vpx_encoder.h revision ba6c59e9d7d7013b3906b6f4230b663422681848
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_VPX_ENCODER_H_ 11#define VPX_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 29extern "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 1 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_FPMB_STATS_PKT, /**< first pass mb statistics for this frame */ 159 VPX_CODEC_PSNR_PKT, /**< PSNR statistics for this frame */ 160#if CONFIG_SPATIAL_SVC 161 VPX_CODEC_SPATIAL_SVC_LAYER_SIZES, /**< Sizes for each layer in this frame*/ 162#endif 163 VPX_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions */ 164 }; 165 166 167 /*!\brief Encoder output packet 168 * 169 * This structure contains the different kinds of output data the encoder 170 * may produce while compressing a frame. 171 */ 172 typedef struct vpx_codec_cx_pkt { 173 enum vpx_codec_cx_pkt_kind kind; /**< packet variant */ 174 union { 175 struct { 176 void *buf; /**< compressed data buffer */ 177 size_t sz; /**< length of compressed data */ 178 vpx_codec_pts_t pts; /**< time stamp to show frame 179 (in timebase units) */ 180 unsigned long duration; /**< duration to show frame 181 (in timebase units) */ 182 vpx_codec_frame_flags_t flags; /**< flags for this frame */ 183 int partition_id; /**< the partition id 184 defines the decoding order 185 of the partitions. Only 186 applicable when "output partition" 187 mode is enabled. First partition 188 has id 0.*/ 189 190 } frame; /**< data for compressed frame packet */ 191 struct vpx_fixed_buf twopass_stats; /**< data for two-pass packet */ 192 struct vpx_fixed_buf firstpass_mb_stats; /**< first pass mb packet */ 193 struct vpx_psnr_pkt { 194 unsigned int samples[4]; /**< Number of samples, total/y/u/v */ 195 uint64_t sse[4]; /**< sum squared error, total/y/u/v */ 196 double psnr[4]; /**< PSNR, total/y/u/v */ 197 } psnr; /**< data for PSNR packet */ 198 struct vpx_fixed_buf raw; /**< data for arbitrary packets */ 199#if CONFIG_SPATIAL_SVC 200 size_t layer_sizes[VPX_SS_MAX_LAYERS]; 201#endif 202 203 /* This packet size is fixed to allow codecs to extend this 204 * interface without having to manage storage for raw packets, 205 * i.e., if it's smaller than 128 bytes, you can store in the 206 * packet list directly. 207 */ 208 char pad[128 - sizeof(enum vpx_codec_cx_pkt_kind)]; /**< fixed sz */ 209 } data; /**< packet data */ 210 } vpx_codec_cx_pkt_t; /**< alias for struct vpx_codec_cx_pkt */ 211 212 213 /*!\brief Rational Number 214 * 215 * This structure holds a fractional value. 216 */ 217 typedef struct vpx_rational { 218 int num; /**< fraction numerator */ 219 int den; /**< fraction denominator */ 220 } vpx_rational_t; /**< alias for struct vpx_rational */ 221 222 223 /*!\brief Multi-pass Encoding Pass */ 224 enum vpx_enc_pass { 225 VPX_RC_ONE_PASS, /**< Single pass mode */ 226 VPX_RC_FIRST_PASS, /**< First pass of multi-pass mode */ 227 VPX_RC_LAST_PASS /**< Final pass of multi-pass mode */ 228 }; 229 230 231 /*!\brief Rate control mode */ 232 enum vpx_rc_mode { 233 VPX_VBR, /**< Variable Bit Rate (VBR) mode */ 234 VPX_CBR, /**< Constant Bit Rate (CBR) mode */ 235 VPX_CQ, /**< Constrained Quality (CQ) mode */ 236 VPX_Q, /**< Constant Quality (Q) mode */ 237 }; 238 239 240 /*!\brief Keyframe placement mode. 241 * 242 * This enumeration determines whether keyframes are placed automatically by 243 * the encoder or whether this behavior is disabled. Older releases of this 244 * SDK were implemented such that VPX_KF_FIXED meant keyframes were disabled. 245 * This name is confusing for this behavior, so the new symbols to be used 246 * are VPX_KF_AUTO and VPX_KF_DISABLED. 247 */ 248 enum vpx_kf_mode { 249 VPX_KF_FIXED, /**< deprecated, implies VPX_KF_DISABLED */ 250 VPX_KF_AUTO, /**< Encoder determines optimal placement automatically */ 251 VPX_KF_DISABLED = 0 /**< Encoder does not place keyframes. */ 252 }; 253 254 255 /*!\brief Encoded Frame Flags 256 * 257 * This type indicates a bitfield to be passed to vpx_codec_encode(), defining 258 * per-frame boolean values. By convention, bits common to all codecs will be 259 * named VPX_EFLAG_*, and bits specific to an algorithm will be named 260 * /algo/_eflag_*. The lower order 16 bits are reserved for common use. 261 */ 262 typedef long vpx_enc_frame_flags_t; 263#define VPX_EFLAG_FORCE_KF (1<<0) /**< Force this frame to be a keyframe */ 264 265 266 /*!\brief Encoder configuration structure 267 * 268 * This structure contains the encoder settings that have common representations 269 * across all codecs. This doesn't imply that all codecs support all features, 270 * however. 271 */ 272 typedef struct vpx_codec_enc_cfg { 273 /* 274 * generic settings (g) 275 */ 276 277 /*!\brief Algorithm specific "usage" value 278 * 279 * Algorithms may define multiple values for usage, which may convey the 280 * intent of how the application intends to use the stream. If this value 281 * is non-zero, consult the documentation for the codec to determine its 282 * meaning. 283 */ 284 unsigned int g_usage; 285 286 287 /*!\brief Maximum number of threads to use 288 * 289 * For multi-threaded implementations, use no more than this number of 290 * threads. The codec may use fewer threads than allowed. The value 291 * 0 is equivalent to the value 1. 292 */ 293 unsigned int g_threads; 294 295 296 /*!\brief Bitstream profile to use 297 * 298 * Some codecs support a notion of multiple bitstream profiles. Typically 299 * this maps to a set of features that are turned on or off. Often the 300 * profile to use is determined by the features of the intended decoder. 301 * Consult the documentation for the codec to determine the valid values 302 * for this parameter, or set to zero for a sane default. 303 */ 304 unsigned int g_profile; /**< profile of bitstream to use */ 305 306 307 308 /*!\brief Width of the frame 309 * 310 * This value identifies the presentation resolution of the frame, 311 * in pixels. Note that the frames passed as input to the encoder must 312 * have this resolution. Frames will be presented by the decoder in this 313 * resolution, independent of any spatial resampling the encoder may do. 314 */ 315 unsigned int g_w; 316 317 318 /*!\brief Height of the frame 319 * 320 * This value identifies the presentation resolution of the frame, 321 * in pixels. Note that the frames passed as input to the encoder must 322 * have this resolution. Frames will be presented by the decoder in this 323 * resolution, independent of any spatial resampling the encoder may do. 324 */ 325 unsigned int g_h; 326 327 328 /*!\brief Stream timebase units 329 * 330 * Indicates the smallest interval of time, in seconds, used by the stream. 331 * For fixed frame rate material, or variable frame rate material where 332 * frames are timed at a multiple of a given clock (ex: video capture), 333 * the \ref RECOMMENDED method is to set the timebase to the reciprocal 334 * of the frame rate (ex: 1001/30000 for 29.970 Hz NTSC). This allows the 335 * pts to correspond to the frame number, which can be handy. For 336 * re-encoding video from containers with absolute time timestamps, the 337 * \ref RECOMMENDED method is to set the timebase to that of the parent 338 * container or multimedia framework (ex: 1/1000 for ms, as in FLV). 339 */ 340 struct vpx_rational g_timebase; 341 342 343 /*!\brief Enable error resilient modes. 344 * 345 * The error resilient bitfield indicates to the encoder which features 346 * it should enable to take measures for streaming over lossy or noisy 347 * links. 348 */ 349 vpx_codec_er_flags_t g_error_resilient; 350 351 352 /*!\brief Multi-pass Encoding Mode 353 * 354 * This value should be set to the current phase for multi-pass encoding. 355 * For single pass, set to #VPX_RC_ONE_PASS. 356 */ 357 enum vpx_enc_pass g_pass; 358 359 360 /*!\brief Allow lagged encoding 361 * 362 * If set, this value allows the encoder to consume a number of input 363 * frames before producing output frames. This allows the encoder to 364 * base decisions for the current frame on future frames. This does 365 * increase the latency of the encoding pipeline, so it is not appropriate 366 * in all situations (ex: realtime encoding). 367 * 368 * Note that this is a maximum value -- the encoder may produce frames 369 * sooner than the given limit. Set this value to 0 to disable this 370 * feature. 371 */ 372 unsigned int g_lag_in_frames; 373 374 375 /* 376 * rate control settings (rc) 377 */ 378 379 /*!\brief Temporal resampling configuration, if supported by the codec. 380 * 381 * Temporal resampling allows the codec to "drop" frames as a strategy to 382 * meet its target data rate. This can cause temporal discontinuities in 383 * the encoded video, which may appear as stuttering during playback. This 384 * trade-off is often acceptable, but for many applications is not. It can 385 * be disabled in these cases. 386 * 387 * Note that not all codecs support this feature. All vpx VPx codecs do. 388 * For other codecs, consult the documentation for that algorithm. 389 * 390 * This threshold is described as a percentage of the target data buffer. 391 * When the data buffer falls below this percentage of fullness, a 392 * dropped frame is indicated. Set the threshold to zero (0) to disable 393 * this feature. 394 */ 395 unsigned int rc_dropframe_thresh; 396 397 398 /*!\brief Enable/disable spatial resampling, if supported by the codec. 399 * 400 * Spatial resampling allows the codec to compress a lower resolution 401 * version of the frame, which is then upscaled by the encoder to the 402 * correct presentation resolution. This increases visual quality at 403 * low data rates, at the expense of CPU time on the encoder/decoder. 404 */ 405 unsigned int rc_resize_allowed; 406 407 /*!\brief Internal coded frame width. 408 * 409 * If spatial resampling is enabled this specifies the width of the 410 * encoded frame. 411 */ 412 unsigned int rc_scaled_width; 413 414 /*!\brief Internal coded frame height. 415 * 416 * If spatial resampling is enabled this specifies the height of the 417 * encoded frame. 418 */ 419 unsigned int rc_scaled_height; 420 421 /*!\brief Spatial resampling up watermark. 422 * 423 * This threshold is described as a percentage of the target data buffer. 424 * When the data buffer rises above this percentage of fullness, the 425 * encoder will step up to a higher resolution version of the frame. 426 */ 427 unsigned int rc_resize_up_thresh; 428 429 430 /*!\brief Spatial resampling down watermark. 431 * 432 * This threshold is described as a percentage of the target data buffer. 433 * When the data buffer falls below this percentage of fullness, the 434 * encoder will step down to a lower resolution version of the frame. 435 */ 436 unsigned int rc_resize_down_thresh; 437 438 439 /*!\brief Rate control algorithm to use. 440 * 441 * Indicates whether the end usage of this stream is to be streamed over 442 * a bandwidth constrained link, indicating that Constant Bit Rate (CBR) 443 * mode should be used, or whether it will be played back on a high 444 * bandwidth link, as from a local disk, where higher variations in 445 * bitrate are acceptable. 446 */ 447 enum vpx_rc_mode rc_end_usage; 448 449 450 /*!\brief Two-pass stats buffer. 451 * 452 * A buffer containing all of the stats packets produced in the first 453 * pass, concatenated. 454 */ 455 struct vpx_fixed_buf rc_twopass_stats_in; 456 457 /*!\brief first pass mb stats buffer. 458 * 459 * A buffer containing all of the first pass mb stats packets produced 460 * in the first pass, concatenated. 461 */ 462 struct vpx_fixed_buf rc_firstpass_mb_stats_in; 463 464 /*!\brief Target data rate 465 * 466 * Target bandwidth to use for this stream, in kilobits per second. 467 */ 468 unsigned int rc_target_bitrate; 469 470 471 /* 472 * quantizer settings 473 */ 474 475 476 /*!\brief Minimum (Best Quality) Quantizer 477 * 478 * The quantizer is the most direct control over the quality of the 479 * encoded image. The range of valid values for the quantizer is codec 480 * specific. Consult the documentation for the codec to determine the 481 * values to use. To determine the range programmatically, call 482 * vpx_codec_enc_config_default() with a usage value of 0. 483 */ 484 unsigned int rc_min_quantizer; 485 486 487 /*!\brief Maximum (Worst Quality) Quantizer 488 * 489 * The quantizer is the most direct control over the quality of the 490 * encoded image. The range of valid values for the quantizer is codec 491 * specific. Consult the documentation for the codec to determine the 492 * values to use. To determine the range programmatically, call 493 * vpx_codec_enc_config_default() with a usage value of 0. 494 */ 495 unsigned int rc_max_quantizer; 496 497 498 /* 499 * bitrate tolerance 500 */ 501 502 503 /*!\brief Rate control adaptation undershoot control 504 * 505 * This value, expressed as a percentage of the target bitrate, 506 * controls the maximum allowed adaptation speed of the codec. 507 * This factor controls the maximum amount of bits that can 508 * be subtracted from the target bitrate in order to compensate 509 * for prior overshoot. 510 * 511 * Valid values in the range 0-1000. 512 */ 513 unsigned int rc_undershoot_pct; 514 515 516 /*!\brief Rate control adaptation overshoot control 517 * 518 * This value, expressed as a percentage of the target bitrate, 519 * controls the maximum allowed adaptation speed of the codec. 520 * This factor controls the maximum amount of bits that can 521 * be added to the target bitrate in order to compensate for 522 * prior undershoot. 523 * 524 * Valid values in the range 0-1000. 525 */ 526 unsigned int rc_overshoot_pct; 527 528 529 /* 530 * decoder buffer model parameters 531 */ 532 533 534 /*!\brief Decoder Buffer Size 535 * 536 * This value indicates the amount of data that may be buffered by the 537 * decoding application. Note that this value is expressed in units of 538 * time (milliseconds). For example, a value of 5000 indicates that the 539 * client will buffer (at least) 5000ms worth of encoded data. Use the 540 * target bitrate (#rc_target_bitrate) to convert to bits/bytes, if 541 * necessary. 542 */ 543 unsigned int rc_buf_sz; 544 545 546 /*!\brief Decoder Buffer Initial Size 547 * 548 * This value indicates the amount of data that will be buffered by the 549 * decoding application prior to beginning playback. This value is 550 * expressed in units of time (milliseconds). Use the target bitrate 551 * (#rc_target_bitrate) to convert to bits/bytes, if necessary. 552 */ 553 unsigned int rc_buf_initial_sz; 554 555 556 /*!\brief Decoder Buffer Optimal Size 557 * 558 * This value indicates the amount of data that the encoder should try 559 * to maintain in the decoder's buffer. This value is expressed in units 560 * of time (milliseconds). Use the target bitrate (#rc_target_bitrate) 561 * to convert to bits/bytes, if necessary. 562 */ 563 unsigned int rc_buf_optimal_sz; 564 565 566 /* 567 * 2 pass rate control parameters 568 */ 569 570 571 /*!\brief Two-pass mode CBR/VBR bias 572 * 573 * Bias, expressed on a scale of 0 to 100, for determining target size 574 * for the current frame. The value 0 indicates the optimal CBR mode 575 * value should be used. The value 100 indicates the optimal VBR mode 576 * value should be used. Values in between indicate which way the 577 * encoder should "lean." 578 */ 579 unsigned int rc_2pass_vbr_bias_pct; /**< RC mode bias between CBR and VBR(0-100: 0->CBR, 100->VBR) */ 580 581 582 /*!\brief Two-pass mode per-GOP minimum bitrate 583 * 584 * This value, expressed as a percentage of the target bitrate, indicates 585 * the minimum bitrate to be used for a single GOP (aka "section") 586 */ 587 unsigned int rc_2pass_vbr_minsection_pct; 588 589 590 /*!\brief Two-pass mode per-GOP maximum bitrate 591 * 592 * This value, expressed as a percentage of the target bitrate, indicates 593 * the maximum bitrate to be used for a single GOP (aka "section") 594 */ 595 unsigned int rc_2pass_vbr_maxsection_pct; 596 597 598 /* 599 * keyframing settings (kf) 600 */ 601 602 /*!\brief Keyframe placement mode 603 * 604 * This value indicates whether the encoder should place keyframes at a 605 * fixed interval, or determine the optimal placement automatically 606 * (as governed by the #kf_min_dist and #kf_max_dist parameters) 607 */ 608 enum vpx_kf_mode kf_mode; 609 610 611 /*!\brief Keyframe minimum interval 612 * 613 * This value, expressed as a number of frames, prevents the encoder from 614 * placing a keyframe nearer than kf_min_dist to the previous keyframe. At 615 * least kf_min_dist frames non-keyframes will be coded before the next 616 * keyframe. Set kf_min_dist equal to kf_max_dist for a fixed interval. 617 */ 618 unsigned int kf_min_dist; 619 620 621 /*!\brief Keyframe maximum interval 622 * 623 * This value, expressed as a number of frames, forces the encoder to code 624 * a keyframe if one has not been coded in the last kf_max_dist frames. 625 * A value of 0 implies all frames will be keyframes. Set kf_min_dist 626 * equal to kf_max_dist for a fixed interval. 627 */ 628 unsigned int kf_max_dist; 629 630 /* 631 * Spatial scalability settings (ss) 632 */ 633 634 /*!\brief Number of spatial coding layers. 635 * 636 * This value specifies the number of spatial coding layers to be used. 637 */ 638 unsigned int ss_number_layers; 639 640 /*!\brief Enable auto alt reference flags for each spatial layer. 641 * 642 * These values specify if auto alt reference frame is enabled for each 643 * spatial layer. 644 */ 645 int ss_enable_auto_alt_ref[VPX_SS_MAX_LAYERS]; 646 647 /*!\brief Target bitrate for each spatial layer. 648 * 649 * These values specify the target coding bitrate to be used for each 650 * spatial layer. 651 */ 652 unsigned int ss_target_bitrate[VPX_SS_MAX_LAYERS]; 653 654 /*!\brief Number of temporal coding layers. 655 * 656 * This value specifies the number of temporal layers to be used. 657 */ 658 unsigned int ts_number_layers; 659 660 /*!\brief Target bitrate for each temporal layer. 661 * 662 * These values specify the target coding bitrate to be used for each 663 * temporal layer. 664 */ 665 unsigned int ts_target_bitrate[VPX_TS_MAX_LAYERS]; 666 667 /*!\brief Frame rate decimation factor for each temporal layer. 668 * 669 * These values specify the frame rate decimation factors to apply 670 * to each temporal layer. 671 */ 672 unsigned int ts_rate_decimator[VPX_TS_MAX_LAYERS]; 673 674 /*!\brief Length of the sequence defining frame temporal layer membership. 675 * 676 * This value specifies the length of the sequence that defines the 677 * membership of frames to temporal layers. For example, if the 678 * ts_periodicity = 8, then the frames are assigned to coding layers with a 679 * repeated sequence of length 8. 680 */ 681 unsigned int ts_periodicity; 682 683 /*!\brief Template defining the membership of frames to temporal layers. 684 * 685 * This array defines the membership of frames to temporal coding layers. 686 * For a 2-layer encoding that assigns even numbered frames to one temporal 687 * layer (0) and odd numbered frames to a second temporal layer (1) with 688 * ts_periodicity=8, then ts_layer_id = (0,1,0,1,0,1,0,1). 689 */ 690 unsigned int ts_layer_id[VPX_TS_MAX_PERIODICITY]; 691 } vpx_codec_enc_cfg_t; /**< alias for struct vpx_codec_enc_cfg */ 692 693 694 /*!\brief Initialize an encoder instance 695 * 696 * Initializes a encoder context using the given interface. Applications 697 * should call the vpx_codec_enc_init convenience macro instead of this 698 * function directly, to ensure that the ABI version number parameter 699 * is properly initialized. 700 * 701 * If the library was configured with --disable-multithread, this call 702 * is not thread safe and should be guarded with a lock if being used 703 * in a multithreaded context. 704 * 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] flags Bitfield of VPX_CODEC_USE_* flags 709 * \param[in] ver ABI version number. Must be set to 710 * VPX_ENCODER_ABI_VERSION 711 * \retval #VPX_CODEC_OK 712 * The decoder algorithm initialized. 713 * \retval #VPX_CODEC_MEM_ERROR 714 * Memory allocation failed. 715 */ 716 vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t *ctx, 717 vpx_codec_iface_t *iface, 718 vpx_codec_enc_cfg_t *cfg, 719 vpx_codec_flags_t flags, 720 int ver); 721 722 723 /*!\brief Convenience macro for vpx_codec_enc_init_ver() 724 * 725 * Ensures the ABI version parameter is properly set. 726 */ 727#define vpx_codec_enc_init(ctx, iface, cfg, flags) \ 728 vpx_codec_enc_init_ver(ctx, iface, cfg, flags, VPX_ENCODER_ABI_VERSION) 729 730 731 /*!\brief Initialize multi-encoder instance 732 * 733 * Initializes multi-encoder context using the given interface. 734 * Applications should call the vpx_codec_enc_init_multi convenience macro 735 * instead of this function directly, to ensure that the ABI version number 736 * parameter is properly initialized. 737 * 738 * \param[in] ctx Pointer to this instance's context. 739 * \param[in] iface Pointer to the algorithm interface to use. 740 * \param[in] cfg Configuration to use, if known. May be NULL. 741 * \param[in] num_enc Total number of encoders. 742 * \param[in] flags Bitfield of VPX_CODEC_USE_* flags 743 * \param[in] dsf Pointer to down-sampling factors. 744 * \param[in] ver ABI version number. Must be set to 745 * VPX_ENCODER_ABI_VERSION 746 * \retval #VPX_CODEC_OK 747 * The decoder algorithm initialized. 748 * \retval #VPX_CODEC_MEM_ERROR 749 * Memory allocation failed. 750 */ 751 vpx_codec_err_t vpx_codec_enc_init_multi_ver(vpx_codec_ctx_t *ctx, 752 vpx_codec_iface_t *iface, 753 vpx_codec_enc_cfg_t *cfg, 754 int num_enc, 755 vpx_codec_flags_t flags, 756 vpx_rational_t *dsf, 757 int ver); 758 759 760 /*!\brief Convenience macro for vpx_codec_enc_init_multi_ver() 761 * 762 * Ensures the ABI version parameter is properly set. 763 */ 764#define vpx_codec_enc_init_multi(ctx, iface, cfg, num_enc, flags, dsf) \ 765 vpx_codec_enc_init_multi_ver(ctx, iface, cfg, num_enc, flags, dsf, \ 766 VPX_ENCODER_ABI_VERSION) 767 768 769 /*!\brief Get a default configuration 770 * 771 * Initializes a encoder configuration structure with default values. Supports 772 * the notion of "usages" so that an algorithm may offer different default 773 * settings depending on the user's intended goal. This function \ref SHOULD 774 * be called by all applications to initialize the configuration structure 775 * before specializing the configuration with application specific values. 776 * 777 * \param[in] iface Pointer to the algorithm interface to use. 778 * \param[out] cfg Configuration buffer to populate 779 * \param[in] usage End usage. Set to 0 or use codec specific values. 780 * 781 * \retval #VPX_CODEC_OK 782 * The configuration was populated. 783 * \retval #VPX_CODEC_INCAPABLE 784 * Interface is not an encoder interface. 785 * \retval #VPX_CODEC_INVALID_PARAM 786 * A parameter was NULL, or the usage value was not recognized. 787 */ 788 vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface, 789 vpx_codec_enc_cfg_t *cfg, 790 unsigned int usage); 791 792 793 /*!\brief Set or change configuration 794 * 795 * Reconfigures an encoder instance according to the given configuration. 796 * 797 * \param[in] ctx Pointer to this instance's context 798 * \param[in] cfg Configuration buffer to use 799 * 800 * \retval #VPX_CODEC_OK 801 * The configuration was populated. 802 * \retval #VPX_CODEC_INCAPABLE 803 * Interface is not an encoder interface. 804 * \retval #VPX_CODEC_INVALID_PARAM 805 * A parameter was NULL, or the usage value was not recognized. 806 */ 807 vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx, 808 const vpx_codec_enc_cfg_t *cfg); 809 810 811 /*!\brief Get global stream headers 812 * 813 * Retrieves a stream level global header packet, if supported by the codec. 814 * 815 * \param[in] ctx Pointer to this instance's context 816 * 817 * \retval NULL 818 * Encoder does not support global header 819 * \retval Non-NULL 820 * Pointer to buffer containing global header packet 821 */ 822 vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx); 823 824 825#define VPX_DL_REALTIME (1) /**< deadline parameter analogous to 826 * VPx REALTIME mode. */ 827#define VPX_DL_GOOD_QUALITY (1000000) /**< deadline parameter analogous to 828 * VPx GOOD QUALITY mode. */ 829#define VPX_DL_BEST_QUALITY (0) /**< deadline parameter analogous to 830 * VPx BEST QUALITY mode. */ 831 /*!\brief Encode a frame 832 * 833 * Encodes a video frame at the given "presentation time." The presentation 834 * time stamp (PTS) \ref MUST be strictly increasing. 835 * 836 * The encoder supports the notion of a soft real-time deadline. Given a 837 * non-zero value to the deadline parameter, the encoder will make a "best 838 * effort" guarantee to return before the given time slice expires. It is 839 * implicit that limiting the available time to encode will degrade the 840 * output quality. The encoder can be given an unlimited time to produce the 841 * best possible frame by specifying a deadline of '0'. This deadline 842 * supercedes the VPx notion of "best quality, good quality, realtime". 843 * Applications that wish to map these former settings to the new deadline 844 * based system can use the symbols #VPX_DL_REALTIME, #VPX_DL_GOOD_QUALITY, 845 * and #VPX_DL_BEST_QUALITY. 846 * 847 * When the last frame has been passed to the encoder, this function should 848 * continue to be called, with the img parameter set to NULL. This will 849 * signal the end-of-stream condition to the encoder and allow it to encode 850 * any held buffers. Encoding is complete when vpx_codec_encode() is called 851 * and vpx_codec_get_cx_data() returns no data. 852 * 853 * \param[in] ctx Pointer to this instance's context 854 * \param[in] img Image data to encode, NULL to flush. 855 * \param[in] pts Presentation time stamp, in timebase units. 856 * \param[in] duration Duration to show frame, in timebase units. 857 * \param[in] flags Flags to use for encoding this frame. 858 * \param[in] deadline Time to spend encoding, in microseconds. (0=infinite) 859 * 860 * \retval #VPX_CODEC_OK 861 * The configuration was populated. 862 * \retval #VPX_CODEC_INCAPABLE 863 * Interface is not an encoder interface. 864 * \retval #VPX_CODEC_INVALID_PARAM 865 * A parameter was NULL, the image format is unsupported, etc. 866 */ 867 vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx, 868 const vpx_image_t *img, 869 vpx_codec_pts_t pts, 870 unsigned long duration, 871 vpx_enc_frame_flags_t flags, 872 unsigned long deadline); 873 874 /*!\brief Set compressed data output buffer 875 * 876 * Sets the buffer that the codec should output the compressed data 877 * into. This call effectively sets the buffer pointer returned in the 878 * next VPX_CODEC_CX_FRAME_PKT packet. Subsequent packets will be 879 * appended into this buffer. The buffer is preserved across frames, 880 * so applications must periodically call this function after flushing 881 * the accumulated compressed data to disk or to the network to reset 882 * the pointer to the buffer's head. 883 * 884 * `pad_before` bytes will be skipped before writing the compressed 885 * data, and `pad_after` bytes will be appended to the packet. The size 886 * of the packet will be the sum of the size of the actual compressed 887 * data, pad_before, and pad_after. The padding bytes will be preserved 888 * (not overwritten). 889 * 890 * Note that calling this function does not guarantee that the returned 891 * compressed data will be placed into the specified buffer. In the 892 * event that the encoded data will not fit into the buffer provided, 893 * the returned packet \ref MAY point to an internal buffer, as it would 894 * if this call were never used. In this event, the output packet will 895 * NOT have any padding, and the application must free space and copy it 896 * to the proper place. This is of particular note in configurations 897 * that may output multiple packets for a single encoded frame (e.g., lagged 898 * encoding) or if the application does not reset the buffer periodically. 899 * 900 * Applications may restore the default behavior of the codec providing 901 * the compressed data buffer by calling this function with a NULL 902 * buffer. 903 * 904 * Applications \ref MUSTNOT call this function during iteration of 905 * vpx_codec_get_cx_data(). 906 * 907 * \param[in] ctx Pointer to this instance's context 908 * \param[in] buf Buffer to store compressed data into 909 * \param[in] pad_before Bytes to skip before writing compressed data 910 * \param[in] pad_after Bytes to skip after writing compressed data 911 * 912 * \retval #VPX_CODEC_OK 913 * The buffer was set successfully. 914 * \retval #VPX_CODEC_INVALID_PARAM 915 * A parameter was NULL, the image format is unsupported, etc. 916 */ 917 vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t *ctx, 918 const vpx_fixed_buf_t *buf, 919 unsigned int pad_before, 920 unsigned int pad_after); 921 922 923 /*!\brief Encoded data iterator 924 * 925 * Iterates over a list of data packets to be passed from the encoder to the 926 * application. The different kinds of packets available are enumerated in 927 * #vpx_codec_cx_pkt_kind. 928 * 929 * #VPX_CODEC_CX_FRAME_PKT packets should be passed to the application's 930 * muxer. Multiple compressed frames may be in the list. 931 * #VPX_CODEC_STATS_PKT packets should be appended to a global buffer. 932 * 933 * The application \ref MUST silently ignore any packet kinds that it does 934 * not recognize or support. 935 * 936 * The data buffers returned from this function are only guaranteed to be 937 * valid until the application makes another call to any vpx_codec_* function. 938 * 939 * \param[in] ctx Pointer to this instance's context 940 * \param[in,out] iter Iterator storage, initialized to NULL 941 * 942 * \return Returns a pointer to an output data packet (compressed frame data, 943 * two-pass statistics, etc.) or NULL to signal end-of-list. 944 * 945 */ 946 const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx, 947 vpx_codec_iter_t *iter); 948 949 950 /*!\brief Get Preview Frame 951 * 952 * Returns an image that can be used as a preview. Shows the image as it would 953 * exist at the decompressor. The application \ref MUST NOT write into this 954 * image buffer. 955 * 956 * \param[in] ctx Pointer to this instance's context 957 * 958 * \return Returns a pointer to a preview image, or NULL if no image is 959 * available. 960 * 961 */ 962 const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t *ctx); 963 964 965 /*!@} - end defgroup encoder*/ 966#ifdef __cplusplus 967} 968#endif 969#endif // VPX_VPX_ENCODER_H_ 970 971