vpx_encoder.h 36.2 KB
Newer Older
John Koleszar's avatar
John Koleszar committed
1
/*
2
 *  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
John Koleszar's avatar
John Koleszar committed
3
 *
4
 *  Use of this source code is governed by a BSD-style license
5 6
 *  that can be found in the LICENSE file in the root of the source
 *  tree. An additional intellectual property rights grant can be found
7
 *  in the file PATENTS.  All contributing project authors may
8
 *  be found in the AUTHORS file in the root of the source tree.
John Koleszar's avatar
John Koleszar committed
9
 */
James Zern's avatar
James Zern committed
10 11
#ifndef VPX_VPX_ENCODER_H_
#define VPX_VPX_ENCODER_H_
John Koleszar's avatar
John Koleszar committed
12 13 14 15 16 17 18 19 20

/*!\defgroup encoder Encoder Algorithm Interface
 * \ingroup codec
 * This abstraction allows applications using this encoder to easily support
 * multiple video formats with minimal code duplication. This section describes
 * the interface common to all encoders.
 * @{
 */

James Zern's avatar
James Zern committed
21
/*!\file
John Koleszar's avatar
John Koleszar committed
22 23 24 25 26 27 28 29 30 31
 * \brief Describes the encoder algorithm interface to applications.
 *
 * This file describes the interface between an application and a
 * video encoder algorithm.
 *
 */
#ifdef __cplusplus
extern "C" {
#endif

32
#include "./vpx_codec.h"
John Koleszar's avatar
John Koleszar committed
33

John Koleszar's avatar
John Koleszar committed
34 35 36 37 38 39 40 41 42 43 44 45 46
  /*! Temporal Scalability: Maximum length of the sequence defining frame
   * layer membership
   */
#define VPX_TS_MAX_PERIODICITY 16

  /*! Temporal Scalability: Maximum number of coding layers */
#define VPX_TS_MAX_LAYERS       5

  /*!\deprecated Use #VPX_TS_MAX_PERIODICITY instead. */
#define MAX_PERIODICITY VPX_TS_MAX_PERIODICITY

  /*!\deprecated Use #VPX_TS_MAX_LAYERS instead. */
#define MAX_LAYERS      VPX_TS_MAX_LAYERS
John Koleszar's avatar
John Koleszar committed
47

48 49 50 51 52 53
/*! Spatial Scalability: Maximum number of coding layers */
#define VPX_SS_MAX_LAYERS       5

/*! Spatial Scalability: Default number of coding layers */
#define VPX_SS_DEFAULT_LAYERS       3

John Koleszar's avatar
John Koleszar committed
54 55 56 57 58 59 60 61
  /*!\brief Current ABI version number
   *
   * \internal
   * If this file is altered in any way that changes the ABI, this value
   * must be bumped.  Examples include, but are not limited to, changing
   * types, removing or reassigning enums, adding/removing/rearranging
   * fields to structures
   */
John Koleszar's avatar
John Koleszar committed
62
#define VPX_ENCODER_ABI_VERSION (3 + VPX_CODEC_ABI_VERSION) /**<\hideinitializer*/
John Koleszar's avatar
John Koleszar committed
63 64


John Koleszar's avatar
John Koleszar committed
65 66 67 68 69 70 71 72 73
  /*! \brief Encoder capabilities bitfield
   *
   *  Each encoder advertises the capabilities it supports as part of its
   *  ::vpx_codec_iface_t interface structure. Capabilities are extra
   *  interfaces or functionality, and are not required to be supported
   *  by an encoder.
   *
   *  The available flags are specified by VPX_CODEC_CAP_* defines.
   */
John Koleszar's avatar
John Koleszar committed
74 75
#define VPX_CODEC_CAP_PSNR  0x10000 /**< Can issue PSNR packets */

John Koleszar's avatar
John Koleszar committed
76 77 78 79 80
  /*! Can output one partition at a time. Each partition is returned in its
   *  own VPX_CODEC_CX_FRAME_PKT, with the FRAME_IS_FRAGMENT flag set for
   *  every partition but the last. In this mode all frames are always
   *  returned partition by partition.
   */
81 82
#define VPX_CODEC_CAP_OUTPUT_PARTITION  0x20000

John Koleszar's avatar
John Koleszar committed
83

John Koleszar's avatar
John Koleszar committed
84 85 86 87 88 89 90
  /*! \brief Initialization-time Feature Enabling
   *
   *  Certain codec features must be known at initialization time, to allow
   *  for proper memory allocation.
   *
   *  The available flags are specified by VPX_CODEC_USE_* defines.
   */
John Koleszar's avatar
John Koleszar committed
91
#define VPX_CODEC_USE_PSNR  0x10000 /**< Calculate PSNR on each frame */
92
#define VPX_CODEC_USE_OUTPUT_PARTITION  0x20000 /**< Make the encoder output one
John Koleszar's avatar
John Koleszar committed
93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121
  partition at a time. */


  /*!\brief Generic fixed size buffer structure
   *
   * This structure is able to hold a reference to any fixed size buffer.
   */
  typedef struct vpx_fixed_buf {
    void          *buf; /**< Pointer to the data */
    size_t         sz;  /**< Length of the buffer, in chars */
  } vpx_fixed_buf_t; /**< alias for struct vpx_fixed_buf */


  /*!\brief Time Stamp Type
   *
   * An integer, which when multiplied by the stream's time base, provides
   * the absolute time of a sample.
   */
  typedef int64_t vpx_codec_pts_t;


  /*!\brief Compressed Frame Flags
   *
   * This type represents a bitfield containing information about a compressed
   * frame that may be useful to an application. The most significant 16 bits
   * can be used by an algorithm to provide additional detail, for example to
   * support frame types that are codec specific (MPEG-1 D-frames for example)
   */
  typedef uint32_t vpx_codec_frame_flags_t;
John Koleszar's avatar
John Koleszar committed
122 123
#define VPX_FRAME_IS_KEY       0x1 /**< frame is the start of a GOP */
#define VPX_FRAME_IS_DROPPABLE 0x2 /**< frame can be dropped without affecting
John Koleszar's avatar
John Koleszar committed
124 125
  the stream (no future frame depends on
              this one) */
John Koleszar's avatar
John Koleszar committed
126
#define VPX_FRAME_IS_INVISIBLE 0x4 /**< frame should be decoded but will not
John Koleszar's avatar
John Koleszar committed
127
  be shown */
128
#define VPX_FRAME_IS_FRAGMENT  0x8 /**< this is a fragment of the encoded
John Koleszar's avatar
John Koleszar committed
129 130 131 132 133 134 135 136 137
  frame */

  /*!\brief Error Resilient flags
   *
   * These flags define which error resilient features to enable in the
   * encoder. The flags are specified through the
   * vpx_codec_enc_cfg::g_error_resilient variable.
   */
  typedef uint32_t vpx_codec_er_flags_t;
138
#define VPX_ERROR_RESILIENT_DEFAULT     0x1 /**< Improve resiliency against
John Koleszar's avatar
John Koleszar committed
139
  losses of whole frames */
140
#define VPX_ERROR_RESILIENT_PARTITIONS  0x2 /**< The frame partitions are
John Koleszar's avatar
John Koleszar committed
141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
  independently decodable by the
  bool decoder, meaning that
  partitions can be decoded even
  though earlier partitions have
  been lost. Note that intra
  predicition is still done over
  the partition boundary. */

  /*!\brief Encoder output packet variants
   *
   * This enumeration lists the different kinds of data packets that can be
   * returned by calls to vpx_codec_get_cx_data(). Algorithms \ref MAY
   * extend this list to provide additional functionality.
   */
  enum vpx_codec_cx_pkt_kind {
    VPX_CODEC_CX_FRAME_PKT,    /**< Compressed video frame */
    VPX_CODEC_STATS_PKT,       /**< Two-pass statistics for this frame */
    VPX_CODEC_PSNR_PKT,        /**< PSNR statistics for this frame */
    VPX_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions  */
  };


  /*!\brief Encoder output packet
   *
   * This structure contains the different kinds of output data the encoder
   * may produce while compressing a frame.
   */
  typedef struct vpx_codec_cx_pkt {
    enum vpx_codec_cx_pkt_kind  kind; /**< packet variant */
    union {
      struct {
        void                    *buf;      /**< compressed data buffer */
        size_t                   sz;       /**< length of compressed data */
        vpx_codec_pts_t          pts;      /**< time stamp to show frame
John Koleszar's avatar
John Koleszar committed
175
                                                    (in timebase units) */
John Koleszar's avatar
John Koleszar committed
176
        unsigned long            duration; /**< duration to show frame
John Koleszar's avatar
John Koleszar committed
177
                                                    (in timebase units) */
John Koleszar's avatar
John Koleszar committed
178 179
        vpx_codec_frame_flags_t  flags;    /**< flags for this frame */
        int                      partition_id; /**< the partition id
180 181 182 183 184 185
                                              defines the decoding order
                                              of the partitions. Only
                                              applicable when "output partition"
                                              mode is enabled. First partition
                                              has id 0.*/

John Koleszar's avatar
John Koleszar committed
186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224
      } frame;  /**< data for compressed frame packet */
      struct vpx_fixed_buf twopass_stats;  /**< data for two-pass packet */
      struct vpx_psnr_pkt {
        unsigned int samples[4];  /**< Number of samples, total/y/u/v */
        uint64_t     sse[4];      /**< sum squared error, total/y/u/v */
        double       psnr[4];     /**< PSNR, total/y/u/v */
      } psnr;                       /**< data for PSNR packet */
      struct vpx_fixed_buf raw;     /**< data for arbitrary packets */

      /* This packet size is fixed to allow codecs to extend this
       * interface without having to manage storage for raw packets,
       * i.e., if it's smaller than 128 bytes, you can store in the
       * packet list directly.
       */
      char pad[128 - sizeof(enum vpx_codec_cx_pkt_kind)]; /**< fixed sz */
    } data; /**< packet data */
  } vpx_codec_cx_pkt_t; /**< alias for struct vpx_codec_cx_pkt */


  /*!\brief Rational Number
   *
   * This structure holds a fractional value.
   */
  typedef struct vpx_rational {
    int num; /**< fraction numerator */
    int den; /**< fraction denominator */
  } vpx_rational_t; /**< alias for struct vpx_rational */


  /*!\brief Multi-pass Encoding Pass */
  enum vpx_enc_pass {
    VPX_RC_ONE_PASS,   /**< Single pass mode */
    VPX_RC_FIRST_PASS, /**< First pass of multi-pass mode */
    VPX_RC_LAST_PASS   /**< Final pass of multi-pass mode */
  };


  /*!\brief Rate control mode */
  enum vpx_rc_mode {
225
    VPX_VBR,  /**< Variable Bit Rate (VBR) mode */
John Koleszar's avatar
John Koleszar committed
226
    VPX_CBR,  /**< Constant Bit Rate (CBR) mode */
227 228
    VPX_CQ,   /**< Constrained Quality (CQ)  mode */
    VPX_Q,    /**< Constant Quality (Q) mode */
John Koleszar's avatar
John Koleszar committed
229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267
  };


  /*!\brief Keyframe placement mode.
   *
   * This enumeration determines whether keyframes are placed automatically by
   * the encoder or whether this behavior is disabled. Older releases of this
   * SDK were implemented such that VPX_KF_FIXED meant keyframes were disabled.
   * This name is confusing for this behavior, so the new symbols to be used
   * are VPX_KF_AUTO and VPX_KF_DISABLED.
   */
  enum vpx_kf_mode {
    VPX_KF_FIXED, /**< deprecated, implies VPX_KF_DISABLED */
    VPX_KF_AUTO,  /**< Encoder determines optimal placement automatically */
    VPX_KF_DISABLED = 0 /**< Encoder does not place keyframes. */
  };


  /*!\brief Encoded Frame Flags
   *
   * This type indicates a bitfield to be passed to vpx_codec_encode(), defining
   * per-frame boolean values. By convention, bits common to all codecs will be
   * named VPX_EFLAG_*, and bits specific to an algorithm will be named
   * /algo/_eflag_*. The lower order 16 bits are reserved for common use.
   */
  typedef long vpx_enc_frame_flags_t;
#define VPX_EFLAG_FORCE_KF (1<<0)  /**< Force this frame to be a keyframe */


  /*!\brief Encoder configuration structure
   *
   * This structure contains the encoder settings that have common representations
   * across all codecs. This doesn't imply that all codecs support all features,
   * however.
   */
  typedef struct vpx_codec_enc_cfg {
    /*
     * generic settings (g)
     */
John Koleszar's avatar
John Koleszar committed
268

John Koleszar's avatar
John Koleszar committed
269 270 271 272 273 274 275 276
    /*!\brief Algorithm specific "usage" value
     *
     * Algorithms may define multiple values for usage, which may convey the
     * intent of how the application intends to use the stream. If this value
     * is non-zero, consult the documentation for the codec to determine its
     * meaning.
     */
    unsigned int           g_usage;
John Koleszar's avatar
John Koleszar committed
277 278


John Koleszar's avatar
John Koleszar committed
279
    /*!\brief Maximum number of threads to use
John Koleszar's avatar
John Koleszar committed
280
     *
John Koleszar's avatar
John Koleszar committed
281 282 283
     * For multi-threaded implementations, use no more than this number of
     * threads. The codec may use fewer threads than allowed. The value
     * 0 is equivalent to the value 1.
John Koleszar's avatar
John Koleszar committed
284
     */
John Koleszar's avatar
John Koleszar committed
285
    unsigned int           g_threads;
John Koleszar's avatar
John Koleszar committed
286 287


John Koleszar's avatar
John Koleszar committed
288 289 290 291 292 293 294 295 296
    /*!\brief Bitstream profile to use
     *
     * Some codecs support a notion of multiple bitstream profiles. Typically
     * this maps to a set of features that are turned on or off. Often the
     * profile to use is determined by the features of the intended decoder.
     * Consult the documentation for the codec to determine the valid values
     * for this parameter, or set to zero for a sane default.
     */
    unsigned int           g_profile;  /**< profile of bitstream to use */
John Koleszar's avatar
John Koleszar committed
297 298


John Koleszar's avatar
John Koleszar committed
299 300 301 302 303 304 305 306 307

    /*!\brief Width of the frame
     *
     * This value identifies the presentation resolution of the frame,
     * in pixels. Note that the frames passed as input to the encoder must
     * have this resolution. Frames will be presented by the decoder in this
     * resolution, independent of any spatial resampling the encoder may do.
     */
    unsigned int           g_w;
John Koleszar's avatar
John Koleszar committed
308 309


John Koleszar's avatar
John Koleszar committed
310
    /*!\brief Height of the frame
John Koleszar's avatar
John Koleszar committed
311
     *
John Koleszar's avatar
John Koleszar committed
312 313 314 315
     * This value identifies the presentation resolution of the frame,
     * in pixels. Note that the frames passed as input to the encoder must
     * have this resolution. Frames will be presented by the decoder in this
     * resolution, independent of any spatial resampling the encoder may do.
John Koleszar's avatar
John Koleszar committed
316
     */
John Koleszar's avatar
John Koleszar committed
317
    unsigned int           g_h;
John Koleszar's avatar
John Koleszar committed
318 319


John Koleszar's avatar
John Koleszar committed
320
    /*!\brief Stream timebase units
John Koleszar's avatar
John Koleszar committed
321
     *
John Koleszar's avatar
John Koleszar committed
322 323 324 325 326 327 328 329 330
     * Indicates the smallest interval of time, in seconds, used by the stream.
     * For fixed frame rate material, or variable frame rate material where
     * frames are timed at a multiple of a given clock (ex: video capture),
     * the \ref RECOMMENDED method is to set the timebase to the reciprocal
     * of the frame rate (ex: 1001/30000 for 29.970 Hz NTSC). This allows the
     * pts to correspond to the frame number, which can be handy. For
     * re-encoding video from containers with absolute time timestamps, the
     * \ref RECOMMENDED method is to set the timebase to that of the parent
     * container or multimedia framework (ex: 1/1000 for ms, as in FLV).
John Koleszar's avatar
John Koleszar committed
331
     */
John Koleszar's avatar
John Koleszar committed
332
    struct vpx_rational    g_timebase;
John Koleszar's avatar
John Koleszar committed
333 334


John Koleszar's avatar
John Koleszar committed
335
    /*!\brief Enable error resilient modes.
John Koleszar's avatar
John Koleszar committed
336
     *
John Koleszar's avatar
John Koleszar committed
337 338 339
     * The error resilient bitfield indicates to the encoder which features
     * it should enable to take measures for streaming over lossy or noisy
     * links.
John Koleszar's avatar
John Koleszar committed
340
     */
John Koleszar's avatar
John Koleszar committed
341
    vpx_codec_er_flags_t   g_error_resilient;
John Koleszar's avatar
John Koleszar committed
342 343


John Koleszar's avatar
John Koleszar committed
344
    /*!\brief Multi-pass Encoding Mode
John Koleszar's avatar
John Koleszar committed
345
     *
John Koleszar's avatar
John Koleszar committed
346 347 348 349 350 351 352
     * This value should be set to the current phase for multi-pass encoding.
     * For single pass, set to #VPX_RC_ONE_PASS.
     */
    enum vpx_enc_pass      g_pass;


    /*!\brief Allow lagged encoding
John Koleszar's avatar
John Koleszar committed
353
     *
John Koleszar's avatar
John Koleszar committed
354 355 356 357 358
     * If set, this value allows the encoder to consume a number of input
     * frames before producing output frames. This allows the encoder to
     * base decisions for the current frame on future frames. This does
     * increase the latency of the encoding pipeline, so it is not appropriate
     * in all situations (ex: realtime encoding).
John Koleszar's avatar
John Koleszar committed
359
     *
John Koleszar's avatar
John Koleszar committed
360 361 362
     * Note that this is a maximum value -- the encoder may produce frames
     * sooner than the given limit. Set this value to 0 to disable this
     * feature.
John Koleszar's avatar
John Koleszar committed
363
     */
John Koleszar's avatar
John Koleszar committed
364 365
    unsigned int           g_lag_in_frames;

John Koleszar's avatar
John Koleszar committed
366

John Koleszar's avatar
John Koleszar committed
367 368 369
    /*
     * rate control settings (rc)
     */
John Koleszar's avatar
John Koleszar committed
370

John Koleszar's avatar
John Koleszar committed
371
    /*!\brief Temporal resampling configuration, if supported by the codec.
John Koleszar's avatar
John Koleszar committed
372
     *
John Koleszar's avatar
John Koleszar committed
373 374 375 376 377
     * Temporal resampling allows the codec to "drop" frames as a strategy to
     * meet its target data rate. This can cause temporal discontinuities in
     * the encoded video, which may appear as stuttering during playback. This
     * trade-off is often acceptable, but for many applications is not. It can
     * be disabled in these cases.
John Koleszar's avatar
John Koleszar committed
378
     *
John Koleszar's avatar
John Koleszar committed
379 380
     * Note that not all codecs support this feature. All vpx VPx codecs do.
     * For other codecs, consult the documentation for that algorithm.
John Koleszar's avatar
John Koleszar committed
381
     *
John Koleszar's avatar
John Koleszar committed
382 383 384 385
     * This threshold is described as a percentage of the target data buffer.
     * When the data buffer falls below this percentage of fullness, a
     * dropped frame is indicated. Set the threshold to zero (0) to disable
     * this feature.
John Koleszar's avatar
John Koleszar committed
386
     */
John Koleszar's avatar
John Koleszar committed
387
    unsigned int           rc_dropframe_thresh;
John Koleszar's avatar
John Koleszar committed
388 389


John Koleszar's avatar
John Koleszar committed
390
    /*!\brief Enable/disable spatial resampling, if supported by the codec.
John Koleszar's avatar
John Koleszar committed
391
     *
John Koleszar's avatar
John Koleszar committed
392 393 394 395 396 397 398 399 400
     * Spatial resampling allows the codec to compress a lower resolution
     * version of the frame, which is then upscaled by the encoder to the
     * correct presentation resolution. This increases visual quality at
     * low data rates, at the expense of CPU time on the encoder/decoder.
     */
    unsigned int           rc_resize_allowed;


    /*!\brief Spatial resampling up watermark.
John Koleszar's avatar
John Koleszar committed
401
     *
John Koleszar's avatar
John Koleszar committed
402 403 404
     * This threshold is described as a percentage of the target data buffer.
     * When the data buffer rises above this percentage of fullness, the
     * encoder will step up to a higher resolution version of the frame.
John Koleszar's avatar
John Koleszar committed
405
     */
John Koleszar's avatar
John Koleszar committed
406
    unsigned int           rc_resize_up_thresh;
John Koleszar's avatar
John Koleszar committed
407 408


John Koleszar's avatar
John Koleszar committed
409 410 411 412 413
    /*!\brief Spatial resampling down watermark.
     *
     * This threshold is described as a percentage of the target data buffer.
     * When the data buffer falls below this percentage of fullness, the
     * encoder will step down to a lower resolution version of the frame.
John Koleszar's avatar
John Koleszar committed
414
     */
John Koleszar's avatar
John Koleszar committed
415
    unsigned int           rc_resize_down_thresh;
John Koleszar's avatar
John Koleszar committed
416 417


John Koleszar's avatar
John Koleszar committed
418
    /*!\brief Rate control algorithm to use.
John Koleszar's avatar
John Koleszar committed
419
     *
John Koleszar's avatar
John Koleszar committed
420 421 422 423 424 425 426 427 428 429
     * Indicates whether the end usage of this stream is to be streamed over
     * a bandwidth constrained link, indicating that Constant Bit Rate (CBR)
     * mode should be used, or whether it will be played back on a high
     * bandwidth link, as from a local disk, where higher variations in
     * bitrate are acceptable.
     */
    enum vpx_rc_mode       rc_end_usage;


    /*!\brief Two-pass stats buffer.
John Koleszar's avatar
John Koleszar committed
430
     *
John Koleszar's avatar
John Koleszar committed
431 432 433 434 435 436 437
     * A buffer containing all of the stats packets produced in the first
     * pass, concatenated.
     */
    struct vpx_fixed_buf   rc_twopass_stats_in;


    /*!\brief Target data rate
John Koleszar's avatar
John Koleszar committed
438
     *
John Koleszar's avatar
John Koleszar committed
439 440 441 442 443 444 445 446 447 448 449
     * Target bandwidth to use for this stream, in kilobits per second.
     */
    unsigned int           rc_target_bitrate;


    /*
     * quantizer settings
     */


    /*!\brief Minimum (Best Quality) Quantizer
John Koleszar's avatar
John Koleszar committed
450
     *
John Koleszar's avatar
John Koleszar committed
451 452 453 454 455 456 457 458 459 460
     * The quantizer is the most direct control over the quality of the
     * encoded image. The range of valid values for the quantizer is codec
     * specific. Consult the documentation for the codec to determine the
     * values to use. To determine the range programmatically, call
     * vpx_codec_enc_config_default() with a usage value of 0.
     */
    unsigned int           rc_min_quantizer;


    /*!\brief Maximum (Worst Quality) Quantizer
John Koleszar's avatar
John Koleszar committed
461
     *
John Koleszar's avatar
John Koleszar committed
462 463 464 465 466 467 468 469 470 471 472 473 474 475 476
     * The quantizer is the most direct control over the quality of the
     * encoded image. The range of valid values for the quantizer is codec
     * specific. Consult the documentation for the codec to determine the
     * values to use. To determine the range programmatically, call
     * vpx_codec_enc_config_default() with a usage value of 0.
     */
    unsigned int           rc_max_quantizer;


    /*
     * bitrate tolerance
     */


    /*!\brief Rate control adaptation undershoot control
John Koleszar's avatar
John Koleszar committed
477
     *
John Koleszar's avatar
John Koleszar committed
478 479 480 481 482
     * This value, expressed as a percentage of the target bitrate,
     * controls the maximum allowed adaptation speed of the codec.
     * This factor controls the maximum amount of bits that can
     * be subtracted from the target bitrate in order to compensate
     * for prior overshoot.
John Koleszar's avatar
John Koleszar committed
483
     *
John Koleszar's avatar
John Koleszar committed
484
     * Valid values in the range 0-1000.
John Koleszar's avatar
John Koleszar committed
485
     */
John Koleszar's avatar
John Koleszar committed
486
    unsigned int           rc_undershoot_pct;
John Koleszar's avatar
John Koleszar committed
487 488


John Koleszar's avatar
John Koleszar committed
489
    /*!\brief Rate control adaptation overshoot control
John Koleszar's avatar
John Koleszar committed
490
     *
John Koleszar's avatar
John Koleszar committed
491 492 493 494 495
     * This value, expressed as a percentage of the target bitrate,
     * controls the maximum allowed adaptation speed of the codec.
     * This factor controls the maximum amount of bits that can
     * be added to the target bitrate in order to compensate for
     * prior undershoot.
John Koleszar's avatar
John Koleszar committed
496
     *
John Koleszar's avatar
John Koleszar committed
497 498 499 500 501 502 503 504 505 506 507
     * Valid values in the range 0-1000.
     */
    unsigned int           rc_overshoot_pct;


    /*
     * decoder buffer model parameters
     */


    /*!\brief Decoder Buffer Size
John Koleszar's avatar
John Koleszar committed
508
     *
John Koleszar's avatar
John Koleszar committed
509 510 511 512 513 514 515 516 517 518 519
     * This value indicates the amount of data that may be buffered by the
     * decoding application. Note that this value is expressed in units of
     * time (milliseconds). For example, a value of 5000 indicates that the
     * client will buffer (at least) 5000ms worth of encoded data. Use the
     * target bitrate (#rc_target_bitrate) to convert to bits/bytes, if
     * necessary.
     */
    unsigned int           rc_buf_sz;


    /*!\brief Decoder Buffer Initial Size
John Koleszar's avatar
John Koleszar committed
520
     *
John Koleszar's avatar
John Koleszar committed
521 522 523 524 525 526 527 528 529
     * This value indicates the amount of data that will be buffered by the
     * decoding application prior to beginning playback. This value is
     * expressed in units of time (milliseconds). Use the target bitrate
     * (#rc_target_bitrate) to convert to bits/bytes, if necessary.
     */
    unsigned int           rc_buf_initial_sz;


    /*!\brief Decoder Buffer Optimal Size
John Koleszar's avatar
John Koleszar committed
530
     *
John Koleszar's avatar
John Koleszar committed
531 532 533 534 535 536 537 538 539 540 541 542 543 544
     * This value indicates the amount of data that the encoder should try
     * to maintain in the decoder's buffer. This value is expressed in units
     * of time (milliseconds). Use the target bitrate (#rc_target_bitrate)
     * to convert to bits/bytes, if necessary.
     */
    unsigned int           rc_buf_optimal_sz;


    /*
     * 2 pass rate control parameters
     */


    /*!\brief Two-pass mode CBR/VBR bias
John Koleszar's avatar
John Koleszar committed
545
     *
John Koleszar's avatar
John Koleszar committed
546 547 548 549 550 551 552 553 554 555
     * Bias, expressed on a scale of 0 to 100, for determining target size
     * for the current frame. The value 0 indicates the optimal CBR mode
     * value should be used. The value 100 indicates the optimal VBR mode
     * value should be used. Values in between indicate which way the
     * encoder should "lean."
     */
    unsigned int           rc_2pass_vbr_bias_pct;       /**< RC mode bias between CBR and VBR(0-100: 0->CBR, 100->VBR)   */


    /*!\brief Two-pass mode per-GOP minimum bitrate
John Koleszar's avatar
John Koleszar committed
556
     *
John Koleszar's avatar
John Koleszar committed
557 558
     * This value, expressed as a percentage of the target bitrate, indicates
     * the minimum bitrate to be used for a single GOP (aka "section")
John Koleszar's avatar
John Koleszar committed
559
     */
John Koleszar's avatar
John Koleszar committed
560
    unsigned int           rc_2pass_vbr_minsection_pct;
John Koleszar's avatar
John Koleszar committed
561 562


John Koleszar's avatar
John Koleszar committed
563
    /*!\brief Two-pass mode per-GOP maximum bitrate
John Koleszar's avatar
John Koleszar committed
564
     *
John Koleszar's avatar
John Koleszar committed
565 566 567 568 569 570 571 572 573 574 575
     * This value, expressed as a percentage of the target bitrate, indicates
     * the maximum bitrate to be used for a single GOP (aka "section")
     */
    unsigned int           rc_2pass_vbr_maxsection_pct;


    /*
     * keyframing settings (kf)
     */

    /*!\brief Keyframe placement mode
John Koleszar's avatar
John Koleszar committed
576
     *
John Koleszar's avatar
John Koleszar committed
577 578 579 580 581 582 583 584
     * This value indicates whether the encoder should place keyframes at a
     * fixed interval, or determine the optimal placement automatically
     * (as governed by the #kf_min_dist and #kf_max_dist parameters)
     */
    enum vpx_kf_mode       kf_mode;


    /*!\brief Keyframe minimum interval
John Koleszar's avatar
John Koleszar committed
585
     *
John Koleszar's avatar
John Koleszar committed
586 587 588 589 590 591 592 593 594
     * This value, expressed as a number of frames, prevents the encoder from
     * placing a keyframe nearer than kf_min_dist to the previous keyframe. At
     * least kf_min_dist frames non-keyframes will be coded before the next
     * keyframe. Set kf_min_dist equal to kf_max_dist for a fixed interval.
     */
    unsigned int           kf_min_dist;


    /*!\brief Keyframe maximum interval
John Koleszar's avatar
John Koleszar committed
595
     *
John Koleszar's avatar
John Koleszar committed
596 597 598 599
     * This value, expressed as a number of frames, forces the encoder to code
     * a keyframe if one has not been coded in the last kf_max_dist frames.
     * A value of 0 implies all frames will be keyframes. Set kf_min_dist
     * equal to kf_max_dist for a fixed interval.
John Koleszar's avatar
John Koleszar committed
600
     */
John Koleszar's avatar
John Koleszar committed
601
    unsigned int           kf_max_dist;
John Koleszar's avatar
John Koleszar committed
602

John Koleszar's avatar
John Koleszar committed
603
    /*
604 605 606
     * Spatial scalability settings (ss)
     */

607
    /*!\brief Number of spatial coding layers.
608
     *
609
     * This value specifies the number of spatial coding layers to be used.
John Koleszar's avatar
John Koleszar committed
610
     */
611
    unsigned int           ss_number_layers;
John Koleszar's avatar
John Koleszar committed
612

613
    /*!\brief Number of temporal coding layers.
John Koleszar's avatar
John Koleszar committed
614
     *
615
     * This value specifies the number of temporal layers to be used.
John Koleszar's avatar
John Koleszar committed
616 617 618
     */
    unsigned int           ts_number_layers;

619
    /*!\brief Target bitrate for each temporal layer.
John Koleszar's avatar
John Koleszar committed
620
     *
621 622
     * These values specify the target coding bitrate to be used for each
     * temporal layer.
John Koleszar's avatar
John Koleszar committed
623 624 625
     */
    unsigned int           ts_target_bitrate[VPX_TS_MAX_LAYERS];

626
    /*!\brief Frame rate decimation factor for each temporal layer.
John Koleszar's avatar
John Koleszar committed
627 628
     *
     * These values specify the frame rate decimation factors to apply
629
     * to each temporal layer.
John Koleszar's avatar
John Koleszar committed
630 631
     */
    unsigned int           ts_rate_decimator[VPX_TS_MAX_LAYERS];
John Koleszar's avatar
John Koleszar committed
632

633 634 635 636 637 638 639
    /*!\brief Length of the sequence defining frame temporal layer membership.
    *
    * This value specifies the length of the sequence that defines the
    * membership of frames to temporal layers. For example, if ts_periodicity=8
    * then frames are assigned to coding layers with a repeated sequence of
    * length 8.
    */
John Koleszar's avatar
John Koleszar committed
640 641
    unsigned int           ts_periodicity;

642 643 644 645 646 647 648
    /*!\brief Template defining the membership of frames to temporal layers.
    *
    * This array defines the membership of frames to temporal coding layers.
    * For a 2-layer encoding that assigns even numbered frames to one temporal
    * layer (0) and odd numbered frames to a second temporal layer (1) with
    * ts_periodicity=8, then ts_layer_id = (0,1,0,1,0,1,0,1).
    */
John Koleszar's avatar
John Koleszar committed
649
    unsigned int           ts_layer_id[VPX_TS_MAX_PERIODICITY];
John Koleszar's avatar
John Koleszar committed
650 651 652 653 654 655 656 657 658 659
  } vpx_codec_enc_cfg_t; /**< alias for struct vpx_codec_enc_cfg */


  /*!\brief Initialize an encoder instance
   *
   * Initializes a encoder context using the given interface. Applications
   * should call the vpx_codec_enc_init convenience macro instead of this
   * function directly, to ensure that the ABI version number parameter
   * is properly initialized.
   *
John Koleszar's avatar
John Koleszar committed
660 661 662 663
   * If the library was configured with --disable-multithread, this call
   * is not thread safe and should be guarded with a lock if being used
   * in a multithreaded context.
   *
John Koleszar's avatar
John Koleszar committed
664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693
   * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags
   * parameter), the storage pointed to by the cfg parameter must be
   * kept readable and stable until all memory maps have been set.
   *
   * \param[in]    ctx     Pointer to this instance's context.
   * \param[in]    iface   Pointer to the algorithm interface to use.
   * \param[in]    cfg     Configuration to use, if known. May be NULL.
   * \param[in]    flags   Bitfield of VPX_CODEC_USE_* flags
   * \param[in]    ver     ABI version number. Must be set to
   *                       VPX_ENCODER_ABI_VERSION
   * \retval #VPX_CODEC_OK
   *     The decoder algorithm initialized.
   * \retval #VPX_CODEC_MEM_ERROR
   *     Memory allocation failed.
   */
  vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t      *ctx,
                                         vpx_codec_iface_t    *iface,
                                         vpx_codec_enc_cfg_t  *cfg,
                                         vpx_codec_flags_t     flags,
                                         int                   ver);


  /*!\brief Convenience macro for vpx_codec_enc_init_ver()
   *
   * Ensures the ABI version parameter is properly set.
   */
#define vpx_codec_enc_init(ctx, iface, cfg, flags) \
  vpx_codec_enc_init_ver(ctx, iface, cfg, flags, VPX_ENCODER_ABI_VERSION)


John Koleszar's avatar
John Koleszar committed
694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735
  /*!\brief Initialize multi-encoder instance
   *
   * Initializes multi-encoder context using the given interface.
   * Applications should call the vpx_codec_enc_init_multi convenience macro
   * instead of this function directly, to ensure that the ABI version number
   * parameter is properly initialized.
   *
   * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags
   * parameter), the storage pointed to by the cfg parameter must be
   * kept readable and stable until all memory maps have been set.
   *
   * \param[in]    ctx     Pointer to this instance's context.
   * \param[in]    iface   Pointer to the algorithm interface to use.
   * \param[in]    cfg     Configuration to use, if known. May be NULL.
   * \param[in]    num_enc Total number of encoders.
   * \param[in]    flags   Bitfield of VPX_CODEC_USE_* flags
   * \param[in]    dsf     Pointer to down-sampling factors.
   * \param[in]    ver     ABI version number. Must be set to
   *                       VPX_ENCODER_ABI_VERSION
   * \retval #VPX_CODEC_OK
   *     The decoder algorithm initialized.
   * \retval #VPX_CODEC_MEM_ERROR
   *     Memory allocation failed.
   */
  vpx_codec_err_t vpx_codec_enc_init_multi_ver(vpx_codec_ctx_t      *ctx,
                                               vpx_codec_iface_t    *iface,
                                               vpx_codec_enc_cfg_t  *cfg,
                                               int                   num_enc,
                                               vpx_codec_flags_t     flags,
                                               vpx_rational_t       *dsf,
                                               int                   ver);


  /*!\brief Convenience macro for vpx_codec_enc_init_multi_ver()
   *
   * Ensures the ABI version parameter is properly set.
   */
#define vpx_codec_enc_init_multi(ctx, iface, cfg, num_enc, flags, dsf) \
  vpx_codec_enc_init_multi_ver(ctx, iface, cfg, num_enc, flags, dsf, \
                               VPX_ENCODER_ABI_VERSION)


John Koleszar's avatar
John Koleszar committed
736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932
  /*!\brief Get a default configuration
   *
   * Initializes a encoder configuration structure with default values. Supports
   * the notion of "usages" so that an algorithm may offer different default
   * settings depending on the user's intended goal. This function \ref SHOULD
   * be called by all applications to initialize the configuration structure
   * before specializing the configuration with application specific values.
   *
   * \param[in]    iface   Pointer to the algorithm interface to use.
   * \param[out]   cfg     Configuration buffer to populate
   * \param[in]    usage   End usage. Set to 0 or use codec specific values.
   *
   * \retval #VPX_CODEC_OK
   *     The configuration was populated.
   * \retval #VPX_CODEC_INCAPABLE
   *     Interface is not an encoder interface.
   * \retval #VPX_CODEC_INVALID_PARAM
   *     A parameter was NULL, or the usage value was not recognized.
   */
  vpx_codec_err_t  vpx_codec_enc_config_default(vpx_codec_iface_t    *iface,
                                                vpx_codec_enc_cfg_t  *cfg,
                                                unsigned int          usage);


  /*!\brief Set or change configuration
   *
   * Reconfigures an encoder instance according to the given configuration.
   *
   * \param[in]    ctx     Pointer to this instance's context
   * \param[in]    cfg     Configuration buffer to use
   *
   * \retval #VPX_CODEC_OK
   *     The configuration was populated.
   * \retval #VPX_CODEC_INCAPABLE
   *     Interface is not an encoder interface.
   * \retval #VPX_CODEC_INVALID_PARAM
   *     A parameter was NULL, or the usage value was not recognized.
   */
  vpx_codec_err_t  vpx_codec_enc_config_set(vpx_codec_ctx_t            *ctx,
                                            const vpx_codec_enc_cfg_t  *cfg);


  /*!\brief Get global stream headers
   *
   * Retrieves a stream level global header packet, if supported by the codec.
   *
   * \param[in]    ctx     Pointer to this instance's context
   *
   * \retval NULL
   *     Encoder does not support global header
   * \retval Non-NULL
   *     Pointer to buffer containing global header packet
   */
  vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t   *ctx);


#define VPX_DL_REALTIME     (1)        /**< deadline parameter analogous to
  *   VPx REALTIME mode. */
#define VPX_DL_GOOD_QUALITY (1000000)  /**< deadline parameter analogous to
  *   VPx GOOD QUALITY mode. */
#define VPX_DL_BEST_QUALITY (0)        /**< deadline parameter analogous to
  *   VPx BEST QUALITY mode. */
  /*!\brief Encode a frame
   *
   * Encodes a video frame at the given "presentation time." The presentation
   * time stamp (PTS) \ref MUST be strictly increasing.
   *
   * The encoder supports the notion of a soft real-time deadline. Given a
   * non-zero value to the deadline parameter, the encoder will make a "best
   * effort" guarantee to  return before the given time slice expires. It is
   * implicit that limiting the available time to encode will degrade the
   * output quality. The encoder can be given an unlimited time to produce the
   * best possible frame by specifying a deadline of '0'. This deadline
   * supercedes the VPx notion of "best quality, good quality, realtime".
   * Applications that wish to map these former settings to the new deadline
   * based system can use the symbols #VPX_DL_REALTIME, #VPX_DL_GOOD_QUALITY,
   * and #VPX_DL_BEST_QUALITY.
   *
   * When the last frame has been passed to the encoder, this function should
   * continue to be called, with the img parameter set to NULL. This will
   * signal the end-of-stream condition to the encoder and allow it to encode
   * any held buffers. Encoding is complete when vpx_codec_encode() is called
   * and vpx_codec_get_cx_data() returns no data.
   *
   * \param[in]    ctx       Pointer to this instance's context
   * \param[in]    img       Image data to encode, NULL to flush.
   * \param[in]    pts       Presentation time stamp, in timebase units.
   * \param[in]    duration  Duration to show frame, in timebase units.
   * \param[in]    flags     Flags to use for encoding this frame.
   * \param[in]    deadline  Time to spend encoding, in microseconds. (0=infinite)
   *
   * \retval #VPX_CODEC_OK
   *     The configuration was populated.
   * \retval #VPX_CODEC_INCAPABLE
   *     Interface is not an encoder interface.
   * \retval #VPX_CODEC_INVALID_PARAM
   *     A parameter was NULL, the image format is unsupported, etc.
   */
  vpx_codec_err_t  vpx_codec_encode(vpx_codec_ctx_t            *ctx,
                                    const vpx_image_t          *img,
                                    vpx_codec_pts_t             pts,
                                    unsigned long               duration,
                                    vpx_enc_frame_flags_t       flags,
                                    unsigned long               deadline);

  /*!\brief Set compressed data output buffer
   *
   * Sets the buffer that the codec should output the compressed data
   * into. This call effectively sets the buffer pointer returned in the
   * next VPX_CODEC_CX_FRAME_PKT packet. Subsequent packets will be
   * appended into this buffer. The buffer is preserved across frames,
   * so applications must periodically call this function after flushing
   * the accumulated compressed data to disk or to the network to reset
   * the pointer to the buffer's head.
   *
   * `pad_before` bytes will be skipped before writing the compressed
   * data, and `pad_after` bytes will be appended to the packet. The size
   * of the packet will be the sum of the size of the actual compressed
   * data, pad_before, and pad_after. The padding bytes will be preserved
   * (not overwritten).
   *
   * Note that calling this function does not guarantee that the returned
   * compressed data will be placed into the specified buffer. In the
   * event that the encoded data will not fit into the buffer provided,
   * the returned packet \ref MAY point to an internal buffer, as it would
   * if this call were never used. In this event, the output packet will
   * NOT have any padding, and the application must free space and copy it
   * to the proper place. This is of particular note in configurations
   * that may output multiple packets for a single encoded frame (e.g., lagged
   * encoding) or if the application does not reset the buffer periodically.
   *
   * Applications may restore the default behavior of the codec providing
   * the compressed data buffer by calling this function with a NULL
   * buffer.
   *
   * Applications \ref MUSTNOT call this function during iteration of
   * vpx_codec_get_cx_data().
   *
   * \param[in]    ctx         Pointer to this instance's context
   * \param[in]    buf         Buffer to store compressed data into
   * \param[in]    pad_before  Bytes to skip before writing compressed data
   * \param[in]    pad_after   Bytes to skip after writing compressed data
   *
   * \retval #VPX_CODEC_OK
   *     The buffer was set successfully.
   * \retval #VPX_CODEC_INVALID_PARAM
   *     A parameter was NULL, the image format is unsupported, etc.
   */
  vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t       *ctx,
                                            const vpx_fixed_buf_t *buf,
                                            unsigned int           pad_before,
                                            unsigned int           pad_after);


  /*!\brief Encoded data iterator
   *
   * Iterates over a list of data packets to be passed from the encoder to the
   * application. The different kinds of packets available are enumerated in
   * #vpx_codec_cx_pkt_kind.
   *
   * #VPX_CODEC_CX_FRAME_PKT packets should be passed to the application's
   * muxer. Multiple compressed frames may be in the list.
   * #VPX_CODEC_STATS_PKT packets should be appended to a global buffer.
   *
   * The application \ref MUST silently ignore any packet kinds that it does
   * not recognize or support.
   *
   * The data buffers returned from this function are only guaranteed to be
   * valid until the application makes another call to any vpx_codec_* function.
   *
   * \param[in]     ctx      Pointer to this instance's context
   * \param[in,out] iter     Iterator storage, initialized to NULL
   *
   * \return Returns a pointer to an output data packet (compressed frame data,
   *         two-pass statistics, etc.) or NULL to signal end-of-list.
   *
   */
  const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t   *ctx,
                                                  vpx_codec_iter_t  *iter);


  /*!\brief Get Preview Frame
   *
   * Returns an image that can be used as a preview. Shows the image as it would
   * exist at the decompressor. The application \ref MUST NOT write into this
   * image buffer.
   *
   * \param[in]     ctx      Pointer to this instance's context
   *
   * \return Returns a pointer to a preview image, or NULL if no image is
   *         available.
   *
   */
  const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t   *ctx);


  /*!@} - end defgroup encoder*/
John Koleszar's avatar
John Koleszar committed
933 934 935
#ifdef __cplusplus
}
#endif
James Zern's avatar
James Zern committed
936
#endif  // VPX_VPX_ENCODER_H_
Adrian Grange's avatar
Adrian Grange committed
937