|
1 /* |
|
2 * Copyright (c) 2010 The WebM project authors. All Rights Reserved. |
|
3 * |
|
4 * Use of this source code is governed by a BSD-style license |
|
5 * that can be found in the LICENSE file in the root of the source |
|
6 * 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. |
|
9 */ |
|
10 #ifndef VPX_ENCODER_H |
|
11 #define VPX_ENCODER_H |
|
12 |
|
13 /*!\defgroup encoder Encoder Algorithm Interface |
|
14 * \ingroup codec |
|
15 * This abstraction allows applications using this encoder to easily support |
|
16 * multiple video formats with minimal code duplication. This section describes |
|
17 * the interface common to all encoders. |
|
18 * @{ |
|
19 */ |
|
20 |
|
21 /*!\file |
|
22 * \brief Describes the encoder algorithm interface to applications. |
|
23 * |
|
24 * This file describes the interface between an application and a |
|
25 * video encoder algorithm. |
|
26 * |
|
27 */ |
|
28 #ifdef __cplusplus |
|
29 extern "C" { |
|
30 #endif |
|
31 |
|
32 #include "vpx_codec.h" |
|
33 |
|
34 /*! Temporal Scalability: Maximum length of the sequence defining frame |
|
35 * layer membership |
|
36 */ |
|
37 #define VPX_TS_MAX_PERIODICITY 16 |
|
38 |
|
39 /*! Temporal Scalability: Maximum number of coding layers */ |
|
40 #define VPX_TS_MAX_LAYERS 5 |
|
41 |
|
42 /*!\deprecated Use #VPX_TS_MAX_PERIODICITY instead. */ |
|
43 #define MAX_PERIODICITY VPX_TS_MAX_PERIODICITY |
|
44 |
|
45 /*!\deprecated Use #VPX_TS_MAX_LAYERS instead. */ |
|
46 #define MAX_LAYERS VPX_TS_MAX_LAYERS |
|
47 |
|
48 /*! Spatial Scalability: Maximum number of coding layers */ |
|
49 #define VPX_SS_MAX_LAYERS 5 |
|
50 |
|
51 /*! Spatial Scalability: Default number of coding layers */ |
|
52 #define VPX_SS_DEFAULT_LAYERS 3 |
|
53 |
|
54 /*!\brief Current ABI version number |
|
55 * |
|
56 * \internal |
|
57 * If this file is altered in any way that changes the ABI, this value |
|
58 * must be bumped. Examples include, but are not limited to, changing |
|
59 * types, removing or reassigning enums, adding/removing/rearranging |
|
60 * fields to structures |
|
61 */ |
|
62 #define VPX_ENCODER_ABI_VERSION (3 + VPX_CODEC_ABI_VERSION) /**<\hideinitializer*/ |
|
63 |
|
64 |
|
65 /*! \brief Encoder capabilities bitfield |
|
66 * |
|
67 * Each encoder advertises the capabilities it supports as part of its |
|
68 * ::vpx_codec_iface_t interface structure. Capabilities are extra |
|
69 * interfaces or functionality, and are not required to be supported |
|
70 * by an encoder. |
|
71 * |
|
72 * The available flags are specified by VPX_CODEC_CAP_* defines. |
|
73 */ |
|
74 #define VPX_CODEC_CAP_PSNR 0x10000 /**< Can issue PSNR packets */ |
|
75 |
|
76 /*! Can output one partition at a time. Each partition is returned in its |
|
77 * own VPX_CODEC_CX_FRAME_PKT, with the FRAME_IS_FRAGMENT flag set for |
|
78 * every partition but the last. In this mode all frames are always |
|
79 * returned partition by partition. |
|
80 */ |
|
81 #define VPX_CODEC_CAP_OUTPUT_PARTITION 0x20000 |
|
82 |
|
83 |
|
84 /*! \brief Initialization-time Feature Enabling |
|
85 * |
|
86 * Certain codec features must be known at initialization time, to allow |
|
87 * for proper memory allocation. |
|
88 * |
|
89 * The available flags are specified by VPX_CODEC_USE_* defines. |
|
90 */ |
|
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 |
|
93 partition at a time. */ |
|
94 |
|
95 |
|
96 /*!\brief Generic fixed size buffer structure |
|
97 * |
|
98 * This structure is able to hold a reference to any fixed size buffer. |
|
99 */ |
|
100 typedef struct vpx_fixed_buf { |
|
101 void *buf; /**< Pointer to the data */ |
|
102 size_t sz; /**< Length of the buffer, in chars */ |
|
103 } vpx_fixed_buf_t; /**< alias for struct vpx_fixed_buf */ |
|
104 |
|
105 |
|
106 /*!\brief Time Stamp Type |
|
107 * |
|
108 * An integer, which when multiplied by the stream's time base, provides |
|
109 * the absolute time of a sample. |
|
110 */ |
|
111 typedef int64_t vpx_codec_pts_t; |
|
112 |
|
113 |
|
114 /*!\brief Compressed Frame Flags |
|
115 * |
|
116 * This type represents a bitfield containing information about a compressed |
|
117 * frame that may be useful to an application. The most significant 16 bits |
|
118 * can be used by an algorithm to provide additional detail, for example to |
|
119 * support frame types that are codec specific (MPEG-1 D-frames for example) |
|
120 */ |
|
121 typedef uint32_t vpx_codec_frame_flags_t; |
|
122 #define VPX_FRAME_IS_KEY 0x1 /**< frame is the start of a GOP */ |
|
123 #define VPX_FRAME_IS_DROPPABLE 0x2 /**< frame can be dropped without affecting |
|
124 the stream (no future frame depends on |
|
125 this one) */ |
|
126 #define VPX_FRAME_IS_INVISIBLE 0x4 /**< frame should be decoded but will not |
|
127 be shown */ |
|
128 #define VPX_FRAME_IS_FRAGMENT 0x8 /**< this is a fragment of the encoded |
|
129 frame */ |
|
130 |
|
131 /*!\brief Error Resilient flags |
|
132 * |
|
133 * These flags define which error resilient features to enable in the |
|
134 * encoder. The flags are specified through the |
|
135 * vpx_codec_enc_cfg::g_error_resilient variable. |
|
136 */ |
|
137 typedef uint32_t vpx_codec_er_flags_t; |
|
138 #define VPX_ERROR_RESILIENT_DEFAULT 0x1 /**< Improve resiliency against |
|
139 losses of whole frames */ |
|
140 #define VPX_ERROR_RESILIENT_PARTITIONS 0x2 /**< The frame partitions are |
|
141 independently decodable by the |
|
142 bool decoder, meaning that |
|
143 partitions can be decoded even |
|
144 though earlier partitions have |
|
145 been lost. Note that intra |
|
146 predicition is still done over |
|
147 the partition boundary. */ |
|
148 |
|
149 /*!\brief Encoder output packet variants |
|
150 * |
|
151 * This enumeration lists the different kinds of data packets that can be |
|
152 * returned by calls to vpx_codec_get_cx_data(). Algorithms \ref MAY |
|
153 * extend this list to provide additional functionality. |
|
154 */ |
|
155 enum vpx_codec_cx_pkt_kind { |
|
156 VPX_CODEC_CX_FRAME_PKT, /**< Compressed video frame */ |
|
157 VPX_CODEC_STATS_PKT, /**< Two-pass statistics for this frame */ |
|
158 VPX_CODEC_PSNR_PKT, /**< PSNR statistics for this frame */ |
|
159 VPX_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions */ |
|
160 }; |
|
161 |
|
162 |
|
163 /*!\brief Encoder output packet |
|
164 * |
|
165 * This structure contains the different kinds of output data the encoder |
|
166 * may produce while compressing a frame. |
|
167 */ |
|
168 typedef struct vpx_codec_cx_pkt { |
|
169 enum vpx_codec_cx_pkt_kind kind; /**< packet variant */ |
|
170 union { |
|
171 struct { |
|
172 void *buf; /**< compressed data buffer */ |
|
173 size_t sz; /**< length of compressed data */ |
|
174 vpx_codec_pts_t pts; /**< time stamp to show frame |
|
175 (in timebase units) */ |
|
176 unsigned long duration; /**< duration to show frame |
|
177 (in timebase units) */ |
|
178 vpx_codec_frame_flags_t flags; /**< flags for this frame */ |
|
179 int partition_id; /**< the partition id |
|
180 defines the decoding order |
|
181 of the partitions. Only |
|
182 applicable when "output partition" |
|
183 mode is enabled. First partition |
|
184 has id 0.*/ |
|
185 |
|
186 } frame; /**< data for compressed frame packet */ |
|
187 struct vpx_fixed_buf twopass_stats; /**< data for two-pass packet */ |
|
188 struct vpx_psnr_pkt { |
|
189 unsigned int samples[4]; /**< Number of samples, total/y/u/v */ |
|
190 uint64_t sse[4]; /**< sum squared error, total/y/u/v */ |
|
191 double psnr[4]; /**< PSNR, total/y/u/v */ |
|
192 } psnr; /**< data for PSNR packet */ |
|
193 struct vpx_fixed_buf raw; /**< data for arbitrary packets */ |
|
194 |
|
195 /* This packet size is fixed to allow codecs to extend this |
|
196 * interface without having to manage storage for raw packets, |
|
197 * i.e., if it's smaller than 128 bytes, you can store in the |
|
198 * packet list directly. |
|
199 */ |
|
200 char pad[128 - sizeof(enum vpx_codec_cx_pkt_kind)]; /**< fixed sz */ |
|
201 } data; /**< packet data */ |
|
202 } vpx_codec_cx_pkt_t; /**< alias for struct vpx_codec_cx_pkt */ |
|
203 |
|
204 |
|
205 /*!\brief Rational Number |
|
206 * |
|
207 * This structure holds a fractional value. |
|
208 */ |
|
209 typedef struct vpx_rational { |
|
210 int num; /**< fraction numerator */ |
|
211 int den; /**< fraction denominator */ |
|
212 } vpx_rational_t; /**< alias for struct vpx_rational */ |
|
213 |
|
214 |
|
215 /*!\brief Multi-pass Encoding Pass */ |
|
216 enum vpx_enc_pass { |
|
217 VPX_RC_ONE_PASS, /**< Single pass mode */ |
|
218 VPX_RC_FIRST_PASS, /**< First pass of multi-pass mode */ |
|
219 VPX_RC_LAST_PASS /**< Final pass of multi-pass mode */ |
|
220 }; |
|
221 |
|
222 |
|
223 /*!\brief Rate control mode */ |
|
224 enum vpx_rc_mode { |
|
225 VPX_VBR, /**< Variable Bit Rate (VBR) mode */ |
|
226 VPX_CBR, /**< Constant Bit Rate (CBR) mode */ |
|
227 VPX_CQ, /**< Constrained Quality (CQ) mode */ |
|
228 VPX_Q, /**< Constant Quality (Q) mode */ |
|
229 }; |
|
230 |
|
231 |
|
232 /*!\brief Keyframe placement mode. |
|
233 * |
|
234 * This enumeration determines whether keyframes are placed automatically by |
|
235 * the encoder or whether this behavior is disabled. Older releases of this |
|
236 * SDK were implemented such that VPX_KF_FIXED meant keyframes were disabled. |
|
237 * This name is confusing for this behavior, so the new symbols to be used |
|
238 * are VPX_KF_AUTO and VPX_KF_DISABLED. |
|
239 */ |
|
240 enum vpx_kf_mode { |
|
241 VPX_KF_FIXED, /**< deprecated, implies VPX_KF_DISABLED */ |
|
242 VPX_KF_AUTO, /**< Encoder determines optimal placement automatically */ |
|
243 VPX_KF_DISABLED = 0 /**< Encoder does not place keyframes. */ |
|
244 }; |
|
245 |
|
246 |
|
247 /*!\brief Encoded Frame Flags |
|
248 * |
|
249 * This type indicates a bitfield to be passed to vpx_codec_encode(), defining |
|
250 * per-frame boolean values. By convention, bits common to all codecs will be |
|
251 * named VPX_EFLAG_*, and bits specific to an algorithm will be named |
|
252 * /algo/_eflag_*. The lower order 16 bits are reserved for common use. |
|
253 */ |
|
254 typedef long vpx_enc_frame_flags_t; |
|
255 #define VPX_EFLAG_FORCE_KF (1<<0) /**< Force this frame to be a keyframe */ |
|
256 |
|
257 |
|
258 /*!\brief Encoder configuration structure |
|
259 * |
|
260 * This structure contains the encoder settings that have common representations |
|
261 * across all codecs. This doesn't imply that all codecs support all features, |
|
262 * however. |
|
263 */ |
|
264 typedef struct vpx_codec_enc_cfg { |
|
265 /* |
|
266 * generic settings (g) |
|
267 */ |
|
268 |
|
269 /*!\brief Algorithm specific "usage" value |
|
270 * |
|
271 * Algorithms may define multiple values for usage, which may convey the |
|
272 * intent of how the application intends to use the stream. If this value |
|
273 * is non-zero, consult the documentation for the codec to determine its |
|
274 * meaning. |
|
275 */ |
|
276 unsigned int g_usage; |
|
277 |
|
278 |
|
279 /*!\brief Maximum number of threads to use |
|
280 * |
|
281 * For multi-threaded implementations, use no more than this number of |
|
282 * threads. The codec may use fewer threads than allowed. The value |
|
283 * 0 is equivalent to the value 1. |
|
284 */ |
|
285 unsigned int g_threads; |
|
286 |
|
287 |
|
288 /*!\brief Bitstream profile to use |
|
289 * |
|
290 * Some codecs support a notion of multiple bitstream profiles. Typically |
|
291 * this maps to a set of features that are turned on or off. Often the |
|
292 * profile to use is determined by the features of the intended decoder. |
|
293 * Consult the documentation for the codec to determine the valid values |
|
294 * for this parameter, or set to zero for a sane default. |
|
295 */ |
|
296 unsigned int g_profile; /**< profile of bitstream to use */ |
|
297 |
|
298 |
|
299 |
|
300 /*!\brief Width of the frame |
|
301 * |
|
302 * This value identifies the presentation resolution of the frame, |
|
303 * in pixels. Note that the frames passed as input to the encoder must |
|
304 * have this resolution. Frames will be presented by the decoder in this |
|
305 * resolution, independent of any spatial resampling the encoder may do. |
|
306 */ |
|
307 unsigned int g_w; |
|
308 |
|
309 |
|
310 /*!\brief Height of the frame |
|
311 * |
|
312 * This value identifies the presentation resolution of the frame, |
|
313 * in pixels. Note that the frames passed as input to the encoder must |
|
314 * have this resolution. Frames will be presented by the decoder in this |
|
315 * resolution, independent of any spatial resampling the encoder may do. |
|
316 */ |
|
317 unsigned int g_h; |
|
318 |
|
319 |
|
320 /*!\brief Stream timebase units |
|
321 * |
|
322 * Indicates the smallest interval of time, in seconds, used by the stream. |
|
323 * For fixed frame rate material, or variable frame rate material where |
|
324 * frames are timed at a multiple of a given clock (ex: video capture), |
|
325 * the \ref RECOMMENDED method is to set the timebase to the reciprocal |
|
326 * of the frame rate (ex: 1001/30000 for 29.970 Hz NTSC). This allows the |
|
327 * pts to correspond to the frame number, which can be handy. For |
|
328 * re-encoding video from containers with absolute time timestamps, the |
|
329 * \ref RECOMMENDED method is to set the timebase to that of the parent |
|
330 * container or multimedia framework (ex: 1/1000 for ms, as in FLV). |
|
331 */ |
|
332 struct vpx_rational g_timebase; |
|
333 |
|
334 |
|
335 /*!\brief Enable error resilient modes. |
|
336 * |
|
337 * The error resilient bitfield indicates to the encoder which features |
|
338 * it should enable to take measures for streaming over lossy or noisy |
|
339 * links. |
|
340 */ |
|
341 vpx_codec_er_flags_t g_error_resilient; |
|
342 |
|
343 |
|
344 /*!\brief Multi-pass Encoding Mode |
|
345 * |
|
346 * This value should be set to the current phase for multi-pass encoding. |
|
347 * For single pass, set to #VPX_RC_ONE_PASS. |
|
348 */ |
|
349 enum vpx_enc_pass g_pass; |
|
350 |
|
351 |
|
352 /*!\brief Allow lagged encoding |
|
353 * |
|
354 * If set, this value allows the encoder to consume a number of input |
|
355 * frames before producing output frames. This allows the encoder to |
|
356 * base decisions for the current frame on future frames. This does |
|
357 * increase the latency of the encoding pipeline, so it is not appropriate |
|
358 * in all situations (ex: realtime encoding). |
|
359 * |
|
360 * Note that this is a maximum value -- the encoder may produce frames |
|
361 * sooner than the given limit. Set this value to 0 to disable this |
|
362 * feature. |
|
363 */ |
|
364 unsigned int g_lag_in_frames; |
|
365 |
|
366 |
|
367 /* |
|
368 * rate control settings (rc) |
|
369 */ |
|
370 |
|
371 /*!\brief Temporal resampling configuration, if supported by the codec. |
|
372 * |
|
373 * Temporal resampling allows the codec to "drop" frames as a strategy to |
|
374 * meet its target data rate. This can cause temporal discontinuities in |
|
375 * the encoded video, which may appear as stuttering during playback. This |
|
376 * trade-off is often acceptable, but for many applications is not. It can |
|
377 * be disabled in these cases. |
|
378 * |
|
379 * Note that not all codecs support this feature. All vpx VPx codecs do. |
|
380 * For other codecs, consult the documentation for that algorithm. |
|
381 * |
|
382 * This threshold is described as a percentage of the target data buffer. |
|
383 * When the data buffer falls below this percentage of fullness, a |
|
384 * dropped frame is indicated. Set the threshold to zero (0) to disable |
|
385 * this feature. |
|
386 */ |
|
387 unsigned int rc_dropframe_thresh; |
|
388 |
|
389 |
|
390 /*!\brief Enable/disable spatial resampling, if supported by the codec. |
|
391 * |
|
392 * Spatial resampling allows the codec to compress a lower resolution |
|
393 * version of the frame, which is then upscaled by the encoder to the |
|
394 * correct presentation resolution. This increases visual quality at |
|
395 * low data rates, at the expense of CPU time on the encoder/decoder. |
|
396 */ |
|
397 unsigned int rc_resize_allowed; |
|
398 |
|
399 |
|
400 /*!\brief Spatial resampling up watermark. |
|
401 * |
|
402 * This threshold is described as a percentage of the target data buffer. |
|
403 * When the data buffer rises above this percentage of fullness, the |
|
404 * encoder will step up to a higher resolution version of the frame. |
|
405 */ |
|
406 unsigned int rc_resize_up_thresh; |
|
407 |
|
408 |
|
409 /*!\brief Spatial resampling down watermark. |
|
410 * |
|
411 * This threshold is described as a percentage of the target data buffer. |
|
412 * When the data buffer falls below this percentage of fullness, the |
|
413 * encoder will step down to a lower resolution version of the frame. |
|
414 */ |
|
415 unsigned int rc_resize_down_thresh; |
|
416 |
|
417 |
|
418 /*!\brief Rate control algorithm to use. |
|
419 * |
|
420 * Indicates whether the end usage of this stream is to be streamed over |
|
421 * a bandwidth constrained link, indicating that Constant Bit Rate (CBR) |
|
422 * mode should be used, or whether it will be played back on a high |
|
423 * bandwidth link, as from a local disk, where higher variations in |
|
424 * bitrate are acceptable. |
|
425 */ |
|
426 enum vpx_rc_mode rc_end_usage; |
|
427 |
|
428 |
|
429 /*!\brief Two-pass stats buffer. |
|
430 * |
|
431 * A buffer containing all of the stats packets produced in the first |
|
432 * pass, concatenated. |
|
433 */ |
|
434 struct vpx_fixed_buf rc_twopass_stats_in; |
|
435 |
|
436 |
|
437 /*!\brief Target data rate |
|
438 * |
|
439 * Target bandwidth to use for this stream, in kilobits per second. |
|
440 */ |
|
441 unsigned int rc_target_bitrate; |
|
442 |
|
443 |
|
444 /* |
|
445 * quantizer settings |
|
446 */ |
|
447 |
|
448 |
|
449 /*!\brief Minimum (Best Quality) Quantizer |
|
450 * |
|
451 * The quantizer is the most direct control over the quality of the |
|
452 * encoded image. The range of valid values for the quantizer is codec |
|
453 * specific. Consult the documentation for the codec to determine the |
|
454 * values to use. To determine the range programmatically, call |
|
455 * vpx_codec_enc_config_default() with a usage value of 0. |
|
456 */ |
|
457 unsigned int rc_min_quantizer; |
|
458 |
|
459 |
|
460 /*!\brief Maximum (Worst Quality) Quantizer |
|
461 * |
|
462 * The quantizer is the most direct control over the quality of the |
|
463 * encoded image. The range of valid values for the quantizer is codec |
|
464 * specific. Consult the documentation for the codec to determine the |
|
465 * values to use. To determine the range programmatically, call |
|
466 * vpx_codec_enc_config_default() with a usage value of 0. |
|
467 */ |
|
468 unsigned int rc_max_quantizer; |
|
469 |
|
470 |
|
471 /* |
|
472 * bitrate tolerance |
|
473 */ |
|
474 |
|
475 |
|
476 /*!\brief Rate control adaptation undershoot control |
|
477 * |
|
478 * This value, expressed as a percentage of the target bitrate, |
|
479 * controls the maximum allowed adaptation speed of the codec. |
|
480 * This factor controls the maximum amount of bits that can |
|
481 * be subtracted from the target bitrate in order to compensate |
|
482 * for prior overshoot. |
|
483 * |
|
484 * Valid values in the range 0-1000. |
|
485 */ |
|
486 unsigned int rc_undershoot_pct; |
|
487 |
|
488 |
|
489 /*!\brief Rate control adaptation overshoot control |
|
490 * |
|
491 * This value, expressed as a percentage of the target bitrate, |
|
492 * controls the maximum allowed adaptation speed of the codec. |
|
493 * This factor controls the maximum amount of bits that can |
|
494 * be added to the target bitrate in order to compensate for |
|
495 * prior undershoot. |
|
496 * |
|
497 * Valid values in the range 0-1000. |
|
498 */ |
|
499 unsigned int rc_overshoot_pct; |
|
500 |
|
501 |
|
502 /* |
|
503 * decoder buffer model parameters |
|
504 */ |
|
505 |
|
506 |
|
507 /*!\brief Decoder Buffer Size |
|
508 * |
|
509 * This value indicates the amount of data that may be buffered by the |
|
510 * decoding application. Note that this value is expressed in units of |
|
511 * time (milliseconds). For example, a value of 5000 indicates that the |
|
512 * client will buffer (at least) 5000ms worth of encoded data. Use the |
|
513 * target bitrate (#rc_target_bitrate) to convert to bits/bytes, if |
|
514 * necessary. |
|
515 */ |
|
516 unsigned int rc_buf_sz; |
|
517 |
|
518 |
|
519 /*!\brief Decoder Buffer Initial Size |
|
520 * |
|
521 * This value indicates the amount of data that will be buffered by the |
|
522 * decoding application prior to beginning playback. This value is |
|
523 * expressed in units of time (milliseconds). Use the target bitrate |
|
524 * (#rc_target_bitrate) to convert to bits/bytes, if necessary. |
|
525 */ |
|
526 unsigned int rc_buf_initial_sz; |
|
527 |
|
528 |
|
529 /*!\brief Decoder Buffer Optimal Size |
|
530 * |
|
531 * This value indicates the amount of data that the encoder should try |
|
532 * to maintain in the decoder's buffer. This value is expressed in units |
|
533 * of time (milliseconds). Use the target bitrate (#rc_target_bitrate) |
|
534 * to convert to bits/bytes, if necessary. |
|
535 */ |
|
536 unsigned int rc_buf_optimal_sz; |
|
537 |
|
538 |
|
539 /* |
|
540 * 2 pass rate control parameters |
|
541 */ |
|
542 |
|
543 |
|
544 /*!\brief Two-pass mode CBR/VBR bias |
|
545 * |
|
546 * Bias, expressed on a scale of 0 to 100, for determining target size |
|
547 * for the current frame. The value 0 indicates the optimal CBR mode |
|
548 * value should be used. The value 100 indicates the optimal VBR mode |
|
549 * value should be used. Values in between indicate which way the |
|
550 * encoder should "lean." |
|
551 */ |
|
552 unsigned int rc_2pass_vbr_bias_pct; /**< RC mode bias between CBR and VBR(0-100: 0->CBR, 100->VBR) */ |
|
553 |
|
554 |
|
555 /*!\brief Two-pass mode per-GOP minimum bitrate |
|
556 * |
|
557 * This value, expressed as a percentage of the target bitrate, indicates |
|
558 * the minimum bitrate to be used for a single GOP (aka "section") |
|
559 */ |
|
560 unsigned int rc_2pass_vbr_minsection_pct; |
|
561 |
|
562 |
|
563 /*!\brief Two-pass mode per-GOP maximum bitrate |
|
564 * |
|
565 * This value, expressed as a percentage of the target bitrate, indicates |
|
566 * the maximum bitrate to be used for a single GOP (aka "section") |
|
567 */ |
|
568 unsigned int rc_2pass_vbr_maxsection_pct; |
|
569 |
|
570 |
|
571 /* |
|
572 * keyframing settings (kf) |
|
573 */ |
|
574 |
|
575 /*!\brief Keyframe placement mode |
|
576 * |
|
577 * This value indicates whether the encoder should place keyframes at a |
|
578 * fixed interval, or determine the optimal placement automatically |
|
579 * (as governed by the #kf_min_dist and #kf_max_dist parameters) |
|
580 */ |
|
581 enum vpx_kf_mode kf_mode; |
|
582 |
|
583 |
|
584 /*!\brief Keyframe minimum interval |
|
585 * |
|
586 * This value, expressed as a number of frames, prevents the encoder from |
|
587 * placing a keyframe nearer than kf_min_dist to the previous keyframe. At |
|
588 * least kf_min_dist frames non-keyframes will be coded before the next |
|
589 * keyframe. Set kf_min_dist equal to kf_max_dist for a fixed interval. |
|
590 */ |
|
591 unsigned int kf_min_dist; |
|
592 |
|
593 |
|
594 /*!\brief Keyframe maximum interval |
|
595 * |
|
596 * This value, expressed as a number of frames, forces the encoder to code |
|
597 * a keyframe if one has not been coded in the last kf_max_dist frames. |
|
598 * A value of 0 implies all frames will be keyframes. Set kf_min_dist |
|
599 * equal to kf_max_dist for a fixed interval. |
|
600 */ |
|
601 unsigned int kf_max_dist; |
|
602 |
|
603 /* |
|
604 * Spatial scalability settings (ss) |
|
605 */ |
|
606 |
|
607 /*!\brief Number of coding layers (spatial) |
|
608 * |
|
609 * This value specifies the number of coding layers to be used. |
|
610 */ |
|
611 unsigned int ss_number_layers; |
|
612 |
|
613 /*!\brief Number of coding layers |
|
614 * |
|
615 * This value specifies the number of coding layers to be used. |
|
616 */ |
|
617 unsigned int ts_number_layers; |
|
618 |
|
619 /*!\brief Target bitrate for each layer |
|
620 * |
|
621 * These values specify the target coding bitrate for each coding layer. |
|
622 */ |
|
623 unsigned int ts_target_bitrate[VPX_TS_MAX_LAYERS]; |
|
624 |
|
625 /*!\brief Frame rate decimation factor for each layer |
|
626 * |
|
627 * These values specify the frame rate decimation factors to apply |
|
628 * to each layer. |
|
629 */ |
|
630 unsigned int ts_rate_decimator[VPX_TS_MAX_LAYERS]; |
|
631 |
|
632 /*!\brief Length of the sequence defining frame layer membership |
|
633 * |
|
634 * This value specifies the length of the sequence that defines the |
|
635 * membership of frames to layers. For example, if ts_periodicity=8 then |
|
636 * frames are assigned to coding layers with a repeated sequence of |
|
637 * length 8. |
|
638 */ |
|
639 unsigned int ts_periodicity; |
|
640 |
|
641 /*!\brief Template defining the membership of frames to coding layers |
|
642 * |
|
643 * This array defines the membership of frames to coding layers. For a |
|
644 * 2-layer encoding that assigns even numbered frames to one layer (0) |
|
645 * and odd numbered frames to a second layer (1) with ts_periodicity=8, |
|
646 * then ts_layer_id = (0,1,0,1,0,1,0,1). |
|
647 */ |
|
648 unsigned int ts_layer_id[VPX_TS_MAX_PERIODICITY]; |
|
649 } vpx_codec_enc_cfg_t; /**< alias for struct vpx_codec_enc_cfg */ |
|
650 |
|
651 |
|
652 /*!\brief Initialize an encoder instance |
|
653 * |
|
654 * Initializes a encoder context using the given interface. Applications |
|
655 * should call the vpx_codec_enc_init convenience macro instead of this |
|
656 * function directly, to ensure that the ABI version number parameter |
|
657 * is properly initialized. |
|
658 * |
|
659 * If the library was configured with --disable-multithread, this call |
|
660 * is not thread safe and should be guarded with a lock if being used |
|
661 * in a multithreaded context. |
|
662 * |
|
663 * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags |
|
664 * parameter), the storage pointed to by the cfg parameter must be |
|
665 * kept readable and stable until all memory maps have been set. |
|
666 * |
|
667 * \param[in] ctx Pointer to this instance's context. |
|
668 * \param[in] iface Pointer to the algorithm interface to use. |
|
669 * \param[in] cfg Configuration to use, if known. May be NULL. |
|
670 * \param[in] flags Bitfield of VPX_CODEC_USE_* flags |
|
671 * \param[in] ver ABI version number. Must be set to |
|
672 * VPX_ENCODER_ABI_VERSION |
|
673 * \retval #VPX_CODEC_OK |
|
674 * The decoder algorithm initialized. |
|
675 * \retval #VPX_CODEC_MEM_ERROR |
|
676 * Memory allocation failed. |
|
677 */ |
|
678 vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t *ctx, |
|
679 vpx_codec_iface_t *iface, |
|
680 vpx_codec_enc_cfg_t *cfg, |
|
681 vpx_codec_flags_t flags, |
|
682 int ver); |
|
683 |
|
684 |
|
685 /*!\brief Convenience macro for vpx_codec_enc_init_ver() |
|
686 * |
|
687 * Ensures the ABI version parameter is properly set. |
|
688 */ |
|
689 #define vpx_codec_enc_init(ctx, iface, cfg, flags) \ |
|
690 vpx_codec_enc_init_ver(ctx, iface, cfg, flags, VPX_ENCODER_ABI_VERSION) |
|
691 |
|
692 |
|
693 /*!\brief Initialize multi-encoder instance |
|
694 * |
|
695 * Initializes multi-encoder context using the given interface. |
|
696 * Applications should call the vpx_codec_enc_init_multi convenience macro |
|
697 * instead of this function directly, to ensure that the ABI version number |
|
698 * parameter is properly initialized. |
|
699 * |
|
700 * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags |
|
701 * parameter), the storage pointed to by the cfg parameter must be |
|
702 * kept readable and stable until all memory maps have been set. |
|
703 * |
|
704 * \param[in] ctx Pointer to this instance's context. |
|
705 * \param[in] iface Pointer to the algorithm interface to use. |
|
706 * \param[in] cfg Configuration to use, if known. May be NULL. |
|
707 * \param[in] num_enc Total number of encoders. |
|
708 * \param[in] flags Bitfield of VPX_CODEC_USE_* flags |
|
709 * \param[in] dsf Pointer to down-sampling factors. |
|
710 * \param[in] ver ABI version number. Must be set to |
|
711 * VPX_ENCODER_ABI_VERSION |
|
712 * \retval #VPX_CODEC_OK |
|
713 * The decoder algorithm initialized. |
|
714 * \retval #VPX_CODEC_MEM_ERROR |
|
715 * Memory allocation failed. |
|
716 */ |
|
717 vpx_codec_err_t vpx_codec_enc_init_multi_ver(vpx_codec_ctx_t *ctx, |
|
718 vpx_codec_iface_t *iface, |
|
719 vpx_codec_enc_cfg_t *cfg, |
|
720 int num_enc, |
|
721 vpx_codec_flags_t flags, |
|
722 vpx_rational_t *dsf, |
|
723 int ver); |
|
724 |
|
725 |
|
726 /*!\brief Convenience macro for vpx_codec_enc_init_multi_ver() |
|
727 * |
|
728 * Ensures the ABI version parameter is properly set. |
|
729 */ |
|
730 #define vpx_codec_enc_init_multi(ctx, iface, cfg, num_enc, flags, dsf) \ |
|
731 vpx_codec_enc_init_multi_ver(ctx, iface, cfg, num_enc, flags, dsf, \ |
|
732 VPX_ENCODER_ABI_VERSION) |
|
733 |
|
734 |
|
735 /*!\brief Get a default configuration |
|
736 * |
|
737 * Initializes a encoder configuration structure with default values. Supports |
|
738 * the notion of "usages" so that an algorithm may offer different default |
|
739 * settings depending on the user's intended goal. This function \ref SHOULD |
|
740 * be called by all applications to initialize the configuration structure |
|
741 * before specializing the configuration with application specific values. |
|
742 * |
|
743 * \param[in] iface Pointer to the algorithm interface to use. |
|
744 * \param[out] cfg Configuration buffer to populate |
|
745 * \param[in] usage End usage. Set to 0 or use codec specific values. |
|
746 * |
|
747 * \retval #VPX_CODEC_OK |
|
748 * The configuration was populated. |
|
749 * \retval #VPX_CODEC_INCAPABLE |
|
750 * Interface is not an encoder interface. |
|
751 * \retval #VPX_CODEC_INVALID_PARAM |
|
752 * A parameter was NULL, or the usage value was not recognized. |
|
753 */ |
|
754 vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface, |
|
755 vpx_codec_enc_cfg_t *cfg, |
|
756 unsigned int usage); |
|
757 |
|
758 |
|
759 /*!\brief Set or change configuration |
|
760 * |
|
761 * Reconfigures an encoder instance according to the given configuration. |
|
762 * |
|
763 * \param[in] ctx Pointer to this instance's context |
|
764 * \param[in] cfg Configuration buffer to use |
|
765 * |
|
766 * \retval #VPX_CODEC_OK |
|
767 * The configuration was populated. |
|
768 * \retval #VPX_CODEC_INCAPABLE |
|
769 * Interface is not an encoder interface. |
|
770 * \retval #VPX_CODEC_INVALID_PARAM |
|
771 * A parameter was NULL, or the usage value was not recognized. |
|
772 */ |
|
773 vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx, |
|
774 const vpx_codec_enc_cfg_t *cfg); |
|
775 |
|
776 |
|
777 /*!\brief Get global stream headers |
|
778 * |
|
779 * Retrieves a stream level global header packet, if supported by the codec. |
|
780 * |
|
781 * \param[in] ctx Pointer to this instance's context |
|
782 * |
|
783 * \retval NULL |
|
784 * Encoder does not support global header |
|
785 * \retval Non-NULL |
|
786 * Pointer to buffer containing global header packet |
|
787 */ |
|
788 vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx); |
|
789 |
|
790 |
|
791 #define VPX_DL_REALTIME (1) /**< deadline parameter analogous to |
|
792 * VPx REALTIME mode. */ |
|
793 #define VPX_DL_GOOD_QUALITY (1000000) /**< deadline parameter analogous to |
|
794 * VPx GOOD QUALITY mode. */ |
|
795 #define VPX_DL_BEST_QUALITY (0) /**< deadline parameter analogous to |
|
796 * VPx BEST QUALITY mode. */ |
|
797 /*!\brief Encode a frame |
|
798 * |
|
799 * Encodes a video frame at the given "presentation time." The presentation |
|
800 * time stamp (PTS) \ref MUST be strictly increasing. |
|
801 * |
|
802 * The encoder supports the notion of a soft real-time deadline. Given a |
|
803 * non-zero value to the deadline parameter, the encoder will make a "best |
|
804 * effort" guarantee to return before the given time slice expires. It is |
|
805 * implicit that limiting the available time to encode will degrade the |
|
806 * output quality. The encoder can be given an unlimited time to produce the |
|
807 * best possible frame by specifying a deadline of '0'. This deadline |
|
808 * supercedes the VPx notion of "best quality, good quality, realtime". |
|
809 * Applications that wish to map these former settings to the new deadline |
|
810 * based system can use the symbols #VPX_DL_REALTIME, #VPX_DL_GOOD_QUALITY, |
|
811 * and #VPX_DL_BEST_QUALITY. |
|
812 * |
|
813 * When the last frame has been passed to the encoder, this function should |
|
814 * continue to be called, with the img parameter set to NULL. This will |
|
815 * signal the end-of-stream condition to the encoder and allow it to encode |
|
816 * any held buffers. Encoding is complete when vpx_codec_encode() is called |
|
817 * and vpx_codec_get_cx_data() returns no data. |
|
818 * |
|
819 * \param[in] ctx Pointer to this instance's context |
|
820 * \param[in] img Image data to encode, NULL to flush. |
|
821 * \param[in] pts Presentation time stamp, in timebase units. |
|
822 * \param[in] duration Duration to show frame, in timebase units. |
|
823 * \param[in] flags Flags to use for encoding this frame. |
|
824 * \param[in] deadline Time to spend encoding, in microseconds. (0=infinite) |
|
825 * |
|
826 * \retval #VPX_CODEC_OK |
|
827 * The configuration was populated. |
|
828 * \retval #VPX_CODEC_INCAPABLE |
|
829 * Interface is not an encoder interface. |
|
830 * \retval #VPX_CODEC_INVALID_PARAM |
|
831 * A parameter was NULL, the image format is unsupported, etc. |
|
832 */ |
|
833 vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx, |
|
834 const vpx_image_t *img, |
|
835 vpx_codec_pts_t pts, |
|
836 unsigned long duration, |
|
837 vpx_enc_frame_flags_t flags, |
|
838 unsigned long deadline); |
|
839 |
|
840 /*!\brief Set compressed data output buffer |
|
841 * |
|
842 * Sets the buffer that the codec should output the compressed data |
|
843 * into. This call effectively sets the buffer pointer returned in the |
|
844 * next VPX_CODEC_CX_FRAME_PKT packet. Subsequent packets will be |
|
845 * appended into this buffer. The buffer is preserved across frames, |
|
846 * so applications must periodically call this function after flushing |
|
847 * the accumulated compressed data to disk or to the network to reset |
|
848 * the pointer to the buffer's head. |
|
849 * |
|
850 * `pad_before` bytes will be skipped before writing the compressed |
|
851 * data, and `pad_after` bytes will be appended to the packet. The size |
|
852 * of the packet will be the sum of the size of the actual compressed |
|
853 * data, pad_before, and pad_after. The padding bytes will be preserved |
|
854 * (not overwritten). |
|
855 * |
|
856 * Note that calling this function does not guarantee that the returned |
|
857 * compressed data will be placed into the specified buffer. In the |
|
858 * event that the encoded data will not fit into the buffer provided, |
|
859 * the returned packet \ref MAY point to an internal buffer, as it would |
|
860 * if this call were never used. In this event, the output packet will |
|
861 * NOT have any padding, and the application must free space and copy it |
|
862 * to the proper place. This is of particular note in configurations |
|
863 * that may output multiple packets for a single encoded frame (e.g., lagged |
|
864 * encoding) or if the application does not reset the buffer periodically. |
|
865 * |
|
866 * Applications may restore the default behavior of the codec providing |
|
867 * the compressed data buffer by calling this function with a NULL |
|
868 * buffer. |
|
869 * |
|
870 * Applications \ref MUSTNOT call this function during iteration of |
|
871 * vpx_codec_get_cx_data(). |
|
872 * |
|
873 * \param[in] ctx Pointer to this instance's context |
|
874 * \param[in] buf Buffer to store compressed data into |
|
875 * \param[in] pad_before Bytes to skip before writing compressed data |
|
876 * \param[in] pad_after Bytes to skip after writing compressed data |
|
877 * |
|
878 * \retval #VPX_CODEC_OK |
|
879 * The buffer was set successfully. |
|
880 * \retval #VPX_CODEC_INVALID_PARAM |
|
881 * A parameter was NULL, the image format is unsupported, etc. |
|
882 */ |
|
883 vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t *ctx, |
|
884 const vpx_fixed_buf_t *buf, |
|
885 unsigned int pad_before, |
|
886 unsigned int pad_after); |
|
887 |
|
888 |
|
889 /*!\brief Encoded data iterator |
|
890 * |
|
891 * Iterates over a list of data packets to be passed from the encoder to the |
|
892 * application. The different kinds of packets available are enumerated in |
|
893 * #vpx_codec_cx_pkt_kind. |
|
894 * |
|
895 * #VPX_CODEC_CX_FRAME_PKT packets should be passed to the application's |
|
896 * muxer. Multiple compressed frames may be in the list. |
|
897 * #VPX_CODEC_STATS_PKT packets should be appended to a global buffer. |
|
898 * |
|
899 * The application \ref MUST silently ignore any packet kinds that it does |
|
900 * not recognize or support. |
|
901 * |
|
902 * The data buffers returned from this function are only guaranteed to be |
|
903 * valid until the application makes another call to any vpx_codec_* function. |
|
904 * |
|
905 * \param[in] ctx Pointer to this instance's context |
|
906 * \param[in,out] iter Iterator storage, initialized to NULL |
|
907 * |
|
908 * \return Returns a pointer to an output data packet (compressed frame data, |
|
909 * two-pass statistics, etc.) or NULL to signal end-of-list. |
|
910 * |
|
911 */ |
|
912 const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx, |
|
913 vpx_codec_iter_t *iter); |
|
914 |
|
915 |
|
916 /*!\brief Get Preview Frame |
|
917 * |
|
918 * Returns an image that can be used as a preview. Shows the image as it would |
|
919 * exist at the decompressor. The application \ref MUST NOT write into this |
|
920 * image buffer. |
|
921 * |
|
922 * \param[in] ctx Pointer to this instance's context |
|
923 * |
|
924 * \return Returns a pointer to a preview image, or NULL if no image is |
|
925 * available. |
|
926 * |
|
927 */ |
|
928 const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t *ctx); |
|
929 |
|
930 |
|
931 /*!@} - end defgroup encoder*/ |
|
932 #ifdef __cplusplus |
|
933 } |
|
934 #endif |
|
935 #endif |
|
936 |