FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
hevc.h
Go to the documentation of this file.
1 /*
2  * HEVC video decoder
3  *
4  * Copyright (C) 2012 - 2013 Guillaume Martres
5  *
6  * This file is part of FFmpeg.
7  *
8  * FFmpeg is free software; you can redistribute it and/or
9  * modify it under the terms of the GNU Lesser General Public
10  * License as published by the Free Software Foundation; either
11  * version 2.1 of the License, or (at your option) any later version.
12  *
13  * FFmpeg is distributed in the hope that it will be useful,
14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16  * Lesser General Public License for more details.
17  *
18  * You should have received a copy of the GNU Lesser General Public
19  * License along with FFmpeg; if not, write to the Free Software
20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21  */
22 
23 #ifndef AVCODEC_HEVC_H
24 #define AVCODEC_HEVC_H
25 
26 #include "libavutil/buffer.h"
27 #include "libavutil/md5.h"
28 
29 #include "avcodec.h"
30 #include "bswapdsp.h"
31 #include "cabac.h"
32 #include "get_bits.h"
33 #include "hevcpred.h"
34 #include "hevcdsp.h"
35 #include "internal.h"
36 #include "thread.h"
37 #include "videodsp.h"
38 
39 #define MAX_DPB_SIZE 16 // A.4.1
40 #define MAX_REFS 16
41 
42 #define MAX_NB_THREADS 16
43 #define SHIFT_CTB_WPP 2
44 
45 /**
46  * 7.4.2.1
47  */
48 #define MAX_SUB_LAYERS 7
49 #define MAX_VPS_COUNT 16
50 #define MAX_SPS_COUNT 32
51 #define MAX_PPS_COUNT 256
52 #define MAX_SHORT_TERM_RPS_COUNT 64
53 #define MAX_CU_SIZE 128
54 
55 //TODO: check if this is really the maximum
56 #define MAX_TRANSFORM_DEPTH 5
57 
58 #define MAX_TB_SIZE 32
59 #define MAX_LOG2_CTB_SIZE 6
60 #define MAX_QP 51
61 #define DEFAULT_INTRA_TC_OFFSET 2
62 
63 #define HEVC_CONTEXTS 199
64 
65 #define MRG_MAX_NUM_CANDS 5
66 
67 #define L0 0
68 #define L1 1
69 
70 #define EPEL_EXTRA_BEFORE 1
71 #define EPEL_EXTRA_AFTER 2
72 #define EPEL_EXTRA 3
73 #define QPEL_EXTRA_BEFORE 3
74 #define QPEL_EXTRA_AFTER 4
75 #define QPEL_EXTRA 7
76 
77 #define EDGE_EMU_BUFFER_STRIDE 80
78 
79 /**
80  * Value of the luma sample at position (x, y) in the 2D array tab.
81  */
82 #define SAMPLE(tab, x, y) ((tab)[(y) * s->sps->width + (x)])
83 #define SAMPLE_CTB(tab, x, y) ((tab)[(y) * min_cb_width + (x)])
84 
85 #define IS_IDR(s) ((s)->nal_unit_type == NAL_IDR_W_RADL || (s)->nal_unit_type == NAL_IDR_N_LP)
86 #define IS_BLA(s) ((s)->nal_unit_type == NAL_BLA_W_RADL || (s)->nal_unit_type == NAL_BLA_W_LP || \
87  (s)->nal_unit_type == NAL_BLA_N_LP)
88 #define IS_IRAP(s) ((s)->nal_unit_type >= 16 && (s)->nal_unit_type <= 23)
89 
90 /**
91  * Table 7-3: NAL unit type codes
92  */
96  NAL_TSA_N = 2,
97  NAL_TSA_R = 3,
110  NAL_VPS = 32,
111  NAL_SPS = 33,
112  NAL_PPS = 34,
113  NAL_AUD = 35,
119 };
120 
121 enum RPSType {
128 };
129 
130 enum SliceType {
131  B_SLICE = 0,
132  P_SLICE = 1,
133  I_SLICE = 2,
134 };
135 
186 };
187 
188 enum PartMode {
192  PART_NxN = 3,
197 };
198 
199 enum PredMode {
203 };
204 
206  PRED_L0 = 0,
209 };
210 
211 enum PredFlag {
212  PF_INTRA = 0,
216 };
217 
254 };
255 
256 enum SAOType {
261 };
262 
268 };
269 
270 enum ScanType {
274 };
275 
276 typedef struct ShortTermRPS {
277  unsigned int num_negative_pics;
281 } ShortTermRPS;
282 
283 typedef struct LongTermRPS {
284  int poc[32];
287 } LongTermRPS;
288 
289 typedef struct RefPicList {
293  int nb_refs;
294 } RefPicList;
295 
296 typedef struct RefPicListTab {
298 } RefPicListTab;
299 
300 typedef struct HEVCWindow {
305 } HEVCWindow;
306 
307 typedef struct VUI {
309 
312 
320 
325 
328 
331 
334  uint32_t vui_time_scale;
338 
348 } VUI;
349 
350 typedef struct PTLCommon {
360 } PTLCommon;
361 
362 typedef struct PTL {
365 
368 } PTL;
369 
370 typedef struct HEVCVPS {
373  int vps_max_sub_layers; ///< vps_max_temporal_layers_minus1 + 1
374 
381  int vps_num_layer_sets; ///< vps_num_layer_sets_minus1 + 1
384  uint32_t vps_time_scale;
386  int vps_num_ticks_poc_diff_one; ///< vps_num_ticks_poc_diff_one_minus1 + 1
388 } HEVCVPS;
389 
390 typedef struct ScalingList {
391  /* This is a little wasteful, since sizeID 0 only needs 8 coeffs,
392  * and size ID 3 only has 2 arrays, not 6. */
393  uint8_t sl[4][6][64];
394  uint8_t sl_dc[2][6];
395 } ScalingList;
396 
397 typedef struct HEVCSPS {
398  unsigned vps_id;
401 
402  ///< output (i.e. cropped) values
405 
407 
411 
412  unsigned int log2_max_poc_lsb;
414 
416  struct {
421 
424 
427 
428  unsigned int nb_st_rps;
430 
433 
438 
439  struct {
442  unsigned int log2_min_pcm_cb_size;
443  unsigned int log2_max_pcm_cb_size;
445  } pcm;
448 
449  unsigned int log2_min_cb_size;
451  unsigned int log2_min_tb_size;
452  unsigned int log2_max_trafo_size;
453  unsigned int log2_ctb_size;
454  unsigned int log2_min_pu_size;
455 
458 
465 
466  ///< coded frame dimension in various units
467  int width;
468  int height;
471  int ctb_size;
478  int tb_mask;
479 
480  int hshift[3];
481  int vshift[3];
482 
484 } HEVCSPS;
485 
486 typedef struct HEVCPPS {
487  unsigned int sps_id; ///< seq_parameter_set_id
488 
490 
492 
493  int num_ref_idx_l0_default_active; ///< num_ref_idx_l0_default_active_minus1 + 1
494  int num_ref_idx_l1_default_active; ///< num_ref_idx_l1_default_active_minus1 + 1
496 
499 
502 
510 
514 
515  int num_tile_columns; ///< num_tile_columns_minus1 + 1
516  int num_tile_rows; ///< num_tile_rows_minus1 + 1
519 
521 
525  int beta_offset; ///< beta_offset_div2 * 2
526  int tc_offset; ///< tc_offset_div2 * 2
527 
530 
532  int log2_parallel_merge_level; ///< log2_parallel_merge_level_minus2 + 2
540  int8_t cb_qp_offset_list[5];
541  int8_t cr_qp_offset_list[5];
544 
545  // Inferred parameters
546  unsigned int *column_width; ///< ColumnWidth
547  unsigned int *row_height; ///< RowHeight
548  unsigned int *col_bd; ///< ColBd
549  unsigned int *row_bd; ///< RowBd
550  int *col_idxX;
551 
552  int *ctb_addr_rs_to_ts; ///< CtbAddrRSToTS
553  int *ctb_addr_ts_to_rs; ///< CtbAddrTSToRS
554  int *tile_id; ///< TileId
555  int *tile_pos_rs; ///< TilePosRS
556  int *min_tb_addr_zs; ///< MinTbAddrZS
557  int *min_tb_addr_zs_tab;///< MinTbAddrZS
558 } HEVCPPS;
559 
560 typedef struct SliceHeader {
561  unsigned int pps_id;
562 
563  ///< address (in raster order) of the first block in the current slice segment
564  unsigned int slice_segment_addr;
565  ///< address (in raster order) of the first block in the current slice
566  unsigned int slice_addr;
567 
569 
571 
576 
577  ///< RPS coded in the slice header itself is stored here
581  unsigned int list_entry_lx[2][32];
582 
586 
587  unsigned int nb_refs[2];
588 
591 
593  uint8_t disable_deblocking_filter_flag; ///< slice_header_disable_deblocking_filter_flag
596 
597  unsigned int collocated_ref_idx;
598 
602 
604 
605  int beta_offset; ///< beta_offset_div2 * 2
606  int tc_offset; ///< tc_offset_div2 * 2
607 
608  unsigned int max_num_merge_cand; ///< 5 - 5_minus_max_num_merge_cand
609 
611  int * offset;
612  int * size;
614 
615  int8_t slice_qp;
616 
619 
620  int16_t luma_weight_l0[16];
621  int16_t chroma_weight_l0[16][2];
622  int16_t chroma_weight_l1[16][2];
623  int16_t luma_weight_l1[16];
624 
625  int16_t luma_offset_l0[16];
626  int16_t chroma_offset_l0[16][2];
627 
628  int16_t luma_offset_l1[16];
629  int16_t chroma_offset_l1[16][2];
630 
632 } SliceHeader;
633 
634 typedef struct CodingUnit {
635  int x;
636  int y;
637 
638  enum PredMode pred_mode; ///< PredMode
639  enum PartMode part_mode; ///< PartMode
640 
641  // Inferred parameters
642  uint8_t intra_split_flag; ///< IntraSplitFlag
643  uint8_t max_trafo_depth; ///< MaxTrafoDepth
645 } CodingUnit;
646 
647 typedef struct Mv {
648  int16_t x; ///< horizontal component of motion vector
649  int16_t y; ///< vertical component of motion vector
650 } Mv;
651 
652 typedef struct MvField {
654  int8_t ref_idx[2];
655  int8_t pred_flag;
656 } MvField;
657 
658 typedef struct NeighbourAvailable {
661  int cand_up;
666 
667 typedef struct PredictionUnit {
668  int mpm_idx;
676 
677 typedef struct TransformUnit {
679 
681 
682  // Inferred parameters;
691 } TransformUnit;
692 
693 typedef struct DBParams {
696 } DBParams;
697 
698 #define HEVC_FRAME_FLAG_OUTPUT (1 << 0)
699 #define HEVC_FRAME_FLAG_SHORT_REF (1 << 1)
700 #define HEVC_FRAME_FLAG_LONG_REF (1 << 2)
701 #define HEVC_FRAME_FLAG_BUMPING (1 << 3)
702 
703 typedef struct HEVCFrame {
710  int poc;
712 
714 
718 
719  /**
720  * A sequence counter, so that old frames are output first
721  * after a POC reset
722  */
723  uint16_t sequence;
724 
725  /**
726  * A combination of HEVC_FRAME_FLAG_*
727  */
729 } HEVCFrame;
730 
731 typedef struct HEVCNAL {
734 
735  int size;
736  const uint8_t *data;
737 } HEVCNAL;
738 
739 typedef struct HEVCLocalContext {
741 
743 
745 
748 
749  int8_t qp_y;
750  int8_t curr_qp_y;
751 
752  int qPy_pred;
753 
755 
762  /* +7 is for subpixel interpolation, *2 for high bit depths */
766 
767  int ct_depth;
771 
772 #define BOUNDARY_LEFT_SLICE (1 << 0)
773 #define BOUNDARY_LEFT_TILE (1 << 1)
774 #define BOUNDARY_UPPER_SLICE (1 << 2)
775 #define BOUNDARY_UPPER_TILE (1 << 3)
776  /* properties of the boundary of the current CTB for the purposes
777  * of the deblocking filter */
780 
781 typedef struct HEVCContext {
782  const AVClass *c; // needed by private avoptions
784 
786 
789 
792 
793  int width;
794  int height;
795 
797 
798  /** 1 if the independent slice segment header was successfully parsed */
800 
805 
806  const HEVCVPS *vps;
807  const HEVCSPS *sps;
808  const HEVCPPS *pps;
812 
814 
817 
818  ///< candidate references for the current frame
820 
825  int temporal_id; ///< temporal_id_plus1 - 1
828  int poc;
829  int pocTid0;
830  int slice_idx; ///< number of the slice being currently decoded
831  int eos; ///< current packet contains an EOS/EOB NAL
832  int last_eos; ///< last packet contains an EOS/EOB NAL
833  int max_ra;
834  int bs_width;
836 
838 
843  int8_t *qp_y_tab;
846 
848 
849  // CU
852  // PU
854 
855  uint8_t *cbf_luma; // cbf_luma of colocated TU
857 
858  // CTB-level flags affecting loop filter operation
860 
861  /** used on BE to byteswap the lines for checksumming */
864 
865  /**
866  * Sequence counters for decoded and output frames, so that old
867  * frames are output first after a POC reset
868  */
869  uint16_t seq_decode;
870  uint16_t seq_output;
871 
873  int wpp_err;
877 
881 
882  const uint8_t *data;
883 
885  int nb_nals;
887  // type of the first VCL NAL of the current frame
889 
890  // for checking the frame checksums
891  struct AVMD5 *md5_ctx;
892  uint8_t md5[3][16];
894 
896  uint8_t is_nalff; ///< this flag is != 0 if bitstream is encapsulated
897  ///< as a format defined in 14496-15
899 
901 
902  int nal_length_size; ///< Number of bytes used for nal length (1, 2 or 4)
904 
905  /** frame packing arrangement variables */
910 
911  /** display orientation */
915 
917 } HEVCContext;
918 
920  const HEVCSPS *sps, int is_slice_header);
925 
927  HEVCNAL *nal);
928 
929 /**
930  * Mark all frames in DPB as unused for reference.
931  */
933 
934 /**
935  * Drop all frames currently in DPB.
936  */
938 
939 /**
940  * Compute POC of the current frame and return it.
941  */
942 int ff_hevc_compute_poc(HEVCContext *s, int poc_lsb);
943 
945  int x0, int y0);
946 
947 /**
948  * Construct the reference picture sets for the current frame.
949  */
951 
952 /**
953  * Construct the reference picture list(s) for the current slice.
954  */
956 
957 void ff_hevc_save_states(HEVCContext *s, int ctb_addr_ts);
958 void ff_hevc_cabac_init(HEVCContext *s, int ctb_addr_ts);
967 int ff_hevc_skip_flag_decode(HEVCContext *s, int x0, int y0,
968  int x_cb, int y_cb);
971  int x0, int y0);
972 int ff_hevc_part_mode_decode(HEVCContext *s, int log2_cb_size);
980 int ff_hevc_inter_pred_idc_decode(HEVCContext *s, int nPbW, int nPbH);
981 int ff_hevc_ref_idx_lx_decode(HEVCContext *s, int num_ref_idx_lx);
984 int ff_hevc_split_transform_flag_decode(HEVCContext *s, int log2_trafo_size);
985 int ff_hevc_cbf_cb_cr_decode(HEVCContext *s, int trafo_depth);
986 int ff_hevc_cbf_luma_decode(HEVCContext *s, int trafo_depth);
989 
990 /**
991  * Get the number of candidate references for the current frame.
992  */
994 
995 int ff_hevc_set_new_ref(HEVCContext *s, AVFrame **frame, int poc);
996 
997 /**
998  * Find next frame in output order and put a reference to it in frame.
999  * @return 1 if a frame was output, 0 otherwise
1000  */
1002 
1004 
1006 
1007 void ff_hevc_set_neighbour_available(HEVCContext *s, int x0, int y0,
1008  int nPbW, int nPbH);
1009 void ff_hevc_luma_mv_merge_mode(HEVCContext *s, int x0, int y0,
1010  int nPbW, int nPbH, int log2_cb_size,
1011  int part_idx, int merge_idx, MvField *mv);
1012 void ff_hevc_luma_mv_mvp_mode(HEVCContext *s, int x0, int y0,
1013  int nPbW, int nPbH, int log2_cb_size,
1014  int part_idx, int merge_idx,
1015  MvField *mv, int mvp_lx_flag, int LX);
1016 void ff_hevc_set_qPy(HEVCContext *s, int xBase, int yBase,
1017  int log2_cb_size);
1019  int log2_trafo_size);
1024 void ff_hevc_hls_filter(HEVCContext *s, int x, int y, int ctb_size);
1025 void ff_hevc_hls_filters(HEVCContext *s, int x_ctb, int y_ctb, int ctb_size);
1026 void ff_hevc_hls_residual_coding(HEVCContext *s, int x0, int y0,
1027  int log2_trafo_size, enum ScanType scan_idx,
1028  int c_idx);
1029 
1030 void ff_hevc_hls_mvd_coding(HEVCContext *s, int x0, int y0, int log2_cb_size);
1031 
1032 
1033 extern const uint8_t ff_hevc_qpel_extra_before[4];
1034 extern const uint8_t ff_hevc_qpel_extra_after[4];
1035 extern const uint8_t ff_hevc_qpel_extra[4];
1036 
1037 extern const uint8_t ff_hevc_diag_scan4x4_x[16];
1038 extern const uint8_t ff_hevc_diag_scan4x4_y[16];
1039 extern const uint8_t ff_hevc_diag_scan8x8_x[64];
1040 extern const uint8_t ff_hevc_diag_scan8x8_y[64];
1041 
1042 #endif /* AVCODEC_HEVC_H */