932a6f2b34c575c5818e3a2213ffa659e60e46c6
[libav.git] / libavcodec / mpegvideo.h
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 Libav.
7 *
8 * Libav 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 * Libav 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 Libav; 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 "h263dsp.h"
39 #include "hpeldsp.h"
40 #include "idctdsp.h"
41 #include "internal.h"
42 #include "me_cmp.h"
43 #include "motion_est.h"
44 #include "mpegpicture.h"
45 #include "mpegvideodsp.h"
46 #include "mpegvideoencdsp.h"
47 #include "pixblockdsp.h"
48 #include "put_bits.h"
49 #include "ratecontrol.h"
50 #include "parser.h"
51 #include "mpegutils.h"
52 #include "mpeg12data.h"
53 #include "qpeldsp.h"
54 #include "thread.h"
55 #include "videodsp.h"
56
57 #include "libavutil/opt.h"
58
59 #define MAX_THREADS 16
60
61 #define MAX_B_FRAMES 16
62
63 /* Start codes. */
64 #define SEQ_END_CODE 0x000001b7
65 #define SEQ_START_CODE 0x000001b3
66 #define GOP_START_CODE 0x000001b8
67 #define PICTURE_START_CODE 0x00000100
68 #define SLICE_MIN_START_CODE 0x00000101
69 #define SLICE_MAX_START_CODE 0x000001af
70 #define EXT_START_CODE 0x000001b5
71 #define USER_START_CODE 0x000001b2
72
73 /**
74 * MpegEncContext.
75 */
76 typedef struct MpegEncContext {
77 AVClass *class;
78
79 int y_dc_scale, c_dc_scale;
80 int ac_pred;
81 int block_last_index[12]; ///< last non zero coefficient in block
82 int h263_aic; ///< Advanced INTRA Coding (AIC)
83
84 /* scantables */
85 ScanTable inter_scantable; ///< if inter == intra then intra should be used to reduce the cache usage
86 ScanTable intra_scantable;
87 ScanTable intra_h_scantable;
88 ScanTable intra_v_scantable;
89
90 /* WARNING: changes above this line require updates to hardcoded
91 * offsets used in ASM. */
92
93 struct AVCodecContext *avctx;
94 /* the following parameters must be initialized before encoding */
95 int width, height;///< picture size. must be a multiple of 16
96 int gop_size;
97 int intra_only; ///< if true, only intra pictures are generated
98 int bit_rate; ///< wanted bit rate
99 enum OutputFormat out_format; ///< output format
100 int h263_pred; ///< use MPEG-4/H.263 ac/dc predictions
101 int pb_frame; ///< PB-frame mode (0 = none, 1 = base, 2 = improved)
102
103 /* the following codec id fields are deprecated in favor of codec_id */
104 int h263_plus; ///< H.263+ headers
105 int h263_flv; ///< use flv H.263 header
106
107 enum AVCodecID codec_id; /* see AV_CODEC_ID_xxx */
108 int fixed_qscale; ///< fixed qscale if non zero
109 int encoding; ///< true if we are encoding (vs decoding)
110 int max_b_frames; ///< max number of B-frames for encoding
111 int luma_elim_threshold;
112 int chroma_elim_threshold;
113 int strict_std_compliance; ///< strictly follow the std (MPEG-4, ...)
114 int workaround_bugs; ///< workaround bugs in encoders which cannot be detected automatically
115 int codec_tag; ///< internal codec_tag upper case converted from avctx codec_tag
116 /* the following fields are managed internally by the encoder */
117
118 /* sequence parameters */
119 int context_initialized;
120 int input_picture_number; ///< used to set pic->display_picture_number, should not be used for/by anything else
121 int coded_picture_number; ///< used to set pic->coded_picture_number, should not be used for/by anything else
122 int picture_number; //FIXME remove, unclear definition
123 int picture_in_gop_number; ///< 0-> first pic in gop, ...
124 int mb_width, mb_height; ///< number of MBs horizontally & vertically
125 int mb_stride; ///< mb_width+1 used for some arrays to allow simple addressing of left & top MBs without sig11
126 int b8_stride; ///< 2*mb_width+1 used for some 8x8 block arrays to allow simple addressing
127 int h_edge_pos, v_edge_pos;///< horizontal / vertical position of the right/bottom edge (pixel replication)
128 int mb_num; ///< number of MBs of a picture
129 ptrdiff_t linesize; ///< line size, in bytes, may be different from width
130 ptrdiff_t uvlinesize; ///< line size, for chroma in bytes, may be different from width
131 Picture *picture; ///< main picture buffer
132 Picture **input_picture; ///< next pictures on display order for encoding
133 Picture **reordered_input_picture; ///< pointer to the next pictures in coded order for encoding
134
135 int64_t user_specified_pts; ///< last non-zero pts from AVFrame which was passed into avcodec_encode_video2()
136 /**
137 * pts difference between the first and second input frame, used for
138 * calculating dts of the first frame when there's a delay */
139 int64_t dts_delta;
140 /**
141 * reordered pts to be used as dts for the next output frame when there's
142 * a delay */
143 int64_t reordered_pts;
144
145 /** bit output */
146 PutBitContext pb;
147
148 int start_mb_y; ///< start mb_y of this thread (so current thread should process start_mb_y <= row < end_mb_y)
149 int end_mb_y; ///< end mb_y of this thread (so current thread should process start_mb_y <= row < end_mb_y)
150 struct MpegEncContext *thread_context[MAX_THREADS];
151 int slice_context_count; ///< number of used thread_contexts
152
153 /**
154 * copy of the previous picture structure.
155 * note, linesize & data, might not match the previous picture (for field pictures)
156 */
157 Picture last_picture;
158
159 /**
160 * copy of the next picture structure.
161 * note, linesize & data, might not match the next picture (for field pictures)
162 */
163 Picture next_picture;
164
165 /**
166 * copy of the source picture structure for encoding.
167 * note, linesize & data, might not match the source picture (for field pictures)
168 */
169 Picture new_picture;
170
171 /**
172 * copy of the current picture structure.
173 * note, linesize & data, might not match the current picture (for field pictures)
174 */
175 Picture current_picture; ///< buffer to store the decompressed current picture
176
177 Picture *last_picture_ptr; ///< pointer to the previous picture.
178 Picture *next_picture_ptr; ///< pointer to the next picture (for bidir pred)
179 Picture *current_picture_ptr; ///< pointer to the current picture
180 int last_dc[3]; ///< last DC values for MPEG-1
181 int16_t *dc_val_base;
182 int16_t *dc_val[3]; ///< used for MPEG-4 DC prediction, all 3 arrays must be continuous
183 const uint8_t *y_dc_scale_table; ///< qscale -> y_dc_scale table
184 const uint8_t *c_dc_scale_table; ///< qscale -> c_dc_scale table
185 const uint8_t *chroma_qscale_table; ///< qscale -> chroma_qscale (H.263)
186 uint8_t *coded_block_base;
187 uint8_t *coded_block; ///< used for coded block pattern prediction (msmpeg4v3, wmv1)
188 int16_t (*ac_val_base)[16];
189 int16_t (*ac_val[3])[16]; ///< used for for MPEG-4 AC prediction, all 3 arrays must be continuous
190 int mb_skipped; ///< MUST BE SET only during DECODING
191 uint8_t *mbskip_table; /**< used to avoid copy if macroblock skipped (for black regions for example)
192 and used for B-frame encoding & decoding (contains skip table of next P-frame) */
193 uint8_t *mbintra_table; ///< used to avoid setting {ac, dc, cbp}-pred stuff to zero on inter MB decoding
194 uint8_t *cbp_table; ///< used to store cbp, ac_pred for partitioned decoding
195 uint8_t *pred_dir_table; ///< used to store pred_dir for partitioned decoding
196
197 ScratchpadContext sc;
198
199 int qscale; ///< QP
200 int chroma_qscale; ///< chroma QP
201 unsigned int lambda; ///< Lagrange multiplier used in rate distortion
202 unsigned int lambda2; ///< (lambda*lambda) >> FF_LAMBDA_SHIFT
203 int *lambda_table;
204 int adaptive_quant; ///< use adaptive quantization
205 int dquant; ///< qscale difference to prev qscale
206 int pict_type; ///< AV_PICTURE_TYPE_I, AV_PICTURE_TYPE_P, AV_PICTURE_TYPE_B, ...
207 int last_pict_type; //FIXME removes
208 int last_non_b_pict_type; ///< used for MPEG-4 gmc B-frames & ratecontrol
209 int droppable;
210 int frame_rate_index;
211 int last_lambda_for[5]; ///< last lambda for a specific pict type
212 int skipdct; ///< skip dct and code zero residual
213
214 /* motion compensation */
215 int unrestricted_mv; ///< mv can point outside of the coded picture
216 int h263_long_vectors; ///< use horrible H.263v1 long vector mode
217
218 BlockDSPContext bdsp;
219 FDCTDSPContext fdsp;
220 HpelDSPContext hdsp;
221 IDCTDSPContext idsp;
222 MECmpContext mecc;
223 MpegVideoDSPContext mdsp;
224 MpegvideoEncDSPContext mpvencdsp;
225 PixblockDSPContext pdsp;
226 QpelDSPContext qdsp;
227 VideoDSPContext vdsp;
228 H263DSPContext h263dsp;
229 int f_code; ///< forward MV resolution
230 int b_code; ///< backward MV resolution for B-frames (MPEG-4)
231 int16_t (*p_mv_table_base)[2];
232 int16_t (*b_forw_mv_table_base)[2];
233 int16_t (*b_back_mv_table_base)[2];
234 int16_t (*b_bidir_forw_mv_table_base)[2];
235 int16_t (*b_bidir_back_mv_table_base)[2];
236 int16_t (*b_direct_mv_table_base)[2];
237 int16_t (*p_field_mv_table_base[2][2])[2];
238 int16_t (*b_field_mv_table_base[2][2][2])[2];
239 int16_t (*p_mv_table)[2]; ///< MV table (1MV per MB) P-frame encoding
240 int16_t (*b_forw_mv_table)[2]; ///< MV table (1MV per MB) forward mode B-frame encoding
241 int16_t (*b_back_mv_table)[2]; ///< MV table (1MV per MB) backward mode B-frame encoding
242 int16_t (*b_bidir_forw_mv_table)[2]; ///< MV table (1MV per MB) bidir mode B-frame encoding
243 int16_t (*b_bidir_back_mv_table)[2]; ///< MV table (1MV per MB) bidir mode B-frame encoding
244 int16_t (*b_direct_mv_table)[2]; ///< MV table (1MV per MB) direct mode B-frame encoding
245 int16_t (*p_field_mv_table[2][2])[2]; ///< MV table (2MV per MB) interlaced P-frame encoding
246 int16_t (*b_field_mv_table[2][2][2])[2];///< MV table (4MV per MB) interlaced B-frame encoding
247 uint8_t (*p_field_select_table[2]);
248 uint8_t (*b_field_select_table[2][2]);
249 #if FF_API_MOTION_EST
250 int me_method; ///< ME algorithm
251 #endif
252 int motion_est; ///< ME algorithm
253 int me_penalty_compensation;
254 int me_pre; ///< prepass for motion estimation
255 int mv_dir;
256 #define MV_DIR_FORWARD 1
257 #define MV_DIR_BACKWARD 2
258 #define MV_DIRECT 4 ///< bidirectional mode where the difference equals the MV of the last P/S/I-Frame (MPEG-4)
259 int mv_type;
260 #define MV_TYPE_16X16 0 ///< 1 vector for the whole mb
261 #define MV_TYPE_8X8 1 ///< 4 vectors (H.263, MPEG-4 4MV)
262 #define MV_TYPE_16X8 2 ///< 2 vectors, one per 16x8 block
263 #define MV_TYPE_FIELD 3 ///< 2 vectors, one per field
264 #define MV_TYPE_DMV 4 ///< 2 vectors, special mpeg2 Dual Prime Vectors
265 /**motion vectors for a macroblock
266 first coordinate : 0 = forward 1 = backward
267 second " : depend on type
268 third " : 0 = x, 1 = y
269 */
270 int mv[2][4][2];
271 int field_select[2][2];
272 int last_mv[2][2][2]; ///< last MV, used for MV prediction in MPEG-1 & B-frame MPEG-4
273 uint8_t *fcode_tab; ///< smallest fcode needed for each MV
274 int16_t direct_scale_mv[2][64]; ///< precomputed to avoid divisions in ff_mpeg4_set_direct_mv
275
276 MotionEstContext me;
277
278 int no_rounding; /**< apply no rounding to motion compensation (MPEG-4, msmpeg4, ...)
279 for B-frames rounding mode is always 0 */
280
281 /* macroblock layer */
282 int mb_x, mb_y;
283 int mb_skip_run;
284 int mb_intra;
285 uint16_t *mb_type; ///< Table for candidate MB types for encoding (defines in mpegutils.h)
286
287 int block_index[6]; ///< index to current MB in block based arrays with edges
288 int block_wrap[6];
289 uint8_t *dest[3];
290
291 int *mb_index2xy; ///< mb_index -> mb_x + mb_y*mb_stride
292
293 /** matrix transmitted in the bitstream */
294 uint16_t intra_matrix[64];
295 uint16_t chroma_intra_matrix[64];
296 uint16_t inter_matrix[64];
297 uint16_t chroma_inter_matrix[64];
298
299 int intra_quant_bias; ///< bias for the quantizer
300 int inter_quant_bias; ///< bias for the quantizer
301 int min_qcoeff; ///< minimum encodable coefficient
302 int max_qcoeff; ///< maximum encodable coefficient
303 int ac_esc_length; ///< num of bits needed to encode the longest esc
304 uint8_t *intra_ac_vlc_length;
305 uint8_t *intra_ac_vlc_last_length;
306 uint8_t *inter_ac_vlc_length;
307 uint8_t *inter_ac_vlc_last_length;
308 uint8_t *luma_dc_vlc_length;
309 #define UNI_AC_ENC_INDEX(run,level) ((run)*128 + (level))
310
311 int coded_score[8];
312
313 /** precomputed matrix (combine qscale and DCT renorm) */
314 int (*q_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_inter_matrix16)[2][64];
319
320 /* noise reduction */
321 int (*dct_error_sum)[64];
322 int dct_count[2];
323 uint16_t (*dct_offset)[64];
324
325 /* bit rate control */
326 int64_t total_bits;
327 int frame_bits; ///< bits used for the current frame
328 int next_lambda; ///< next lambda used for retrying to encode a frame
329 RateControlContext rc_context; ///< contains stuff only accessed in ratecontrol.c
330
331 /* statistics, used for 2-pass encoding */
332 int mv_bits;
333 int header_bits;
334 int i_tex_bits;
335 int p_tex_bits;
336 int i_count;
337 int f_count;
338 int b_count;
339 int skip_count;
340 int misc_bits; ///< cbp, mb_type
341 int last_bits; ///< temp var used for calculating the above vars
342
343 /* error concealment / resync */
344 int resync_mb_x; ///< x position of last resync marker
345 int resync_mb_y; ///< y position of last resync marker
346 GetBitContext last_resync_gb; ///< used to search for the next resync marker
347 int mb_num_left; ///< number of MBs left in this video packet (for partitioned Slices only)
348 int next_p_frame_damaged; ///< set if the next p frame is damaged, to avoid showing trashed B-frames
349
350 ParseContext parse_context;
351
352 /* H.263 specific */
353 int gob_index;
354 int obmc; ///< overlapped block motion compensation
355 int mb_info; ///< interval for outputting info about mb offsets as side data
356 int prev_mb_info, last_mb_info;
357 uint8_t *mb_info_ptr;
358 int mb_info_size;
359 int rc_strategy;
360
361 /* H.263+ specific */
362 int umvplus; ///< == H.263+ && unrestricted_mv
363 int h263_aic_dir; ///< AIC direction: 0 = left, 1 = top
364 int h263_slice_structured;
365 int alt_inter_vlc; ///< alternative inter vlc
366 int modified_quant;
367 int loop_filter;
368 int custom_pcf;
369
370 /* MPEG-4 specific */
371 ///< number of bits to represent the fractional part of time (encoder only)
372 int time_increment_bits;
373 int last_time_base;
374 int time_base; ///< time in seconds of last I,P,S Frame
375 int64_t time; ///< time of current frame
376 int64_t last_non_b_time;
377 uint16_t pp_time; ///< time distance between the last 2 p,s,i frames
378 uint16_t pb_time; ///< time distance between the last b and p,s,i frame
379 uint16_t pp_field_time;
380 uint16_t pb_field_time; ///< like above, just for interlaced
381 int real_sprite_warping_points;
382 int sprite_offset[2][2]; ///< sprite offset[isChroma][isMVY]
383 int sprite_delta[2][2]; ///< sprite_delta [isY][isMVY]
384 int mcsel;
385 int quant_precision;
386 int quarter_sample; ///< 1->qpel, 0->half pel ME/MC
387 int aspect_ratio_info; //FIXME remove
388 int sprite_warping_accuracy;
389 int data_partitioning; ///< data partitioning flag from header
390 int partitioned_frame; ///< is current frame partitioned
391 int low_delay; ///< no reordering needed / has no B-frames
392 int vo_type;
393 PutBitContext tex_pb; ///< used for data partitioned VOPs
394 PutBitContext pb2; ///< used for data partitioned VOPs
395 int mpeg_quant;
396 int padding_bug_score; ///< used to detect the VERY common padding bug in MPEG-4
397
398 /* divx specific, used to workaround (many) bugs in divx5 */
399 int divx_packed;
400 uint8_t *bitstream_buffer; //Divx 5.01 puts several frames in a single one, this is used to reorder them
401 int bitstream_buffer_size;
402 unsigned int allocated_bitstream_buffer_size;
403
404 /* RV10 specific */
405 int rv10_version; ///< RV10 version: 0 or 3
406 int rv10_first_dc_coded[3];
407
408 /* MJPEG specific */
409 struct MJpegContext *mjpeg_ctx;
410 int pred;
411
412 /* MSMPEG4 specific */
413 int mv_table_index;
414 int rl_table_index;
415 int rl_chroma_table_index;
416 int dc_table_index;
417 int use_skip_mb_code;
418 int slice_height; ///< in macroblocks
419 int first_slice_line; ///< used in MPEG-4 too to handle resync markers
420 int flipflop_rounding;
421 int msmpeg4_version; ///< 0=not msmpeg4, 1=mp41, 2=mp42, 3=mp43/divx3 4=wmv1/7 5=wmv2/8
422 int per_mb_rl_table;
423 int esc3_level_length;
424 int esc3_run_length;
425 /** [mb_intra][isChroma][level][run][last] */
426 int (*ac_stats)[2][MAX_LEVEL+1][MAX_RUN+1][2];
427 int inter_intra_pred;
428 int mspel;
429
430 /* decompression specific */
431 GetBitContext gb;
432
433 /* MPEG-1 specific */
434 int gop_picture_number; ///< index of the first picture of a GOP based on fake_pic_num & MPEG-1 specific
435 int last_mv_dir; ///< last mv_dir, used for B-frame encoding
436 uint8_t *vbv_delay_ptr; ///< pointer to vbv_delay in the bitstream
437
438 /* MPEG-2-specific - I wished not to have to support this mess. */
439 int progressive_sequence;
440 int mpeg_f_code[2][2];
441
442 // picture structure defines are loaded from mpegutils.h
443 int picture_structure;
444
445 int64_t timecode_frame_start; ///< GOP timecode frame start number, in non drop frame format
446 int intra_dc_precision;
447 int frame_pred_frame_dct;
448 int top_field_first;
449 int concealment_motion_vectors;
450 int q_scale_type;
451 int brd_scale;
452 int intra_vlc_format;
453 int alternate_scan;
454 int repeat_first_field;
455 int chroma_420_type;
456 int chroma_format;
457 #define CHROMA_420 1
458 #define CHROMA_422 2
459 int chroma_x_shift;//depend on pix_format, that depend on chroma_format
460 int chroma_y_shift;
461
462 int progressive_frame;
463 int full_pel[2];
464 int interlaced_dct;
465 int first_field; ///< is 1 for the first field of a field picture 0 otherwise
466 int drop_frame_timecode; ///< timecode is in drop frame format.
467 int scan_offset; ///< reserve space for SVCD scan offset user data.
468
469 /* RTP specific */
470 int rtp_mode;
471 int rtp_payload_size;
472
473 uint8_t *ptr_lastgob;
474 int16_t (*pblocks[12])[64];
475
476 int16_t (*block)[64]; ///< points to one of the following blocks
477 int16_t (*blocks)[12][64]; // for HQ mode we need to keep the best block
478 int (*decode_mb)(struct MpegEncContext *s, int16_t block[6][64]); // used by some codecs to avoid a switch()
479 #define SLICE_OK 0
480 #define SLICE_ERROR -1
481 #define SLICE_END -2 ///<end marker found
482 #define SLICE_NOEND -3 ///<no end marker or error found but mb count exceeded
483
484 void (*dct_unquantize_mpeg1_intra)(struct MpegEncContext *s,
485 int16_t *block/*align 16*/, int n, int qscale);
486 void (*dct_unquantize_mpeg1_inter)(struct MpegEncContext *s,
487 int16_t *block/*align 16*/, int n, int qscale);
488 void (*dct_unquantize_mpeg2_intra)(struct MpegEncContext *s,
489 int16_t *block/*align 16*/, int n, int qscale);
490 void (*dct_unquantize_mpeg2_inter)(struct MpegEncContext *s,
491 int16_t *block/*align 16*/, int n, int qscale);
492 void (*dct_unquantize_h263_intra)(struct MpegEncContext *s,
493 int16_t *block/*align 16*/, int n, int qscale);
494 void (*dct_unquantize_h263_inter)(struct MpegEncContext *s,
495 int16_t *block/*align 16*/, int n, int qscale);
496 void (*dct_unquantize_intra)(struct MpegEncContext *s, // unquantizer to use (MPEG-4 can use both)
497 int16_t *block/*align 16*/, int n, int qscale);
498 void (*dct_unquantize_inter)(struct MpegEncContext *s, // unquantizer to use (MPEG-4 can use both)
499 int16_t *block/*align 16*/, int n, int qscale);
500 int (*dct_quantize)(struct MpegEncContext *s, int16_t *block/*align 16*/, int n, int qscale, int *overflow);
501 int (*fast_dct_quantize)(struct MpegEncContext *s, int16_t *block/*align 16*/, int n, int qscale, int *overflow);
502 void (*denoise_dct)(struct MpegEncContext *s, int16_t *block);
503
504 int mpv_flags; ///< flags set by private options
505 int quantizer_noise_shaping;
506
507 /**
508 * ratecontrol qmin qmax limiting method
509 * 0-> clipping, 1-> use a nice continuous function to limit qscale within qmin/qmax.
510 */
511 float rc_qsquish;
512 float rc_qmod_amp;
513 int rc_qmod_freq;
514 float rc_initial_cplx;
515 float rc_buffer_aggressivity;
516 float border_masking;
517 int lmin, lmax;
518
519 char *rc_eq;
520
521 /* temp buffers for rate control */
522 float *cplx_tab, *bits_tab;
523
524 /* flag to indicate a reinitialization is required, e.g. after
525 * a frame size change */
526 int context_reinit;
527
528 ERContext er;
529
530 int error_rate;
531
532 /* temporary frames used by b_frame_strategy = 2 */
533 AVFrame *tmp_frames[MAX_B_FRAMES + 2];
534 int b_frame_strategy;
535 int b_sensitivity;
536
537 /* frame skip options for encoding */
538 int frame_skip_threshold;
539 int frame_skip_factor;
540 int frame_skip_exp;
541 int frame_skip_cmp;
542
543 int scenechange_threshold;
544 int noise_reduction;
545 } MpegEncContext;
546
547 /* mpegvideo_enc common options */
548 #define FF_MPV_FLAG_SKIP_RD 0x0001
549 #define FF_MPV_FLAG_STRICT_GOP 0x0002
550 #define FF_MPV_FLAG_QP_RD 0x0004
551 #define FF_MPV_FLAG_CBP_RD 0x0008
552 #define FF_MPV_FLAG_NAQ 0x0010
553 #define FF_MPV_FLAG_MV0 0x0020
554
555 #define FF_MPV_OPT_CMP_FUNC \
556 { "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" }, \
557 { "sse", "Sum of squared errors", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_SSE }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
558 { "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" }, \
559 { "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" }, \
560 { "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" }, \
561 { "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" }, \
562 { "rd", "Rate distortion optimal, slow", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_RD }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
563 { "zero", "Zero", 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_ZERO }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
564 { "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" }, \
565 { "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" }, \
566 { "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" }, \
567 { "dct264", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_DCT264 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
568 { "dctmax", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_DCTMAX }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }, \
569 { "chroma", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_CMP_CHROMA }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS, "cmp_func" }
570
571 #define FF_MPV_OFFSET(x) offsetof(MpegEncContext, x)
572 #define FF_MPV_OPT_FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
573 #define FF_MPV_COMMON_OPTS \
574 FF_MPV_OPT_CMP_FUNC, \
575 { "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" },\
576 { "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" },\
577 { "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" },\
578 { "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" },\
579 { "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" },\
580 { "naq", "normalize adaptive quantization", 0, AV_OPT_TYPE_CONST, { .i64 = FF_MPV_FLAG_NAQ }, 0, 0, FF_MPV_OPT_FLAGS, "mpv_flags" },\
581 { "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" },\
582 { "luma_elim_threshold", "single coefficient elimination threshold for luminance (negative values also consider dc coefficient)",\
583 FF_MPV_OFFSET(luma_elim_threshold), AV_OPT_TYPE_INT, { .i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS },\
584 { "chroma_elim_threshold", "single coefficient elimination threshold for chrominance (negative values also consider dc coefficient)",\
585 FF_MPV_OFFSET(chroma_elim_threshold), AV_OPT_TYPE_INT, { .i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS },\
586 { "quantizer_noise_shaping", NULL, FF_MPV_OFFSET(quantizer_noise_shaping), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, FF_MPV_OPT_FLAGS },\
587 { "error_rate", "Simulate errors in the bitstream to test error concealment.", \
588 FF_MPV_OFFSET(error_rate), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, FF_MPV_OPT_FLAGS },\
589 {"qsquish", "how to keep quantizer between qmin and qmax (0 = clip, 1 = use differentiable function)", \
590 FF_MPV_OFFSET(rc_qsquish), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, 0, 99, FF_MPV_OPT_FLAGS}, \
591 {"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}, \
592 {"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}, \
593 {"rc_eq", "Set rate control equation. When computing the expression, besides the standard functions " \
594 "defined in the section 'Expression Evaluation', the following functions are available: " \
595 "bits2qp(bits), qp2bits(qp). Also the following constants are available: iTex pTex tex mv " \
596 "fCode iCount mcVar var isI isP isB avgQP qComp avgIITex avgPITex avgPPTex avgBPTex avgTex.", \
597 FF_MPV_OFFSET(rc_eq), AV_OPT_TYPE_STRING, .flags = FF_MPV_OPT_FLAGS }, \
598 {"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}, \
599 {"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}, \
600 {"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}, \
601 {"lmin", "minimum Lagrange factor (VBR)", FF_MPV_OFFSET(lmin), AV_OPT_TYPE_INT, {.i64 = 2*FF_QP2LAMBDA }, 0, INT_MAX, FF_MPV_OPT_FLAGS }, \
602 {"lmax", "maximum Lagrange factor (VBR)", FF_MPV_OFFSET(lmax), AV_OPT_TYPE_INT, {.i64 = 31*FF_QP2LAMBDA }, 0, INT_MAX, FF_MPV_OPT_FLAGS }, \
603 {"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 }, \
604 {"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 }, \
605 {"rc_strategy", "ratecontrol method", FF_MPV_OFFSET(rc_strategy), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 1, FF_MPV_OPT_FLAGS }, \
606 {"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" }, \
607 { "zero", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FF_ME_ZERO }, 0, 0, FF_MPV_OPT_FLAGS, "motion_est" }, \
608 { "epzs", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FF_ME_EPZS }, 0, 0, FF_MPV_OPT_FLAGS, "motion_est" }, \
609 { "xone", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FF_ME_XONE }, 0, 0, FF_MPV_OPT_FLAGS, "motion_est" }, \
610 {"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 }, \
611 {"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 }, \
612 {"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 }, \
613 {"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 }, \
614 {"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 }, \
615 {"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 }, \
616 {"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" }, \
617 {"sc_threshold", "Scene change threshold", FF_MPV_OFFSET(scenechange_threshold), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
618 {"noise_reduction", "Noise reduction", FF_MPV_OFFSET(noise_reduction), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
619 {"mpeg_quant", "Use MPEG quantizers instead of H.263", FF_MPV_OFFSET(mpeg_quant), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 1, FF_MPV_OPT_FLAGS }, \
620 {"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 }, \
621 {"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 }, \
622 {"mepre", "pre motion estimation", FF_MPV_OFFSET(me_pre), AV_OPT_TYPE_INT, {.i64 = 0 }, INT_MIN, INT_MAX, FF_MPV_OPT_FLAGS }, \
623
624 extern const AVOption ff_mpv_generic_options[];
625
626 /**
627 * Set the given MpegEncContext to common defaults (same for encoding
628 * and decoding). The changed fields will not depend upon the prior
629 * state of the MpegEncContext.
630 */
631 void ff_mpv_common_defaults(MpegEncContext *s);
632
633 int ff_mpv_common_init(MpegEncContext *s);
634 void ff_mpv_common_init_arm(MpegEncContext *s);
635 void ff_mpv_common_init_neon(MpegEncContext *s);
636 void ff_mpv_common_init_ppc(MpegEncContext *s);
637 void ff_mpv_common_init_x86(MpegEncContext *s);
638
639 int ff_mpv_common_frame_size_change(MpegEncContext *s);
640 void ff_mpv_common_end(MpegEncContext *s);
641
642 void ff_mpv_decode_defaults(MpegEncContext *s);
643 void ff_mpv_decode_mb(MpegEncContext *s, int16_t block[12][64]);
644 void ff_mpv_report_decode_progress(MpegEncContext *s);
645
646 int ff_mpv_frame_start(MpegEncContext *s, AVCodecContext *avctx);
647 void ff_mpv_frame_end(MpegEncContext *s);
648
649 int ff_mpv_encode_init(AVCodecContext *avctx);
650 void ff_mpv_encode_init_x86(MpegEncContext *s);
651
652 int ff_mpv_encode_end(AVCodecContext *avctx);
653 int ff_mpv_encode_picture(AVCodecContext *avctx, AVPacket *pkt,
654 const AVFrame *frame, int *got_packet);
655
656 void ff_clean_intra_table_entries(MpegEncContext *s);
657 void ff_mpeg_draw_horiz_band(MpegEncContext *s, int y, int h);
658 void ff_mpeg_flush(AVCodecContext *avctx);
659 void ff_print_debug_info(MpegEncContext *s, Picture *p);
660 void ff_write_quant_matrix(PutBitContext *pb, uint16_t *matrix);
661
662 int ff_update_duplicate_context(MpegEncContext *dst, MpegEncContext *src);
663 int ff_mpeg_update_thread_context(AVCodecContext *dst, const AVCodecContext *src);
664 void ff_set_qscale(MpegEncContext * s, int qscale);
665
666 void ff_mpv_idct_init(MpegEncContext *s);
667 void ff_convert_matrix(MpegEncContext *s, int (*qmat)[64], uint16_t (*qmat16)[2][64],
668 const uint16_t *quant_matrix, int bias, int qmin, int qmax, int intra);
669 int ff_dct_quantize_c(MpegEncContext *s, int16_t *block, int n, int qscale, int *overflow);
670
671 void ff_init_block_index(MpegEncContext *s);
672
673 void ff_mpv_motion(MpegEncContext *s,
674 uint8_t *dest_y, uint8_t *dest_cb,
675 uint8_t *dest_cr, int dir,
676 uint8_t **ref_picture,
677 op_pixels_func (*pix_op)[4],
678 qpel_mc_func (*qpix_op)[16]);
679
680 static inline void ff_update_block_index(MpegEncContext *s){
681 const int block_size = 8;
682
683 s->block_index[0]+=2;
684 s->block_index[1]+=2;
685 s->block_index[2]+=2;
686 s->block_index[3]+=2;
687 s->block_index[4]++;
688 s->block_index[5]++;
689 s->dest[0]+= 2*block_size;
690 s->dest[1]+= block_size;
691 s->dest[2]+= block_size;
692 }
693
694 static inline int get_bits_diff(MpegEncContext *s){
695 const int bits= put_bits_count(&s->pb);
696 const int last= s->last_bits;
697
698 s->last_bits = bits;
699
700 return bits - last;
701 }
702
703 #endif /* AVCODEC_MPEGVIDEO_H */