This implements the limited DM metadata compression scheme described in chapter 9 of the dolby vision bitstream specification. The spec is a bit unclear about how to handle the presence of static metadata inside compressed frames; in that it doesn't explicitly forbid an encoder from repeating redundant metadata. In theory, we would need to detect this case and then strip the corresponding duplicate metadata from the existing set of static metadata. However, this is difficult to implement - esspecially for the case of metadata blocks which may be internally repeated (e.g. level 10). That said, the spec states outright that static metadata should be constant throughout the entire sequence, so a sane bitstream should not have any static metadata values changing from one frame to the next (at least up to a keyframe boundary), and therefore they should never be present in compressed frames. As a consequence, it makes sense to treat this as an error state regardless. (Ignoring them by default, or erroring if either AV_EF_EXPLODE or AV_EF_AGGRESSIVE are set) I was not able to find such samples in the wild (outside of artificially produced test cases for this exact scenario), so I don't think we need to worry about it until somebody produces one.
Unified · Split
| @@ -281,7 +281,8 @@ static int parse_ext_v2(DOVIContext *s, GetBitContext *gb, AVDOVIDmData *dm, · 280 unmodified lines | |||
| 281 | return 0; | 281 | return 0; |
| 282 | } | 282 | } |
| 283 | 283 | ||
| 284 | static int parse_ext_blocks(DOVIContext *s, GetBitContext *gb, int ver) | 284 | static int parse_ext_blocks(DOVIContext *s, GetBitContext *gb, int ver, |
| 285 | int compression, int err_recognition) | ||
| 285 | { | 286 | { |
| 286 | int num_ext_blocks, ext_block_length, start_pos, parsed_bits, ret; | 287 | int num_ext_blocks, ext_block_length, start_pos, parsed_bits, ret; |
| 287 | DOVIExt *ext = s->ext_blocks; | 288 | DOVIExt *ext = s->ext_blocks; |
| @@ -296,6 +297,7 @@ static int parse_ext_blocks(DOVIContext *s, GetBitContext *gb, int ver) · 8 unmodified lines | |||
| 296 | } | 297 | } |
| 297 | 298 | ||
| 298 | while (num_ext_blocks--) { | 299 | while (num_ext_blocks--) { |
| 300 | AVDOVIDmData dummy; | ||
| 299 | AVDOVIDmData *dm; | 301 | AVDOVIDmData *dm; |
| 300 | uint8_t level; | 302 | uint8_t level; |
| 301 | 303 | ||
| @@ -304,9 +306,17 @@ static int parse_ext_blocks(DOVIContext *s, GetBitContext *gb, int ver) · 2 unmodified lines | |||
| 304 | start_pos = get_bits_count(gb); | 306 | start_pos = get_bits_count(gb); |
| 305 | 307 | ||
| 306 | if (ff_dovi_rpu_extension_is_static(level)) { | 308 | if (ff_dovi_rpu_extension_is_static(level)) { |
| 307 | if (ext->num_static >= FF_ARRAY_ELEMS(ext->dm_static)) | 309 | if (compression) { |
| 308 | return AVERROR_INVALIDDATA; | 310 | av_log(s->logctx, AV_LOG_WARNING, "Compressed DM RPU contains " |
| 309 | dm = &ext->dm_static[ext->num_static++]; | 311 | "static extension block level %d\n", level); |
| 312 | if (err_recognition & (AV_EF_AGGRESSIVE | AV_EF_EXPLODE)) | ||
| 313 | return AVERROR_INVALIDDATA; | ||
| 314 | dm = &dummy; | ||
| 315 | } else { | ||
| 316 | if (ext->num_static >= FF_ARRAY_ELEMS(ext->dm_static)) | ||
| 317 | return AVERROR_INVALIDDATA; | ||
| 318 | dm = &ext->dm_static[ext->num_static++]; | ||
| 319 | } | ||
| 310 | } else { | 320 | } else { |
| 311 | if (ext->num_dynamic >= FF_ARRAY_ELEMS(ext->dm_dynamic)) | 321 | if (ext->num_dynamic >= FF_ARRAY_ELEMS(ext->dm_dynamic)) |
| 312 | return AVERROR_INVALIDDATA; | 322 | return AVERROR_INVALIDDATA; |
| @@ -342,6 +352,7 @@ int ff_dovi_rpu_parse(DOVIContext *s, const uint8_t *rpu, size_t rpu_size, · 29 unmodified lines | |||
| 342 | uint8_t rpu_type; | 352 | uint8_t rpu_type; |
| 343 | uint8_t vdr_seq_info_present; | 353 | uint8_t vdr_seq_info_present; |
| 344 | uint8_t vdr_dm_metadata_present; | 354 | uint8_t vdr_dm_metadata_present; |
| 355 | uint8_t dm_compression = 0; | ||
| 345 | uint8_t use_prev_vdr_rpu; | 356 | uint8_t use_prev_vdr_rpu; |
| 346 | uint8_t use_nlq; | 357 | uint8_t use_nlq; |
| 347 | uint8_t profile; | 358 | uint8_t profile; |
| @@ -454,7 +465,6 @@ int ff_dovi_rpu_parse(DOVIContext *s, const uint8_t *rpu, size_t rpu_size, · 106 unmodified lines | |||
| 454 | int bl_bit_depth_minus8 = get_ue_golomb_31(gb); | 465 | int bl_bit_depth_minus8 = get_ue_golomb_31(gb); |
| 455 | int el_bit_depth_minus8 = get_ue_golomb_long(gb); | 466 | int el_bit_depth_minus8 = get_ue_golomb_long(gb); |
| 456 | int vdr_bit_depth_minus8 = get_ue_golomb_31(gb); | 467 | int vdr_bit_depth_minus8 = get_ue_golomb_31(gb); |
| 457 | int reserved_zero_3bits; | ||
| 458 | /* ext_mapping_idc is in the upper 8 bits of el_bit_depth_minus8 */ | 468 | /* ext_mapping_idc is in the upper 8 bits of el_bit_depth_minus8 */ |
| 459 | int ext_mapping_idc = el_bit_depth_minus8 >> 8; | 469 | int ext_mapping_idc = el_bit_depth_minus8 >> 8; |
| 460 | el_bit_depth_minus8 = el_bit_depth_minus8 & 0xFF; | 470 | el_bit_depth_minus8 = el_bit_depth_minus8 & 0xFF; |
| @@ -468,8 +478,7 @@ int ff_dovi_rpu_parse(DOVIContext *s, const uint8_t *rpu, size_t rpu_size, · 7 unmodified lines | |||
| 468 | hdr->ext_mapping_idc_5_7 = ext_mapping_idc >> 5; | 478 | hdr->ext_mapping_idc_5_7 = ext_mapping_idc >> 5; |
| 469 | hdr->vdr_bit_depth = vdr_bit_depth_minus8 + 8; | 479 | hdr->vdr_bit_depth = vdr_bit_depth_minus8 + 8; |
| 470 | hdr->spatial_resampling_filter_flag = get_bits1(gb); | 480 | hdr->spatial_resampling_filter_flag = get_bits1(gb); |
| 471 | reserved_zero_3bits = get_bits(gb, 3); | 481 | dm_compression = get_bits(gb, 3); |
| 472 | VALIDATE(reserved_zero_3bits, 0, 0); | ||
| 473 | hdr->el_spatial_resampling_filter_flag = get_bits1(gb); | 482 | hdr->el_spatial_resampling_filter_flag = get_bits1(gb); |
| 474 | hdr->disable_residual_flag = get_bits1(gb); | 483 | hdr->disable_residual_flag = get_bits1(gb); |
| 475 | } | 484 | } |
| @@ -481,6 +490,17 @@ int ff_dovi_rpu_parse(DOVIContext *s, const uint8_t *rpu, size_t rpu_size, · 5 unmodified lines | |||
| 481 | } | 490 | } |
| 482 | 491 | ||
| 483 | vdr_dm_metadata_present = get_bits1(gb); | 492 | vdr_dm_metadata_present = get_bits1(gb); |
| 493 | if (dm_compression > 1) { | ||
| 494 | /* It seems no device supports this */ | ||
| 495 | av_log(s->logctx, AV_LOG_ERROR, "Dynamic metadata compression is not " | ||
| 496 | "yet implemented"); | ||
| 497 | return AVERROR_PATCHWELCOME; | ||
| 498 | } else if (dm_compression && !vdr_dm_metadata_present) { | ||
| 499 | av_log(s->logctx, AV_LOG_ERROR, "Nonzero DM metadata compression method " | ||
| 500 | "but no DM metadata present"); | ||
| 501 | return AVERROR_INVALIDDATA; | ||
| 502 | } | ||
| 503 | |||
| 484 | use_prev_vdr_rpu = get_bits1(gb); | 504 | use_prev_vdr_rpu = get_bits1(gb); |
| 485 | use_nlq = (hdr->rpu_format & 0x700) == 0 && !hdr->disable_residual_flag; | 505 | use_nlq = (hdr->rpu_format & 0x700) == 0 && !hdr->disable_residual_flag; |
| 486 | 506 | ||
| @@ -503,6 +523,12 @@ int ff_dovi_rpu_parse(DOVIContext *s, const uint8_t *rpu, size_t rpu_size, · 16 unmodified lines | |||
| 503 | "use_prev_vdr_rpu=1\n"); | 523 | "use_prev_vdr_rpu=1\n"); |
| 504 | return AVERROR_INVALIDDATA; | 524 | return AVERROR_INVALIDDATA; |
| 505 | } | 525 | } |
| 526 | |||
| 527 | if (dm_compression && !compression) { | ||
| 528 | av_log(s->logctx, AV_LOG_ERROR, "Uncompressed RPUs should not use " | ||
| 529 | "dm_compression=%d\n", dm_compression); | ||
| 530 | return AVERROR_INVALIDDATA; | ||
| 531 | } | ||
| 506 | } | 532 | } |
| 507 | 533 | ||
| 508 | if (use_prev_vdr_rpu) { | 534 | if (use_prev_vdr_rpu) { |
| @@ -654,46 +680,50 @@ int ff_dovi_rpu_parse(DOVIContext *s, const uint8_t *rpu, size_t rpu_size, · 145 unmodified lines | |||
| 654 | s->color = color = s->dm; | 680 | s->color = color = s->dm; |
| 655 | color->dm_metadata_id = affected_dm_id; | 681 | color->dm_metadata_id = affected_dm_id; |
| 656 | color->scene_refresh_flag = get_ue_golomb_31(gb); | 682 | color->scene_refresh_flag = get_ue_golomb_31(gb); |
| 657 | for (int i = 0; i < 9; i++) | 683 | if (!dm_compression) { |
| 658 | color->ycc_to_rgb_matrix[i] = av_make_q(get_sbits(gb, 16), 1 << 13); | 684 | for (int i = 0; i < 9; i++) |
| 659 | for (int i = 0; i < 3; i++) { | 685 | color->ycc_to_rgb_matrix[i] = av_make_q(get_sbits(gb, 16), 1 << 13); |
| 660 | int denom = profile == 4 ? (1 << 30) : (1 << 28); | 686 | for (int i = 0; i < 3; i++) { |
| 661 | unsigned offset = get_bits_long(gb, 32); | 687 | int denom = profile == 4 ? (1 << 30) : (1 << 28); |
| 662 | if (offset > INT_MAX) { | 688 | unsigned offset = get_bits_long(gb, 32); |
| 663 | /* Ensure the result fits inside AVRational */ | 689 | if (offset > INT_MAX) { |
| 664 | offset >>= 1; | 690 | /* Ensure the result fits inside AVRational */ |
| 665 | denom >>= 1; | 691 | offset >>= 1; |
| 692 | denom >>= 1; | ||
| 693 | } | ||
| 694 | color->ycc_to_rgb_offset[i] = av_make_q(offset, denom); | ||
| 666 | } | 695 | } |
| 667 | color->ycc_to_rgb_offset[i] = av_make_q(offset, denom); | 696 | for (int i = 0; i < 9; i++) |
| 697 | color->rgb_to_lms_matrix[i] = av_make_q(get_sbits(gb, 16), 1 << 14); | ||
| 698 | |||
| 699 | color->signal_eotf = get_bits(gb, 16); | ||
| 700 | color->signal_eotf_param0 = get_bits(gb, 16); | ||
| 701 | color->signal_eotf_param1 = get_bits(gb, 16); | ||
| 702 | color->signal_eotf_param2 = get_bits_long(gb, 32); | ||
| 703 | color->signal_bit_depth = get_bits(gb, 5); | ||
| 704 | VALIDATE(color->signal_bit_depth, 8, 16); | ||
| 705 | color->signal_color_space = get_bits(gb, 2); | ||
| 706 | color->signal_chroma_format = get_bits(gb, 2); | ||
| 707 | color->signal_full_range_flag = get_bits(gb, 2); | ||
| 708 | color->source_min_pq = get_bits(gb, 12); | ||
| 709 | color->source_max_pq = get_bits(gb, 12); | ||
| 710 | color->source_diagonal = get_bits(gb, 10); | ||
| 668 | } | 711 | } |
| 669 | for (int i = 0; i < 9; i++) | ||
| 670 | color->rgb_to_lms_matrix[i] = av_make_q(get_sbits(gb, 16), 1 << 14); | ||
| 671 | |||
| 672 | color->signal_eotf = get_bits(gb, 16); | ||
| 673 | color->signal_eotf_param0 = get_bits(gb, 16); | ||
| 674 | color->signal_eotf_param1 = get_bits(gb, 16); | ||
| 675 | color->signal_eotf_param2 = get_bits_long(gb, 32); | ||
| 676 | color->signal_bit_depth = get_bits(gb, 5); | ||
| 677 | VALIDATE(color->signal_bit_depth, 8, 16); | ||
| 678 | color->signal_color_space = get_bits(gb, 2); | ||
| 679 | color->signal_chroma_format = get_bits(gb, 2); | ||
| 680 | color->signal_full_range_flag = get_bits(gb, 2); | ||
| 681 | color->source_min_pq = get_bits(gb, 12); | ||
| 682 | color->source_max_pq = get_bits(gb, 12); | ||
| 683 | color->source_diagonal = get_bits(gb, 10); | ||
| 684 | 712 | ||
| 685 | /* Parse extension blocks */ | 713 | /* Parse extension blocks */ |
| 686 | if (s->ext_blocks) { | 714 | if (s->ext_blocks) { |
| 687 | DOVIExt *ext = s->ext_blocks; | 715 | DOVIExt *ext = s->ext_blocks; |
| 688 | ext->num_static = ext->num_dynamic = 0; | 716 | if (!dm_compression) |
| 717 | ext->num_static = 0; | ||
| 718 | ext->num_dynamic = 0; | ||
| 689 | } | 719 | } |
| 690 | if ((ret = parse_ext_blocks(s, gb, 1)) < 0) { | 720 | if ((ret = parse_ext_blocks(s, gb, 1, dm_compression, err_recognition)) < 0) { |
| 691 | ff_dovi_ctx_unref(s); | 721 | ff_dovi_ctx_unref(s); |
| 692 | return ret; | 722 | return ret; |
| 693 | } | 723 | } |
| 694 | 724 | ||
| 695 | if (get_bits_left(gb) > 48 /* padding + CRC32 + terminator */) { | 725 | if (get_bits_left(gb) > 48 /* padding + CRC32 + terminator */) { |
| 696 | if ((ret = parse_ext_blocks(s, gb, 2)) < 0) { | 726 | if ((ret = parse_ext_blocks(s, gb, 2, dm_compression, err_recognition)) < 0) { |
| 697 | ff_dovi_ctx_unref(s); | 727 | ff_dovi_ctx_unref(s); |
| 698 | return ret; | 728 | return ret; |
| 699 | } | 729 | } |