2 * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
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.
10 #ifndef VPX_VPX_ENCODER_H_
11 #define VPX_VPX_ENCODER_H_
13 /*!\defgroup encoder Encoder Algorithm Interface
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.
22 * \brief Describes the encoder algorithm interface to applications.
24 * This file describes the interface between an application and a
25 * video encoder algorithm.
32 #include "./vpx_codec.h"
34 /*! Temporal Scalability: Maximum length of the sequence defining frame
37 #define VPX_TS_MAX_PERIODICITY 16
39 /*! Temporal Scalability: Maximum number of coding layers */
40 #define VPX_TS_MAX_LAYERS 5
42 /*!\deprecated Use #VPX_TS_MAX_PERIODICITY instead. */
43 #define MAX_PERIODICITY VPX_TS_MAX_PERIODICITY
45 /*!\deprecated Use #VPX_TS_MAX_LAYERS instead. */
46 #define MAX_LAYERS VPX_TS_MAX_LAYERS
48 /*! Spatial Scalability: Maximum number of coding layers */
49 #define VPX_SS_MAX_LAYERS 5
51 /*! Spatial Scalability: Default number of coding layers */
52 #define VPX_SS_DEFAULT_LAYERS 3
54 /*!\brief Current ABI version number
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
62 #define VPX_ENCODER_ABI_VERSION (3 + VPX_CODEC_ABI_VERSION) /**<\hideinitializer*/
65 /*! \brief Encoder capabilities bitfield
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
72 * The available flags are specified by VPX_CODEC_CAP_* defines.
74 #define VPX_CODEC_CAP_PSNR 0x10000 /**< Can issue PSNR packets */
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.
81 #define VPX_CODEC_CAP_OUTPUT_PARTITION 0x20000
84 /*! \brief Initialization-time Feature Enabling
86 * Certain codec features must be known at initialization time, to allow
87 * for proper memory allocation.
89 * The available flags are specified by VPX_CODEC_USE_* defines.
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. */
96 /*!\brief Generic fixed size buffer structure
98 * This structure is able to hold a reference to any fixed size buffer.
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 */
106 /*!\brief Time Stamp Type
108 * An integer, which when multiplied by the stream's time base, provides
109 * the absolute time of a sample.
111 typedef int64_t vpx_codec_pts_t;
114 /*!\brief Compressed Frame Flags
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)
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
126 #define VPX_FRAME_IS_INVISIBLE 0x4 /**< frame should be decoded but will not
128 #define VPX_FRAME_IS_FRAGMENT 0x8 /**< this is a fragment of the encoded
131 /*!\brief Error Resilient flags
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.
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. */
149 /*!\brief Encoder output packet variants
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.
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 */
163 /*!\brief Encoder output packet
165 * This structure contains the different kinds of output data the encoder
166 * may produce while compressing a frame.
168 typedef struct vpx_codec_cx_pkt {
169 enum vpx_codec_cx_pkt_kind kind; /**< packet variant */
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
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 */
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.
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 */
205 /*!\brief Rational Number
207 * This structure holds a fractional value.
209 typedef struct vpx_rational {
210 int num; /**< fraction numerator */
211 int den; /**< fraction denominator */
212 } vpx_rational_t; /**< alias for struct vpx_rational */
215 /*!\brief Multi-pass Encoding 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 */
223 /*!\brief Rate control 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 */
232 /*!\brief Keyframe placement mode.
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.
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. */
247 /*!\brief Encoded Frame Flags
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.
254 typedef long vpx_enc_frame_flags_t;
255 #define VPX_EFLAG_FORCE_KF (1<<0) /**< Force this frame to be a keyframe */
258 /*!\brief Encoder configuration structure
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,
264 typedef struct vpx_codec_enc_cfg {
266 * generic settings (g)
269 /*!\brief Algorithm specific "usage" value
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
276 unsigned int g_usage;
279 /*!\brief Maximum number of threads to use
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.
285 unsigned int g_threads;
288 /*!\brief Bitstream profile to use
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.
296 unsigned int g_profile; /**< profile of bitstream to use */
300 /*!\brief Width of the frame
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.
310 /*!\brief Height of the frame
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.
320 /*!\brief Stream timebase units
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).
332 struct vpx_rational g_timebase;
335 /*!\brief Enable error resilient modes.
337 * The error resilient bitfield indicates to the encoder which features
338 * it should enable to take measures for streaming over lossy or noisy
341 vpx_codec_er_flags_t g_error_resilient;
344 /*!\brief Multi-pass Encoding Mode
346 * This value should be set to the current phase for multi-pass encoding.
347 * For single pass, set to #VPX_RC_ONE_PASS.
349 enum vpx_enc_pass g_pass;
352 /*!\brief Allow lagged encoding
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).
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
364 unsigned int g_lag_in_frames;
368 * rate control settings (rc)
371 /*!\brief Temporal resampling configuration, if supported by the codec.
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.
379 * Note that not all codecs support this feature. All vpx VPx codecs do.
380 * For other codecs, consult the documentation for that algorithm.
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
387 unsigned int rc_dropframe_thresh;
390 /*!\brief Enable/disable spatial resampling, if supported by the codec.
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.
397 unsigned int rc_resize_allowed;
400 /*!\brief Spatial resampling up watermark.
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.
406 unsigned int rc_resize_up_thresh;
409 /*!\brief Spatial resampling down watermark.
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.
415 unsigned int rc_resize_down_thresh;
418 /*!\brief Rate control algorithm to use.
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.
426 enum vpx_rc_mode rc_end_usage;
429 /*!\brief Two-pass stats buffer.
431 * A buffer containing all of the stats packets produced in the first
432 * pass, concatenated.
434 struct vpx_fixed_buf rc_twopass_stats_in;
437 /*!\brief Target data rate
439 * Target bandwidth to use for this stream, in kilobits per second.
441 unsigned int rc_target_bitrate;
449 /*!\brief Minimum (Best Quality) Quantizer
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.
457 unsigned int rc_min_quantizer;
460 /*!\brief Maximum (Worst Quality) Quantizer
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.
468 unsigned int rc_max_quantizer;
476 /*!\brief Rate control adaptation undershoot control
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.
484 * Valid values in the range 0-1000.
486 unsigned int rc_undershoot_pct;
489 /*!\brief Rate control adaptation overshoot control
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
497 * Valid values in the range 0-1000.
499 unsigned int rc_overshoot_pct;
503 * decoder buffer model parameters
507 /*!\brief Decoder Buffer Size
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
516 unsigned int rc_buf_sz;
519 /*!\brief Decoder Buffer Initial Size
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.
526 unsigned int rc_buf_initial_sz;
529 /*!\brief Decoder Buffer Optimal Size
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.
536 unsigned int rc_buf_optimal_sz;
540 * 2 pass rate control parameters
544 /*!\brief Two-pass mode CBR/VBR bias
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."
552 unsigned int rc_2pass_vbr_bias_pct; /**< RC mode bias between CBR and VBR(0-100: 0->CBR, 100->VBR) */
555 /*!\brief Two-pass mode per-GOP minimum bitrate
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")
560 unsigned int rc_2pass_vbr_minsection_pct;
563 /*!\brief Two-pass mode per-GOP maximum bitrate
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")
568 unsigned int rc_2pass_vbr_maxsection_pct;
572 * keyframing settings (kf)
575 /*!\brief Keyframe placement mode
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)
581 enum vpx_kf_mode kf_mode;
584 /*!\brief Keyframe minimum interval
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.
591 unsigned int kf_min_dist;
594 /*!\brief Keyframe maximum interval
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.
601 unsigned int kf_max_dist;
604 * Spatial scalability settings (ss)
607 /*!\brief Number of spatial coding layers.
609 * This value specifies the number of spatial coding layers to be used.
611 unsigned int ss_number_layers;
613 /*!\brief Number of temporal coding layers.
615 * This value specifies the number of temporal layers to be used.
617 unsigned int ts_number_layers;
619 /*!\brief Target bitrate for each temporal layer.
621 * These values specify the target coding bitrate to be used for each
624 unsigned int ts_target_bitrate[VPX_TS_MAX_LAYERS];
626 /*!\brief Frame rate decimation factor for each temporal layer.
628 * These values specify the frame rate decimation factors to apply
629 * to each temporal layer.
631 unsigned int ts_rate_decimator[VPX_TS_MAX_LAYERS];
633 /*!\brief Length of the sequence defining frame temporal layer membership.
635 * This value specifies the length of the sequence that defines the
636 * membership of frames to temporal layers. For example, if the
637 * ts_periodicity = 8, then the frames are assigned to coding layers with a
638 * repeated sequence of length 8.
640 unsigned int ts_periodicity;
642 /*!\brief Template defining the membership of frames to temporal layers.
644 * This array defines the membership of frames to temporal coding layers.
645 * For a 2-layer encoding that assigns even numbered frames to one temporal
646 * layer (0) and odd numbered frames to a second temporal layer (1) with
647 * ts_periodicity=8, then ts_layer_id = (0,1,0,1,0,1,0,1).
649 unsigned int ts_layer_id[VPX_TS_MAX_PERIODICITY];
650 } vpx_codec_enc_cfg_t; /**< alias for struct vpx_codec_enc_cfg */
653 /*!\brief Initialize an encoder instance
655 * Initializes a encoder context using the given interface. Applications
656 * should call the vpx_codec_enc_init convenience macro instead of this
657 * function directly, to ensure that the ABI version number parameter
658 * is properly initialized.
660 * If the library was configured with --disable-multithread, this call
661 * is not thread safe and should be guarded with a lock if being used
662 * in a multithreaded context.
664 * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags
665 * parameter), the storage pointed to by the cfg parameter must be
666 * kept readable and stable until all memory maps have been set.
668 * \param[in] ctx Pointer to this instance's context.
669 * \param[in] iface Pointer to the algorithm interface to use.
670 * \param[in] cfg Configuration to use, if known. May be NULL.
671 * \param[in] flags Bitfield of VPX_CODEC_USE_* flags
672 * \param[in] ver ABI version number. Must be set to
673 * VPX_ENCODER_ABI_VERSION
674 * \retval #VPX_CODEC_OK
675 * The decoder algorithm initialized.
676 * \retval #VPX_CODEC_MEM_ERROR
677 * Memory allocation failed.
679 vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t *ctx,
680 vpx_codec_iface_t *iface,
681 vpx_codec_enc_cfg_t *cfg,
682 vpx_codec_flags_t flags,
686 /*!\brief Convenience macro for vpx_codec_enc_init_ver()
688 * Ensures the ABI version parameter is properly set.
690 #define vpx_codec_enc_init(ctx, iface, cfg, flags) \
691 vpx_codec_enc_init_ver(ctx, iface, cfg, flags, VPX_ENCODER_ABI_VERSION)
694 /*!\brief Initialize multi-encoder instance
696 * Initializes multi-encoder context using the given interface.
697 * Applications should call the vpx_codec_enc_init_multi convenience macro
698 * instead of this function directly, to ensure that the ABI version number
699 * parameter is properly initialized.
701 * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags
702 * parameter), the storage pointed to by the cfg parameter must be
703 * kept readable and stable until all memory maps have been set.
705 * \param[in] ctx Pointer to this instance's context.
706 * \param[in] iface Pointer to the algorithm interface to use.
707 * \param[in] cfg Configuration to use, if known. May be NULL.
708 * \param[in] num_enc Total number of encoders.
709 * \param[in] flags Bitfield of VPX_CODEC_USE_* flags
710 * \param[in] dsf Pointer to down-sampling factors.
711 * \param[in] ver ABI version number. Must be set to
712 * VPX_ENCODER_ABI_VERSION
713 * \retval #VPX_CODEC_OK
714 * The decoder algorithm initialized.
715 * \retval #VPX_CODEC_MEM_ERROR
716 * Memory allocation failed.
718 vpx_codec_err_t vpx_codec_enc_init_multi_ver(vpx_codec_ctx_t *ctx,
719 vpx_codec_iface_t *iface,
720 vpx_codec_enc_cfg_t *cfg,
722 vpx_codec_flags_t flags,
727 /*!\brief Convenience macro for vpx_codec_enc_init_multi_ver()
729 * Ensures the ABI version parameter is properly set.
731 #define vpx_codec_enc_init_multi(ctx, iface, cfg, num_enc, flags, dsf) \
732 vpx_codec_enc_init_multi_ver(ctx, iface, cfg, num_enc, flags, dsf, \
733 VPX_ENCODER_ABI_VERSION)
736 /*!\brief Get a default configuration
738 * Initializes a encoder configuration structure with default values. Supports
739 * the notion of "usages" so that an algorithm may offer different default
740 * settings depending on the user's intended goal. This function \ref SHOULD
741 * be called by all applications to initialize the configuration structure
742 * before specializing the configuration with application specific values.
744 * \param[in] iface Pointer to the algorithm interface to use.
745 * \param[out] cfg Configuration buffer to populate
746 * \param[in] usage End usage. Set to 0 or use codec specific values.
748 * \retval #VPX_CODEC_OK
749 * The configuration was populated.
750 * \retval #VPX_CODEC_INCAPABLE
751 * Interface is not an encoder interface.
752 * \retval #VPX_CODEC_INVALID_PARAM
753 * A parameter was NULL, or the usage value was not recognized.
755 vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface,
756 vpx_codec_enc_cfg_t *cfg,
760 /*!\brief Set or change configuration
762 * Reconfigures an encoder instance according to the given configuration.
764 * \param[in] ctx Pointer to this instance's context
765 * \param[in] cfg Configuration buffer to use
767 * \retval #VPX_CODEC_OK
768 * The configuration was populated.
769 * \retval #VPX_CODEC_INCAPABLE
770 * Interface is not an encoder interface.
771 * \retval #VPX_CODEC_INVALID_PARAM
772 * A parameter was NULL, or the usage value was not recognized.
774 vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx,
775 const vpx_codec_enc_cfg_t *cfg);
778 /*!\brief Get global stream headers
780 * Retrieves a stream level global header packet, if supported by the codec.
782 * \param[in] ctx Pointer to this instance's context
785 * Encoder does not support global header
787 * Pointer to buffer containing global header packet
789 vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx);
792 #define VPX_DL_REALTIME (1) /**< deadline parameter analogous to
793 * VPx REALTIME mode. */
794 #define VPX_DL_GOOD_QUALITY (1000000) /**< deadline parameter analogous to
795 * VPx GOOD QUALITY mode. */
796 #define VPX_DL_BEST_QUALITY (0) /**< deadline parameter analogous to
797 * VPx BEST QUALITY mode. */
798 /*!\brief Encode a frame
800 * Encodes a video frame at the given "presentation time." The presentation
801 * time stamp (PTS) \ref MUST be strictly increasing.
803 * The encoder supports the notion of a soft real-time deadline. Given a
804 * non-zero value to the deadline parameter, the encoder will make a "best
805 * effort" guarantee to return before the given time slice expires. It is
806 * implicit that limiting the available time to encode will degrade the
807 * output quality. The encoder can be given an unlimited time to produce the
808 * best possible frame by specifying a deadline of '0'. This deadline
809 * supercedes the VPx notion of "best quality, good quality, realtime".
810 * Applications that wish to map these former settings to the new deadline
811 * based system can use the symbols #VPX_DL_REALTIME, #VPX_DL_GOOD_QUALITY,
812 * and #VPX_DL_BEST_QUALITY.
814 * When the last frame has been passed to the encoder, this function should
815 * continue to be called, with the img parameter set to NULL. This will
816 * signal the end-of-stream condition to the encoder and allow it to encode
817 * any held buffers. Encoding is complete when vpx_codec_encode() is called
818 * and vpx_codec_get_cx_data() returns no data.
820 * \param[in] ctx Pointer to this instance's context
821 * \param[in] img Image data to encode, NULL to flush.
822 * \param[in] pts Presentation time stamp, in timebase units.
823 * \param[in] duration Duration to show frame, in timebase units.
824 * \param[in] flags Flags to use for encoding this frame.
825 * \param[in] deadline Time to spend encoding, in microseconds. (0=infinite)
827 * \retval #VPX_CODEC_OK
828 * The configuration was populated.
829 * \retval #VPX_CODEC_INCAPABLE
830 * Interface is not an encoder interface.
831 * \retval #VPX_CODEC_INVALID_PARAM
832 * A parameter was NULL, the image format is unsupported, etc.
834 vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx,
835 const vpx_image_t *img,
837 unsigned long duration,
838 vpx_enc_frame_flags_t flags,
839 unsigned long deadline);
841 /*!\brief Set compressed data output buffer
843 * Sets the buffer that the codec should output the compressed data
844 * into. This call effectively sets the buffer pointer returned in the
845 * next VPX_CODEC_CX_FRAME_PKT packet. Subsequent packets will be
846 * appended into this buffer. The buffer is preserved across frames,
847 * so applications must periodically call this function after flushing
848 * the accumulated compressed data to disk or to the network to reset
849 * the pointer to the buffer's head.
851 * `pad_before` bytes will be skipped before writing the compressed
852 * data, and `pad_after` bytes will be appended to the packet. The size
853 * of the packet will be the sum of the size of the actual compressed
854 * data, pad_before, and pad_after. The padding bytes will be preserved
857 * Note that calling this function does not guarantee that the returned
858 * compressed data will be placed into the specified buffer. In the
859 * event that the encoded data will not fit into the buffer provided,
860 * the returned packet \ref MAY point to an internal buffer, as it would
861 * if this call were never used. In this event, the output packet will
862 * NOT have any padding, and the application must free space and copy it
863 * to the proper place. This is of particular note in configurations
864 * that may output multiple packets for a single encoded frame (e.g., lagged
865 * encoding) or if the application does not reset the buffer periodically.
867 * Applications may restore the default behavior of the codec providing
868 * the compressed data buffer by calling this function with a NULL
871 * Applications \ref MUSTNOT call this function during iteration of
872 * vpx_codec_get_cx_data().
874 * \param[in] ctx Pointer to this instance's context
875 * \param[in] buf Buffer to store compressed data into
876 * \param[in] pad_before Bytes to skip before writing compressed data
877 * \param[in] pad_after Bytes to skip after writing compressed data
879 * \retval #VPX_CODEC_OK
880 * The buffer was set successfully.
881 * \retval #VPX_CODEC_INVALID_PARAM
882 * A parameter was NULL, the image format is unsupported, etc.
884 vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t *ctx,
885 const vpx_fixed_buf_t *buf,
886 unsigned int pad_before,
887 unsigned int pad_after);
890 /*!\brief Encoded data iterator
892 * Iterates over a list of data packets to be passed from the encoder to the
893 * application. The different kinds of packets available are enumerated in
894 * #vpx_codec_cx_pkt_kind.
896 * #VPX_CODEC_CX_FRAME_PKT packets should be passed to the application's
897 * muxer. Multiple compressed frames may be in the list.
898 * #VPX_CODEC_STATS_PKT packets should be appended to a global buffer.
900 * The application \ref MUST silently ignore any packet kinds that it does
901 * not recognize or support.
903 * The data buffers returned from this function are only guaranteed to be
904 * valid until the application makes another call to any vpx_codec_* function.
906 * \param[in] ctx Pointer to this instance's context
907 * \param[in,out] iter Iterator storage, initialized to NULL
909 * \return Returns a pointer to an output data packet (compressed frame data,
910 * two-pass statistics, etc.) or NULL to signal end-of-list.
913 const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx,
914 vpx_codec_iter_t *iter);
917 /*!\brief Get Preview Frame
919 * Returns an image that can be used as a preview. Shows the image as it would
920 * exist at the decompressor. The application \ref MUST NOT write into this
923 * \param[in] ctx Pointer to this instance's context
925 * \return Returns a pointer to a preview image, or NULL if no image is
929 const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t *ctx);
932 /*!@} - end defgroup encoder*/
936 #endif // VPX_VPX_ENCODER_H_