2 * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder
3 * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
5 * This file is part of FFmpeg.
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.
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.
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
23 * @file libavcodec/h264.h
24 * H.264 / AVC / MPEG4 part10 codec.
25 * @author Michael Niedermayer <michaelni@gmx.at>
28 #ifndef AVCODEC_H264_H
29 #define AVCODEC_H264_H
33 #include "mpegvideo.h"
36 #define interlaced_dct interlaced_dct_is_a_bad_name
37 #define mb_intra mb_intra_is_not_initialized_see_mb_type
39 #define LUMA_DC_BLOCK_INDEX 25
40 #define CHROMA_DC_BLOCK_INDEX 26
42 #define CHROMA_DC_COEFF_TOKEN_VLC_BITS 8
43 #define COEFF_TOKEN_VLC_BITS 8
44 #define TOTAL_ZEROS_VLC_BITS 9
45 #define CHROMA_DC_TOTAL_ZEROS_VLC_BITS 3
46 #define RUN_VLC_BITS 3
47 #define RUN7_VLC_BITS 6
49 #define MAX_SPS_COUNT 32
50 #define MAX_PPS_COUNT 256
52 #define MAX_MMCO_COUNT 66
54 #define MAX_DELAYED_PIC_COUNT 16
56 /* Compiling in interlaced support reduces the speed
57 * of progressive decoding by about 2%. */
58 #define ALLOW_INTERLACE
60 #define ALLOW_NOCHROMA
63 * The maximum number of slices supported by the decoder.
64 * must be a power of 2
68 #ifdef ALLOW_INTERLACE
69 #define MB_MBAFF h->mb_mbaff
70 #define MB_FIELD h->mb_field_decoding_flag
71 #define FRAME_MBAFF h->mb_aff_frame
72 #define FIELD_PICTURE (s->picture_structure != PICT_FRAME)
77 #define FIELD_PICTURE 0
79 #define IS_INTERLACED(mb_type) 0
81 #define FIELD_OR_MBAFF_PICTURE (FRAME_MBAFF || FIELD_PICTURE)
84 #define CHROMA h->sps.chroma_format_idc
89 #define EXTENDED_SAR 255
91 #define MB_TYPE_REF0 MB_TYPE_ACPRED //dirty but it fits in 16 bit
92 #define MB_TYPE_8x8DCT 0x01000000
93 #define IS_REF0(a) ((a) & MB_TYPE_REF0)
94 #define IS_8x8DCT(a) ((a) & MB_TYPE_8x8DCT)
111 NAL_AUXILIARY_SLICE=19
118 SEI_TYPE_PIC_TIMING = 1, ///< picture timing
119 SEI_TYPE_USER_DATA_UNREGISTERED = 5, ///< unregistered user data
120 SEI_TYPE_RECOVERY_POINT = 6 ///< recovery point (frame # to decoder sync)
124 * pic_struct in picture timing SEI message
127 SEI_PIC_STRUCT_FRAME = 0, ///< 0: %frame
128 SEI_PIC_STRUCT_TOP_FIELD = 1, ///< 1: top field
129 SEI_PIC_STRUCT_BOTTOM_FIELD = 2, ///< 2: bottom field
130 SEI_PIC_STRUCT_TOP_BOTTOM = 3, ///< 3: top field, bottom field, in that order
131 SEI_PIC_STRUCT_BOTTOM_TOP = 4, ///< 4: bottom field, top field, in that order
132 SEI_PIC_STRUCT_TOP_BOTTOM_TOP = 5, ///< 5: top field, bottom field, top field repeated, in that order
133 SEI_PIC_STRUCT_BOTTOM_TOP_BOTTOM = 6, ///< 6: bottom field, top field, bottom field repeated, in that order
134 SEI_PIC_STRUCT_FRAME_DOUBLING = 7, ///< 7: %frame doubling
135 SEI_PIC_STRUCT_FRAME_TRIPLING = 8 ///< 8: %frame tripling
139 * Sequence parameter set
145 int chroma_format_idc;
146 int transform_bypass; ///< qpprime_y_zero_transform_bypass_flag
147 int log2_max_frame_num; ///< log2_max_frame_num_minus4 + 4
148 int poc_type; ///< pic_order_cnt_type
149 int log2_max_poc_lsb; ///< log2_max_pic_order_cnt_lsb_minus4
150 int delta_pic_order_always_zero_flag;
151 int offset_for_non_ref_pic;
152 int offset_for_top_to_bottom_field;
153 int poc_cycle_length; ///< num_ref_frames_in_pic_order_cnt_cycle
154 int ref_frame_count; ///< num_ref_frames
155 int gaps_in_frame_num_allowed_flag;
156 int mb_width; ///< pic_width_in_mbs_minus1 + 1
157 int mb_height; ///< pic_height_in_map_units_minus1 + 1
158 int frame_mbs_only_flag;
159 int mb_aff; ///<mb_adaptive_frame_field_flag
160 int direct_8x8_inference_flag;
161 int crop; ///< frame_cropping_flag
162 unsigned int crop_left; ///< frame_cropping_rect_left_offset
163 unsigned int crop_right; ///< frame_cropping_rect_right_offset
164 unsigned int crop_top; ///< frame_cropping_rect_top_offset
165 unsigned int crop_bottom; ///< frame_cropping_rect_bottom_offset
166 int vui_parameters_present_flag;
168 int timing_info_present_flag;
169 uint32_t num_units_in_tick;
171 int fixed_frame_rate_flag;
172 short offset_for_ref_frame[256]; //FIXME dyn aloc?
173 int bitstream_restriction_flag;
174 int num_reorder_frames;
175 int scaling_matrix_present;
176 uint8_t scaling_matrix4[6][16];
177 uint8_t scaling_matrix8[2][64];
178 int nal_hrd_parameters_present_flag;
179 int vcl_hrd_parameters_present_flag;
180 int pic_struct_present_flag;
181 int time_offset_length;
182 int cpb_removal_delay_length; ///< cpb_removal_delay_length_minus1 + 1
183 int dpb_output_delay_length; ///< dpb_output_delay_length_minus1 + 1
184 int bit_depth_luma; ///< bit_depth_luma_minus8 + 8
185 int bit_depth_chroma; ///< bit_depth_chroma_minus8 + 8
186 int residual_color_transform_flag; ///< residual_colour_transform_flag
190 * Picture parameter set
194 int cabac; ///< entropy_coding_mode_flag
195 int pic_order_present; ///< pic_order_present_flag
196 int slice_group_count; ///< num_slice_groups_minus1 + 1
197 int mb_slice_group_map_type;
198 unsigned int ref_count[2]; ///< num_ref_idx_l0/1_active_minus1 + 1
199 int weighted_pred; ///< weighted_pred_flag
200 int weighted_bipred_idc;
201 int init_qp; ///< pic_init_qp_minus26 + 26
202 int init_qs; ///< pic_init_qs_minus26 + 26
203 int chroma_qp_index_offset[2];
204 int deblocking_filter_parameters_present; ///< deblocking_filter_parameters_present_flag
205 int constrained_intra_pred; ///< constrained_intra_pred_flag
206 int redundant_pic_cnt_present; ///< redundant_pic_cnt_present_flag
207 int transform_8x8_mode; ///< transform_8x8_mode_flag
208 uint8_t scaling_matrix4[6][16];
209 uint8_t scaling_matrix8[2][64];
210 uint8_t chroma_qp_table[2][64]; ///< pre-scaled (with chroma_qp_index_offset) version of qp_table
215 * Memory management control operation opcode.
217 typedef enum MMCOOpcode{
228 * Memory management control operation.
232 int short_pic_num; ///< pic_num without wrapping (pic_num & max_pic_num)
233 int long_arg; ///< index, pic_num, or num long refs depending on opcode
239 typedef struct H264Context{
243 uint8_t *rbsp_buffer[2];
244 unsigned int rbsp_buffer_size[2];
247 * Used to parse AVC variant of h264
249 int is_avc; ///< this flag is != 0 if codec is avc1
250 int got_avcC; ///< flag used to parse avcC data only once
251 int nal_length_size; ///< Number of bytes used for nal length (1, 2 or 4)
253 int chroma_qp[2]; //QPc
259 int chroma_pred_mode;
260 int intra16x16_pred_mode;
265 int8_t intra4x4_pred_mode_cache[5*8];
266 int8_t (*intra4x4_pred_mode)[8];
268 unsigned int topleft_samples_available;
269 unsigned int top_samples_available;
270 unsigned int topright_samples_available;
271 unsigned int left_samples_available;
272 uint8_t (*top_borders[2])[16+2*8];
273 uint8_t left_border[2*(17+2*9)];
276 * non zero coeff count cache.
277 * is 64 if not available.
279 DECLARE_ALIGNED_8(uint8_t, non_zero_count_cache[6*8]);
280 uint8_t (*non_zero_count)[16];
283 * Motion vector cache.
285 DECLARE_ALIGNED_8(int16_t, mv_cache[2][5*8][2]);
286 DECLARE_ALIGNED_8(int8_t, ref_cache[2][5*8]);
287 #define LIST_NOT_USED -1 //FIXME rename?
288 #define PART_NOT_AVAILABLE -2
291 * is 1 if the specific list MV&references are set to 0,0,-2.
293 int mv_cache_clean[2];
296 * number of neighbors (top and/or left) that used 8x8 dct
298 int neighbor_transform_size;
301 * block_offset[ 0..23] for frame macroblocks
302 * block_offset[24..47] for field macroblocks
304 int block_offset[2*(16+8)];
306 uint32_t *mb2b_xy; //FIXME are these 4 a good idea?
308 int b_stride; //FIXME use s->b4_stride
311 int mb_linesize; ///< may be equal to s->linesize or s->linesize*2, for mbaff
320 int unknown_svq3_flag;
321 int next_slice_index;
323 SPS *sps_buffers[MAX_SPS_COUNT];
324 SPS sps; ///< current sps
326 PPS *pps_buffers[MAX_PPS_COUNT];
330 PPS pps; //FIXME move to Picture perhaps? (->no) do we need that?
332 uint32_t dequant4_buffer[6][52][16];
333 uint32_t dequant8_buffer[2][52][64];
334 uint32_t (*dequant4_coeff[6])[16];
335 uint32_t (*dequant8_coeff[2])[64];
336 int dequant_coeff_pps; ///< reinit tables when pps changes
339 uint16_t *slice_table_base;
340 uint16_t *slice_table; ///< slice_table_base + 2*mb_stride + 1
342 int slice_type_nos; ///< S free slice type (SI/SP are remapped to I/P)
343 int slice_type_fixed;
345 //interlacing specific flags
347 int mb_field_decoding_flag;
348 int mb_mbaff; ///< mb_aff_frame && mb_field_decoding_flag
350 uint16_t sub_mb_type[4];
355 int delta_poc_bottom;
358 int prev_poc_msb; ///< poc_msb of the last reference pic for POC type 0
359 int prev_poc_lsb; ///< poc_lsb of the last reference pic for POC type 0
360 int frame_num_offset; ///< for POC type 2
361 int prev_frame_num_offset; ///< for POC type 2
362 int prev_frame_num; ///< frame_num of the last pic for POC type 1/2
365 * frame_num for frames or 2*frame_num+1 for field pics.
370 * max_frame_num or 2*max_frame_num for field pics.
374 //Weighted pred stuff
376 int use_weight_chroma;
377 int luma_log2_weight_denom;
378 int chroma_log2_weight_denom;
379 int luma_weight[2][48];
380 int luma_offset[2][48];
381 int chroma_weight[2][48][2];
382 int chroma_offset[2][48][2];
383 int implicit_weight[48][48];
386 int deblocking_filter; ///< disable_deblocking_filter_idc with 1<->0
387 int slice_alpha_c0_offset;
388 int slice_beta_offset;
390 int redundant_pic_count;
392 int direct_spatial_mv_pred;
393 int dist_scale_factor[16];
394 int dist_scale_factor_field[2][32];
395 int map_col_to_list0[2][16+32];
396 int map_col_to_list0_field[2][2][16+32];
399 * num_ref_idx_l0/1_active_minus1 + 1
401 unsigned int ref_count[2]; ///< counts frames or fields, depending on current mb mode
402 unsigned int list_count;
403 Picture *short_ref[32];
404 Picture *long_ref[32];
405 Picture default_ref_list[2][32]; ///< base reference list for all slices of a coded picture
406 Picture ref_list[2][48]; /**< 0..15: frame refs, 16..47: mbaff field refs.
407 Reordered version of default_ref_list
408 according to picture reordering in slice header */
409 int ref2frm[MAX_SLICES][2][64]; ///< reference to frame number lists, used in the loop filter, the first 2 are for -2,-1
410 Picture *delayed_pic[MAX_DELAYED_PIC_COUNT+2]; //FIXME size?
414 * memory management control operations buffer.
416 MMCO mmco[MAX_MMCO_COUNT];
419 int long_ref_count; ///< number of actual long term references
420 int short_ref_count; ///< number of actual short term references
423 GetBitContext intra_gb;
424 GetBitContext inter_gb;
425 GetBitContext *intra_gb_ptr;
426 GetBitContext *inter_gb_ptr;
428 DECLARE_ALIGNED_16(DCTELEM, mb[16*24]);
429 DCTELEM mb_padding[256]; ///< as mb is addressed by scantable[i] and scantable is uint8_t we can either check that i is not too large or ensure that there is some unused stuff after mb
435 uint8_t cabac_state[460];
438 /* 0x100 -> non null luma_dc, 0x80/0x40 -> non null chroma_dc (cb/cr), 0x?0 -> chroma_cbp(0,1,2), 0x0? luma_cbp */
443 /* chroma_pred_mode for i4x4 or i16x16, else 0 */
444 uint8_t *chroma_pred_mode_table;
445 int last_qscale_diff;
446 int16_t (*mvd_table[2])[2];
447 DECLARE_ALIGNED_8(int16_t, mvd_cache[2][5*8][2]);
448 uint8_t *direct_table;
449 uint8_t direct_cache[5*8];
451 uint8_t zigzag_scan[16];
452 uint8_t zigzag_scan8x8[64];
453 uint8_t zigzag_scan8x8_cavlc[64];
454 uint8_t field_scan[16];
455 uint8_t field_scan8x8[64];
456 uint8_t field_scan8x8_cavlc[64];
457 const uint8_t *zigzag_scan_q0;
458 const uint8_t *zigzag_scan8x8_q0;
459 const uint8_t *zigzag_scan8x8_cavlc_q0;
460 const uint8_t *field_scan_q0;
461 const uint8_t *field_scan8x8_q0;
462 const uint8_t *field_scan8x8_cavlc_q0;
467 * @defgroup multithreading Members for slice based multithreading
470 struct H264Context *thread_context[MAX_THREADS];
473 * current slice number, used to initalize slice_num of each thread/context
478 * Max number of threads / contexts.
479 * This is equal to AVCodecContext.thread_count unless
480 * multithreaded decoding is impossible, in which case it is
486 * 1 if the single thread fallback warning has already been
487 * displayed, 0 otherwise.
489 int single_decode_warning;
496 uint32_t svq3_watermark_key;
499 * pic_struct in picture timing SEI message
501 SEI_PicStructType sei_pic_struct;
504 * recovery_frame_cnt from SEI message
506 * Set to -1 if no recovery point SEI message found or to number of frames
507 * before playback synchronizes. Frames having recovery point are key
510 int sei_recovery_frame_cnt;
514 int luma_weight_flag[2]; ///< 7.4.3.2 luma_weight_lX_flag
515 int chroma_weight_flag[2]; ///< 7.4.3.2 chroma_weight_lX_flag
518 #endif /* AVCODEC_H264_H */