x264.h

   1 /*****************************************************************************
   2  * x264.h: h264 encoder library
   3  *****************************************************************************
   4  * Copyright (C) 2003-2008 x264 Project
   5  *
   6  * Authors: Laurent Aimar <fenrir@via.ecp.fr>
   7  *          Loren Merritt <lorenm@u.washington.edu>
   8  *
   9  * This program is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * This program is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License
  20  * along with this program; if not, write to the Free Software
  21  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  22  *****************************************************************************/
  23
  24 #ifndef X264_X264_H
  25 #define X264_X264_H
  26
  27 #if !defined(_STDINT_H) && !defined(_STDINT_H_) && \
  28     !defined(_INTTYPES_H) && !defined(_INTTYPES_H_)
  29 # ifdef _MSC_VER
  30 #  pragma message("You must include stdint.h or inttypes.h before x264.h")
  31 # else
  32 #  warning You must include stdint.h or inttypes.h before x264.h
  33 # endif
  34 #endif
  35
  36 #include <stdarg.h>
  37
  38 #define X264_BUILD 65
  39
  40 /* x264_t:
  41  *      opaque handler for encoder */
  42 typedef struct x264_t x264_t;
  43
  44 /****************************************************************************
  45  * Initialisation structure and function.
  46  ****************************************************************************/
  47 /* CPU flags
  48  */
  49 #define X264_CPU_CACHELINE_32   0x000001  /* avoid memory loads that span the border between two cachelines */
  50 #define X264_CPU_CACHELINE_64   0x000002  /* 32/64 is the size of a cacheline in bytes */
  51 #define X264_CPU_ALTIVEC        0x000004
  52 #define X264_CPU_MMX            0x000008
  53 #define X264_CPU_MMXEXT         0x000010  /* MMX2 aka MMXEXT aka ISSE */
  54 #define X264_CPU_SSE            0x000020
  55 #define X264_CPU_SSE2           0x000040
  56 #define X264_CPU_SSE2_IS_SLOW   0x000080  /* avoid most SSE2 functions on Athlon64 */
  57 #define X264_CPU_SSE2_IS_FAST   0x000100  /* a few functions are only faster on Core2 and Phenom */
  58 #define X264_CPU_SSE3           0x000200
  59 #define X264_CPU_SSSE3          0x000400
  60 #define X264_CPU_PHADD_IS_FAST  0x000800  /* pre-Penryn Core2 have a uselessly slow PHADD instruction */
  61 #define X264_CPU_STACK_MOD4     0x001000  /* if stack is only mod4 and not mod16 */
  62 #define X264_CPU_SSE4           0x002000  /* SSE4.1 */
  63 #define X264_CPU_SSE42          0x004000  /* SSE4.2 */
  64 #define X264_CPU_SSE_MISALIGN   0x008000  /* Phenom support for misaligned SSE instruction arguments */
  65
  66 /* Analyse flags
  67  */
  68 #define X264_ANALYSE_I4x4       0x0001  /* Analyse i4x4 */
  69 #define X264_ANALYSE_I8x8       0x0002  /* Analyse i8x8 (requires 8x8 transform) */
  70 #define X264_ANALYSE_PSUB16x16  0x0010  /* Analyse p16x8, p8x16 and p8x8 */
  71 #define X264_ANALYSE_PSUB8x8    0x0020  /* Analyse p8x4, p4x8, p4x4 */
  72 #define X264_ANALYSE_BSUB16x16  0x0100  /* Analyse b16x8, b8x16 and b8x8 */
  73 #define X264_DIRECT_PRED_NONE        0
  74 #define X264_DIRECT_PRED_SPATIAL     1
  75 #define X264_DIRECT_PRED_TEMPORAL    2
  76 #define X264_DIRECT_PRED_AUTO        3
  77 #define X264_ME_DIA                  0
  78 #define X264_ME_HEX                  1
  79 #define X264_ME_UMH                  2
  80 #define X264_ME_ESA                  3
  81 #define X264_ME_TESA                 4
  82 #define X264_CQM_FLAT                0
  83 #define X264_CQM_JVT                 1
  84 #define X264_CQM_CUSTOM              2
  85 #define X264_RC_NONE                 -1
  86 #define X264_RC_CQP                  0
  87 #define X264_RC_CRF                  1
  88 #define X264_RC_ABR                  2
  89 #define X264_AQ_NONE                 0
  90 #define X264_AQ_VARIANCE             1
  91 #define X264_B_ADAPT_NONE            0
  92 #define X264_B_ADAPT_FAST            1
  93 #define X264_B_ADAPT_TRELLIS         2
  94
  95 static const char * const x264_direct_pred_names[] = { "none", "spatial", "temporal", "auto", 0 };
  96 static const char * const x264_motion_est_names[] = { "dia", "hex", "umh", "esa", "tesa", 0 };
  97 static const char * const x264_overscan_names[] = { "undef", "show", "crop", 0 };
  98 static const char * const x264_vidformat_names[] = { "component", "pal", "ntsc", "secam", "mac", "undef", 0 };
  99 static const char * const x264_fullrange_names[] = { "off", "on", 0 };
 100 static const char * const x264_colorprim_names[] = { "", "bt709", "undef", "", "bt470m", "bt470bg", "smpte170m", "smpte240m", "film", 0 };
 101 static const char * const x264_transfer_names[] = { "", "bt709", "undef", "", "bt470m", "bt470bg", "smpte170m", "smpte240m", "linear", "log100", "log316", 0 };
 102 static const char * const x264_colmatrix_names[] = { "GBR", "bt709", "undef", "", "fcc", "bt470bg", "smpte170m", "smpte240m", "YCgCo", 0 };
 103
 104 /* Colorspace type
 105  * legacy only; nothing other than I420 is really supported.
 106  */
 107 #define X264_CSP_MASK           0x00ff  /* */
 108 #define X264_CSP_NONE           0x0000  /* Invalid mode     */
 109 #define X264_CSP_I420           0x0001  /* yuv 4:2:0 planar */
 110 #define X264_CSP_I422           0x0002  /* yuv 4:2:2 planar */
 111 #define X264_CSP_I444           0x0003  /* yuv 4:4:4 planar */
 112 #define X264_CSP_YV12           0x0004  /* yuv 4:2:0 planar */
 113 #define X264_CSP_YUYV           0x0005  /* yuv 4:2:2 packed */
 114 #define X264_CSP_RGB            0x0006  /* rgb 24bits       */
 115 #define X264_CSP_BGR            0x0007  /* bgr 24bits       */
 116 #define X264_CSP_BGRA           0x0008  /* bgr 32bits       */
 117 #define X264_CSP_MAX            0x0009  /* end of list */
 118 #define X264_CSP_VFLIP          0x1000  /* */
 119
 120 /* Slice type
 121  */
 122 #define X264_TYPE_AUTO          0x0000  /* Let x264 choose the right type */
 123 #define X264_TYPE_IDR           0x0001
 124 #define X264_TYPE_I             0x0002
 125 #define X264_TYPE_P             0x0003
 126 #define X264_TYPE_BREF          0x0004  /* Non-disposable B-frame */
 127 #define X264_TYPE_B             0x0005
 128 #define IS_X264_TYPE_I(x) ((x)==X264_TYPE_I || (x)==X264_TYPE_IDR)
 129 #define IS_X264_TYPE_B(x) ((x)==X264_TYPE_B || (x)==X264_TYPE_BREF)
 130
 131 /* Log level
 132  */
 133 #define X264_LOG_NONE          (-1)
 134 #define X264_LOG_ERROR          0
 135 #define X264_LOG_WARNING        1
 136 #define X264_LOG_INFO           2
 137 #define X264_LOG_DEBUG          3
 138
 139 /* Zones: override ratecontrol or other options for specific sections of the video.
 140  * See x264_encoder_reconfig() for which options can be changed.
 141  * If zones overlap, whichever comes later in the list takes precedence. */
 142 typedef struct
 143 {
 144     int i_start, i_end; /* range of frame numbers */
 145     int b_force_qp; /* whether to use qp vs bitrate factor */
 146     int i_qp;
 147     float f_bitrate_factor;
 148     struct x264_param_t *param;
 149 } x264_zone_t;
 150
 151 typedef struct x264_param_t
 152 {
 153     /* CPU flags */
 154     unsigned int cpu;
 155     int         i_threads;       /* encode multiple frames in parallel */
 156     int         b_deterministic; /* whether to allow non-deterministic optimizations when threaded */
 157
 158     /* Video Properties */
 159     int         i_width;
 160     int         i_height;
 161     int         i_csp;  /* CSP of encoded bitstream, only i420 supported */
 162     int         i_level_idc;
 163     int         i_frame_total; /* number of frames to encode if known, else 0 */
 164
 165     struct
 166     {
 167         /* they will be reduced to be 0 < x <= 65535 and prime */
 168         int         i_sar_height;
 169         int         i_sar_width;
 170
 171         int         i_overscan;    /* 0=undef, 1=no overscan, 2=overscan */
 172
 173         /* see h264 annex E for the values of the following */
 174         int         i_vidformat;
 175         int         b_fullrange;
 176         int         i_colorprim;
 177         int         i_transfer;
 178         int         i_colmatrix;
 179         int         i_chroma_loc;    /* both top & bottom */
 180     } vui;
 181
 182     int         i_fps_num;
 183     int         i_fps_den;
 184
 185     /* Bitstream parameters */
 186     int         i_frame_reference;  /* Maximum number of reference frames */
 187     int         i_keyint_max;       /* Force an IDR keyframe at this interval */
 188     int         i_keyint_min;       /* Scenecuts closer together than this are coded as I, not IDR. */
 189     int         i_scenecut_threshold; /* how aggressively to insert extra I frames */
 190     int         b_pre_scenecut;     /* compute scenecut on lowres frames */
 191     int         i_bframe;   /* how many b-frame between 2 references pictures */
 192     int         i_bframe_adaptive;
 193     int         i_bframe_bias;
 194     int         b_bframe_pyramid;   /* Keep some B-frames as references */
 195
 196     int         b_deblocking_filter;
 197     int         i_deblocking_filter_alphac0;    /* [-6, 6] -6 light filter, 6 strong */
 198     int         i_deblocking_filter_beta;       /* [-6, 6]  idem */
 199
 200     int         b_cabac;
 201     int         i_cabac_init_idc;
 202
 203     int         b_interlaced;
 204
 205     int         i_cqm_preset;
 206     char        *psz_cqm_file;      /* JM format */
 207     uint8_t     cqm_4iy[16];        /* used only if i_cqm_preset == X264_CQM_CUSTOM */
 208     uint8_t     cqm_4ic[16];
 209     uint8_t     cqm_4py[16];
 210     uint8_t     cqm_4pc[16];
 211     uint8_t     cqm_8iy[64];
 212     uint8_t     cqm_8py[64];
 213
 214     /* Log */
 215     void        (*pf_log)( void *, int i_level, const char *psz, va_list );
 216     void        *p_log_private;
 217     int         i_log_level;
 218     int         b_visualize;
 219     char        *psz_dump_yuv;  /* filename for reconstructed frames */
 220
 221     /* Encoder analyser parameters */
 222     struct
 223     {
 224         unsigned int intra;     /* intra partitions */
 225         unsigned int inter;     /* inter partitions */
 226
 227         int          b_transform_8x8;
 228         int          b_weighted_bipred; /* implicit weighting for B-frames */
 229         int          i_direct_mv_pred; /* spatial vs temporal mv prediction */
 230         int          i_direct_8x8_inference; /* forbid 4x4 direct partitions. -1 = auto, based on level */
 231         int          i_chroma_qp_offset;
 232
 233         int          i_me_method; /* motion estimation algorithm to use (X264_ME_*) */
 234         int          i_me_range; /* integer pixel motion estimation search range (from predicted mv) */
 235         int          i_mv_range; /* maximum length of a mv (in pixels). -1 = auto, based on level */
 236         int          i_mv_range_thread; /* minimum space between threads. -1 = auto, based on number of threads. */
 237         int          i_subpel_refine; /* subpixel motion estimation quality */
 238         int          b_chroma_me; /* chroma ME for subpel and mode decision in P-frames */
 239         int          b_mixed_references; /* allow each mb partition in P-frames to have it's own reference number */
 240         int          i_trellis;  /* trellis RD quantization */
 241         int          b_fast_pskip; /* early SKIP detection on P-frames */
 242         int          b_dct_decimate; /* transform coefficient thresholding on P-frames */
 243         int          i_noise_reduction; /* adaptive pseudo-deadzone */
 244         float        f_psy_rd; /* Psy RD strength */
 245         float        f_psy_trellis; /* Psy trellis strength */
 246
 247         /* the deadzone size that will be used in luma quantization */
 248         int          i_luma_deadzone[2]; /* {inter, intra} */
 249
 250         int          b_psnr;    /* compute and print PSNR stats */
 251         int          b_ssim;    /* compute and print SSIM stats */
 252     } analyse;
 253
 254     /* Rate control parameters */
 255     struct
 256     {
 257         int         i_rc_method;    /* X264_RC_* */
 258
 259         int         i_qp_constant;  /* 0-51 */
 260         int         i_qp_min;       /* min allowed QP value */
 261         int         i_qp_max;       /* max allowed QP value */
 262         int         i_qp_step;      /* max QP step between frames */
 263
 264         int         i_bitrate;
 265         float       f_rf_constant;  /* 1pass VBR, nominal QP */
 266         float       f_rate_tolerance;
 267         int         i_vbv_max_bitrate;
 268         int         i_vbv_buffer_size;
 269         float       f_vbv_buffer_init; /* <=1: fraction of buffer_size. >1: kbit */
 270         float       f_ip_factor;
 271         float       f_pb_factor;
 272
 273         int         i_aq_mode;      /* psy adaptive QP. (X264_AQ_*) */
 274         float       f_aq_strength;
 275
 276         /* 2pass */
 277         int         b_stat_write;   /* Enable stat writing in psz_stat_out */
 278         char        *psz_stat_out;
 279         int         b_stat_read;    /* Read stat from psz_stat_in and use it */
 280         char        *psz_stat_in;
 281
 282         /* 2pass params (same as ffmpeg ones) */
 283         float       f_qcompress;    /* 0.0 => cbr, 1.0 => constant qp */
 284         float       f_qblur;        /* temporally blur quants */
 285         float       f_complexity_blur; /* temporally blur complexity */
 286         x264_zone_t *zones;         /* ratecontrol overrides */
 287         int         i_zones;        /* number of zone_t's */
 288         char        *psz_zones;     /* alternate method of specifying zones */
 289     } rc;
 290
 291     /* Muxing parameters */
 292     int b_aud;                  /* generate access unit delimiters */
 293     int b_repeat_headers;       /* put SPS/PPS before each keyframe */
 294     int i_sps_id;               /* SPS and PPS id number */
 295 } x264_param_t;
 296
 297 typedef struct {
 298     int level_idc;
 299     int mbps;        /* max macroblock processing rate (macroblocks/sec) */
 300     int frame_size;  /* max frame size (macroblocks) */
 301     int dpb;         /* max decoded picture buffer (bytes) */
 302     int bitrate;     /* max bitrate (kbit/sec) */
 303     int cpb;         /* max vbv buffer (kbit) */
 304     int mv_range;    /* max vertical mv component range (pixels) */
 305     int mvs_per_2mb; /* max mvs per 2 consecutive mbs. */
 306     int slice_rate;  /* ?? */
 307     int bipred8x8;   /* limit bipred to >=8x8 */
 308     int direct8x8;   /* limit b_direct to >=8x8 */
 309     int frame_only;  /* forbid interlacing */
 310 } x264_level_t;
 311
 312 /* all of the levels defined in the standard, terminated by .level_idc=0 */
 313 extern const x264_level_t x264_levels[];
 314
 315 /* x264_param_default:
 316  *      fill x264_param_t with default values and do CPU detection */
 317 void    x264_param_default( x264_param_t * );
 318
 319 /* x264_param_parse:
 320  *  set one parameter by name.
 321  *  returns 0 on success, or returns one of the following errors.
 322  *  note: BAD_VALUE occurs only if it can't even parse the value,
 323  *  numerical range is not checked until x264_encoder_open() or
 324  *  x264_encoder_reconfig().
 325  *  value=NULL means "true" for boolean options, but is a BAD_VALUE for non-booleans. */
 326 #define X264_PARAM_BAD_NAME  (-1)
 327 #define X264_PARAM_BAD_VALUE (-2)
 328 int x264_param_parse( x264_param_t *, const char *name, const char *value );
 329
 330 /****************************************************************************
 331  * Picture structures and functions.
 332  ****************************************************************************/
 333 typedef struct
 334 {
 335     int     i_csp;
 336
 337     int     i_plane;
 338     int     i_stride[4];
 339     uint8_t *plane[4];
 340 } x264_image_t;
 341
 342 typedef struct
 343 {
 344     /* In: force picture type (if not auto) XXX: ignored for now
 345      * Out: type of the picture encoded */
 346     int     i_type;
 347     /* In: force quantizer for > 0 */
 348     int     i_qpplus1;
 349     /* In: user pts, Out: pts of encoded picture (user)*/
 350     int64_t i_pts;
 351
 352     /* In: raw data */
 353     x264_image_t img;
 354 } x264_picture_t;
 355
 356 /* x264_picture_alloc:
 357  *  alloc data for a picture. You must call x264_picture_clean on it. */
 358 void x264_picture_alloc( x264_picture_t *pic, int i_csp, int i_width, int i_height );
 359
 360 /* x264_picture_clean:
 361  *  free associated resource for a x264_picture_t allocated with
 362  *  x264_picture_alloc ONLY */
 363 void x264_picture_clean( x264_picture_t *pic );
 364
 365 /****************************************************************************
 366  * NAL structure and functions:
 367  ****************************************************************************/
 368 /* nal */
 369 enum nal_unit_type_e
 370 {
 371     NAL_UNKNOWN = 0,
 372     NAL_SLICE   = 1,
 373     NAL_SLICE_DPA   = 2,
 374     NAL_SLICE_DPB   = 3,
 375     NAL_SLICE_DPC   = 4,
 376     NAL_SLICE_IDR   = 5,    /* ref_idc != 0 */
 377     NAL_SEI         = 6,    /* ref_idc == 0 */
 378     NAL_SPS         = 7,
 379     NAL_PPS         = 8,
 380     NAL_AUD         = 9,
 381     /* ref_idc == 0 for 6,9,10,11,12 */
 382 };
 383 enum nal_priority_e
 384 {
 385     NAL_PRIORITY_DISPOSABLE = 0,
 386     NAL_PRIORITY_LOW        = 1,
 387     NAL_PRIORITY_HIGH       = 2,
 388     NAL_PRIORITY_HIGHEST    = 3,
 389 };
 390
 391 typedef struct
 392 {
 393     int i_ref_idc;  /* nal_priority_e */
 394     int i_type;     /* nal_unit_type_e */
 395
 396     /* This data are raw payload */
 397     int     i_payload;
 398     uint8_t *p_payload;
 399 } x264_nal_t;
 400
 401 /* x264_nal_encode:
 402  *      encode a nal into a buffer, setting the size.
 403  *      if b_annexeb then a long synch work is added
 404  *      XXX: it currently doesn't check for overflow */
 405 int x264_nal_encode( void *, int *, int b_annexeb, x264_nal_t *nal );
 406
 407 /* x264_nal_decode:
 408  *      decode a buffer nal into a x264_nal_t */
 409 int x264_nal_decode( x264_nal_t *nal, void *, int );
 410
 411 /****************************************************************************
 412  * Encoder functions:
 413  ****************************************************************************/
 414
 415 /* x264_encoder_open:
 416  *      create a new encoder handler, all parameters from x264_param_t are copied */
 417 x264_t *x264_encoder_open   ( x264_param_t * );
 418 /* x264_encoder_reconfig:
 419  *      change encoder options while encoding,
 420  *      analysis-related parameters from x264_param_t are copied */
 421 int     x264_encoder_reconfig( x264_t *, x264_param_t * );
 422 /* x264_encoder_headers:
 423  *      return the SPS and PPS that will be used for the whole stream */
 424 int     x264_encoder_headers( x264_t *, x264_nal_t **, int * );
 425 /* x264_encoder_encode:
 426  *      encode one picture */
 427 int     x264_encoder_encode ( x264_t *, x264_nal_t **, int *, x264_picture_t *, x264_picture_t * );
 428 /* x264_encoder_close:
 429  *      close an encoder handler */
 430 void    x264_encoder_close  ( x264_t * );
 431
 432 #endif