1 /*****************************************************************************
2 * x264.h: x264 public header
3 *****************************************************************************
4 * Copyright (C) 2003-2013 x264 project
6 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
7 * Loren Merritt <lorenm@u.washington.edu>
8 * Fiona Glaser <fiona@x264.com>
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation; either version 2 of the License, or
13 * (at your option) any later version.
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with this program; if not, write to the Free Software
22 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
24 * This program is also available under a commercial proprietary license.
25 * For more information, contact us at licensing@x264.com.
26 *****************************************************************************/
31 #if !defined(_STDINT_H) && !defined(_STDINT_H_) && !defined(_STDINT_H_INCLUDED) &&\
32 !defined(_INTTYPES_H) && !defined(_INTTYPES_H_)
34 # pragma message("You must include stdint.h or inttypes.h before x264.h")
36 # warning You must include stdint.h or inttypes.h before x264.h
42 #include "x264_config.h"
44 #define X264_BUILD 137
46 /* Application developers planning to link against a shared library version of
47 * libx264 from a Microsoft Visual Studio or similar development environment
48 * will need to define X264_API_IMPORTS before including this header.
49 * This clause does not apply to MinGW, similar development environments, or non
50 * Windows platforms. */
51 #ifdef X264_API_IMPORTS
52 #define X264_API __declspec(dllimport)
58 * opaque handler for encoder */
59 typedef struct x264_t x264_t;
61 /****************************************************************************
62 * NAL structure and functions
63 ****************************************************************************/
72 NAL_SLICE_IDR = 5, /* ref_idc != 0 */
73 NAL_SEI = 6, /* ref_idc == 0 */
78 /* ref_idc == 0 for 6,9,10,11,12 */
82 NAL_PRIORITY_DISPOSABLE = 0,
84 NAL_PRIORITY_HIGH = 2,
85 NAL_PRIORITY_HIGHEST = 3,
88 /* The data within the payload is already NAL-encapsulated; the ref_idc and type
89 * are merely in the struct for easy access by the calling application.
90 * All data returned in an x264_nal_t, including the data in p_payload, is no longer
91 * valid after the next call to x264_encoder_encode. Thus it must be used or copied
92 * before calling x264_encoder_encode or x264_encoder_headers again. */
95 int i_ref_idc; /* nal_priority_e */
96 int i_type; /* nal_unit_type_e */
98 int i_first_mb; /* If this NAL is a slice, the index of the first MB in the slice. */
99 int i_last_mb; /* If this NAL is a slice, the index of the last MB in the slice. */
101 /* Size of payload (including any padding) in bytes. */
103 /* If param->b_annexb is set, Annex-B bytestream with startcode.
104 * Otherwise, startcode is replaced with a 4-byte size.
105 * This size is the size used in mp4/similar muxing; it is equal to i_payload-4 */
108 /* Size of padding in bytes. */
112 /****************************************************************************
114 ****************************************************************************/
118 #define X264_CPU_CMOV 0x0000001
119 #define X264_CPU_MMX 0x0000002
120 #define X264_CPU_MMX2 0x0000004 /* MMX2 aka MMXEXT aka ISSE */
121 #define X264_CPU_MMXEXT X264_CPU_MMX2
122 #define X264_CPU_SSE 0x0000008
123 #define X264_CPU_SSE2 0x0000010
124 #define X264_CPU_SSE3 0x0000020
125 #define X264_CPU_SSSE3 0x0000040
126 #define X264_CPU_SSE4 0x0000080 /* SSE4.1 */
127 #define X264_CPU_SSE42 0x0000100 /* SSE4.2 */
128 #define X264_CPU_LZCNT 0x0000200 /* Phenom support for "leading zero count" instruction. */
129 #define X264_CPU_AVX 0x0000400 /* AVX support: requires OS support even if YMM registers aren't used. */
130 #define X264_CPU_XOP 0x0000800 /* AMD XOP */
131 #define X264_CPU_FMA4 0x0001000 /* AMD FMA4 */
132 #define X264_CPU_AVX2 0x0002000 /* AVX2 */
133 #define X264_CPU_FMA3 0x0004000 /* Intel FMA3 */
134 #define X264_CPU_BMI1 0x0008000 /* BMI1 */
135 #define X264_CPU_BMI2 0x0010000 /* BMI2 */
137 #define X264_CPU_CACHELINE_32 0x0020000 /* avoid memory loads that span the border between two cachelines */
138 #define X264_CPU_CACHELINE_64 0x0040000 /* 32/64 is the size of a cacheline in bytes */
139 #define X264_CPU_SSE2_IS_SLOW 0x0080000 /* avoid most SSE2 functions on Athlon64 */
140 #define X264_CPU_SSE2_IS_FAST 0x0100000 /* a few functions are only faster on Core2 and Phenom */
141 #define X264_CPU_SLOW_SHUFFLE 0x0200000 /* The Conroe has a slow shuffle unit (relative to overall SSE performance) */
142 #define X264_CPU_STACK_MOD4 0x0400000 /* if stack is only mod4 and not mod16 */
143 #define X264_CPU_SLOW_CTZ 0x0800000 /* BSR/BSF x86 instructions are really slow on some CPUs */
144 #define X264_CPU_SLOW_ATOM 0x1000000 /* The Atom is terrible: slow SSE unaligned loads, slow
145 * SIMD multiplies, slow SIMD variable shifts, slow pshufb,
146 * cacheline split penalties -- gather everything here that
147 * isn't shared by other CPUs to avoid making half a dozen
149 #define X264_CPU_SLOW_PSHUFB 0x2000000 /* such as on the Intel Atom */
150 #define X264_CPU_SLOW_PALIGNR 0x4000000 /* such as on the AMD Bobcat */
153 #define X264_CPU_ALTIVEC 0x0000001
156 #define X264_CPU_ARMV6 0x0000001
157 #define X264_CPU_NEON 0x0000002 /* ARM NEON */
158 #define X264_CPU_FAST_NEON_MRC 0x0000004 /* Transfer from NEON to ARM register is fast (Cortex-A9) */
161 #define X264_ANALYSE_I4x4 0x0001 /* Analyse i4x4 */
162 #define X264_ANALYSE_I8x8 0x0002 /* Analyse i8x8 (requires 8x8 transform) */
163 #define X264_ANALYSE_PSUB16x16 0x0010 /* Analyse p16x8, p8x16 and p8x8 */
164 #define X264_ANALYSE_PSUB8x8 0x0020 /* Analyse p8x4, p4x8, p4x4 */
165 #define X264_ANALYSE_BSUB16x16 0x0100 /* Analyse b16x8, b8x16 and b8x8 */
166 #define X264_DIRECT_PRED_NONE 0
167 #define X264_DIRECT_PRED_SPATIAL 1
168 #define X264_DIRECT_PRED_TEMPORAL 2
169 #define X264_DIRECT_PRED_AUTO 3
170 #define X264_ME_DIA 0
171 #define X264_ME_HEX 1
172 #define X264_ME_UMH 2
173 #define X264_ME_ESA 3
174 #define X264_ME_TESA 4
175 #define X264_CQM_FLAT 0
176 #define X264_CQM_JVT 1
177 #define X264_CQM_CUSTOM 2
178 #define X264_RC_CQP 0
179 #define X264_RC_CRF 1
180 #define X264_RC_ABR 2
181 #define X264_QP_AUTO 0
182 #define X264_AQ_NONE 0
183 #define X264_AQ_VARIANCE 1
184 #define X264_AQ_AUTOVARIANCE 2
185 #define X264_B_ADAPT_NONE 0
186 #define X264_B_ADAPT_FAST 1
187 #define X264_B_ADAPT_TRELLIS 2
188 #define X264_WEIGHTP_NONE 0
189 #define X264_WEIGHTP_SIMPLE 1
190 #define X264_WEIGHTP_SMART 2
191 #define X264_B_PYRAMID_NONE 0
192 #define X264_B_PYRAMID_STRICT 1
193 #define X264_B_PYRAMID_NORMAL 2
194 #define X264_KEYINT_MIN_AUTO 0
195 #define X264_KEYINT_MAX_INFINITE (1<<30)
197 static const char * const x264_direct_pred_names[] = { "none", "spatial", "temporal", "auto", 0 };
198 static const char * const x264_motion_est_names[] = { "dia", "hex", "umh", "esa", "tesa", 0 };
199 static const char * const x264_b_pyramid_names[] = { "none", "strict", "normal", 0 };
200 static const char * const x264_overscan_names[] = { "undef", "show", "crop", 0 };
201 static const char * const x264_vidformat_names[] = { "component", "pal", "ntsc", "secam", "mac", "undef", 0 };
202 static const char * const x264_fullrange_names[] = { "off", "on", 0 };
203 static const char * const x264_colorprim_names[] = { "", "bt709", "undef", "", "bt470m", "bt470bg", "smpte170m", "smpte240m", "film", "bt2020", 0 };
204 static const char * const x264_transfer_names[] = { "", "bt709", "undef", "", "bt470m", "bt470bg", "smpte170m", "smpte240m", "linear", "log100", "log316",
205 "iec61966-2-4", "bt1361e", "iec61966-2-1", "bt2020-10", "bt2020-12", 0 };
206 static const char * const x264_colmatrix_names[] = { "GBR", "bt709", "undef", "", "fcc", "bt470bg", "smpte170m", "smpte240m", "YCgCo", "bt2020nc", "bt2020c", 0 };
207 static const char * const x264_nal_hrd_names[] = { "none", "vbr", "cbr", 0 };
209 /* Colorspace type */
210 #define X264_CSP_MASK 0x00ff /* */
211 #define X264_CSP_NONE 0x0000 /* Invalid mode */
212 #define X264_CSP_I420 0x0001 /* yuv 4:2:0 planar */
213 #define X264_CSP_YV12 0x0002 /* yvu 4:2:0 planar */
214 #define X264_CSP_NV12 0x0003 /* yuv 4:2:0, with one y plane and one packed u+v */
215 #define X264_CSP_I422 0x0004 /* yuv 4:2:2 planar */
216 #define X264_CSP_YV16 0x0005 /* yvu 4:2:2 planar */
217 #define X264_CSP_NV16 0x0006 /* yuv 4:2:2, with one y plane and one packed u+v */
218 #define X264_CSP_I444 0x0007 /* yuv 4:4:4 planar */
219 #define X264_CSP_YV24 0x0008 /* yvu 4:4:4 planar */
220 #define X264_CSP_BGR 0x0009 /* packed bgr 24bits */
221 #define X264_CSP_BGRA 0x000a /* packed bgr 32bits */
222 #define X264_CSP_RGB 0x000b /* packed rgb 24bits */
223 #define X264_CSP_MAX 0x000c /* end of list */
224 #define X264_CSP_VFLIP 0x1000 /* the csp is vertically flipped */
225 #define X264_CSP_HIGH_DEPTH 0x2000 /* the csp has a depth of 16 bits per pixel component */
228 #define X264_TYPE_AUTO 0x0000 /* Let x264 choose the right type */
229 #define X264_TYPE_IDR 0x0001
230 #define X264_TYPE_I 0x0002
231 #define X264_TYPE_P 0x0003
232 #define X264_TYPE_BREF 0x0004 /* Non-disposable B-frame */
233 #define X264_TYPE_B 0x0005
234 #define X264_TYPE_KEYFRAME 0x0006 /* IDR or I depending on b_open_gop option */
235 #define IS_X264_TYPE_I(x) ((x)==X264_TYPE_I || (x)==X264_TYPE_IDR)
236 #define IS_X264_TYPE_B(x) ((x)==X264_TYPE_B || (x)==X264_TYPE_BREF)
239 #define X264_LOG_NONE (-1)
240 #define X264_LOG_ERROR 0
241 #define X264_LOG_WARNING 1
242 #define X264_LOG_INFO 2
243 #define X264_LOG_DEBUG 3
246 #define X264_THREADS_AUTO 0 /* Automatically select optimal number of threads */
247 #define X264_SYNC_LOOKAHEAD_AUTO (-1) /* Automatically select optimal lookahead thread buffer size */
250 #define X264_NAL_HRD_NONE 0
251 #define X264_NAL_HRD_VBR 1
252 #define X264_NAL_HRD_CBR 2
254 /* Zones: override ratecontrol or other options for specific sections of the video.
255 * See x264_encoder_reconfig() for which options can be changed.
256 * If zones overlap, whichever comes later in the list takes precedence. */
259 int i_start, i_end; /* range of frame numbers */
260 int b_force_qp; /* whether to use qp vs bitrate factor */
262 float f_bitrate_factor;
263 struct x264_param_t *param;
266 typedef struct x264_param_t
270 int i_threads; /* encode multiple frames in parallel */
271 int i_lookahead_threads; /* multiple threads for lookahead analysis */
272 int b_sliced_threads; /* Whether to use slice-based threading. */
273 int b_deterministic; /* whether to allow non-deterministic optimizations when threaded */
274 int b_cpu_independent; /* force canonical behavior rather than cpu-dependent optimal algorithms */
275 int i_sync_lookahead; /* threaded lookahead buffer */
277 /* Video Properties */
280 int i_csp; /* CSP of encoded bitstream */
282 int i_frame_total; /* number of frames to encode if known, else 0 */
285 * Uses Buffering and Picture Timing SEIs to signal HRD
286 * The HRD in H.264 was not designed with VFR in mind.
287 * It is therefore not recommendeded to use NAL HRD with VFR.
288 * Furthermore, reconfiguring the VBV (via x264_encoder_reconfig)
289 * will currently generate invalid HRD. */
294 /* they will be reduced to be 0 < x <= 65535 and prime */
298 int i_overscan; /* 0=undef, 1=no overscan, 2=overscan */
300 /* see h264 annex E for the values of the following */
306 int i_chroma_loc; /* both top & bottom */
309 /* Bitstream parameters */
310 int i_frame_reference; /* Maximum number of reference frames */
311 int i_dpb_size; /* Force a DPB size larger than that implied by B-frames and reference frames.
312 * Useful in combination with interactive error resilience. */
313 int i_keyint_max; /* Force an IDR keyframe at this interval */
314 int i_keyint_min; /* Scenecuts closer together than this are coded as I, not IDR. */
315 int i_scenecut_threshold; /* how aggressively to insert extra I frames */
316 int b_intra_refresh; /* Whether or not to use periodic intra refresh instead of IDR frames. */
318 int i_bframe; /* how many b-frame between 2 references pictures */
319 int i_bframe_adaptive;
321 int i_bframe_pyramid; /* Keep some B-frames as references: 0=off, 1=strict hierarchical, 2=normal */
324 int b_avcintra_compat;
326 int b_deblocking_filter;
327 int i_deblocking_filter_alphac0; /* [-6, 6] -6 light filter, 6 strong */
328 int i_deblocking_filter_beta; /* [-6, 6] idem */
331 int i_cabac_init_idc;
334 int b_constrained_intra;
337 char *psz_cqm_file; /* JM format */
338 uint8_t cqm_4iy[16]; /* used only if i_cqm_preset == X264_CQM_CUSTOM */
348 void (*pf_log)( void *, int i_level, const char *psz, va_list );
352 int b_full_recon; /* fully reconstruct frames, even when not necessary for encoding. Implied by psz_dump_yuv */
353 char *psz_dump_yuv; /* filename for reconstructed frames */
355 /* Encoder analyser parameters */
358 unsigned int intra; /* intra partitions */
359 unsigned int inter; /* inter partitions */
362 int i_weighted_pred; /* weighting for P-frames */
363 int b_weighted_bipred; /* implicit weighting for B-frames */
364 int i_direct_mv_pred; /* spatial vs temporal mv prediction */
365 int i_chroma_qp_offset;
367 int i_me_method; /* motion estimation algorithm to use (X264_ME_*) */
368 int i_me_range; /* integer pixel motion estimation search range (from predicted mv) */
369 int i_mv_range; /* maximum length of a mv (in pixels). -1 = auto, based on level */
370 int i_mv_range_thread; /* minimum space between threads. -1 = auto, based on number of threads. */
371 int i_subpel_refine; /* subpixel motion estimation quality */
372 int b_chroma_me; /* chroma ME for subpel and mode decision in P-frames */
373 int b_mixed_references; /* allow each mb partition to have its own reference number */
374 int i_trellis; /* trellis RD quantization */
375 int b_fast_pskip; /* early SKIP detection on P-frames */
376 int b_dct_decimate; /* transform coefficient thresholding on P-frames */
377 int i_noise_reduction; /* adaptive pseudo-deadzone */
378 float f_psy_rd; /* Psy RD strength */
379 float f_psy_trellis; /* Psy trellis strength */
380 int b_psy; /* Toggle all psy optimizations */
382 int b_mb_info; /* Use input mb_info data in x264_picture_t */
383 int b_mb_info_update; /* Update the values in mb_info according to the results of encoding. */
385 /* the deadzone size that will be used in luma quantization */
386 int i_luma_deadzone[2]; /* {inter, intra} */
388 int b_psnr; /* compute and print PSNR stats */
389 int b_ssim; /* compute and print SSIM stats */
392 /* Rate control parameters */
395 int i_rc_method; /* X264_RC_* */
397 int i_qp_constant; /* 0 to (51 + 6*(x264_bit_depth-8)). 0=lossless */
398 int i_qp_min; /* min allowed QP value */
399 int i_qp_max; /* max allowed QP value */
400 int i_qp_step; /* max QP step between frames */
403 float f_rf_constant; /* 1pass VBR, nominal QP */
404 float f_rf_constant_max; /* In CRF mode, maximum CRF as caused by VBV */
405 float f_rate_tolerance;
406 int i_vbv_max_bitrate;
407 int i_vbv_buffer_size;
408 float f_vbv_buffer_init; /* <=1: fraction of buffer_size. >1: kbit */
412 int i_aq_mode; /* psy adaptive QP. (X264_AQ_*) */
414 int b_mb_tree; /* Macroblock-tree ratecontrol. */
418 int b_stat_write; /* Enable stat writing in psz_stat_out */
420 int b_stat_read; /* Read stat from psz_stat_in and use it */
423 /* 2pass params (same as ffmpeg ones) */
424 float f_qcompress; /* 0.0 => cbr, 1.0 => constant qp */
425 float f_qblur; /* temporally blur quants */
426 float f_complexity_blur; /* temporally blur complexity */
427 x264_zone_t *zones; /* ratecontrol overrides */
428 int i_zones; /* number of zone_t's */
429 char *psz_zones; /* alternate method of specifying zones */
432 /* Cropping Rectangle parameters: added to those implicitly defined by
433 non-mod16 video resolutions. */
438 unsigned int i_right;
439 unsigned int i_bottom;
442 /* frame packing arrangement flag */
445 /* Muxing parameters */
446 int b_aud; /* generate access unit delimiters */
447 int b_repeat_headers; /* put SPS/PPS before each keyframe */
448 int b_annexb; /* if set, place start codes (4 bytes) before NAL units,
449 * otherwise place size (4 bytes) before NAL units. */
450 int i_sps_id; /* SPS and PPS id number */
451 int b_vfr_input; /* VFR input. If 1, use timebase and timestamps for ratecontrol purposes.
452 * If 0, use fps only. */
453 int b_pulldown; /* use explicity set timebase for CFR */
456 uint32_t i_timebase_num; /* Timebase numerator */
457 uint32_t i_timebase_den; /* Timebase denominator */
462 * The correct pic_struct must be passed with each input frame.
463 * The input timebase should be the timebase corresponding to the output framerate. This should be constant.
464 * e.g. for 3:2 pulldown timebase should be 1001/30000
465 * The PTS passed with each frame must be the PTS of the frame after pulldown is applied.
466 * Frame doubling and tripling require b_vfr_input set to zero (see H.264 Table D-1)
468 * Pulldown changes are not clearly defined in H.264. Therefore, it is the calling app's responsibility to manage this.
475 * Used only when b_interlaced=0. Setting this flag makes it possible to flag the stream as PAFF interlaced yet
476 * encode all frames progessively. It is useful for encoding 25p and 30p Blu-Ray streams.
479 int b_fake_interlaced;
481 /* Don't optimize header parameters based on video content, e.g. ensure that splitting an input video, compressing
482 * each part, and stitching them back together will result in identical SPS/PPS. This is necessary for stitching
483 * with container formats that don't allow multiple SPS/PPS. */
486 int b_opencl; /* use OpenCL when available */
487 int i_opencl_device; /* specify count of GPU devices to skip, for CLI users */
488 void *opencl_device_id; /* pass explicit cl_device_id as void*, for API users */
489 char *psz_clbin_file; /* compiled OpenCL kernel cache file */
491 /* Slicing parameters */
492 int i_slice_max_size; /* Max size per slice in bytes; includes estimated NAL overhead. */
493 int i_slice_max_mbs; /* Max number of MBs per slice; overrides i_slice_count. */
494 int i_slice_min_mbs; /* Min number of MBs per slice */
495 int i_slice_count; /* Number of slices per frame: forces rectangular slices. */
496 int i_slice_count_max; /* Absolute cap on slices per frame; stops applying slice-max-size
497 * and slice-max-mbs if this is reached. */
499 /* Optional callback for freeing this x264_param_t when it is done being used.
500 * Only used when the x264_param_t sits in memory for an indefinite period of time,
501 * i.e. when an x264_param_t is passed to x264_t in an x264_picture_t or in zones.
502 * Not used when x264_encoder_reconfig is called directly. */
503 void (*param_free)( void* );
505 /* Optional low-level callback for low-latency encoding. Called for each output NAL unit
506 * immediately after the NAL unit is finished encoding. This allows the calling application
507 * to begin processing video data (e.g. by sending packets over a network) before the frame
510 * This callback MUST do the following in order to work correctly:
511 * 1) Have available an output buffer of at least size nal->i_payload*3/2 + 5 + 64.
512 * 2) Call x264_nal_encode( h, dst, nal ), where dst is the output buffer.
513 * After these steps, the content of nal is valid and can be used in the same way as if
514 * the NAL unit were output by x264_encoder_encode.
516 * This does not need to be synchronous with the encoding process: the data pointed to
517 * by nal (both before and after x264_nal_encode) will remain valid until the next
518 * x264_encoder_encode call. The callback must be re-entrant.
520 * This callback does not work with frame-based threads; threads must be disabled
521 * or sliced-threads enabled. This callback also does not work as one would expect
522 * with HRD -- since the buffering period SEI cannot be calculated until the frame
523 * is finished encoding, it will not be sent via this callback.
525 * Note also that the NALs are not necessarily returned in order when sliced threads is
526 * enabled. Accordingly, the variable i_first_mb and i_last_mb are available in
527 * x264_nal_t to help the calling application reorder the slices if necessary.
529 * When this callback is enabled, x264_encoder_encode does not return valid NALs;
530 * the calling application is expected to acquire all output NALs through the callback.
532 * It is generally sensible to combine this callback with a use of slice-max-mbs or
535 * The opaque pointer is the opaque pointer from the input frame associated with this
536 * NAL unit. This helps distinguish between nalu_process calls from different sources,
537 * e.g. if doing multiple encodes in one process.
539 void (*nalu_process) ( x264_t *h, x264_nal_t *nal, void *opaque );
542 void x264_nal_encode( x264_t *h, uint8_t *dst, x264_nal_t *nal );
544 /****************************************************************************
545 * H.264 level restriction information
546 ****************************************************************************/
551 int mbps; /* max macroblock processing rate (macroblocks/sec) */
552 int frame_size; /* max frame size (macroblocks) */
553 int dpb; /* max decoded picture buffer (mbs) */
554 int bitrate; /* max bitrate (kbit/sec) */
555 int cpb; /* max vbv buffer (kbit) */
556 int mv_range; /* max vertical mv component range (pixels) */
557 int mvs_per_2mb; /* max mvs per 2 consecutive mbs. */
558 int slice_rate; /* ?? */
559 int mincr; /* min compression ratio */
560 int bipred8x8; /* limit bipred to >=8x8 */
561 int direct8x8; /* limit b_direct to >=8x8 */
562 int frame_only; /* forbid interlacing */
565 /* all of the levels defined in the standard, terminated by .level_idc=0 */
566 X264_API extern const x264_level_t x264_levels[];
568 /****************************************************************************
569 * Basic parameter handling functions
570 ****************************************************************************/
572 /* x264_param_default:
573 * fill x264_param_t with default values and do CPU detection */
574 void x264_param_default( x264_param_t * );
577 * set one parameter by name.
578 * returns 0 on success, or returns one of the following errors.
579 * note: BAD_VALUE occurs only if it can't even parse the value,
580 * numerical range is not checked until x264_encoder_open() or
581 * x264_encoder_reconfig().
582 * value=NULL means "true" for boolean options, but is a BAD_VALUE for non-booleans. */
583 #define X264_PARAM_BAD_NAME (-1)
584 #define X264_PARAM_BAD_VALUE (-2)
585 int x264_param_parse( x264_param_t *, const char *name, const char *value );
587 /****************************************************************************
588 * Advanced parameter handling functions
589 ****************************************************************************/
591 /* These functions expose the full power of x264's preset-tune-profile system for
592 * easy adjustment of large numbers of internal parameters.
594 * In order to replicate x264CLI's option handling, these functions MUST be called
595 * in the following order:
596 * 1) x264_param_default_preset
597 * 2) Custom user options (via param_parse or directly assigned variables)
598 * 3) x264_param_apply_fastfirstpass
599 * 4) x264_param_apply_profile
601 * Additionally, x264CLI does not apply step 3 if the preset chosen is "placebo"
602 * or --slow-firstpass is set. */
604 /* x264_param_default_preset:
605 * The same as x264_param_default, but also use the passed preset and tune
606 * to modify the default settings.
607 * (either can be NULL, which implies no preset or no tune, respectively)
609 * Currently available presets are, ordered from fastest to slowest: */
610 static const char * const x264_preset_names[] = { "ultrafast", "superfast", "veryfast", "faster", "fast", "medium", "slow", "slower", "veryslow", "placebo", 0 };
612 /* The presets can also be indexed numerically, as in:
613 * x264_param_default_preset( ¶m, "3", ... )
614 * with ultrafast mapping to "0" and placebo mapping to "9". This mapping may
615 * of course change if new presets are added in between, but will always be
616 * ordered from fastest to slowest.
618 * Warning: the speed of these presets scales dramatically. Ultrafast is a full
619 * 100 times faster than placebo!
621 * Currently available tunings are: */
622 static const char * const x264_tune_names[] = { "film", "animation", "grain", "stillimage", "psnr", "ssim", "fastdecode", "zerolatency", 0 };
624 /* Multiple tunings can be used if separated by a delimiter in ",./-+",
625 * however multiple psy tunings cannot be used.
626 * film, animation, grain, stillimage, psnr, and ssim are psy tunings.
628 * returns 0 on success, negative on failure (e.g. invalid preset/tune name). */
629 int x264_param_default_preset( x264_param_t *, const char *preset, const char *tune );
631 /* x264_param_apply_fastfirstpass:
632 * If first-pass mode is set (rc.b_stat_read == 0, rc.b_stat_write == 1),
633 * modify the encoder settings to disable options generally not useful on
635 void x264_param_apply_fastfirstpass( x264_param_t * );
637 /* x264_param_apply_profile:
638 * Applies the restrictions of the given profile.
639 * Currently available profiles are, from most to least restrictive: */
640 static const char * const x264_profile_names[] = { "baseline", "main", "high", "high10", "high422", "high444", 0 };
642 /* (can be NULL, in which case the function will do nothing)
644 * Does NOT guarantee that the given profile will be used: if the restrictions
645 * of "High" are applied to settings that are already Baseline-compatible, the
646 * stream will remain baseline. In short, it does not increase settings, only
649 * returns 0 on success, negative on failure (e.g. invalid profile name). */
650 int x264_param_apply_profile( x264_param_t *, const char *profile );
652 /****************************************************************************
653 * Picture structures and functions
654 ****************************************************************************/
657 * Specifies the number of bits per pixel that x264 uses. This is also the
658 * bit depth that x264 encodes in. If this value is > 8, x264 will read
659 * two bytes of input data for each pixel sample, and expect the upper
660 * (16-x264_bit_depth) bits to be zero.
661 * Note: The flag X264_CSP_HIGH_DEPTH must be used to specify the
662 * colorspace depth as well. */
663 X264_API extern const int x264_bit_depth;
665 /* x264_chroma_format:
666 * Specifies the chroma formats that x264 supports encoding. When this
667 * value is non-zero, then it represents a X264_CSP_* that is the only
668 * chroma format that x264 supports encoding. If the value is 0 then
669 * there are no restrictions. */
670 X264_API extern const int x264_chroma_format;
674 PIC_STRUCT_AUTO = 0, // automatically decide (default)
675 PIC_STRUCT_PROGRESSIVE = 1, // progressive frame
676 // "TOP" and "BOTTOM" are not supported in x264 (PAFF only)
677 PIC_STRUCT_TOP_BOTTOM = 4, // top field followed by bottom
678 PIC_STRUCT_BOTTOM_TOP = 5, // bottom field followed by top
679 PIC_STRUCT_TOP_BOTTOM_TOP = 6, // top field, bottom field, top field repeated
680 PIC_STRUCT_BOTTOM_TOP_BOTTOM = 7, // bottom field, top field, bottom field repeated
681 PIC_STRUCT_DOUBLE = 8, // double frame
682 PIC_STRUCT_TRIPLE = 9, // triple frame
687 double cpb_initial_arrival_time;
688 double cpb_final_arrival_time;
689 double cpb_removal_time;
691 double dpb_output_time;
694 /* Arbitrary user SEI:
695 * Payload size is in bytes and the payload pointer must be valid.
696 * Payload types and syntax can be found in Annex D of the H.264 Specification.
697 * SEI payload alignment bits as described in Annex D must be included at the
698 * end of the payload if needed.
699 * The payload should not be NAL-encapsulated.
700 * Payloads are written first in order of input, apart from in the case when HRD
701 * is enabled where payloads are written after the Buffering Period SEI. */
708 } x264_sei_payload_t;
713 x264_sei_payload_t *payloads;
714 /* In: optional callback to free each payload AND x264_sei_payload_t when used. */
715 void (*sei_free)( void* );
720 int i_csp; /* Colorspace */
721 int i_plane; /* Number of image planes */
722 int i_stride[4]; /* Strides for each plane */
723 uint8_t *plane[4]; /* Pointers to each plane */
728 /* All arrays of data here are ordered as follows:
729 * each array contains one offset per macroblock, in raster scan order. In interlaced
730 * mode, top-field MBs and bottom-field MBs are interleaved at the row level.
731 * Macroblocks are 16x16 blocks of pixels (with respect to the luma plane). For the
732 * purposes of calculating the number of macroblocks, width and height are rounded up to
733 * the nearest 16. If in interlaced mode, height is rounded up to the nearest 32 instead. */
735 /* In: an array of quantizer offsets to be applied to this image during encoding.
736 * These are added on top of the decisions made by x264.
737 * Offsets can be fractional; they are added before QPs are rounded to integer.
738 * Adaptive quantization must be enabled to use this feature. Behavior if quant
739 * offsets differ between encoding passes is undefined. */
740 float *quant_offsets;
741 /* In: optional callback to free quant_offsets when used.
742 * Useful if one wants to use a different quant_offset array for each frame. */
743 void (*quant_offsets_free)( void* );
745 /* In: optional array of flags for each macroblock.
746 * Allows specifying additional information for the encoder such as which macroblocks
747 * remain unchanged. Usable flags are listed below.
748 * x264_param_t.analyse.b_mb_info must be set to use this, since x264 needs to track
749 * extra data internally to make full use of this information.
751 * Out: if b_mb_info_update is set, x264 will update this array as a result of encoding.
753 * For "MBINFO_CONSTANT", it will remove this flag on any macroblock whose decoded
754 * pixels have changed. This can be useful for e.g. noting which areas of the
755 * frame need to actually be blitted. Note: this intentionally ignores the effects
756 * of deblocking for the current frame, which should be fine unless one needs exact
757 * pixel-perfect accuracy.
759 * Results for MBINFO_CONSTANT are currently only set for P-frames, and are not
760 * guaranteed to enumerate all blocks which haven't changed. (There may be false
761 * negatives, but no false positives.)
764 /* In: optional callback to free mb_info when used. */
765 void (*mb_info_free)( void* );
767 /* The macroblock is constant and remains unchanged from the previous frame. */
768 #define X264_MBINFO_CONSTANT (1<<0)
769 /* More flags may be added in the future. */
771 /* Out: SSIM of the the frame luma (if x264_param_t.b_ssim is set) */
773 /* Out: Average PSNR of the frame (if x264_param_t.b_psnr is set) */
775 /* Out: PSNR of Y, U, and V (if x264_param_t.b_psnr is set) */
778 /* Out: Average effective CRF of the encoded frame */
780 } x264_image_properties_t;
784 /* In: force picture type (if not auto)
785 * If x264 encoding parameters are violated in the forcing of picture types,
786 * x264 will correct the input picture type and log a warning.
787 * The quality of frametype decisions may suffer if a great deal of fine-grained
788 * mixing of auto and forced frametypes is done.
789 * Out: type of the picture encoded */
791 /* In: force quantizer for != X264_QP_AUTO */
793 /* In: pic_struct, for pulldown/doubling/etc...used only if b_pic_struct=1.
794 * use pic_struct_e for pic_struct inputs
795 * Out: pic_struct element associated with frame */
797 /* Out: whether this frame is a keyframe. Important when using modes that result in
798 * SEI recovery points being used instead of IDR frames. */
800 /* In: user pts, Out: pts of encoded picture (user)*/
802 /* Out: frame dts. When the pts of the first frame is close to zero,
803 * initial frames may have a negative dts which must be dealt with by any muxer */
805 /* In: custom encoding parameters to be set from this frame forwards
806 (in coded order, not display order). If NULL, continue using
807 parameters from the previous frame. Some parameters, such as
808 aspect ratio, can only be changed per-GOP due to the limitations
809 of H.264 itself; in this case, the caller must force an IDR frame
810 if it needs the changed parameter to apply immediately. */
812 /* In: raw image data */
813 /* Out: reconstructed image data. x264 may skip part of the reconstruction process,
814 e.g. deblocking, in frames where it isn't necessary. To force complete
815 reconstruction, at a small speed cost, set b_full_recon. */
817 /* In: optional information to modify encoder decisions for this frame
818 * Out: information about the encoded frame */
819 x264_image_properties_t prop;
820 /* Out: HRD timing information. Output only when i_nal_hrd is set. */
821 x264_hrd_t hrd_timing;
822 /* In: arbitrary user SEI (e.g subtitles, AFDs) */
823 x264_sei_t extra_sei;
824 /* private user data. copied from input to output frames. */
828 /* x264_picture_init:
829 * initialize an x264_picture_t. Needs to be done if the calling application
830 * allocates its own x264_picture_t as opposed to using x264_picture_alloc. */
831 void x264_picture_init( x264_picture_t *pic );
833 /* x264_picture_alloc:
834 * alloc data for a picture. You must call x264_picture_clean on it.
835 * returns 0 on success, or -1 on malloc failure or invalid colorspace. */
836 int x264_picture_alloc( x264_picture_t *pic, int i_csp, int i_width, int i_height );
838 /* x264_picture_clean:
839 * free associated resource for a x264_picture_t allocated with
840 * x264_picture_alloc ONLY */
841 void x264_picture_clean( x264_picture_t *pic );
843 /****************************************************************************
845 ****************************************************************************/
847 /* Force a link error in the case of linking against an incompatible API version.
848 * Glue #defines exist to force correct macro expansion; the final output of the macro
849 * is x264_encoder_open_##X264_BUILD (for purposes of dlopen). */
850 #define x264_encoder_glue1(x,y) x##y
851 #define x264_encoder_glue2(x,y) x264_encoder_glue1(x,y)
852 #define x264_encoder_open x264_encoder_glue2(x264_encoder_open_,X264_BUILD)
854 /* x264_encoder_open:
855 * create a new encoder handler, all parameters from x264_param_t are copied */
856 x264_t *x264_encoder_open( x264_param_t * );
858 /* x264_encoder_reconfig:
859 * various parameters from x264_param_t are copied.
860 * this takes effect immediately, on whichever frame is encoded next;
861 * due to delay, this may not be the next frame passed to encoder_encode.
862 * if the change should apply to some particular frame, use x264_picture_t->param instead.
863 * returns 0 on success, negative on parameter validation error.
864 * not all parameters can be changed; see the actual function for a detailed breakdown.
866 * since not all parameters can be changed, moving from preset to preset may not always
867 * fully copy all relevant parameters, but should still work usably in practice. however,
868 * more so than for other presets, many of the speed shortcuts used in ultrafast cannot be
869 * switched out of; using reconfig to switch between ultrafast and other presets is not
870 * recommended without a more fine-grained breakdown of parameters to take this into account. */
871 int x264_encoder_reconfig( x264_t *, x264_param_t * );
872 /* x264_encoder_parameters:
873 * copies the current internal set of parameters to the pointer provided
874 * by the caller. useful when the calling application needs to know
875 * how x264_encoder_open has changed the parameters, or the current state
876 * of the encoder after multiple x264_encoder_reconfig calls.
877 * note that the data accessible through pointers in the returned param struct
878 * (e.g. filenames) should not be modified by the calling application. */
879 void x264_encoder_parameters( x264_t *, x264_param_t * );
880 /* x264_encoder_headers:
881 * return the SPS and PPS that will be used for the whole stream.
882 * *pi_nal is the number of NAL units outputted in pp_nal.
883 * returns negative on error.
884 * the payloads of all output NALs are guaranteed to be sequential in memory. */
885 int x264_encoder_headers( x264_t *, x264_nal_t **pp_nal, int *pi_nal );
886 /* x264_encoder_encode:
887 * encode one picture.
888 * *pi_nal is the number of NAL units outputted in pp_nal.
889 * returns negative on error, zero if no NAL units returned.
890 * the payloads of all output NALs are guaranteed to be sequential in memory. */
891 int x264_encoder_encode( x264_t *, x264_nal_t **pp_nal, int *pi_nal, x264_picture_t *pic_in, x264_picture_t *pic_out );
892 /* x264_encoder_close:
893 * close an encoder handler */
894 void x264_encoder_close ( x264_t * );
895 /* x264_encoder_delayed_frames:
896 * return the number of currently delayed (buffered) frames
897 * this should be used at the end of the stream, to know when you have all the encoded frames. */
898 int x264_encoder_delayed_frames( x264_t * );
899 /* x264_encoder_maximum_delayed_frames( x264_t *h ):
900 * return the maximum number of delayed (buffered) frames that can occur with the current
902 int x264_encoder_maximum_delayed_frames( x264_t *h );
903 /* x264_encoder_intra_refresh:
904 * If an intra refresh is not in progress, begin one with the next P-frame.
905 * If an intra refresh is in progress, begin one as soon as the current one finishes.
906 * Requires that b_intra_refresh be set.
908 * Useful for interactive streaming where the client can tell the server that packet loss has
909 * occurred. In this case, keyint can be set to an extremely high value so that intra refreshes
910 * only occur when calling x264_encoder_intra_refresh.
912 * In multi-pass encoding, if x264_encoder_intra_refresh is called differently in each pass,
913 * behavior is undefined.
915 * Should not be called during an x264_encoder_encode. */
916 void x264_encoder_intra_refresh( x264_t * );
917 /* x264_encoder_invalidate_reference:
918 * An interactive error resilience tool, designed for use in a low-latency one-encoder-few-clients
919 * system. When the client has packet loss or otherwise incorrectly decodes a frame, the encoder
920 * can be told with this command to "forget" the frame and all frames that depend on it, referencing
921 * only frames that occurred before the loss. This will force a keyframe if no frames are left to
922 * reference after the aforementioned "forgetting".
924 * It is strongly recommended to use a large i_dpb_size in this case, which allows the encoder to
925 * keep around extra, older frames to fall back on in case more recent frames are all invalidated.
926 * Unlike increasing i_frame_reference, this does not increase the number of frames used for motion
927 * estimation and thus has no speed impact. It is also recommended to set a very large keyframe
928 * interval, so that keyframes are not used except as necessary for error recovery.
930 * x264_encoder_invalidate_reference is not currently compatible with the use of B-frames or intra
933 * In multi-pass encoding, if x264_encoder_invalidate_reference is called differently in each pass,
934 * behavior is undefined.
936 * Should not be called during an x264_encoder_encode, but multiple calls can be made simultaneously.
938 * Returns 0 on success, negative on failure. */
939 int x264_encoder_invalidate_reference( x264_t *, int64_t pts );