FFmpeg coverage

Directory: ../../../ffmpeg/
File: src/libavcodec/mpegvideo.h
Date: 2021-09-24 03:35:22
Exec Total Coverage
Lines: 23 23 100.0%
Branches: 3 4 75.0%
Line Branch Exec Source
1 /*
2 * Generic DCT based hybrid video encoder
3 * Copyright (c) 2000, 2001, 2002 Fabrice Bellard
4 * Copyright (c) 2002-2004 Michael Niedermayer
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 /**
24 * @file
25 * mpegvideo header.
26 */
27
28 #ifndef AVCODEC_MPEGVIDEO_H
29 #define AVCODEC_MPEGVIDEO_H
30
31 #include <float.h>
32
33 #include "avcodec.h"
34 #include "blockdsp.h"
35 #include "error_resilience.h"
36 #include "fdctdsp.h"
37 #include "get_bits.h"
38 #include "h264chroma.h"
39 #include "h263dsp.h"
40 #include "hpeldsp.h"
41 #include "idctdsp.h"
42 #include "internal.h"
43 #include "me_cmp.h"
44 #include "motion_est.h"
45 #include "mpegpicture.h"
46 #include "mpegvideodsp.h"
47 #include "mpegvideoencdsp.h"
48 #include "mpegvideodata.h"
49 #include "pixblockdsp.h"
50 #include "put_bits.h"
51 #include "ratecontrol.h"
52 #if FF_API_FLAG_TRUNCATED
53 #include "parser.h"
54 #endif
55 #include "mpegutils.h"
56 #include "mpeg12data.h"
57 #include "qpeldsp.h"
58 #include "thread.h"
59 #include "videodsp.h"
60
61 #include "libavutil/opt.h"
62 #include "libavutil/timecode.h"
63
64 #define MAX_THREADS 32
65
66 #define MAX_B_FRAMES 16
67
68 /**
69 * MpegEncContext.
70 */
71 typedef struct MpegEncContext {
72 AVClass *class;
73
74 int y_dc_scale, c_dc_scale;
75 int ac_pred;
76 int block_last_index[12]; ///< last non zero coefficient in block
77 int h263_aic; ///< Advanced INTRA Coding (AIC)
78
79 /* scantables */
80 ScanTable inter_scantable; ///< if inter == intra then intra should be used to reduce the cache usage
81 ScanTable intra_scantable;
82 ScanTable intra_h_scantable;
83 ScanTable intra_v_scantable;
84
85 /* WARNING: changes above this line require updates to hardcoded
86 * offsets used in ASM. */
87
88 struct AVCodecContext *avctx;
89 /* the following parameters must be initialized before encoding */
90 int width, height;///< picture size. must be a multiple of 16
91 int gop_size;
92 int intra_only; ///< if true, only intra pictures are generated
93 int64_t bit_rate; ///< wanted bit rate
94 enum OutputFormat out_format; ///< output format
95 int h263_pred; ///< use MPEG-4/H.263 ac/dc predictions
96 int pb_frame; ///< PB-frame mode (0 = none, 1 = base, 2 = improved)
97
98 /* the following codec id fields are deprecated in favor of codec_id */
99 int h263_plus; ///< H.263+ headers
100 int h263_flv; ///< use flv H.263 header
101
102 enum AVCodecID codec_id; /* see AV_CODEC_ID_xxx */
103 int fixed_qscale; ///< fixed qscale if non zero
104 int encoding; ///< true if we are encoding (vs decoding)
105 int max_b_frames; ///< max number of B-frames for encoding
106 int luma_elim_threshold;
107 int chroma_elim_threshold;
108 int strict_std_compliance; ///< strictly follow the std (MPEG-4, ...)
109 int workaround_bugs; ///< workaround bugs in encoders which cannot be detected automatically
110 int codec_tag; ///< internal codec_tag upper case converted from avctx codec_tag
111 /* the following fields are managed internally by the encoder */
112
113 /* sequence parameters */
114 int context_initialized;
115 int input_picture_number; ///< used to set pic->display_picture_number, should not be used for/by anything else
116 int coded_picture_number; ///< used to set pic->coded_picture_number, should not be used for/by anything else
117 int picture_number; //FIXME remove, unclear definition
118 int picture_in_gop_number; ///< 0-> first pic in gop, ...
119 int mb_width, mb_height; ///< number of MBs horizontally & vertically
120 int mb_stride; ///< mb_width+1 used for some arrays to allow simple addressing of left & top MBs without sig11
121 int b8_stride; ///< 2*mb_width+1 used for some 8x8 block arrays to allow simple addressing
122 int h_edge_pos, v_edge_pos;///< horizontal / vertical position of the right/bottom edge (pixel replication)
123 int mb_num; ///< number of MBs of a picture
124 ptrdiff_t linesize; ///< line size, in bytes, may be different from width
125 ptrdiff_t uvlinesize; ///< line size, for chroma in bytes, may be different from width
126 Picture *picture; ///< main picture buffer
127 Picture **input_picture; ///< next pictures on display order for encoding
128 Picture **reordered_input_picture; ///< pointer to the next pictures in coded order for encoding
129
130 int64_t user_specified_pts; ///< last non-zero pts from AVFrame which was passed into avcodec_send_frame()
131 /**
132 * pts difference between the first and second input frame, used for
133 * calculating dts of the first frame when there's a delay */
134 int64_t dts_delta;
135 /**
136 * reordered pts to be used as dts for the next output frame when there's
137 * a delay */
138 int64_t reordered_pts;
139
140 /** bit output */
141 PutBitContext pb;
142
143 int start_mb_y; ///< start mb_y of this thread (so current thread should process start_mb_y <= row < end_mb_y)
144 int end_mb_y; ///< end mb_y of this thread (so current thread should process start_mb_y <= row < end_mb_y)
145 struct MpegEncContext *thread_context[MAX_THREADS];
146 int slice_context_count; ///< number of used thread_contexts
147
148 /**
149 * copy of the previous picture structure.
150 * note, linesize & data, might not match the previous picture (for field pictures)
151 */
152 Picture last_picture;
153
154 /**
155 * copy of the next picture structure.
156 * note, linesize & data, might not match the next picture (for field pictures)
157 */
158 Picture next_picture;
159
160 /**
161 * copy of the source picture structure for encoding.
162 * note, linesize & data, might not match the source picture (for field pictures)
163 */
164 Picture new_picture;
165
166 /**
167 * copy of the current picture structure.
168 * note, linesize & data, might not match the current picture (for field pictures)
169 */
170 Picture current_picture; ///< buffer to store the decompressed current picture
171
172 Picture *last_picture_ptr; ///< pointer to the previous picture.
173 Picture *next_picture_ptr; ///< pointer to the next picture (for bidir pred)
174 Picture *current_picture_ptr; ///< pointer to the current picture
175 int last_dc[3]; ///< last DC values for MPEG-1
176 int16_t *dc_val_base;
177 int16_t *dc_val[3]; ///< used for MPEG-4 DC prediction, all 3 arrays must be continuous
178 const uint8_t *y_dc_scale_table; ///< qscale -> y_dc_scale table
179 const uint8_t *c_dc_scale_table; ///< qscale -> c_dc_scale table
180 const uint8_t *chroma_qscale_table; ///< qscale -> chroma_qscale (H.263)
181 uint8_t *coded_block_base;
182 uint8_t *coded_block; ///< used for coded block pattern prediction (msmpeg4v3, wmv1)
183 int16_t (*ac_val_base)[16];
184 int16_t (*ac_val[3])[16]; ///< used for MPEG-4 AC prediction, all 3 arrays must be continuous
185 int mb_skipped; ///< MUST BE SET only during DECODING
186 uint8_t *mbskip_table; /**< used to avoid copy if macroblock skipped (for black regions for example)
187 and used for B-frame encoding & decoding (contains skip table of next P-frame) */
188 uint8_t *mbintra_table; ///< used to avoid setting {ac, dc, cbp}-pred stuff to zero on inter MB decoding
189 uint8_t *cbp_table; ///< used to store cbp, ac_pred for partitioned decoding
190 uint8_t *pred_dir_table; ///< used to store pred_dir for partitioned decoding
191
192 ScratchpadContext sc;
193
194 int qscale; ///< QP
195 int chroma_qscale; ///< chroma QP
196 unsigned int lambda; ///< Lagrange multiplier used in rate distortion
197 unsigned int lambda2; ///< (lambda*lambda) >> FF_LAMBDA_SHIFT
198 int *lambda_table;
199 int adaptive_quant; ///< use adaptive quantization
200 int dquant; ///< qscale difference to prev qscale
201 int closed_gop; ///< MPEG1/2 GOP is closed
202 int pict_type; ///< AV_PICTURE_TYPE_I, AV_PICTURE_TYPE_P, AV_PICTURE_TYPE_B, ...
203 int vbv_delay;
204 int last_pict_type; //FIXME removes
205 int last_non_b_pict_type; ///< used for MPEG-4 gmc B-frames & ratecontrol
206 int droppable;
207 int frame_rate_index;
208 AVRational mpeg2_frame_rate_ext;
209 int last_lambda_for[5]; ///< last lambda for a specific pict type
210 int skipdct; ///< skip dct and code zero residual
211
212 /* motion compensation */
213 int unrestricted_mv; ///< mv can point outside of the coded picture
214 int h263_long_vectors; ///< use horrible H.263v1 long vector mode
215
216 BlockDSPContext bdsp;
217 FDCTDSPContext fdsp;
218 H264ChromaContext h264chroma;
219 HpelDSPContext hdsp;
220 IDCTDSPContext idsp;
221 MECmpContext mecc;
222 MpegVideoDSPContext mdsp;
223 MpegvideoEncDSPContext mpvencdsp;
224 PixblockDSPContext pdsp;
225 QpelDSPContext qdsp;
226 VideoDSPContext vdsp;
227 H263DSPContext h263dsp;
228 int f_code; ///< forward MV resolution
229 int b_code; ///< backward MV resolution for B-frames (MPEG-4)
230 int16_t (*p_mv_table_base)[2];
231 int16_t (*b_forw_mv_table_base)[2];
232 int16_t (*b_back_mv_table_base)[2];
233 int16_t (*b_bidir_forw_mv_table_base)[2];
234 int16_t (*b_bidir_back_mv_table_base)[2];
235 int16_t (*b_direct_mv_table_base)[2];
236 int16_t (*p_field_mv_table_base[2][2])[2];
237 int16_t (*b_field_mv_table_base[2][2][2])[2];
238 int16_t (*p_mv_table)[2]; ///< MV table (1MV per MB) P-frame encoding
239 int16_t (*b_forw_mv_table)[2]; ///< MV table (1MV per MB) forward mode B-frame encoding
240 int16_t (*b_back_mv_table)[2]; ///< MV table (1MV per MB) backward mode B-frame encoding
241 int16_t (*b_bidir_forw_mv_table)[2]; ///< MV table (1MV per MB) bidir mode B-frame encoding
242 int16_t (*b_bidir_back_mv_table)[2]; ///< MV table (1MV per MB) bidir mode B-frame encoding
243 int16_t (*b_direct_mv_table)[2]; ///< MV table (1MV per MB) direct mode B-frame encoding
244 int16_t (*p_field_mv_table[2][2])[2]; ///< MV table (2MV per MB) interlaced P-frame encoding
245 int16_t (*b_field_mv_table[2][2][2])[2];///< MV table (4MV per MB) interlaced B-frame encoding
246 uint8_t (*p_field_select_table[2]);
247 uint8_t (*b_field_select_table[2][2]);
248 int motion_est; ///< ME algorithm
249 int me_penalty_compensation;
250 int me_pre; ///< prepass for motion estimation
251 int mv_dir;
252 #define MV_DIR_FORWARD 1
253 #define MV_DIR_BACKWARD 2
254 #define MV_DIRECT 4 ///< bidirectional mode where the difference equals the MV of the last P/S/I-Frame (MPEG-4)
255 int mv_type;
256 #define MV_TYPE_16X16 0 ///< 1 vector for the whole mb
257 #define MV_TYPE_8X8 1 ///< 4 vectors (H.263, MPEG-4 4MV)
258 #define MV_TYPE_16X8 2 ///< 2 vectors, one per 16x8 block
259 #define MV_TYPE_FIELD 3 ///< 2 vectors, one per field
260 #define MV_TYPE_DMV 4 ///< 2 vectors, special mpeg2 Dual Prime Vectors
261 /**motion vectors for a macroblock
262 first coordinate : 0 = forward 1 = backward
263 second " : depend on type
264 third " : 0 = x, 1 = y
265 */
266 int mv[2][4][2];
267 int field_select[2][2];
268 int last_mv[2][2][2]; ///< last MV, used for MV prediction in MPEG-1 & B-frame MPEG-4
269 const uint8_t *fcode_tab; ///< smallest fcode needed for each MV
270 int16_t direct_scale_mv[2][64]; ///< precomputed to avoid divisions in ff_mpeg4_set_direct_mv
271
272 MotionEstContext me;
273
274 int no_rounding; /**< apply no rounding to motion compensation (MPEG-4, msmpeg4, ...)
275 for B-frames rounding mode is always 0 */
276
277 /* macroblock layer */
278 int mb_x, mb_y;
279 int mb_skip_run;
280 int mb_intra;
281 uint16_t *mb_type; ///< Table for candidate MB types for encoding (defines in mpegutils.h)
282
283 int block_index[6]; ///< index to current MB in block based arrays with edges
284 int block_wrap[6];
285 uint8_t *dest[3];
286
287 int *mb_index2xy; ///< mb_index -> mb_x + mb_y*mb_stride
288
289 /** matrix transmitted in the bitstream */
290 uint16_t intra_matrix[64];
291 uint16_t chroma_intra_matrix[64];
292 uint16_t inter_matrix[64];
293 uint16_t chroma_inter_matrix[64];
294 int force_duplicated_matrix; ///< Force duplication of mjpeg matrices, useful for rtp streaming
295
296 int intra_quant_bias; ///< bias for the quantizer
297 int inter_quant_bias; ///< bias for the quantizer
298 int min_qcoeff; ///< minimum encodable coefficient
299 int max_qcoeff; ///< maximum encodable coefficient
300 int ac_esc_length; ///< num of bits needed to encode the longest esc
301 uint8_t *intra_ac_vlc_length;
302 uint8_t *intra_ac_vlc_last_length;
303 uint8_t *intra_chroma_ac_vlc_length;
304 uint8_t *intra_chroma_ac_vlc_last_length;
305 uint8_t *inter_ac_vlc_length;
306 uint8_t *inter_ac_vlc_last_length;
307 uint8_t *luma_dc_vlc_length;
308 #define UNI_AC_ENC_INDEX(run,level) ((run)*128 + (level))
309
310 int coded_score[12];
311
312 /** precomputed matrix (combine qscale and DCT renorm) */
313 int (*q_intra_matrix)[64];
314 int (*q_chroma_intra_matrix)[64];
315 int (*q_inter_matrix)[64];
316 /** identical to the above but for MMX & these are not permutated, second 64 entries are bias*/
317 uint16_t (*q_intra_matrix16)[2][64];
318 uint16_t (*q_chroma_intra_matrix16)[2][64];
319 uint16_t (*q_inter_matrix16)[2][64];
320
321 /* noise reduction */
322 int (*dct_error_sum)[64];
323 int dct_count[2];
324 uint16_t (*dct_offset)[64];
325
326 /* bit rate control */
327 int64_t total_bits;
328 int frame_bits; ///< bits used for the current frame
329 int stuffing_bits; ///< bits used for stuffing
330 int next_lambda; ///< next lambda used for retrying to encode a frame
331 RateControlContext rc_context; ///< contains stuff only accessed in ratecontrol.c
332
333 /* statistics, used for 2-pass encoding */
334 int mv_bits;
335 int header_bits;
336 int i_tex_bits;
337 int p_tex_bits;
338 int i_count;
339 int f_count;
340 int b_count;
341 int skip_count;
342 int misc_bits; ///< cbp, mb_type
343 int last_bits; ///< temp var used for calculating the above vars
344
345 /* error concealment / resync */
346 int resync_mb_x; ///< x position of last resync marker
347 int resync_mb_y; ///< y position of last resync marker
348 GetBitContext last_resync_gb; ///< used to search for the next resync marker
349 int mb_num_left; ///< number of MBs left in this video packet (for partitioned Slices only)
350 int next_p_frame_damaged; ///< set if the next p frame is damaged, to avoid showing trashed B-frames
351
352 #if FF_API_FLAG_TRUNCATED
353 ParseContext parse_context;
354 #endif
355
356 /* H.263 specific */
357 int gob_index;
358 int obmc; ///< overlapped block motion compensation
359 int mb_info; ///< interval for outputting info about mb offsets as side data
360 int prev_mb_info, last_mb_info;
361 uint8_t *mb_info_ptr;
362 int mb_info_size;
363 int ehc_mode;
364
365 /* H.263+ specific */
366 int umvplus; ///< == H.263+ && unrestricted_mv
367 int h263_aic_dir; ///< AIC direction: 0 = left, 1 = top
368 int h263_slice_structured;
369 int alt_inter_vlc; ///< alternative inter vlc
370 int modified_quant;
371 int loop_filter;
372 int custom_pcf;
373
374 /* MPEG-4 specific */
375 int studio_profile;
376 int dct_precision;
377 ///< number of bits to represent the fractional part of time (encoder only)
378 int time_increment_bits;
379 int last_time_base;
380 int time_base; ///< time in seconds of last I,P,S Frame
381 int64_t time; ///< time of current frame
382 int64_t last_non_b_time;
383 uint16_t pp_time; ///< time distance between the last 2 p,s,i frames
384 uint16_t pb_time; ///< time distance between the last b and p,s,i frame
385 uint16_t pp_field_time;
386 uint16_t pb_field_time; ///< like above, just for interlaced
387 int real_sprite_warping_points;
388 int sprite_offset[2][2]; ///< sprite offset[isChroma][isMVY]
389 int sprite_delta[2][2]; ///< sprite_delta [isY][isMVY]
390 int mcsel;
391 int quant_precision;
392 int quarter_sample; ///< 1->qpel, 0->half pel ME/MC
393 int aspect_ratio_info; //FIXME remove
394 int sprite_warping_accuracy;
395 int data_partitioning; ///< data partitioning flag from header
396 int partitioned_frame; ///< is current frame partitioned
397 int low_delay; ///< no reordering needed / has no B-frames
398 int vo_type;
399 PutBitContext tex_pb; ///< used for data partitioned VOPs
400 PutBitContext pb2; ///< used for data partitioned VOPs
401 int mpeg_quant;
402 int padding_bug_score; ///< used to detect the VERY common padding bug in MPEG-4
403
404 /* divx specific, used to workaround (many) bugs in divx5 */
405 int divx_packed;
406 uint8_t *bitstream_buffer; //Divx 5.01 puts several frames in a single one, this is used to reorder them
407 int bitstream_buffer_size;
408 unsigned int allocated_bitstream_buffer_size;
409
410 /* RV10 specific */
411 int rv10_version; ///< RV10 version: 0 or 3
412 int rv10_first_dc_coded[3];
413
414 /* MJPEG specific */
415 struct MJpegContext *mjpeg_ctx;
416 int esc_pos;
417 int pred;
418 int huffman;
419
420 /* MSMPEG4 specific */
421 int mv_table_index;
422 int rl_table_index;
423 int rl_chroma_table_index;
424 int dc_table_index;
425 int use_skip_mb_code;
426 int slice_height; ///< in macroblocks
427 int first_slice_line; ///< used in MPEG-4 too to handle resync markers
428 int flipflop_rounding;
429 int msmpeg4_version; ///< 0=not msmpeg4, 1=mp41, 2=mp42, 3=mp43/divx3 4=wmv1/7 5=wmv2/8
430 int per_mb_rl_table;
431 int esc3_level_length;
432 int esc3_run_length;
433 /** [mb_intra][isChroma][level][run][last] */
434 int (*ac_stats)[2][MAX_LEVEL+1][MAX_RUN+1][2];
435 int inter_intra_pred;
436 int mspel;
437
438 /* SpeedHQ specific */
439 int slice_start;
440
441 /* decompression specific */
442 GetBitContext gb;
443
444 /* MPEG-1 specific */
445 int gop_picture_number; ///< index of the first picture of a GOP based on fake_pic_num & MPEG-1 specific
446 int last_mv_dir; ///< last mv_dir, used for B-frame encoding
447 uint8_t *vbv_delay_ptr; ///< pointer to vbv_delay in the bitstream
448
449 /* MPEG-2-specific - I wished not to have to support this mess. */
450 int progressive_sequence;
451 int mpeg_f_code[2][2];
452 int a53_cc;
453
454 // picture structure defines are loaded from mpegutils.h
455 int picture_structure;
456
457 int64_t timecode_frame_start; ///< GOP timecode frame start number, in non drop frame format
458 int intra_dc_precision;
459 int frame_pred_frame_dct;
460 int top_field_first;
461 int concealment_motion_vectors;
462 int q_scale_type;
463 int brd_scale;
464 int intra_vlc_format;
465 int alternate_scan;
466 int seq_disp_ext;
467 int video_format;
468 #define VIDEO_FORMAT_COMPONENT 0
469 #define VIDEO_FORMAT_PAL 1
470 #define VIDEO_FORMAT_NTSC 2
471 #define VIDEO_FORMAT_SECAM 3
472 #define VIDEO_FORMAT_MAC 4
473 #define VIDEO_FORMAT_UNSPECIFIED 5
474 int repeat_first_field;
475 int chroma_420_type;
476 int chroma_format;
477 #define CHROMA_420 1
478 #define CHROMA_422 2
479 #define CHROMA_444 3
480 int chroma_x_shift;//depend on pix_format, that depend on chroma_format
481 int chroma_y_shift;
482
483 int progressive_frame;
484 int full_pel[2];
485 int interlaced_dct;
486 int first_field; ///< is 1 for the first field of a field picture 0 otherwise
487 int drop_frame_timecode; ///< timecode is in drop frame format.
488 int scan_offset; ///< reserve space for SVCD scan offset user data.
489
490 /* RTP specific */
491 int rtp_mode;
492 int rtp_payload_size;
493
494 char *tc_opt_str; ///< timecode option string
495 AVTimecode tc; ///< timecode context
496
497 uint8_t *ptr_lastgob;
498 int swap_uv; //vcr2 codec is an MPEG-2 variant with U and V swapped
499 int pack_pblocks; //xvmc needs to keep blocks without gaps.
500 int16_t (*pblocks[12])[64];
501
502 int16_t (*block)[64]; ///< points to one of the following blocks
503 int16_t (*blocks)[12][64]; // for HQ mode we need to keep the best block
504 int (*decode_mb)(struct MpegEncContext *s, int16_t block[12][64]); // used by some codecs to avoid a switch()
505
506 int32_t (*block32)[12][64];
507 int dpcm_direction; // 0 = DCT, 1 = DPCM top to bottom scan, -1 = DPCM bottom to top scan
508 int16_t (*dpcm_macroblock)[3][256];
509
510 #define SLICE_OK 0
511 #define SLICE_ERROR -1
512 #define SLICE_END -2 ///<end marker found
513 #define SLICE_NOEND -3 ///<no end marker or error found but mb count exceeded
514
515 void (*dct_unquantize_mpeg1_intra)(struct MpegEncContext *s,
516 int16_t *block/*align 16*/, int n, int qscale);
517 void (*dct_unquantize_mpeg1_inter)(struct MpegEncContext *s,
518 int16_t *block/*align 16*/, int n, int qscale);
519 void (*dct_unquantize_mpeg2_intra)(struct MpegEncContext *s,
520 int16_t *block/*align 16*/, int n, int qscale);
521 void (*dct_unquantize_mpeg2_inter)(struct MpegEncContext *s,
522 int16_t *block/*align 16*/, int n, int qscale);
523 void (*dct_unquantize_h263_intra)(struct MpegEncContext *s,
524 int16_t *block/*align 16*/, int n, int qscale);
525 void (*dct_unquantize_h263_inter)(struct MpegEncContext *s,
526 int16_t *block/*align 16*/, int n, int qscale);
527 void (*dct_unquantize_intra)(struct MpegEncContext *s, // unquantizer to use (MPEG-4 can use both)
528 int16_t *block/*align 16*/, int n, int qscale);
529 void (*dct_unquantize_inter)(struct MpegEncContext *s, // unquantizer to use (MPEG-4 can use both)
530 int16_t *block/*align 16*/, int n, int qscale);
531 int (*dct_quantize)(struct MpegEncContext *s, int16_t *block/*align 16*/, int n, int qscale, int *overflow);
532 int (*fast_dct_quantize)(struct MpegEncContext *s, int16_t *block/*align 16*/, int n, int qscale, int *overflow);
533 void (*denoise_dct)(struct MpegEncContext *s, int16_t *block);
534
535 int mpv_flags; ///< flags set by private options
536 int quantizer_noise_shaping;
537
538 /**
539 * ratecontrol qmin qmax limiting method
540 * 0-> clipping, 1-> use a nice continuous function to limit qscale within qmin/qmax.
541 */
542 float rc_qsquish;
543 float rc_qmod_amp;
544 int rc_qmod_freq;
545 float rc_initial_cplx;
546 float rc_buffer_aggressivity;
547 float border_masking;
548 int lmin, lmax;
549 int vbv_ignore_qmax;
550
551 char *rc_eq;
552
553 /* temp buffers for rate control */
554 float *cplx_tab, *bits_tab;
555
556 /* flag to indicate a reinitialization is required, e.g. after
557 * a frame size change */
558 int context_reinit;
559
560 ERContext er;
561
562 int error_rate;
563
564 /* temporary frames used by b_frame_strategy = 2 */
565 AVFrame *tmp_frames[MAX_B_FRAMES + 2];
566 int b_frame_strategy;
567 int b_sensitivity;
568
569 /* frame skip options for encoding */
570 int frame_skip_threshold;
571 int frame_skip_factor;
572 int frame_skip_exp;
573 int frame_skip_cmp;
574
575 int scenechange_threshold;
576 int noise_reduction;
577
578 int intra_penalty;
579 } MpegEncContext;
580
581 /* mpegvideo_enc common options */
582 #define FF_MPV_FLAG_SKIP_RD 0x0001
583 #define FF_MPV_FLAG_STRICT_GOP 0x0002
584 #define FF_MPV_FLAG_QP_RD 0x0004
585 #define FF_MPV_FLAG_CBP_RD 0x0008
586 #define FF_MPV_FLAG_NAQ 0x0010
587 #define FF_MPV_FLAG_MV0 0x0020
588
589 #define FF_MPV_OPT_CMP_FUNC \
590 { "sad", "Sum of absolute differences, fast", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_SAD }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
591 { "sse", "Sum of squared errors", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_SSE }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
592 { "satd", "Sum of absolute Hadamard transformed differences", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_SATD }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
593 { "dct", "Sum of absolute DCT transformed differences", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_DCT }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
594 { "psnr", "Sum of squared quantization errors, low quality", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_PSNR }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
595 { "bit", "Number of bits needed for the block", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_BIT }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
596 { "rd", "Rate distortion optimal, slow", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_RD }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
597 { "zero", "Zero", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_ZERO }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
598 { "vsad", "Sum of absolute vertical differences", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_VSAD }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
599 { "vsse", "Sum of squared vertical differences", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_VSSE }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
600 { "nsse", "Noise preserving sum of squared differences", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_NSSE }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
601 { "dct264", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_DCT264 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
602 { "dctmax", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_DCTMAX }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
603 { "chroma", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_CHROMA }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
604 { "msad", "Sum of absolute differences, median predicted", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_MEDIAN_SAD }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }
605
606 #ifndef FF_MPV_OFFSET
607 #define FF_MPV_OFFSET(x) offsetof(MpegEncContext, x)
608 #endif
609 #define FF_MPV_OPT_FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
610 #define FF_MPV_COMMON_OPTS \
611 FF_MPV_OPT_CMP_FUNC, \
612 { "mpv_flags", "Flags common for all mpegvideo-based encoders.", FF_MPV_OFFSET(mpv_flags), AV_OPT_TYPE_FLAGS, { .i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "mpv_flags" },\
613 { "skip_rd", "RD optimal MB level residual skipping", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_SKIP_RD }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
614 { "strict_gop", "Strictly enforce gop size", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_STRICT_GOP }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
615 { "qp_rd", "Use rate distortion optimization for qp selection", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_QP_RD }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
616 { "cbp_rd", "use rate distortion optimization for CBP", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_CBP_RD }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
617 { "naq", "normalize adaptive quantization", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_NAQ }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
618 { "mv0", "always try a mb with mv=<0,0>", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_MV0 }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
619 { "luma_elim_threshold", "single coefficient elimination threshold for luminance (negative values also consider dc coefficient)",\
620 FF_MPV_OFFSET(luma_elim_threshold), AV_OPT_TYPE_INT, { .i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS },\
621 { "chroma_elim_threshold", "single coefficient elimination threshold for chrominance (negative values also consider dc coefficient)",\
622 FF_MPV_OFFSET(chroma_elim_threshold), AV_OPT_TYPE_INT, { .i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS },\
623 { "quantizer_noise_shaping", NULL, FF_MPV_OFFSET(quantizer_noise_shaping), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, FF_MPV_OPT_FLAGS },\
624 { "error_rate", "Simulate errors in the bitstream to test error concealment.", \
625 FF_MPV_OFFSET(error_rate), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, FF_MPV_OPT_FLAGS },\
626 {"qsquish", "how to keep quantizer between qmin and qmax (0 = clip, 1 = use differentiable function)", \
627 FF_MPV_OFFSET(rc_qsquish), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, 0, 99, FF_MPV_OPT_FLAGS}, \
628 {"rc_qmod_amp", "experimental quantizer modulation", FF_MPV_OFFSET(rc_qmod_amp), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, -FLT_MAX, FLT_MAX, FF_MPV_OPT_FLAGS}, \
629 {"rc_qmod_freq", "experimental quantizer modulation", FF_MPV_OFFSET(rc_qmod_freq), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS}, \
630 {"rc_eq", "Set rate control equation. When computing the expression, besides the standard functions " \
631 "defined in the section 'Expression Evaluation', the following functions are available: " \
632 "bits2qp(bits), qp2bits(qp). Also the following constants are available: iTex pTex tex mv " \
633 "fCode iCount mcVar var isI isP isB avgQP qComp avgIITex avgPITex avgPPTex avgBPTex avgTex.", \
634 FF_MPV_OFFSET(rc_eq), AV_OPT_TYPE_STRING, .flags = FF_MPV_OPT_FLAGS }, \
635 {"rc_init_cplx", "initial complexity for 1-pass encoding", FF_MPV_OFFSET(rc_initial_cplx), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, -FLT_MAX, FLT_MAX, FF_MPV_OPT_FLAGS}, \
636 {"rc_buf_aggressivity", "currently useless", FF_MPV_OFFSET(rc_buffer_aggressivity), AV_OPT_TYPE_FLOAT, {.dbl = 1.0 }, -FLT_MAX, FLT_MAX, FF_MPV_OPT_FLAGS}, \
637 {"border_mask", "increase the quantizer for macroblocks close to borders", FF_MPV_OFFSET(border_masking), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, -FLT_MAX, FLT_MAX, FF_MPV_OPT_FLAGS}, \
638 {"lmin", "minimum Lagrange factor (VBR)", FF_MPV_OFFSET(lmin), AV_OPT_TYPE_INT, {.i64 = 2*FF_QP2LAMBDA }, 0, INT_MAX, FF_MPV_OPT_FLAGS }, \
639 {"lmax", "maximum Lagrange factor (VBR)", FF_MPV_OFFSET(lmax), AV_OPT_TYPE_INT, {.i64 = 31*FF_QP2LAMBDA }, 0, INT_MAX, FF_MPV_OPT_FLAGS }, \
640 {"ibias", "intra quant bias", FF_MPV_OFFSET(intra_quant_bias), AV_OPT_TYPE_INT, {.i64 = FF_DEFAULT_QUANT_BIAS }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
641 {"pbias", "inter quant bias", FF_MPV_OFFSET(inter_quant_bias), AV_OPT_TYPE_INT, {.i64 = FF_DEFAULT_QUANT_BIAS }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
642 {"motion_est", "motion estimation algorithm", FF_MPV_OFFSET(motion_est), AV_OPT_TYPE_INT, {.i64 = FF_ME_EPZS }, FF_ME_ZERO, FF_ME_XONE, FF_MPV_OPT_FLAGS, "motion_est" }, \
643 { "zero", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FF_ME_ZERO }, 0, 0, FF_MPV_OPT_FLAGS, "motion_est" }, \
644 { "epzs", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FF_ME_EPZS }, 0, 0, FF_MPV_OPT_FLAGS, "motion_est" }, \
645 { "xone", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FF_ME_XONE }, 0, 0, FF_MPV_OPT_FLAGS, "motion_est" }, \
646 {"skip_threshold", "Frame skip threshold", FF_MPV_OFFSET(frame_skip_threshold), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
647 {"skip_factor", "Frame skip factor", FF_MPV_OFFSET(frame_skip_factor), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
648 {"skip_exp", "Frame skip exponent", FF_MPV_OFFSET(frame_skip_exp), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
649 {"skip_cmp", "Frame skip compare function", FF_MPV_OFFSET(frame_skip_cmp), AV_OPT_TYPE_INT, {.i64 = FF_CMP_DCTMAX }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
650 {"sc_threshold", "Scene change threshold", FF_MPV_OFFSET(scenechange_threshold), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
651 {"noise_reduction", "Noise reduction", FF_MPV_OFFSET(noise_reduction), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
652 {"ps", "RTP payload size in bytes", FF_MPV_OFFSET(rtp_payload_size), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
653 {"mepc", "Motion estimation bitrate penalty compensation (1.0 = 256)", FF_MPV_OFFSET(me_penalty_compensation), AV_OPT_TYPE_INT, {.i64 = 256 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
654 {"mepre", "pre motion estimation", FF_MPV_OFFSET(me_pre), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
655 {"intra_penalty", "Penalty for intra blocks in block decision", FF_MPV_OFFSET(intra_penalty), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, INT_MAX/2, FF_MPV_OPT_FLAGS }, \
656
657 #define FF_MPV_COMMON_BFRAME_OPTS \
658 {"b_strategy", "Strategy to choose between I/P/B-frames", FF_MPV_OFFSET(b_frame_strategy), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 2, FF_MPV_OPT_FLAGS }, \
659 {"b_sensitivity", "Adjust sensitivity of b_frame_strategy 1", FF_MPV_OFFSET(b_sensitivity), AV_OPT_TYPE_INT, {.i64 = 40 }, 1, INT_MAX, FF_MPV_OPT_FLAGS }, \
660 {"brd_scale", "Downscale frames for dynamic B-frame decision", FF_MPV_OFFSET(brd_scale), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 3, FF_MPV_OPT_FLAGS },
661
662 #if FF_API_MPEGVIDEO_OPTS
663 #define FF_MPV_DEPRECATED_MPEG_QUANT_OPT \
664 { "mpeg_quant", "Deprecated, does nothing", FF_MPV_OFFSET(mpeg_quant), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 0, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED },
665 #define FF_MPV_DEPRECATED_A53_CC_OPT \
666 { "a53cc", "Deprecated, does nothing", FF_MPV_OFFSET(a53_cc), AV_OPT_TYPE_BOOL, { .i64 = 1 }, 0, 1, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED },
667 #define FF_MPV_DEPRECATED_MATRIX_OPT \
668 { "force_duplicated_matrix", "Deprecated, does nothing", FF_MPV_OFFSET(force_duplicated_matrix), AV_OPT_TYPE_BOOL, {.i64 = 0 }, 0, 1, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED },
669 #define FF_MPV_DEPRECATED_BFRAME_OPTS \
670 { "b_strategy", "Deprecated, does nothing", FF_MPV_OFFSET(b_frame_strategy), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 2, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED }, \
671 { "b_sensitivity", "Deprecated, does nothing", FF_MPV_OFFSET(b_sensitivity), AV_OPT_TYPE_INT, { .i64 = 40 }, 1, INT_MAX, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED }, \
672 { "brd_scale", "Deprecated, does nothing", FF_MPV_OFFSET(brd_scale), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 3, FF_MPV_OPT_FLAGS | AV_OPT_FLAG_DEPRECATED },
673 #endif
674
675 extern const AVClass ff_mpv_enc_class;
676
677 /**
678 * Set the given MpegEncContext to common defaults (same for encoding
679 * and decoding). The changed fields will not depend upon the prior
680 * state of the MpegEncContext.
681 */
682 void ff_mpv_common_defaults(MpegEncContext *s);
683
684 void ff_dct_encode_init_x86(MpegEncContext *s);
685
686 int ff_mpv_common_init(MpegEncContext *s);
687 void ff_mpv_common_init_arm(MpegEncContext *s);
688 void ff_mpv_common_init_axp(MpegEncContext *s);
689 void ff_mpv_common_init_neon(MpegEncContext *s);
690 void ff_mpv_common_init_ppc(MpegEncContext *s);
691 void ff_mpv_common_init_x86(MpegEncContext *s);
692 void ff_mpv_common_init_mips(MpegEncContext *s);
693
694 int ff_mpv_common_frame_size_change(MpegEncContext *s);
695 void ff_mpv_common_end(MpegEncContext *s);
696
697 void ff_mpv_decode_init(MpegEncContext *s, AVCodecContext *avctx);
698 void ff_mpv_reconstruct_mb(MpegEncContext *s, int16_t block[12][64]);
699 void ff_mpv_report_decode_progress(MpegEncContext *s);
700
701 int ff_mpv_frame_start(MpegEncContext *s, AVCodecContext *avctx);
702 void ff_mpv_frame_end(MpegEncContext *s);
703
704 int ff_mpv_encode_init(AVCodecContext *avctx);
705 void ff_mpv_encode_init_x86(MpegEncContext *s);
706
707 int ff_mpv_encode_end(AVCodecContext *avctx);
708 int ff_mpv_encode_picture(AVCodecContext *avctx, AVPacket *pkt,
709 const AVFrame *frame, int *got_packet);
710 int ff_mpv_reallocate_putbitbuffer(MpegEncContext *s, size_t threshold, size_t size_increase);
711
712 void ff_clean_intra_table_entries(MpegEncContext *s);
713 void ff_mpeg_draw_horiz_band(MpegEncContext *s, int y, int h);
714 void ff_mpeg_flush(AVCodecContext *avctx);
715
716 void ff_print_debug_info(MpegEncContext *s, Picture *p, AVFrame *pict);
717
718 int ff_mpv_export_qp_table(MpegEncContext *s, AVFrame *f, Picture *p, int qp_type);
719
720 void ff_write_quant_matrix(PutBitContext *pb, uint16_t *matrix);
721
722 int ff_update_duplicate_context(MpegEncContext *dst, MpegEncContext *src);
723 int ff_mpeg_update_thread_context(AVCodecContext *dst, const AVCodecContext *src);
724 void ff_set_qscale(MpegEncContext * s, int qscale);
725
726 void ff_mpv_idct_init(MpegEncContext *s);
727 int ff_dct_encode_init(MpegEncContext *s);
728 void ff_convert_matrix(MpegEncContext *s, int (*qmat)[64], uint16_t (*qmat16)[2][64],
729 const uint16_t *quant_matrix, int bias, int qmin, int qmax, int intra);
730 int ff_dct_quantize_c(MpegEncContext *s, int16_t *block, int n, int qscale, int *overflow);
731 void ff_block_permute(int16_t *block, uint8_t *permutation,
732 const uint8_t *scantable, int last);
733 void ff_init_block_index(MpegEncContext *s);
734
735 void ff_mpv_motion(MpegEncContext *s,
736 uint8_t *dest_y, uint8_t *dest_cb,
737 uint8_t *dest_cr, int dir,
738 uint8_t **ref_picture,
739 op_pixels_func (*pix_op)[4],
740 qpel_mc_func (*qpix_op)[16]);
741
742 6616657 static inline void ff_update_block_index(MpegEncContext *s){
743
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 6616657 times.
 6616657 const int bytes_per_pixel = 1 + (s->avctx->bits_per_raw_sample > 8);
744 6616657 const int block_size= (8*bytes_per_pixel) >> s->avctx->lowres;
745
746 6616657 s->block_index[0]+=2;
747 6616657 s->block_index[1]+=2;
748 6616657 s->block_index[2]+=2;
749 6616657 s->block_index[3]+=2;
750 6616657 s->block_index[4]++;
751 6616657 s->block_index[5]++;
752 6616657 s->dest[0]+= 2*block_size;
753 6616657 s->dest[1]+= (2 >> s->chroma_x_shift) * block_size;
754 6616657 s->dest[2]+= (2 >> s->chroma_x_shift) * block_size;
755 6616657 }
756
757 4992717 static inline int get_bits_diff(MpegEncContext *s){
758 4992717 const int bits= put_bits_count(&s->pb);
759 4992717 const int last= s->last_bits;
760
761 4992717 s->last_bits = bits;
762
763 4992717 return bits - last;
764 }
765
766 271837 static inline int mpeg_get_qscale(MpegEncContext *s)
767 {
768 271837 int qscale = get_bits(&s->gb, 5);
769
2/2
✓ Branch 0 taken 210729 times.
✓ Branch 1 taken 61108 times.
 271837 if (s->q_scale_type)
770 210729 return ff_mpeg2_non_linear_qscale[qscale];
771 else
772 61108 return qscale << 1;
773 }
774
775 #endif /* AVCODEC_MPEGVIDEO_H */
776