aom_encoder.h 37.7 KB
Newer Older
John Koleszar's avatar
John Koleszar committed
1
/*
2
 * Copyright (c) 2016, Alliance for Open Media. All rights reserved
John Koleszar's avatar
John Koleszar committed
3
 *
4
5
6
7
8
9
 * This source code is subject to the terms of the BSD 2 Clause License and
 * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
 * was not distributed with this source code in the LICENSE file, you can
 * obtain it at www.aomedia.org/license/software. If the Alliance for Open
 * Media Patent License 1.0 was not distributed with this source code in the
 * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
John Koleszar's avatar
John Koleszar committed
10
 */
Adrian Grange's avatar
Adrian Grange committed
11
12
#ifndef AOM_AOM_ENCODER_H_
#define AOM_AOM_ENCODER_H_
John Koleszar's avatar
John Koleszar committed
13
14
15
16
17
18
19
20
21

/*!\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
22
/*!\file
John Koleszar's avatar
John Koleszar committed
23
24
25
26
27
28
29
30
31
32
 * \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

Adrian Grange's avatar
Adrian Grange committed
33
#include "./aom_codec.h"
John Koleszar's avatar
John Koleszar committed
34

clang-format's avatar
clang-format committed
35
36
37
/*! Temporal Scalability: Maximum length of the sequence defining frame
 * layer membership
 */
Adrian Grange's avatar
Adrian Grange committed
38
#define AOM_TS_MAX_PERIODICITY 16
John Koleszar's avatar
John Koleszar committed
39

clang-format's avatar
clang-format committed
40
/*! Temporal Scalability: Maximum number of coding layers */
Adrian Grange's avatar
Adrian Grange committed
41
#define AOM_TS_MAX_LAYERS 5
John Koleszar's avatar
John Koleszar committed
42

Adrian Grange's avatar
Adrian Grange committed
43
44
/*!\deprecated Use #AOM_TS_MAX_PERIODICITY instead. */
#define MAX_PERIODICITY AOM_TS_MAX_PERIODICITY
John Koleszar's avatar
John Koleszar committed
45

46
/*! Temporal+Spatial Scalability: Maximum number of coding layers */
Adrian Grange's avatar
Adrian Grange committed
47
#define AOM_MAX_LAYERS 12  // 3 temporal + 4 spatial layers are allowed.
48

Adrian Grange's avatar
Adrian Grange committed
49
50
/*!\deprecated Use #AOM_MAX_LAYERS instead. */
#define MAX_LAYERS AOM_MAX_LAYERS  // 3 temporal + 4 spatial layers allowed.
John Koleszar's avatar
John Koleszar committed
51

52
/*! Spatial Scalability: Maximum number of coding layers */
Adrian Grange's avatar
Adrian Grange committed
53
#define AOM_SS_MAX_LAYERS 5
54
55

/*! Spatial Scalability: Default number of coding layers */
Adrian Grange's avatar
Adrian Grange committed
56
#define AOM_SS_DEFAULT_LAYERS 1
John Koleszar's avatar
John Koleszar committed
57

clang-format's avatar
clang-format committed
58
59
60
61
62
63
64
65
/*!\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
 */
Adrian Grange's avatar
Adrian Grange committed
66
67
#define AOM_ENCODER_ABI_VERSION \
  (5 + AOM_CODEC_ABI_VERSION) /**<\hideinitializer*/
John Koleszar's avatar
John Koleszar committed
68

clang-format's avatar
clang-format committed
69
70
71
/*! \brief Encoder capabilities bitfield
 *
 *  Each encoder advertises the capabilities it supports as part of its
Adrian Grange's avatar
Adrian Grange committed
72
 *  ::aom_codec_iface_t interface structure. Capabilities are extra
clang-format's avatar
clang-format committed
73
74
75
 *  interfaces or functionality, and are not required to be supported
 *  by an encoder.
 *
Adrian Grange's avatar
Adrian Grange committed
76
 *  The available flags are specified by AOM_CODEC_CAP_* defines.
clang-format's avatar
clang-format committed
77
 */
Adrian Grange's avatar
Adrian Grange committed
78
#define AOM_CODEC_CAP_PSNR 0x10000 /**< Can issue PSNR packets */
John Koleszar's avatar
John Koleszar committed
79

clang-format's avatar
clang-format committed
80
/*! Can output one partition at a time. Each partition is returned in its
Adrian Grange's avatar
Adrian Grange committed
81
 *  own AOM_CODEC_CX_FRAME_PKT, with the FRAME_IS_FRAGMENT flag set for
clang-format's avatar
clang-format committed
82
83
84
 *  every partition but the last. In this mode all frames are always
 *  returned partition by partition.
 */
Adrian Grange's avatar
Adrian Grange committed
85
#define AOM_CODEC_CAP_OUTPUT_PARTITION 0x20000
86

87
88
/*! Can support input images at greater than 8 bitdepth.
 */
Adrian Grange's avatar
Adrian Grange committed
89
#define AOM_CODEC_CAP_HIGHBITDEPTH 0x40000
John Koleszar's avatar
John Koleszar committed
90

clang-format's avatar
clang-format committed
91
92
93
94
95
/*! \brief Initialization-time Feature Enabling
 *
 *  Certain codec features must be known at initialization time, to allow
 *  for proper memory allocation.
 *
Adrian Grange's avatar
Adrian Grange committed
96
 *  The available flags are specified by AOM_CODEC_USE_* defines.
clang-format's avatar
clang-format committed
97
 */
Adrian Grange's avatar
Adrian Grange committed
98
99
#define AOM_CODEC_USE_PSNR 0x10000 /**< Calculate PSNR on each frame */
#define AOM_CODEC_USE_OUTPUT_PARTITION 0x20000
clang-format's avatar
clang-format committed
100
/**< Make the encoder output one partition at a time. */
Adrian Grange's avatar
Adrian Grange committed
101
#define AOM_CODEC_USE_HIGHBITDEPTH 0x40000 /**< Use high bitdepth */
John Koleszar's avatar
John Koleszar committed
102

clang-format's avatar
clang-format committed
103
104
105
106
/*!\brief Generic fixed size buffer structure
 *
 * This structure is able to hold a reference to any fixed size buffer.
 */
Adrian Grange's avatar
Adrian Grange committed
107
typedef struct aom_fixed_buf {
clang-format's avatar
clang-format committed
108
109
  void *buf;       /**< Pointer to the data */
  size_t sz;       /**< Length of the buffer, in chars */
Adrian Grange's avatar
Adrian Grange committed
110
} aom_fixed_buf_t; /**< alias for struct aom_fixed_buf */
John Koleszar's avatar
John Koleszar committed
111

clang-format's avatar
clang-format committed
112
113
114
115
116
/*!\brief Time Stamp Type
 *
 * An integer, which when multiplied by the stream's time base, provides
 * the absolute time of a sample.
 */
Adrian Grange's avatar
Adrian Grange committed
117
typedef int64_t aom_codec_pts_t;
John Koleszar's avatar
John Koleszar committed
118

clang-format's avatar
clang-format committed
119
120
121
122
123
124
125
/*!\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)
 */
Adrian Grange's avatar
Adrian Grange committed
126
typedef uint32_t aom_codec_frame_flags_t;
Adrian Grange's avatar
Adrian Grange committed
127
128
#define AOM_FRAME_IS_KEY 0x1 /**< frame is the start of a GOP */
#define AOM_FRAME_IS_DROPPABLE 0x2
clang-format's avatar
clang-format committed
129
130
/**< frame can be dropped without affecting the stream (no future frame depends
 * on this one) */
Adrian Grange's avatar
Adrian Grange committed
131
#define AOM_FRAME_IS_INVISIBLE 0x4
clang-format's avatar
clang-format committed
132
/**< frame should be decoded but will not be shown */
Adrian Grange's avatar
Adrian Grange committed
133
#define AOM_FRAME_IS_FRAGMENT 0x8
clang-format's avatar
clang-format committed
134
135
136
137
138
139
/**< this is a fragment of the encoded frame */

/*!\brief Error Resilient flags
 *
 * These flags define which error resilient features to enable in the
 * encoder. The flags are specified through the
Adrian Grange's avatar
Adrian Grange committed
140
 * aom_codec_enc_cfg::g_error_resilient variable.
clang-format's avatar
clang-format committed
141
 */
Adrian Grange's avatar
Adrian Grange committed
142
typedef uint32_t aom_codec_er_flags_t;
Adrian Grange's avatar
Adrian Grange committed
143
#define AOM_ERROR_RESILIENT_DEFAULT 0x1
clang-format's avatar
clang-format committed
144
/**< Improve resiliency against losses of whole frames */
Adrian Grange's avatar
Adrian Grange committed
145
#define AOM_ERROR_RESILIENT_PARTITIONS 0x2
clang-format's avatar
clang-format committed
146
147
148
149
150
151
152
153
/**< The frame partitions are independently decodable by the bool decoder,
 * meaning that partitions can be decoded even though earlier partitions have
 * been lost. Note that intra prediction is still done over the partition
 * boundary. */

/*!\brief Encoder output packet variants
 *
 * This enumeration lists the different kinds of data packets that can be
Adrian Grange's avatar
Adrian Grange committed
154
 * returned by calls to aom_codec_get_cx_data(). Algorithms \ref MAY
clang-format's avatar
clang-format committed
155
156
 * extend this list to provide additional functionality.
 */
Adrian Grange's avatar
Adrian Grange committed
157
enum aom_codec_cx_pkt_kind {
Adrian Grange's avatar
Adrian Grange committed
158
159
160
161
  AOM_CODEC_CX_FRAME_PKT,   /**< Compressed video frame */
  AOM_CODEC_STATS_PKT,      /**< Two-pass statistics for this frame */
  AOM_CODEC_FPMB_STATS_PKT, /**< first pass mb statistics for this frame */
  AOM_CODEC_PSNR_PKT,       /**< PSNR statistics for this frame */
clang-format's avatar
clang-format committed
162
163
// Spatial SVC is still experimental and may be removed before the next ABI
// bump.
Adrian Grange's avatar
Adrian Grange committed
164
165
166
#if AOM_ENCODER_ABI_VERSION > (5 + AOM_CODEC_ABI_VERSION)
  AOM_CODEC_SPATIAL_SVC_LAYER_SIZES, /**< Sizes for each layer in this frame*/
  AOM_CODEC_SPATIAL_SVC_LAYER_PSNR,  /**< PSNR for each layer in this frame*/
clang-format's avatar
clang-format committed
167
#endif
Adrian Grange's avatar
Adrian Grange committed
168
  AOM_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions  */
clang-format's avatar
clang-format committed
169
};
John Koleszar's avatar
John Koleszar committed
170

clang-format's avatar
clang-format committed
171
172
173
174
175
/*!\brief Encoder output packet
 *
 * This structure contains the different kinds of output data the encoder
 * may produce while compressing a frame.
 */
Adrian Grange's avatar
Adrian Grange committed
176
177
typedef struct aom_codec_cx_pkt {
  enum aom_codec_cx_pkt_kind kind; /**< packet variant */
clang-format's avatar
clang-format committed
178
179
180
181
  union {
    struct {
      void *buf;                        /**< compressed data buffer */
      size_t sz;                        /**< length of compressed data */
Adrian Grange's avatar
Adrian Grange committed
182
      aom_codec_pts_t pts;              /**< time stamp to show frame
clang-format's avatar
clang-format committed
183
184
185
                                                 (in timebase units) */
      unsigned long duration;           /**< duration to show frame
                                                 (in timebase units) */
Adrian Grange's avatar
Adrian Grange committed
186
      aom_codec_frame_flags_t flags;    /**< flags for this frame */
clang-format's avatar
clang-format committed
187
188
189
190
191
192
193
      int partition_id;                 /**< the partition id
                                       defines the decoding order
                                       of the partitions. Only
                                       applicable when "output partition"
                                       mode is enabled. First partition
                                       has id 0.*/
    } frame;                            /**< data for compressed frame packet */
Adrian Grange's avatar
Adrian Grange committed
194
195
196
    aom_fixed_buf_t twopass_stats;      /**< data for two-pass packet */
    aom_fixed_buf_t firstpass_mb_stats; /**< first pass mb packet */
    struct aom_psnr_pkt {
clang-format's avatar
clang-format committed
197
198
199
200
      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 */
Adrian Grange's avatar
Adrian Grange committed
201
    aom_fixed_buf_t raw;       /**< data for arbitrary packets */
clang-format's avatar
clang-format committed
202
203
// Spatial SVC is still experimental and may be removed before the next
// ABI bump.
Adrian Grange's avatar
Adrian Grange committed
204
205
206
#if AOM_ENCODER_ABI_VERSION > (5 + AOM_CODEC_ABI_VERSION)
    size_t layer_sizes[AOM_SS_MAX_LAYERS];
    struct aom_psnr_pkt layer_psnr[AOM_SS_MAX_LAYERS];
clang-format's avatar
clang-format committed
207
#endif
John Koleszar's avatar
John Koleszar committed
208

clang-format's avatar
clang-format committed
209
210
211
212
213
    /* 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.
     */
Adrian Grange's avatar
Adrian Grange committed
214
    char pad[128 - sizeof(enum aom_codec_cx_pkt_kind)]; /**< fixed sz */
clang-format's avatar
clang-format committed
215
  } data;                                               /**< packet data */
Adrian Grange's avatar
Adrian Grange committed
216
} aom_codec_cx_pkt_t; /**< alias for struct aom_codec_cx_pkt */
John Koleszar's avatar
John Koleszar committed
217

clang-format's avatar
clang-format committed
218
219
220
221
222
223
224
/*!\brief Encoder return output buffer callback
 *
 * This callback function, when registered, returns with packets when each
 * spatial layer is encoded.
 */
// putting the definitions here for now. (agrange: find if there
// is a better place for this)
Adrian Grange's avatar
Adrian Grange committed
225
typedef void (*aom_codec_enc_output_cx_pkt_cb_fn_t)(aom_codec_cx_pkt_t *pkt,
clang-format's avatar
clang-format committed
226
227
228
                                                    void *user_data);

/*!\brief Callback function pointer / user data pair storage */
Adrian Grange's avatar
Adrian Grange committed
229
230
typedef struct aom_codec_enc_output_cx_cb_pair {
  aom_codec_enc_output_cx_pkt_cb_fn_t output_cx_pkt; /**< Callback function */
clang-format's avatar
clang-format committed
231
  void *user_priv; /**< Pointer to private data */
Adrian Grange's avatar
Adrian Grange committed
232
} aom_codec_priv_output_cx_pkt_cb_pair_t;
clang-format's avatar
clang-format committed
233
234
235
236
237

/*!\brief Rational Number
 *
 * This structure holds a fractional value.
 */
Adrian Grange's avatar
Adrian Grange committed
238
typedef struct aom_rational {
clang-format's avatar
clang-format committed
239
240
  int num;        /**< fraction numerator */
  int den;        /**< fraction denominator */
Adrian Grange's avatar
Adrian Grange committed
241
} aom_rational_t; /**< alias for struct aom_rational */
clang-format's avatar
clang-format committed
242
243

/*!\brief Multi-pass Encoding Pass */
Adrian Grange's avatar
Adrian Grange committed
244
enum aom_enc_pass {
Adrian Grange's avatar
Adrian Grange committed
245
246
247
  AOM_RC_ONE_PASS,   /**< Single pass mode */
  AOM_RC_FIRST_PASS, /**< First pass of multi-pass mode */
  AOM_RC_LAST_PASS   /**< Final pass of multi-pass mode */
clang-format's avatar
clang-format committed
248
249
250
};

/*!\brief Rate control mode */
Adrian Grange's avatar
Adrian Grange committed
251
enum aom_rc_mode {
Adrian Grange's avatar
Adrian Grange committed
252
253
254
255
  AOM_VBR, /**< Variable Bit Rate (VBR) mode */
  AOM_CBR, /**< Constant Bit Rate (CBR) mode */
  AOM_CQ,  /**< Constrained Quality (CQ)  mode */
  AOM_Q,   /**< Constant Quality (Q) mode */
clang-format's avatar
clang-format committed
256
257
258
259
260
261
};

/*!\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
Adrian Grange's avatar
Adrian Grange committed
262
 * SDK were implemented such that AOM_KF_FIXED meant keyframes were disabled.
clang-format's avatar
clang-format committed
263
 * This name is confusing for this behavior, so the new symbols to be used
Adrian Grange's avatar
Adrian Grange committed
264
 * are AOM_KF_AUTO and AOM_KF_DISABLED.
clang-format's avatar
clang-format committed
265
 */
Adrian Grange's avatar
Adrian Grange committed
266
enum aom_kf_mode {
Adrian Grange's avatar
Adrian Grange committed
267
268
269
  AOM_KF_FIXED,       /**< deprecated, implies AOM_KF_DISABLED */
  AOM_KF_AUTO,        /**< Encoder determines optimal placement automatically */
  AOM_KF_DISABLED = 0 /**< Encoder does not place keyframes. */
clang-format's avatar
clang-format committed
270
271
272
273
};

/*!\brief Encoded Frame Flags
 *
Adrian Grange's avatar
Adrian Grange committed
274
 * This type indicates a bitfield to be passed to aom_codec_encode(), defining
clang-format's avatar
clang-format committed
275
 * per-frame boolean values. By convention, bits common to all codecs will be
Adrian Grange's avatar
Adrian Grange committed
276
 * named AOM_EFLAG_*, and bits specific to an algorithm will be named
clang-format's avatar
clang-format committed
277
278
 * /algo/_eflag_*. The lower order 16 bits are reserved for common use.
 */
Adrian Grange's avatar
Adrian Grange committed
279
typedef long aom_enc_frame_flags_t;
Adrian Grange's avatar
Adrian Grange committed
280
#define AOM_EFLAG_FORCE_KF (1 << 0) /**< Force this frame to be a keyframe */
clang-format's avatar
clang-format committed
281
282
283
284
285
286
287

/*!\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.
 */
Adrian Grange's avatar
Adrian Grange committed
288
typedef struct aom_codec_enc_cfg {
clang-format's avatar
clang-format committed
289
290
  /*
   * generic settings (g)
John Koleszar's avatar
John Koleszar committed
291
   */
clang-format's avatar
clang-format committed
292
293

  /*!\brief Algorithm specific "usage" value
John Koleszar's avatar
John Koleszar committed
294
   *
clang-format's avatar
clang-format committed
295
296
297
298
   * 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.
John Koleszar's avatar
John Koleszar committed
299
   */
clang-format's avatar
clang-format committed
300
301
302
  unsigned int g_usage;

  /*!\brief Maximum number of threads to use
John Koleszar's avatar
John Koleszar committed
303
   *
clang-format's avatar
clang-format committed
304
305
306
   * 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
307
   */
clang-format's avatar
clang-format committed
308
  unsigned int g_threads;
John Koleszar's avatar
John Koleszar committed
309

clang-format's avatar
clang-format committed
310
  /*!\brief Bitstream profile to use
John Koleszar's avatar
John Koleszar committed
311
   *
clang-format's avatar
clang-format committed
312
313
314
315
316
   * 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.
John Koleszar's avatar
John Koleszar committed
317
   */
clang-format's avatar
clang-format committed
318
  unsigned int g_profile; /**< profile of bitstream to use */
John Koleszar's avatar
John Koleszar committed
319

clang-format's avatar
clang-format committed
320
  /*!\brief Width of the frame
321
   *
clang-format's avatar
clang-format committed
322
323
324
325
   * 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.
326
   */
clang-format's avatar
clang-format committed
327
328
329
  unsigned int g_w;

  /*!\brief Height of the frame
John Koleszar's avatar
John Koleszar committed
330
   *
clang-format's avatar
clang-format committed
331
332
333
334
   * 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
335
   */
clang-format's avatar
clang-format committed
336
  unsigned int g_h;
John Koleszar's avatar
John Koleszar committed
337

clang-format's avatar
clang-format committed
338
  /*!\brief Bit-depth of the codec
John Koleszar's avatar
John Koleszar committed
339
   *
clang-format's avatar
clang-format committed
340
341
   * This value identifies the bit_depth of the codec,
   * Only certain bit-depths are supported as identified in the
Adrian Grange's avatar
Adrian Grange committed
342
   * aom_bit_depth_t enum.
John Koleszar's avatar
John Koleszar committed
343
   */
Adrian Grange's avatar
Adrian Grange committed
344
  aom_bit_depth_t g_bit_depth;
John Koleszar's avatar
John Koleszar committed
345

clang-format's avatar
clang-format committed
346
  /*!\brief Bit-depth of the input frames
John Koleszar's avatar
John Koleszar committed
347
   *
clang-format's avatar
clang-format committed
348
349
350
   * This value identifies the bit_depth of the input frames in bits.
   * Note that the frames passed as input to the encoder must have
   * this bit-depth.
John Koleszar's avatar
John Koleszar committed
351
   */
clang-format's avatar
clang-format committed
352
353
354
355
356
357
358
359
360
361
362
363
364
365
  unsigned int g_input_bit_depth;

  /*!\brief Stream timebase units
   *
   * 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).
   */
Adrian Grange's avatar
Adrian Grange committed
366
  struct aom_rational g_timebase;
John Koleszar's avatar
John Koleszar committed
367

clang-format's avatar
clang-format committed
368
  /*!\brief Enable error resilient modes.
John Koleszar's avatar
John Koleszar committed
369
   *
clang-format's avatar
clang-format committed
370
371
372
   * 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
373
   */
Adrian Grange's avatar
Adrian Grange committed
374
  aom_codec_er_flags_t g_error_resilient;
John Koleszar's avatar
John Koleszar committed
375

clang-format's avatar
clang-format committed
376
377
378
  /*!\brief Multi-pass Encoding Mode
   *
   * This value should be set to the current phase for multi-pass encoding.
Adrian Grange's avatar
Adrian Grange committed
379
   * For single pass, set to #AOM_RC_ONE_PASS.
clang-format's avatar
clang-format committed
380
   */
Adrian Grange's avatar
Adrian Grange committed
381
  enum aom_enc_pass g_pass;
John Koleszar's avatar
John Koleszar committed
382

clang-format's avatar
clang-format committed
383
384
385
386
387
388
389
390
391
392
393
394
395
  /*!\brief Allow lagged encoding
   *
   * 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).
   *
   * 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.
   */
  unsigned int g_lag_in_frames;
John Koleszar's avatar
John Koleszar committed
396

clang-format's avatar
clang-format committed
397
398
399
  /*
   * rate control settings (rc)
   */
John Koleszar's avatar
John Koleszar committed
400

clang-format's avatar
clang-format committed
401
402
403
404
405
406
407
408
  /*!\brief Temporal resampling configuration, if supported by the codec.
   *
   * 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.
   *
Adrian Grange's avatar
Adrian Grange committed
409
   * Note that not all codecs support this feature. All aom VPx codecs do.
clang-format's avatar
clang-format committed
410
411
412
413
414
415
416
417
   * For other codecs, consult the documentation for that algorithm.
   *
   * 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.
   */
  unsigned int rc_dropframe_thresh;
John Koleszar's avatar
John Koleszar committed
418

clang-format's avatar
clang-format committed
419
420
421
422
423
424
425
426
  /*!\brief Enable/disable spatial resampling, if supported by the codec.
   *
   * 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;
John Koleszar's avatar
John Koleszar committed
427

clang-format's avatar
clang-format committed
428
429
430
431
432
433
  /*!\brief Internal coded frame width.
   *
   * If spatial resampling is enabled this specifies the width of the
   * encoded frame.
   */
  unsigned int rc_scaled_width;
John Koleszar's avatar
John Koleszar committed
434

clang-format's avatar
clang-format committed
435
436
437
438
439
440
  /*!\brief Internal coded frame height.
   *
   * If spatial resampling is enabled this specifies the height of the
   * encoded frame.
   */
  unsigned int rc_scaled_height;
John Koleszar's avatar
John Koleszar committed
441

clang-format's avatar
clang-format committed
442
443
444
445
446
447
448
  /*!\brief Spatial resampling up watermark.
   *
   * 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.
   */
  unsigned int rc_resize_up_thresh;
John Koleszar's avatar
John Koleszar committed
449

clang-format's avatar
clang-format committed
450
451
452
453
454
455
456
  /*!\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.
   */
  unsigned int rc_resize_down_thresh;
John Koleszar's avatar
John Koleszar committed
457

clang-format's avatar
clang-format committed
458
459
460
461
462
463
464
465
  /*!\brief Rate control algorithm to use.
   *
   * 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.
   */
Adrian Grange's avatar
Adrian Grange committed
466
  enum aom_rc_mode rc_end_usage;
467

clang-format's avatar
clang-format committed
468
469
470
471
472
  /*!\brief Two-pass stats buffer.
   *
   * A buffer containing all of the stats packets produced in the first
   * pass, concatenated.
   */
Adrian Grange's avatar
Adrian Grange committed
473
  aom_fixed_buf_t rc_twopass_stats_in;
John Koleszar's avatar
John Koleszar committed
474

clang-format's avatar
clang-format committed
475
476
477
478
479
  /*!\brief first pass mb stats buffer.
   *
   * A buffer containing all of the first pass mb stats packets produced
   * in the first pass, concatenated.
   */
Adrian Grange's avatar
Adrian Grange committed
480
  aom_fixed_buf_t rc_firstpass_mb_stats_in;
481

clang-format's avatar
clang-format committed
482
483
484
485
486
  /*!\brief Target data rate
   *
   * Target bandwidth to use for this stream, in kilobits per second.
   */
  unsigned int rc_target_bitrate;
487

clang-format's avatar
clang-format committed
488
489
490
  /*
   * quantizer settings
   */
John Koleszar's avatar
John Koleszar committed
491

clang-format's avatar
clang-format committed
492
493
494
495
496
497
  /*!\brief Minimum (Best Quality) Quantizer
   *
   * 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
Adrian Grange's avatar
Adrian Grange committed
498
   * aom_codec_enc_config_default() with a usage value of 0.
clang-format's avatar
clang-format committed
499
500
   */
  unsigned int rc_min_quantizer;
John Koleszar's avatar
John Koleszar committed
501

clang-format's avatar
clang-format committed
502
503
504
505
506
507
  /*!\brief Maximum (Worst Quality) Quantizer
   *
   * 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
Adrian Grange's avatar
Adrian Grange committed
508
   * aom_codec_enc_config_default() with a usage value of 0.
clang-format's avatar
clang-format committed
509
510
   */
  unsigned int rc_max_quantizer;
511

clang-format's avatar
clang-format committed
512
513
514
  /*
   * bitrate tolerance
   */
John Koleszar's avatar
John Koleszar committed
515

clang-format's avatar
clang-format committed
516
  /*!\brief Rate control adaptation undershoot control
517
   *
clang-format's avatar
clang-format committed
518
519
520
521
522
   * 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.
523
   *
clang-format's avatar
clang-format committed
524
   * Valid values in the range 0-1000.
525
   */
clang-format's avatar
clang-format committed
526
  unsigned int rc_undershoot_pct;
John Koleszar's avatar
John Koleszar committed
527

clang-format's avatar
clang-format committed
528
  /*!\brief Rate control adaptation overshoot control
John Koleszar's avatar
John Koleszar committed
529
   *
clang-format's avatar
clang-format committed
530
531
532
533
534
   * 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
535
   *
clang-format's avatar
clang-format committed
536
   * Valid values in the range 0-1000.
John Koleszar's avatar
John Koleszar committed
537
   */
clang-format's avatar
clang-format committed
538
  unsigned int rc_overshoot_pct;
John Koleszar's avatar
John Koleszar committed
539

clang-format's avatar
clang-format committed
540
541
542
  /*
   * decoder buffer model parameters
   */
John Koleszar's avatar
John Koleszar committed
543

clang-format's avatar
clang-format committed
544
  /*!\brief Decoder Buffer Size
John Koleszar's avatar
John Koleszar committed
545
   *
clang-format's avatar
clang-format committed
546
547
548
549
550
551
   * 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.
John Koleszar's avatar
John Koleszar committed
552
   */
clang-format's avatar
clang-format committed
553
  unsigned int rc_buf_sz;
John Koleszar's avatar
John Koleszar committed
554

clang-format's avatar
clang-format committed
555
  /*!\brief Decoder Buffer Initial Size
John Koleszar's avatar
John Koleszar committed
556
   *
clang-format's avatar
clang-format committed
557
558
559
560
   * 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.
John Koleszar's avatar
John Koleszar committed
561
   */
clang-format's avatar
clang-format committed
562
  unsigned int rc_buf_initial_sz;
John Koleszar's avatar
John Koleszar committed
563

clang-format's avatar
clang-format committed
564
  /*!\brief Decoder Buffer Optimal Size
John Koleszar's avatar
John Koleszar committed
565
   *
clang-format's avatar
clang-format committed
566
567
568
569
   * 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.
John Koleszar's avatar
John Koleszar committed
570
   */
clang-format's avatar
clang-format committed
571
  unsigned int rc_buf_optimal_sz;
John Koleszar's avatar
John Koleszar committed
572

clang-format's avatar
clang-format committed
573
574
575
  /*
   * 2 pass rate control parameters
   */
John Koleszar's avatar
John Koleszar committed
576

clang-format's avatar
clang-format committed
577
  /*!\brief Two-pass mode CBR/VBR bias
John Koleszar's avatar
John Koleszar committed
578
   *
clang-format's avatar
clang-format committed
579
580
581
582
583
   * 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."
John Koleszar's avatar
John Koleszar committed
584
   */
clang-format's avatar
clang-format committed
585
586
  unsigned int rc_2pass_vbr_bias_pct; /**< RC mode bias between CBR and
                                         VBR(0-100: 0->CBR, 100->VBR)   */
John Koleszar's avatar
John Koleszar committed
587

clang-format's avatar
clang-format committed
588
  /*!\brief Two-pass mode per-GOP minimum bitrate
John Koleszar's avatar
John Koleszar committed
589
   *
clang-format's avatar
clang-format committed
590
591
592
593
594
595
   * This value, expressed as a percentage of the target bitrate, indicates
   * the minimum bitrate to be used for a single GOP (aka "section")
   */
  unsigned int rc_2pass_vbr_minsection_pct;

  /*!\brief Two-pass mode per-GOP maximum bitrate
John Koleszar's avatar
John Koleszar committed
596
   *
clang-format's avatar
clang-format committed
597
598
   * This value, expressed as a percentage of the target bitrate, indicates
   * the maximum bitrate to be used for a single GOP (aka "section")
John Koleszar's avatar
John Koleszar committed
599
   */
clang-format's avatar
clang-format committed
600
  unsigned int rc_2pass_vbr_maxsection_pct;
John Koleszar's avatar
John Koleszar committed
601

clang-format's avatar
clang-format committed
602
603
604
  /*
   * keyframing settings (kf)
   */
John Koleszar's avatar
John Koleszar committed
605

clang-format's avatar
clang-format committed
606
  /*!\brief Keyframe placement mode
John Koleszar's avatar
John Koleszar committed
607
   *
clang-format's avatar
clang-format committed
608
609
610
611
   * 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)
   */
Adrian Grange's avatar
Adrian Grange committed
612
  enum aom_kf_mode kf_mode;
clang-format's avatar
clang-format committed
613
614

  /*!\brief Keyframe minimum interval
John Koleszar's avatar
John Koleszar committed
615
   *
clang-format's avatar
clang-format committed
616
617
618
619
620
621
622
623
   * 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
624
   *
clang-format's avatar
clang-format committed
625
626
627
628
   * 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
629
   */
clang-format's avatar
clang-format committed
630
  unsigned int kf_max_dist;
John Koleszar's avatar
John Koleszar committed
631

clang-format's avatar
clang-format committed
632
633
634
  /*
   * Spatial scalability settings (ss)
   */
John Koleszar's avatar
John Koleszar committed
635

clang-format's avatar
clang-format committed
636
  /*!\brief Number of spatial coding layers.
John Koleszar's avatar
John Koleszar committed
637
   *
clang-format's avatar
clang-format committed
638
   * This value specifies the number of spatial coding layers to be used.
John Koleszar's avatar
John Koleszar committed
639
   */
clang-format's avatar
clang-format committed
640
641
642
  unsigned int ss_number_layers;

  /*!\brief Enable auto alt reference flags for each spatial layer.
John Koleszar's avatar
John Koleszar committed
643
   *
clang-format's avatar
clang-format committed
644
645
   * These values specify if auto alt reference frame is enabled for each
   * spatial layer.
John Koleszar's avatar
John Koleszar committed
646
   */
Adrian Grange's avatar
Adrian Grange committed
647
  int ss_enable_auto_alt_ref[AOM_SS_MAX_LAYERS];
John Koleszar's avatar
John Koleszar committed
648

clang-format's avatar
clang-format committed
649
650
651
652
653
  /*!\brief Target bitrate for each spatial layer.
   *
   * These values specify the target coding bitrate to be used for each
   * spatial layer.
   */
Adrian Grange's avatar
Adrian Grange committed
654
  unsigned int ss_target_bitrate[AOM_SS_MAX_LAYERS];
John Koleszar's avatar
John Koleszar committed
655

clang-format's avatar
clang-format committed
656
  /*!\brief Number of temporal coding layers.
John Koleszar's avatar
John Koleszar committed
657
   *
clang-format's avatar
clang-format committed
658
659
660
661
662
   * This value specifies the number of temporal layers to be used.
   */
  unsigned int ts_number_layers;

  /*!\brief Target bitrate for each temporal layer.
John Koleszar's avatar
John Koleszar committed
663
   *
clang-format's avatar
clang-format committed
664
665
666
   * These values specify the target coding bitrate to be used for each
   * temporal layer.
   */
Adrian Grange's avatar
Adrian Grange committed
667
  unsigned int ts_target_bitrate[AOM_TS_MAX_LAYERS];
clang-format's avatar
clang-format committed
668
669

  /*!\brief Frame rate decimation factor for each temporal layer.
John Koleszar's avatar
John Koleszar committed
670
   *
clang-format's avatar
clang-format committed
671
672
673
   * These values specify the frame rate decimation factors to apply
   * to each temporal layer.
   */
Adrian Grange's avatar
Adrian Grange committed
674
  unsigned int ts_rate_decimator[AOM_TS_MAX_LAYERS];
clang-format's avatar
clang-format committed
675
676

  /*!\brief Length of the sequence defining frame temporal layer membership.
John Koleszar's avatar
John Koleszar committed
677
   *
clang-format's avatar
clang-format committed
678
679
680
681
682
683
684
685
   * This value specifies the length of the sequence that defines the
   * membership of frames to temporal layers. For example, if the
   * ts_periodicity = 8, then the frames are assigned to coding layers with a
   * repeated sequence of length 8.
  */
  unsigned int ts_periodicity;

  /*!\brief Template defining the membership of frames to temporal layers.
John Koleszar's avatar
John Koleszar committed
686
   *
clang-format's avatar
clang-format committed
687
688
689
690
691
   * 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).
  */
Adrian Grange's avatar
Adrian Grange committed
692
  unsigned int ts_layer_id[AOM_TS_MAX_PERIODICITY];
clang-format's avatar
clang-format committed
693
694

  /*!\brief Target bitrate for each spatial/temporal layer.
John Koleszar's avatar
John Koleszar committed
695
   *
clang-format's avatar
clang-format committed
696
697
   * These values specify the target coding bitrate to be used for each
   * spatial/temporal layer.
John Koleszar's avatar
John Koleszar committed
698
699
   *
   */
Adrian Grange's avatar
Adrian Grange committed
700
  unsigned int layer_target_bitrate[AOM_MAX_LAYERS];
John Koleszar's avatar
John Koleszar committed
701

clang-format's avatar
clang-format committed
702
703
  /*!\brief Temporal layering mode indicating which temporal layering scheme to
   * use.
John Koleszar's avatar
John Koleszar committed
704
   *
Adrian Grange's avatar
Adrian Grange committed
705
   * The value (refer to AV1E_TEMPORAL_LAYERING_MODE) specifies the
clang-format's avatar
clang-format committed
706
   * temporal layering mode to use.
John Koleszar's avatar
John Koleszar committed
707
708
   *
   */
clang-format's avatar
clang-format committed
709
  int temporal_layering_mode;
Adrian Grange's avatar
Adrian Grange committed
710
} aom_codec_enc_cfg_t; /**< alias for struct aom_codec_enc_cfg */
John Koleszar's avatar
John Koleszar committed
711

Adrian Grange's avatar
Adrian Grange committed
712
/*!\brief  av1 svc extra configure parameters
clang-format's avatar
clang-format committed
713
714
715
716
 *
 * This defines max/min quantizers and scale factors for each layer
 *
 */
Adrian Grange's avatar
Adrian Grange committed
717
typedef struct aom_svc_parameters {
Adrian Grange's avatar
Adrian Grange committed
718
719
720
721
  int max_quantizers[AOM_MAX_LAYERS];     /**< Max Q for each layer */
  int min_quantizers[AOM_MAX_LAYERS];     /**< Min Q for each layer */
  int scaling_factor_num[AOM_MAX_LAYERS]; /**< Scaling factor-numerator */
  int scaling_factor_den[AOM_MAX_LAYERS]; /**< Scaling factor-denominator */
clang-format's avatar
clang-format committed
722
  int temporal_layering_mode;             /**< Temporal layering mode */
Adrian Grange's avatar
Adrian Grange committed
723
} aom_svc_extra_cfg_t;
clang-format's avatar
clang-format committed
724
725
726
727

/*!\brief Initialize an encoder instance
 *
 * Initializes a encoder context using the given interface. Applications
Adrian Grange's avatar
Adrian Grange committed
728
 * should call the aom_codec_enc_init convenience macro instead of this
clang-format's avatar
clang-format committed
729
730
731
732
733
734
735
736
737
738
 * function directly, to ensure that the ABI version number parameter
 * is properly initialized.
 *
 * 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.
 *
 * \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.
Adrian Grange's avatar
Adrian Grange committed
739
 * \param[in]    flags   Bitfield of AOM_CODEC_USE_* flags
clang-format's avatar
clang-format committed
740
 * \param[in]    ver     ABI version number. Must be set to
Adrian Grange's avatar
Adrian Grange committed
741
742
 *                       AOM_ENCODER_ABI_VERSION
 * \retval #AOM_CODEC_OK
clang-format's avatar
clang-format committed
743
 *     The decoder algorithm initialized.
Adrian Grange's avatar
Adrian Grange committed
744
 * \retval #AOM_CODEC_MEM_ERROR
clang-format's avatar
clang-format committed
745
746
 *     Memory allocation failed.
 */
Adrian Grange's avatar
Adrian Grange committed
747
748
749
750
aom_codec_err_t aom_codec_enc_init_ver(aom_codec_ctx_t *ctx,
                                       aom_codec_iface_t *iface,
                                       const aom_codec_enc_cfg_t *cfg,
                                       aom_codec_flags_t flags, int ver);
John Koleszar's avatar
John Koleszar committed
751

Adrian Grange's avatar
Adrian Grange committed
752
/*!\brief Convenience macro for aom_codec_enc_init_ver()
clang-format's avatar
clang-format committed
753
754
755
 *
 * Ensures the ABI version parameter is properly set.
 */
Adrian Grange's avatar
Adrian Grange committed
756
#define aom_codec_enc_init(ctx, iface, cfg, flags) \
Adrian Grange's avatar
Adrian Grange committed
757
  aom_codec_enc_init_ver(ctx, iface, cfg, flags, AOM_ENCODER_ABI_VERSION)
clang-format's avatar
clang-format committed
758
759
760
761

/*!\brief Initialize multi-encoder instance
 *
 * Initializes multi-encoder context using the given interface.
Adrian Grange's avatar
Adrian Grange committed
762
 * Applications should call the aom_codec_enc_init_multi convenience macro
clang-format's avatar
clang-format committed
763
764
765
766
767
768
769
 * instead of this function directly, to ensure that the ABI version number
 * parameter is properly initialized.
 *
 * \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.
Adrian Grange's avatar
Adrian Grange committed
770
 * \param[in]    flags   Bitfield of AOM_CODEC_USE_* flags
clang-format's avatar
clang-format committed
771
772
 * \param[in]    dsf     Pointer to down-sampling factors.
 * \param[in]    ver     ABI version number. Must be set to
Adrian Grange's avatar
Adrian Grange committed
773
774
 *                       AOM_ENCODER_ABI_VERSION
 * \retval #AOM_CODEC_OK
clang-format's avatar
clang-format committed
775
 *     The decoder algorithm initialized.
Adrian Grange's avatar
Adrian Grange committed
776
 * \retval #AOM_CODEC_MEM_ERROR
clang-format's avatar
clang-format committed
777
778
 *     Memory allocation failed.
 */
Adrian Grange's avatar
Adrian Grange committed
779
780
781
aom_codec_err_t aom_codec_enc_init_multi_ver(
    aom_codec_ctx_t *ctx, aom_codec_iface_t *iface, aom_codec_enc_cfg_t *cfg,
    int num_enc, aom_codec_flags_t flags, aom_rational_t *dsf, int ver);
clang-format's avatar
clang-format committed
782

Adrian Grange's avatar
Adrian Grange committed
783
/*!\brief Convenience macro for aom_codec_enc_init_multi_ver()
clang-format's avatar
clang-format committed
784
785
786
 *
 * Ensures the ABI version parameter is properly set.
 */
Adrian Grange's avatar
Adrian Grange committed
787
788
#define aom_codec_enc_init_multi(ctx, iface, cfg, num_enc, flags, dsf) \
  aom_codec_enc_init_multi_ver(ctx, iface, cfg, num_enc, flags, dsf,   \
Adrian Grange's avatar
Adrian Grange committed
789
                               AOM_ENCODER_ABI_VERSION)
clang-format's avatar
clang-format committed
790
791
792
793
794
795
796
797
798
799
800

/*!\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.
Adrian Grange's avatar
Adrian Grange committed
801
 * \param[in]    reserved  Must set to 0 for AOM and AV1.
clang-format's avatar
clang-format committed
802
 *
Adrian Grange's avatar
Adrian Grange committed
803
 * \retval #AOM_CODEC_OK
clang-format's avatar
clang-format committed
804
 *     The configuration was populated.
Adrian Grange's avatar
Adrian Grange committed
805
 * \retval #AOM_CODEC_INCAPABLE
clang-format's avatar
clang-format committed
806
 *     Interface is not an encoder interface.
Adrian Grange's avatar
Adrian Grange committed
807
 * \retval #AOM_CODEC_INVALID_PARAM
clang-format's avatar
clang-format committed
808
809
 *     A parameter was NULL, or the usage value was not recognized.
 */
Adrian Grange's avatar
Adrian Grange committed
810
811
aom_codec_err_t aom_codec_enc_config_default(aom_codec_iface_t *iface,
                                             aom_codec_enc_cfg_t *cfg,
clang-format's avatar
clang-format committed
812
813
814
815
816
817
818
819
820
                                             unsigned int reserved);

/*!\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
 *
Adrian Grange's avatar
Adrian Grange committed
821
 * \retval #AOM_CODEC_OK
clang-format's avatar
clang-format committed
822
 *     The configuration was populated.
Adrian Grange's avatar
Adrian Grange committed
823
 * \retval #AOM_CODEC_INCAPABLE
clang-format's avatar
clang-format committed
824
 *     Interface is not an encoder interface.
Adrian Grange's avatar
Adrian Grange committed
825
 * \retval #AOM_CODEC_INVALID_PARAM
clang-format's avatar
clang-format committed
826
827
 *     A parameter was NULL, or the usage value was not recognized.
 */
Adrian Grange's avatar
Adrian Grange committed
828
829
aom_codec_err_t aom_codec_enc_config_set(aom_codec_ctx_t *ctx,
                                         const aom_codec_enc_cfg_t *cfg);
clang-format's avatar
clang-format committed
830
831
832
833
834
835
836
837
838
839
840
841

/*!\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
 */
Adrian Grange's avatar
Adrian Grange committed
842
aom_fixed_buf_t *aom_codec_get_global_headers(aom_codec_ctx_t *ctx);
clang-format's avatar
clang-format committed
843

Adrian Grange's avatar
Adrian Grange committed
844
#define AOM_DL_REALTIME (1)
clang-format's avatar
clang-format committed
845
/**< deadline parameter analogous to VPx REALTIME mode. */
Adrian Grange's avatar
Adrian Grange committed
846
#define AOM_DL_GOOD_QUALITY (1000000)
clang-format's avatar
clang-format committed
847
/**< deadline parameter analogous to VPx GOOD QUALITY mode. */
Adrian Grange's avatar
Adrian Grange committed
848
#define AOM_DL_BEST_QUALITY (0)
clang-format's avatar
clang-format committed
849
850
851
852
853
854
855
856
857
858
859
860
861
862
/**< 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
Adrian Grange's avatar
Adrian Grange committed
863
864
 * based system can use the symbols #AOM_DL_REALTIME, #AOM_DL_GOOD_QUALITY,
 * and #AOM_DL_BEST_QUALITY.
clang-format's avatar
clang-format committed
865
866
867
868
 *
 * 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
Adrian Grange's avatar
Adrian Grange committed
869
870
 * any held buffers. Encoding is complete when aom_codec_encode() is called
 * and aom_codec_get_cx_data() returns no data.
clang-format's avatar
clang-format committed
871
872
873
874
875
876
877
878
 *
 * \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)
 *
Adrian Grange's avatar
Adrian Grange committed
879
 * \retval #AOM_CODEC_OK
clang-format's avatar
clang-format committed
880
 *     The configuration was populated.
Adrian Grange's avatar
Adrian Grange committed
881
 * \retval #AOM_CODEC_INCAPABLE
clang-format's avatar
clang-format committed
882
 *     Interface is not an encoder interface.
Adrian Grange's avatar
Adrian Grange committed
883
 * \retval #AOM_CODEC_INVALID_PARAM
clang-format's avatar
clang-format committed
884
885
 *     A parameter was NULL, the image format is unsupported, etc.
 */
Adrian Grange's avatar
Adrian Grange committed
886
887
888
aom_codec_err_t aom_codec_encode(aom_codec_ctx_t *ctx, const aom_image_t *img,
                                 aom_codec_pts_t pts, unsigned long duration,
                                 aom_enc_frame_flags_t flags,
clang-format's avatar
clang-format committed
889
890
891
892
893
894
                                 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
Adrian Grange's avatar
Adrian Grange committed
895
 * next AOM_CODEC_CX_FRAME_PKT packet. Subsequent packets will be
clang-format's avatar
clang-format committed
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
 * 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
Adrian Grange's avatar
Adrian Grange committed
922
 * aom_codec_get_cx_data().
clang-format's avatar
clang-format committed
923
924
925
926
927
928
 *
 * \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
 *
Adrian Grange's avatar
Adrian Grange committed
929
 * \retval #AOM_CODEC_OK
clang-format's avatar
clang-format committed
930
 *     The buffer was set successfully.
Adrian Grange's avatar
Adrian Grange committed
931
 * \retval #AOM_CODEC_INVALID_PARAM
clang-format's avatar
clang-format committed
932
933
 *     A parameter was NULL, the image format is unsupported, etc.
 */
Adrian Grange's avatar
Adrian Grange committed
934
935
aom_codec_err_t aom_codec_set_cx_data_buf(aom_codec_ctx_t *ctx,
                                          const aom_fixed_buf_t *buf,
clang-format's avatar
clang-format committed
936
937
938
939
940
941
942
                                          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
Adrian Grange's avatar
Adrian Grange committed
943
 * #aom_codec_cx_pkt_kind.
clang-format's avatar
clang-format committed
944
 *
Adrian Grange's avatar
Adrian Grange committed
945
 * #AOM_CODEC_CX_FRAME_PKT packets should be passed to the application's
clang-format's avatar
clang-format committed
946
 * muxer. Multiple compressed frames may be in the list.
Adrian Grange's avatar
Adrian Grange committed
947
 * #AOM_CODEC_STATS_PKT packets should be appended to a global buffer.
clang-format's avatar
clang-format committed
948
949
950
951
952
 *
 * 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
Adrian Grange's avatar
Adrian Grange committed
953
 * valid until the application makes another call to any aom_codec_* function.
clang-format's avatar
clang-format committed
954
955
956
957
958
959
960
961
 *
 * \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.
 *
 */
Adrian Grange's avatar
Adrian Grange committed
962
963
const aom_codec_cx_pkt_t *aom_codec_get_cx_data(aom_codec_ctx_t *ctx,
                                                aom_codec_iter_t *iter);
clang-format's avatar
clang-format committed
964
965
966
967
968
969
970
971
972
973
974
975
976

/*!\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.
 *
 */
Adrian Grange's avatar
Adrian Grange committed
977
const aom_image_t *aom_codec_get_preview_frame(aom_codec_ctx_t *ctx);
clang-format's avatar
clang-format committed
978
979

/*!@} - end defgroup encoder*/
John Koleszar's avatar
John Koleszar committed
980
981
982
#ifdef __cplusplus
}
#endif
Adrian Grange's avatar
Adrian Grange committed
983
#endif  // AOM_AOM_ENCODER_H_