) {
576  SubBand *b = &s->plane[p].band[level][orientation];
577 
578  const int q_idx = quants[level][orientation];
579  const uint64_t q_m = ((uint64_t)s->qmagic_lut[q_idx][0]) << 2;
580  const uint64_t q_a = s->qmagic_lut[q_idx][1];
581  const int q_s = av_log2(ff_dirac_qscale_tab[q_idx]) + 32;
582 
583  const int left = b->width * slice->x / s->num_x;
584  const int right = b->width *(slice->x+1) / s->num_x;
585  const int top = b->height * slice->y / s->num_y;
586  const int bottom = b->height *(slice->y+1) / s->num_y;
587 
588  dwtcoef *buf = b->buf + top * b->stride;
589 
590  for (y = top; y < bottom; y++) {
591  for (x = left; x < right; x++) {
592  uint32_t c_abs = QUANT(FFABS(buf[x]), q_m, q_a, q_s);
593  bits += count_vc2_ue_uint(c_abs);
594  bits += !!c_abs;
595  }
596  buf += b->stride;
597  }
598  }
599  }
600  bits += FFALIGN(bits, 8) - bits;
601  bytes_len = (bits >> 3) - bytes_start - 1;
602  pad_s = FFALIGN(bytes_len, s->size_scaler)/s->size_scaler;
603  pad_c = (pad_s*s->size_scaler) - bytes_len;
604  bits += pad_c*8;
605  }
606 
607  slice->cache[quant_idx] = bits;
608 
609  return bits;
610 }
611 
612 /* Approaches the best possible quantizer asymptotically, its kinda exaustive
613  * but we have a LUT to get the coefficient size in bits. Guaranteed to never
614  * overshoot, which is apparently very important when streaming */
615 static int rate_control(AVCodecContext *avctx, void *arg)
616 {
617  SliceArgs *slice_dat = arg;
618  VC2EncContext *s = slice_dat->ctx;
619  const int top = slice_dat->bits_ceil;
620  const int bottom = slice_dat->bits_floor;
621  int quant_buf[2] = {-1, -1};
622  int quant = slice_dat->quant_idx, step = 1;
623  int bits_last, bits = count_hq_slice(slice_dat, quant);
624  while ((bits > top) || (bits < bottom)) {
625  const int signed_step = bits > top ? +step : -step;
626  quant = av_clip(quant + signed_step, 0, s->q_ceil-1);
627  bits = count_hq_slice(slice_dat, quant);
628  if (quant_buf[1] == quant) {
629  quant = FFMAX(quant_buf[0], quant);
630  bits = quant == quant_buf[0] ? bits_last : bits;
631  break;
632  }
633  step = av_clip(step/2, 1, (s->q_ceil-1)/2);
634  quant_buf[1] = quant_buf[0];
635  quant_buf[0] = quant;
636  bits_last = bits;
637  }
638  slice_dat->quant_idx = av_clip(quant, 0, s->q_ceil-1);
639  slice_dat->bytes = SSIZE_ROUND(bits >> 3);
640  return 0;
641 }
642 
644 {
645  int i, j, slice_x, slice_y, bytes_left = 0;
646  int bytes_top[SLICE_REDIST_TOTAL] = {0};
647  int64_t total_bytes_needed = 0;
648  int slice_redist_range = FFMIN(SLICE_REDIST_TOTAL, s->num_x*s->num_y);
649  SliceArgs *enc_args = s->slice_args;
650  SliceArgs *top_loc[SLICE_REDIST_TOTAL] = {NULL};
651 
653 
654  for (slice_y = 0; slice_y < s->num_y; slice_y++) {
655  for (slice_x = 0; slice_x < s->num_x; slice_x++) {
656  SliceArgs *args = &enc_args[s->num_x*slice_y + slice_x];
657  args->ctx = s;
658  args->x = slice_x;
659  args->y = slice_y;
660  args->bits_ceil = s->slice_max_bytes << 3;
661  args->bits_floor = s->slice_min_bytes << 3;
662  memset(args->cache, 0, s->q_ceil*sizeof(*args->cache));
663  }
664  }
665 
666  /* First pass - determine baseline slice sizes w.r.t. max_slice_size */
667  s->avctx->execute(s->avctx, rate_control, enc_args, NULL, s->num_x*s->num_y,
668  sizeof(SliceArgs));
669 
670  for (i = 0; i < s->num_x*s->num_y; i++) {
671  SliceArgs *args = &enc_args[i];
672  bytes_left += args->bytes;
673  for (j = 0; j < slice_redist_range; j++) {
674  if (args->bytes > bytes_top[j]) {
675  bytes_top[j] = args->bytes;
676  top_loc[j] = args;
677  break;
678  }
679  }
680  }
681 
682  bytes_left = s->frame_max_bytes - bytes_left;
683 
684  /* Second pass - distribute leftover bytes */
685  while (bytes_left > 0) {
686  int distributed = 0;
687  for (i = 0; i < slice_redist_range; i++) {
688  SliceArgs *args;
689  int bits, bytes, diff, prev_bytes, new_idx;
690  if (bytes_left <= 0)
691  break;
692  if (!top_loc[i] || !top_loc[i]->quant_idx)
693  break;
694  args = top_loc[i];
695  prev_bytes = args->bytes;
696  new_idx = FFMAX(args->quant_idx - 1, 0);
697  bits = count_hq_slice(args, new_idx);
698  bytes = SSIZE_ROUND(bits >> 3);
699  diff = bytes - prev_bytes;
700  if ((bytes_left - diff) > 0) {
701  args->quant_idx = new_idx;
702  args->bytes = bytes;
703  bytes_left -= diff;
704  distributed++;
705  }
706  }
707  if (!distributed)
708  break;
709  }
710 
711  for (i = 0; i < s->num_x*s->num_y; i++) {
712  SliceArgs *args = &enc_args[i];
713  total_bytes_needed += args->bytes;
714  s->q_avg = (s->q_avg + args->quant_idx)/2;
715  }
716 
717  return total_bytes_needed;
718 }
719 
720 /* VC-2 13.5.3 - hq_slice */
721 static int encode_hq_slice(AVCodecContext *avctx, void *arg)
722 {
723  SliceArgs *slice_dat = arg;
724  VC2EncContext *s = slice_dat->ctx;
725  PutBitContext *pb = &slice_dat->pb;
726  const int slice_x = slice_dat->x;
727  const int slice_y = slice_dat->y;
728  const int quant_idx = slice_dat->quant_idx;
729  const int slice_bytes_max = slice_dat->bytes;
730  uint8_t quants[MAX_DWT_LEVELS][4];
731  int p, level, orientation;
732 
733  /* The reference decoder ignores it, and its typical length is 0 */
734  memset(put_bits_ptr(pb), 0, s->prefix_bytes);
736 
737  put_bits(pb, 8, quant_idx);
738 
739  /* Slice quantization (slice_quantizers() in the specs) */
740  for (level = 0; level < s->wavelet_depth; level++)
741  for (orientation = !!level; orientation < 4; orientation++)
742  quants[level][orientation] = FFMAX(quant_idx - s->quant[level][orientation], 0);
743 
744  /* Luma + 2 Chroma planes */
745  for (p = 0; p < 3; p++) {
746  int bytes_start, bytes_len, pad_s, pad_c;
747  bytes_start = put_bits_count(pb) >> 3;
748  put_bits(pb, 8, 0);
749  for (level = 0; level < s->wavelet_depth; level++) {
750  for (orientation = !!level; orientation < 4; orientation++) {
751  encode_subband(s, pb, slice_x, slice_y,
752  &s->plane[p].band[level][orientation],
753  quants[level][orientation]);
754  }
755  }
757  bytes_len = (put_bits_count(pb) >> 3) - bytes_start - 1;
758  if (p == 2) {
759  int len_diff = slice_bytes_max - (put_bits_count(pb) >> 3);
760  pad_s = FFALIGN((bytes_len + len_diff), s->size_scaler)/s->size_scaler;
761  pad_c = (pad_s*s->size_scaler) - bytes_len;
762  } else {
763  pad_s = FFALIGN(bytes_len, s->size_scaler)/s->size_scaler;
764  pad_c = (pad_s*s->size_scaler) - bytes_len;
765  }
766  pb->buf[bytes_start] = pad_s;
767  flush_put_bits(pb);
768  /* vc2-reference uses that padding that decodes to '0' coeffs */
769  memset(put_bits_ptr(pb), 0xFF, pad_c);
770  skip_put_bytes(pb, pad_c);
771  }
772 
773  return 0;
774 }
775 
776 /* VC-2 13.5.1 - low_delay_transform_data() */
778 {
779  uint8_t *buf;
780  int slice_x, slice_y, skip = 0;
781  SliceArgs *enc_args = s->slice_args;
782 
783  flush_put_bits(&s->pb);
784  buf = put_bits_ptr(&s->pb);
785 
786  for (slice_y = 0; slice_y < s->num_y; slice_y++) {
787  for (slice_x = 0; slice_x < s->num_x; slice_x++) {
788  SliceArgs *args = &enc_args[s->num_x*slice_y + slice_x];
789  init_put_bits(&args->pb, buf + skip, args->bytes+s->prefix_bytes);
790  skip += args->bytes;
791  }
792  }
793 
794  s->avctx->execute(s->avctx, encode_hq_slice, enc_args, NULL, s->num_x*s->num_y,
795  sizeof(SliceArgs));
796 
797  skip_put_bytes(&s->pb, skip);
798 
799  return 0;
800 }
801 
802 /*
803  * Transform basics for a 3 level transform
804  * |---------------------------------------------------------------------|
805  * | LL-0 | HL-0 | | |
806  * |--------|-------| HL-1 | |
807  * | LH-0 | HH-0 | | |
808  * |----------------|-----------------| HL-2 |
809  * | | | |
810  * | LH-1 | HH-1 | |
811  * | | | |
812  * |----------------------------------|----------------------------------|
813  * | | |
814  * | | |
815  * | | |
816  * | LH-2 | HH-2 |
817  * | | |
818  * | | |
819  * | | |
820  * |---------------------------------------------------------------------|
821  *
822  * DWT transforms are generally applied by splitting the image in two vertically
823  * and applying a low pass transform on the left part and a corresponding high
824  * pass transform on the right hand side. This is known as the horizontal filter
825  * stage.
826  * After that, the same operation is performed except the image is divided
827  * horizontally, with the high pass on the lower and the low pass on the higher
828  * side.
829  * Therefore, you're left with 4 subdivisions - known as low-low, low-high,
830  * high-low and high-high. They're referred to as orientations in the decoder
831  * and encoder.
832  *
833  * The LL (low-low) area contains the original image downsampled by the amount
834  * of levels. The rest of the areas can be thought as the details needed
835  * to restore the image perfectly to its original size.
836  */
837 static int dwt_plane(AVCodecContext *avctx, void *arg)
838 {
839  TransformArgs *transform_dat = arg;
840  VC2EncContext *s = transform_dat->ctx;
841  const void *frame_data = transform_dat->idata;
842  const ptrdiff_t linesize = transform_dat->istride;
843  const int field = transform_dat->field;
844  const Plane *p = transform_dat->plane;
845  VC2TransformContext *t = &transform_dat->t;
846  dwtcoef *buf = p->coef_buf;
847  const int idx = s->wavelet_idx;
848  const int skip = 1 + s->interlaced;
849 
850  int x, y, level, offset;
851  ptrdiff_t pix_stride = linesize >> (s->bpp - 1);
852 
853  if (field == 1) {
854  offset = 0;
855  pix_stride <<= 1;
856  } else if (field == 2) {
857  offset = pix_stride;
858  pix_stride <<= 1;
859  } else {
860  offset = 0;
861  }
862 
863  if (s->bpp == 1) {
864  const uint8_t *pix = (const uint8_t *)frame_data + offset;
865  for (y = 0; y < p->height*skip; y+=skip) {
866  for (x = 0; x < p->width; x++) {
867  buf[x] = pix[x] - s->diff_offset;
868  }
869  memset(&buf[x], 0, (p->coef_stride - p->width)*sizeof(dwtcoef));
870  buf += p->coef_stride;
871  pix += pix_stride;
872  }
873  } else {
874  const uint16_t *pix = (const uint16_t *)frame_data + offset;
875  for (y = 0; y < p->height*skip; y+=skip) {
876  for (x = 0; x < p->width; x++) {
877  buf[x] = pix[x] - s->diff_offset;
878  }
879  memset(&buf[x], 0, (p->coef_stride - p->width)*sizeof(dwtcoef));
880  buf += p->coef_stride;
881  pix += pix_stride;
882  }
883  }
884 
885  memset(buf, 0, p->coef_stride * (p->dwt_height - p->height) * sizeof(dwtcoef));
886 
887  for (level = s->wavelet_depth-1; level >= 0; level--) {
888  const SubBand *b = &p->band[level][0];
889  t->vc2_subband_dwt[idx](t, p->coef_buf, p->coef_stride,
890  b->width, b->height);
891  }
892 
893  return 0;
894 }
895 
896 static int encode_frame(VC2EncContext *s, AVPacket *avpkt, const AVFrame *frame,
897  const char *aux_data, const int header_size, int field)
898 {
899  int i, ret;
900  int64_t max_frame_bytes;
901 
902  /* Threaded DWT transform */
903  for (i = 0; i < 3; i++) {
904  s->transform_args[i].ctx = s;
905  s->transform_args[i].field = field;
906  s->transform_args[i].plane = &s->plane[i];
907  s->transform_args[i].idata = frame->data[i];
908  s->transform_args[i].istride = frame->linesize[i];
909  }
911  sizeof(TransformArgs));
912 
913  /* Calculate per-slice quantizers and sizes */
914  max_frame_bytes = header_size + calc_slice_sizes(s);
915 
916  if (field < 2) {
917  ret = ff_alloc_packet2(s->avctx, avpkt,
918  max_frame_bytes << s->interlaced,
919  max_frame_bytes << s->interlaced);
920  if (ret) {
921  av_log(s->avctx, AV_LOG_ERROR, "Error getting output packet.\n");
922  return ret;
923  }
924  init_put_bits(&s->pb, avpkt->data, avpkt->size);
925  }
926 
927  /* Sequence header */
930 
931  /* Encoder version */
932  if (aux_data) {
934  avpriv_put_string(&s->pb, aux_data, 1);
935  }
936 
937  /* Picture header */
940 
941  /* Encode slices */
942  encode_slices(s);
943 
944  /* End sequence */
946 
947  return 0;
948 }
949 
951  const AVFrame *frame, int *got_packet)
952 {
953  int ret = 0;
954  int slice_ceil, sig_size = 256;
955  VC2EncContext *s = avctx->priv_data;
956  const int bitexact = avctx->flags & AV_CODEC_FLAG_BITEXACT;
957  const char *aux_data = bitexact ? "Lavc" : LIBAVCODEC_IDENT;
958  const int aux_data_size = bitexact ? sizeof("Lavc") : sizeof(LIBAVCODEC_IDENT);
959  const int header_size = 100 + aux_data_size;
960  int64_t r_bitrate = avctx->bit_rate >> (s->interlaced);
961 
962  s->avctx = avctx;
963  s->size_scaler = 2;
964  s->prefix_bytes = 0;
965  s->last_parse_code = 0;
966  s->next_parse_offset = 0;
967 
968  /* Rate control */
969  s->frame_max_bytes = (av_rescale(r_bitrate, s->avctx->time_base.num,
970  s->avctx->time_base.den) >> 3) - header_size;
971  s->slice_max_bytes = slice_ceil = av_rescale(s->frame_max_bytes, 1, s->num_x*s->num_y);
972 
973  /* Find an appropriate size scaler */
974  while (sig_size > 255) {
975  int r_size = SSIZE_ROUND(s->slice_max_bytes);
976  if (r_size > slice_ceil) {
977  s->slice_max_bytes -= r_size - slice_ceil;
978  r_size = SSIZE_ROUND(s->slice_max_bytes);
979  }
980  sig_size = r_size/s->size_scaler; /* Signalled slize size */
981  s->size_scaler <<= 1;
982  }
983 
984  s->slice_min_bytes = s->slice_max_bytes - s->slice_max_bytes*(s->tolerance/100.0f);
985 
986  ret = encode_frame(s, avpkt, frame, aux_data, header_size, s->interlaced);
987  if (ret)
988  return ret;
989  if (s->interlaced) {
990  ret = encode_frame(s, avpkt, frame, aux_data, header_size, 2);
991  if (ret)
992  return ret;
993  }
994 
995  flush_put_bits(&s->pb);
996  avpkt->size = put_bits_count(&s->pb) >> 3;
997 
998  *got_packet = 1;
999 
1000  return 0;
1001 }
1002 
1004 {
1005  int i;
1006  VC2EncContext *s = avctx->priv_data;
1007 
1008  av_log(avctx, AV_LOG_INFO, "Qavg: %i\n", s->q_avg);
1009 
1010  for (i = 0; i < 3; i++) {
1012  av_freep(&s->plane[i].coef_buf);
1013  }
1014 
1015  av_freep(&s->slice_args);
1016 
1017  return 0;
1018 }
1019 
1021 {
1022  Plane *p;
1023  SubBand *b;
1024  int i, level, o, shift, ret;
1025  const AVPixFmtDescriptor *fmt = av_pix_fmt_desc_get(avctx->pix_fmt);
1026  const int depth = fmt->comp[0].depth;
1027  VC2EncContext *s = avctx->priv_data;
1028 
1029  s->picture_number = 0;
1030 
1031  /* Total allowed quantization range */
1033 
1034  s->ver.major = 2;
1035  s->ver.minor = 0;
1036  s->profile = 3;
1037  s->level = 3;
1038 
1039  s->base_vf = -1;
1040  s->strict_compliance = 1;
1041 
1042  s->q_avg = 0;
1043  s->slice_max_bytes = 0;
1044  s->slice_min_bytes = 0;
1045 
1046  /* Mark unknown as progressive */
1047  s->interlaced = !((avctx->field_order == AV_FIELD_UNKNOWN) ||
1048  (avctx->field_order == AV_FIELD_PROGRESSIVE));
1049 
1050  for (i = 0; i < base_video_fmts_len; i++) {
1051  const VC2BaseVideoFormat *fmt = &base_video_fmts[i];
1052  if (avctx->pix_fmt != fmt->pix_fmt)
1053  continue;
1054  if (avctx->time_base.num != fmt->time_base.num)
1055  continue;
1056  if (avctx->time_base.den != fmt->time_base.den)
1057  continue;
1058  if (avctx->width != fmt->width)
1059  continue;
1060  if (avctx->height != fmt->height)
1061  continue;
1062  if (s->interlaced != fmt->interlaced)
1063  continue;
1064  s->base_vf = i;
1065  s->level = base_video_fmts[i].level;
1066  break;
1067  }
1068 
1069  if (s->interlaced)
1070  av_log(avctx, AV_LOG_WARNING, "Interlacing enabled!\n");
1071 
1072  if ((s->slice_width & (s->slice_width - 1)) ||
1073  (s->slice_height & (s->slice_height - 1))) {
1074  av_log(avctx, AV_LOG_ERROR, "Slice size is not a power of two!\n");
1075  return AVERROR_UNKNOWN;
1076  }
1077 
1078  if ((s->slice_width > avctx->width) ||
1079  (s->slice_height > avctx->height)) {
1080  av_log(avctx, AV_LOG_ERROR, "Slice size is bigger than the image!\n");
1081  return AVERROR_UNKNOWN;
1082  }
1083 
1084  if (s->base_vf <= 0) {
1086  s->strict_compliance = s->base_vf = 0;
1087  av_log(avctx, AV_LOG_WARNING, "Format does not strictly comply with VC2 specs\n");
1088  } else {
1089  av_log(avctx, AV_LOG_WARNING, "Given format does not strictly comply with "
1090  "the specifications, decrease strictness to use it.\n");
1091  return AVERROR_UNKNOWN;
1092  }
1093  } else {
1094  av_log(avctx, AV_LOG_INFO, "Selected base video format = %i (%s)\n",
1095  s->base_vf, base_video_fmts[s->base_vf].name);
1096  }
1097 
1098  /* Chroma subsampling */
1100  if (ret)
1101  return ret;
1102 
1103  /* Bit depth and color range index */
1104  if (depth == 8 && avctx->color_range == AVCOL_RANGE_JPEG) {
1105  s->bpp = 1;
1106  s->bpp_idx = 1;
1107  s->diff_offset = 128;
1108  } else if (depth == 8 && (avctx->color_range == AVCOL_RANGE_MPEG ||
1109  avctx->color_range == AVCOL_RANGE_UNSPECIFIED)) {
1110  s->bpp = 1;
1111  s->bpp_idx = 2;
1112  s->diff_offset = 128;
1113  } else if (depth == 10) {
1114  s->bpp = 2;
1115  s->bpp_idx = 3;
1116  s->diff_offset = 512;
1117  } else {
1118  s->bpp = 2;
1119  s->bpp_idx = 4;
1120  s->diff_offset = 2048;
1121  }
1122 
1123  /* Planes initialization */
1124  for (i = 0; i < 3; i++) {
1125  int w, h;
1126  p = &s->plane[i];
1127  p->width = avctx->width >> (i ? s->chroma_x_shift : 0);
1128  p->height = avctx->height >> (i ? s->chroma_y_shift : 0);
1129  if (s->interlaced)
1130  p->height >>= 1;
1131  p->dwt_width = w = FFALIGN(p->width, (1 << s->wavelet_depth));
1132  p->dwt_height = h = FFALIGN(p->height, (1 << s->wavelet_depth));
1133  p->coef_stride = FFALIGN(p->dwt_width, 32);
1134  p->coef_buf = av_mallocz(p->coef_stride*p->dwt_height*sizeof(dwtcoef));
1135  if (!p->coef_buf)
1136  goto alloc_fail;
1137  for (level = s->wavelet_depth-1; level >= 0; level--) {
1138  w = w >> 1;
1139  h = h >> 1;
1140  for (o = 0; o < 4; o++) {
1141  b = &p->band[level][o];
1142  b->width = w;
1143  b->height = h;
1144  b->stride = p->coef_stride;
1145  shift = (o > 1)*b->height*b->stride + (o & 1)*b->width;
1146  b->buf = p->coef_buf + shift;
1147  }
1148  }
1149 
1150  /* DWT init */
1152  s->plane[i].coef_stride,
1153  s->plane[i].dwt_height,
1154  s->slice_width, s->slice_height))
1155  goto alloc_fail;
1156  }
1157 
1158  /* Slices */
1159  s->num_x = s->plane[0].dwt_width/s->slice_width;
1160  s->num_y = s->plane[0].dwt_height/s->slice_height;
1161 
1162  s->slice_args = av_calloc(s->num_x*s->num_y, sizeof(SliceArgs));
1163  if (!s->slice_args)
1164  goto alloc_fail;
1165 
1166  for (i = 0; i < 116; i++) {
1167  const uint64_t qf = ff_dirac_qscale_tab[i];
1168  const uint32_t m = av_log2(qf);
1169  const uint32_t t = (1ULL << (m + 32)) / qf;
1170  const uint32_t r = (t*qf + qf) & UINT32_MAX;
1171  if (!(qf & (qf - 1))) {
1172  s->qmagic_lut[i][0] = 0xFFFFFFFF;
1173  s->qmagic_lut[i][1] = 0xFFFFFFFF;
1174  } else if (r <= 1 << m) {
1175  s->qmagic_lut[i][0] = t + 1;
1176  s->qmagic_lut[i][1] = 0;
1177  } else {
1178  s->qmagic_lut[i][0] = t;
1179  s->qmagic_lut[i][1] = t;
1180  }
1181  }
1182 
1183  return 0;
1184 
1185 alloc_fail:
1186  vc2_encode_end(avctx);
1187  av_log(avctx, AV_LOG_ERROR, "Unable to allocate memory!\n");
1188  return AVERROR(ENOMEM);
1189 }
1190 
1191 #define VC2ENC_FLAGS (AV_OPT_FLAG_ENCODING_PARAM | AV_OPT_FLAG_VIDEO_PARAM)
1192 static const AVOption vc2enc_options[] = {
1193  {"tolerance", "Max undershoot in percent", offsetof(VC2EncContext, tolerance), AV_OPT_TYPE_DOUBLE, {.dbl = 5.0f}, 0.0f, 45.0f, VC2ENC_FLAGS, "tolerance"},
1194  {"slice_width", "Slice width", offsetof(VC2EncContext, slice_width), AV_OPT_TYPE_INT, {.i64 = 32}, 32, 1024, VC2ENC_FLAGS, "slice_width"},
1195  {"slice_height", "Slice height", offsetof(VC2EncContext, slice_height), AV_OPT_TYPE_INT, {.i64 = 16}, 8, 1024, VC2ENC_FLAGS, "slice_height"},
1196  {"wavelet_depth", "Transform depth", offsetof(VC2EncContext, wavelet_depth), AV_OPT_TYPE_INT, {.i64 = 4}, 1, 5, VC2ENC_FLAGS, "wavelet_depth"},
1197  {"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"},
1198  {"9_7", "Deslauriers-Dubuc (9,7)", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_TRANSFORM_9_7}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "wavelet_idx"},
1199  {"5_3", "LeGall (5,3)", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_TRANSFORM_5_3}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "wavelet_idx"},
1200  {"haar", "Haar (with shift)", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_TRANSFORM_HAAR_S}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "wavelet_idx"},
1201  {"haar_noshift", "Haar (without shift)", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_TRANSFORM_HAAR}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "wavelet_idx"},
1202  {"qm", "Custom quantization matrix", offsetof(VC2EncContext, quant_matrix), AV_OPT_TYPE_INT, {.i64 = VC2_QM_DEF}, 0, VC2_QM_NB, VC2ENC_FLAGS, "quant_matrix"},
1203  {"default", "Default from the specifications", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_QM_DEF}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "quant_matrix"},
1204  {"color", "Prevents low bitrate discoloration", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_QM_COL}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "quant_matrix"},
1205  {"flat", "Optimize for PSNR", 0, AV_OPT_TYPE_CONST, {.i64 = VC2_QM_FLAT}, INT_MIN, INT_MAX, VC2ENC_FLAGS, "quant_matrix"},
1206  {NULL}
1207 };
1208 
1209 static const AVClass vc2enc_class = {
1210  .class_name = "SMPTE VC-2 encoder",
1211  .category = AV_CLASS_CATEGORY_ENCODER,
1212  .option = vc2enc_options,
1213  .item_name = av_default_item_name,
1214  .version = LIBAVUTIL_VERSION_INT
1215 };
1216 
1218  { "b", "600000000" },
1219  { NULL },
1220 };
1221 
1222 static const enum AVPixelFormat allowed_pix_fmts[] = {
1227 };
1228 
1230  .name = "vc2",
1231  .long_name = NULL_IF_CONFIG_SMALL("SMPTE VC-2"),
1232  .type = AVMEDIA_TYPE_VIDEO,
1233  .id = AV_CODEC_ID_DIRAC,
1234  .priv_data_size = sizeof(VC2EncContext),
1235  .init = vc2_encode_init,
1236  .close = vc2_encode_end,
1237  .capabilities = AV_CODEC_CAP_SLICE_THREADS,
1238  .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE,
1239  .encode2 = vc2_encode_frame,
1240  .priv_class = &vc2enc_class,
1241  .defaults = vc2enc_defaults,
1243 };
int strict_compliance
Definition: vc2enc.c:172
static void encode_parse_info(VC2EncContext *s, enum DiracParseCodes pcode)
Definition: vc2enc.c:228
static void av_unused put_bits32(PutBitContext *s, uint32_t value)
Write exactly 32 bits into a bitstream.
Definition: put_bits.h:268
#define NULL
Definition: coverity.c:32
const int32_t ff_dirac_qscale_tab[116]
Definition: diractab.c:34
static int encode_frame(VC2EncContext *s, AVPacket *avpkt, const AVFrame *frame, const char *aux_data, const int header_size, int field)
Definition: vc2enc.c:896
static void encode_wavelet_transform(VC2EncContext *s)
Definition: vc2enc.c:509
static void encode_aspect_ratio(VC2EncContext *s)
Definition: vc2enc.c:316
static int shift(int a, int b)
Definition: sonic.c:82
static av_cold int vc2_encode_init(AVCodecContext *avctx)
Definition: vc2enc.c:1020
const AVPixFmtDescriptor * av_pix_fmt_desc_get(enum AVPixelFormat pix_fmt)
Definition: pixdesc.c:2573
This structure describes decoded (raw) audio or video data.
Definition: frame.h:308
av_cold void ff_vc2enc_free_transforms(VC2TransformContext *s)
Definition: vc2enc_dwt.c:277
int base_vf
Definition: vc2enc.c:143
AVOption.
Definition: opt.h:248
int slice_height
Definition: vc2enc.c:173
"Linear transfer characteristics"
Definition: pixfmt.h:492
TransformArgs transform_args[3]
Definition: vc2enc.c:132
uint32_t next_parse_offset
Definition: vc2enc.c:179
planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples)
Definition: pixfmt.h:71
static void put_bits(Jpeg2000EncoderContext *s, int val, int n)
put n times val bit
Definition: j2kenc.c:218
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:200
int64_t bit_rate
the average bitrate
Definition: avcodec.h:576
#define LIBAVUTIL_VERSION_INT
Definition: version.h:85
static void encode_source_params(VC2EncContext *s)
Definition: vc2enc.c:387
static av_cold int init(AVCodecContext *avctx)
Definition: avrndec.c:35
void * ctx
Definition: vc2enc.c:106
also ITU-R BT601-6 625 / ITU-R BT1358 625 / ITU-R BT1700 625 PAL & SECAM / IEC 61966-2-4 xvYCC601 ...
Definition: pixfmt.h:518
int wavelet_depth
Definition: vc2enc.c:171
static av_always_inline void put_vc2_ue_uint(PutBitContext *pb, uint32_t val)
Definition: vc2enc.c:183
enum AVColorRange color_range
MPEG vs JPEG YUV range.
Definition: avcodec.h:1161
int num
Numerator.
Definition: rational.h:59
int size
Definition: packet.h:364
const char * av_default_item_name(void *ptr)
Return the context name.
Definition: log.c:235
int av_log2(unsigned v)
Definition: intmath.c:26
void avpriv_align_put_bits(PutBitContext *s)
Pad the bitstream with zeros up to the next byte boundary.
Definition: bitstream.c:48
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:905
enum AVPixelFormat pix_fmt
Pixel format, see AV_PIX_FMT_xxx.
Definition: avcodec.h:736
#define AV_PIX_FMT_YUV420P12
Definition: pixfmt.h:403
uint32_t picture_number
Definition: vc2enc.c:140
void * av_mallocz(size_t size)
Allocate a memory block with alignment suitable for all memory accesses (including vectors if availab...
Definition: mem.c:237
ptrdiff_t stride
Definition: cfhd.h:112
AVCodec.
Definition: codec.h:190
order of coefficients is actually GBR, also IEC 61966-2-1 (sRGB)
Definition: pixfmt.h:513
AVCodec ff_vc2_encoder
Definition: vc2enc.c:1229
VC2_QM
Definition: vc2enc.c:78
void * av_calloc(size_t nmemb, size_t size)
Non-inlined equivalent of av_mallocz_array().
Definition: mem.c:245
AVRational time_base
This is the fundamental unit of time (in seconds) in terms of which frame timestamps are represented...
Definition: avcodec.h:649
static av_cold int vc2_encode_end(AVCodecContext *avctx)
Definition: vc2enc.c:1003
int width
Definition: cfhd.h:114
#define VC2ENC_FLAGS
Definition: vc2enc.c:1191
const uint8_t vc2_qm_flat_tab[][4]
Definition: vc2enc.c:434
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:72
static void encode_transform_params(VC2EncContext *s)
Definition: vc2enc.c:499
int ff_alloc_packet2(AVCodecContext *avctx, AVPacket *avpkt, int64_t size, int64_t min_size)
Check AVPacket size and/or allocate data.
Definition: encode.c:33
#define FF_CODEC_CAP_INIT_THREADSAFE
The codec does not modify any global variables in the init function, allowing to call the init functi...
Definition: internal.h:40
AVComponentDescriptor comp[4]
Parameters that describe how pixels are packed.
Definition: pixdesc.h:117
uint8_t
#define av_cold
Definition: attributes.h:88
int interlaced
Definition: vc2enc.c:175
enum DiracParseCodes last_parse_code
Definition: vc2enc.c:180
Interface to Dirac Decoder/Encoder.
AVOptions.
dwtcoef * coef_buf
Definition: vc2enc.c:95
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
VC2TransformContext t
Definition: vc2enc.c:121
uint8_t * data
Definition: packet.h:363
int bytes
Definition: vc2enc.c:112
int size_scaler
Definition: vc2enc.c:157
static av_always_inline int count_vc2_ue_uint(uint32_t val)
Definition: vc2enc.c:211
static const int base_video_fmts_len
Definition: vc2enc.c:76
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
#define AV_PIX_FMT_YUV422P12
Definition: pixfmt.h:404
int x
Definition: vc2enc.c:107
static av_cold int vc2_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, const AVFrame *frame, int *got_packet)
Definition: vc2enc.c:950
#define FFALIGN(x, a)
Definition: macros.h:48
#define av_log(a,...)
ptrdiff_t coef_stride
Definition: vc2enc.c:100
Definition: cfhd.h:111
int slice_min_bytes
Definition: vc2enc.c:164
static void init_quant_matrix(VC2EncContext *s)
Definition: vc2enc.c:442
Libavcodec version macros.
static void encode_clean_area(VC2EncContext *s)
Definition: vc2enc.c:328
const uint8_t vc2_qm_col_tab[][4]
Definition: vc2enc.c:426
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:194
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:347
ITU-R BT1361 Extended Colour Gamut.
Definition: pixfmt.h:496
SubBand band[DWT_LEVELS_3D][4]
Definition: cfhd.h:132
int bpp_idx
Definition: vc2enc.c:137
static int count_hq_slice(SliceArgs *slice, int quant_idx)
Definition: vc2enc.c:553
PutBitContext pb
Definition: vc2enc.c:126
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:2601
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
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:153
const char * r
Definition: vf_curves.c:114
static void encode_parse_params(VC2EncContext *s)
Definition: vc2enc.c:260
const char * arg
Definition: jacosubdec.c:66
const uint8_t ff_dirac_default_qmat[7][4][4]
Definition: diractab.c:24
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:606
uint8_t * buf
Definition: put_bits.h:52
const char * name
Name of the codec implementation.
Definition: codec.h:197
#define AV_PIX_FMT_YUV444P10
Definition: pixfmt.h:402
static const AVCodecDefault vc2enc_defaults[]
Definition: vc2enc.c:1217
static const AVCodecDefault defaults[]
Definition: amfenc_h264.c:361
uint8_t bits
Definition: vp3data.h:202
#define DIRAC_MAX_QUANT_INDEX
Definition: diractab.h:41
#define FFMAX(a, b)
Definition: common.h:94
static int put_bits_count(PutBitContext *s)
Definition: put_bits.h:81
planar YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples)
Definition: pixfmt.h:70
#define SSIZE_ROUND(b)
Definition: vc2enc.c:33
static void encode_seq_header(VC2EncContext *s)
Definition: vc2enc.c:400
int chroma_x_shift
Definition: vc2enc.c:158
int profile
Definition: vc2enc.c:145
#define b
Definition: input.c:41
static void skip_put_bytes(PutBitContext *s, int n)
Skip the given number of bytes.
Definition: put_bits.h:356
int dwt_height
Definition: vc2enc.c:99
#define AV_CODEC_FLAG_BITEXACT
Use only bitexact stuff (except (I)DCT).
Definition: avcodec.h:333
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
static void encode_scan_format(VC2EncContext *s)
Definition: vc2enc.c:296
#define FFMIN(a, b)
Definition: common.h:96
#define QUANT(c, mul, add, shift)
Definition: vc2enc.c:524
int diff_offset
Definition: vc2enc.c:135
int custom_quant_matrix
Definition: vc2enc.c:149
int width
picture width / height.
Definition: avcodec.h:699
uint8_t w
Definition: llviddspenc.c:38
int quant_idx
Definition: vc2enc.c:109
also ITU-R BT601-6 625 / ITU-R BT1358 625 / ITU-R BT1700 625 PAL & SECAM
Definition: pixfmt.h:465
enum AVColorPrimaries color_primaries
Chromaticity coordinates of the source primaries.
Definition: avcodec.h:1140
int bits_ceil
Definition: vc2enc.c:110
static void encode_frame_size(VC2EncContext *s)
Definition: vc2enc.c:269
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
#define FFABS(a)
Absolute value, Note, INT_MIN / INT64_MIN result in undefined behavior as they are not representable ...
Definition: common.h:72
#define s(width, name)
Definition: cbs_vp9.c:257
static void encode_quant_matrix(VC2EncContext *s)
Definition: vc2enc.c:484
void * ctx
Definition: vc2enc.c:116
void * idata
Definition: vc2enc.c:118
AVCodecContext * avctx
Definition: vc2enc.c:128
#define FF_ARRAY_ELEMS(a)
the normal 2^n-1 "JPEG" YUV ranges
Definition: pixfmt.h:538
#define SLICE_REDIST_TOTAL
Definition: vc2enc.c:36
static void encode_slice_params(VC2EncContext *s)
Definition: vc2enc.c:417
static void encode_sample_fmt(VC2EncContext *s)
Definition: vc2enc.c:280
#define AV_CODEC_CAP_SLICE_THREADS
Codec supports slice-based (or partition-based) multithreading.
Definition: codec.h:110
uint8_t quant[MAX_DWT_LEVELS][4]
Definition: vc2enc.c:148
int q_ceil
Definition: vc2enc.c:165
static void encode_frame_rate(VC2EncContext *s)
Definition: vc2enc.c:304
int slice_width
Definition: vc2enc.c:174
static void encode_picture_header(VC2EncContext *s)
Definition: vc2enc.c:410
#define AV_LOG_INFO
Standard information.
Definition: log.h:205
static void encode_signal_range(VC2EncContext *s)
Definition: vc2enc.c:334
static int calc_slice_sizes(VC2EncContext *s)
Definition: vc2enc.c:643
static int dwt_plane(AVCodecContext *avctx, void *arg)
Definition: vc2enc.c:837
functionally identical to above
Definition: pixfmt.h:467
#define ff_log2
Definition: intmath.h:50
int linesize[AV_NUM_DATA_POINTERS]
For video, size in bytes of each picture line.
Definition: frame.h:339
Descriptor that unambiguously describes how the bits of a pixel are stored in the up to 4 data planes...
Definition: pixdesc.h:81
AVClass * av_class
Definition: vc2enc.c:125
main external API structure.
Definition: avcodec.h:526
DiracParseCodes
Parse code values:
Definition: dirac.h:57
static void encode_color_spec(VC2EncContext *s)
Definition: vc2enc.c:342
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:206
static int encode_hq_slice(AVCodecContext *avctx, void *arg)
Definition: vc2enc.c:721
#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
static int encode_slices(VC2EncContext *s)
Definition: vc2enc.c:777
#define AV_PIX_FMT_YUV420P10
Definition: pixfmt.h:399
Describe the class of an AVClass context structure.
Definition: log.h:67
#define AV_WB32(p, v)
Definition: intreadwrite.h:419
void(* vc2_subband_dwt[VC2_TRANSFORMS_NB])(struct VC2TransformContext *t, dwtcoef *data, ptrdiff_t stride, int width, int height)
Definition: vc2enc_dwt.h:45
enum AVColorSpace colorspace
YUV colorspace type.
Definition: avcodec.h:1154
Rational number (pair of numerator and denominator).
Definition: rational.h:58
enum AVColorTransferCharacteristic color_trc
Color Transfer Characteristic.
Definition: avcodec.h:1147
double tolerance
Definition: vc2enc.c:169
int frame_max_bytes
Definition: vc2enc.c:162
SliceArgs * slice_args
Definition: vc2enc.c:131
#define FF_COMPLIANCE_STRICT
Strictly conform to all the things in the spec no matter what consequences.
Definition: avcodec.h:1591
Plane * plane
Definition: vc2enc.c:117
static void encode_subband(VC2EncContext *s, PutBitContext *pb, int sx, int sy, SubBand *b, int quant)
Definition: vc2enc.c:527
int cache[DIRAC_MAX_QUANT_INDEX]
Definition: vc2enc.c:105
static enum AVPixelFormat pix_fmts[]
Definition: libkvazaar.c:300
const uint8_t * quant
ptrdiff_t istride
Definition: vc2enc.c:119
Plane plane[3]
Definition: vc2enc.c:127
#define AV_PIX_FMT_YUV422P10
Definition: pixfmt.h:400
#define AV_PIX_FMT_YUV444P12
Definition: pixfmt.h:406
int chroma_y_shift
Definition: vc2enc.c:159
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:322
static enum AVPixelFormat allowed_pix_fmts[]
Definition: vc2enc.c:1222
int dwt_width
Definition: vc2enc.c:98
the normal 219*2^(n-8) "MPEG" YUV ranges
Definition: pixfmt.h:537
DiracVersionInfo ver
Definition: vc2enc.c:129
int wavelet_idx
Definition: vc2enc.c:170
int slice_max_bytes
Definition: vc2enc.c:163
static const VC2BaseVideoFormat base_video_fmts[]
Definition: vc2enc.c:45
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Definition: pixfmt.h:66
common internal api header.
static void flush_put_bits(PutBitContext *s)
Pad the end of the output stream with zeros.
Definition: put_bits.h:115
static const AVOption vc2enc_options[]
Definition: vc2enc.c:1192
static void init_put_bits(PutBitContext *s, uint8_t *buffer, int buffer_size)
Initialize the PutBitContext s.
Definition: put_bits.h:62
int den
Denominator.
Definition: rational.h:60
#define AVERROR_UNKNOWN
Unknown error, typically from an external library.
Definition: error.h:71
dwtcoef * buf
Definition: vc2enc.c:87
int32_t dwtcoef
Definition: vc2enc_dwt.h:28
int y
Definition: vc2enc.c:108
void * priv_data
Definition: avcodec.h:553
static av_always_inline int diff(const uint32_t a, const uint32_t b)
static int rate_control(AVCodecContext *avctx, void *arg)
Definition: vc2enc.c:615
uint32_t qmagic_lut[116][2]
Definition: vc2enc.c:152
int height
Definition: cfhd.h:116
static const double coeff[2][5]
Definition: vf_owdenoise.c:72
static void encode_picture_start(VC2EncContext *s)
Definition: vc2enc.c:516
int width
Definition: cfhd.h:120
#define LIBAVCODEC_IDENT
Definition: version.h:42
void avpriv_put_string(PutBitContext *pb, const char *string, int terminate_string)
Put the string string in the bitstream.
Definition: bitstream.c:53
enum AVPixelFormat pix_fmt
Definition: vc2enc.c:39
also ITU-R BT601-6 525 / ITU-R BT1358 525 / ITU-R BT1700 NTSC
Definition: pixfmt.h:466
DWTELEM * buf
Definition: snow.h:89
#define av_freep(p)
enum AVFieldOrder field_order
Field order.
Definition: avcodec.h:1183
int prefix_bytes
Definition: vc2enc.c:156
#define av_always_inline
Definition: attributes.h:45
#define stride
int height
Definition: cfhd.h:121
AVRational time_base
Definition: vc2enc.c:40
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
int depth
Number of bits in the component.
Definition: pixdesc.h:58
int(* execute)(struct AVCodecContext *c, int(*func)(struct AVCodecContext *c2, void *arg), void *arg2, int *ret, int count, int size)
The codec may call this to execute several independent things.
Definition: avcodec.h:1825
AVPixelFormat
Pixel format.
Definition: pixfmt.h:64
static double val(void *priv, double ch)
Definition: aeval.c:76
This structure stores compressed data.
Definition: packet.h:340
enum VC2_QM quant_matrix
Definition: vc2enc.c:176
int strict_std_compliance
strictly follow the standard (MPEG-4, ...).
Definition: avcodec.h:1589
static const AVClass vc2enc_class
Definition: vc2enc.c:1209
int i
Definition: input.c:407
Definition: cfhd.h:119
int bits_floor
Definition: vc2enc.c:111
const char * name
Definition: vc2enc.c:42
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
PutBitContext pb
Definition: vc2enc.c:104
bitstream writer API