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 11 12 /*!\defgroup encoder Encoder Algorithm Interface 13 * \ingroup codec 14 * This abstraction allows applications using this encoder to easily support 15 * multiple video formats with minimal code duplication. This section describes 16 * the interface common to all encoders. 17 * @{ 18 */ 19 20 /*!\file 21 * \brief Describes the encoder algorithm interface to applications. 22 * 23 * This file describes the interface between an application and a 24 * video encoder algorithm. 25 * 26 */ 27 #ifdef __cplusplus 28 extern "C" { 29 #endif 30 31 #ifndef VPX_ENCODER_H 32 #define VPX_ENCODER_H 33 #include "vpx_codec.h" 34 35 /*! Temporal Scalability: Maximum length of the sequence defining frame 36 * layer membership 37 */ 38 #define VPX_TS_MAX_PERIODICITY 16 39 40 /*! Temporal Scalability: Maximum number of coding layers */ 41 #define VPX_TS_MAX_LAYERS 5 42 43 /*!\deprecated Use #VPX_TS_MAX_PERIODICITY instead. */ 44 #define MAX_PERIODICITY VPX_TS_MAX_PERIODICITY 45 46 /*!\deprecated Use #VPX_TS_MAX_LAYERS instead. */ 47 #define MAX_LAYERS VPX_TS_MAX_LAYERS 48 49 /*! Spatial Scalability: Maximum number of coding layers */ 50 #define VPX_SS_MAX_LAYERS 5 51 52 /*! Spatial Scalability: Default number of coding layers */ 53 #define VPX_SS_DEFAULT_LAYERS 3 54 55 /*!\brief Current ABI version number 56 * 57 * \internal 58 * If this file is altered in any way that changes the ABI, this value 59 * must be bumped. Examples include, but are not limited to, changing 60 * types, removing or reassigning enums, adding/removing/rearranging 61 * fields to structures 62 */ 63 #define VPX_ENCODER_ABI_VERSION (3 + VPX_CODEC_ABI_VERSION) /**<\hideinitializer*/ 64 65 66 /*! \brief Encoder capabilities bitfield 67 * 68 * Each encoder advertises the capabilities it supports as part of its 69 * ::vpx_codec_iface_t interface structure. Capabilities are extra 70 * interfaces or functionality, and are not required to be supported 71 * by an encoder. 72 * 73 * The available flags are specified by VPX_CODEC_CAP_* defines. 74 */ 75 #define VPX_CODEC_CAP_PSNR 0x10000 /**< Can issue PSNR packets */ 76 77 /*! Can output one partition at a time. Each partition is returned in its 78 * own VPX_CODEC_CX_FRAME_PKT, with the FRAME_IS_FRAGMENT flag set for 79 * every partition but the last. In this mode all frames are always 80 * returned partition by partition. 81 */ 82 #define VPX_CODEC_CAP_OUTPUT_PARTITION 0x20000 83 84 85 /*! \brief Initialization-time Feature Enabling 86 * 87 * Certain codec features must be known at initialization time, to allow 88 * for proper memory allocation. 89 * 90 * The available flags are specified by VPX_CODEC_USE_* defines. 91 */ 92 #define VPX_CODEC_USE_PSNR 0x10000 /**< Calculate PSNR on each frame */ 93 #define VPX_CODEC_USE_OUTPUT_PARTITION 0x20000 /**< Make the encoder output one 94 partition at a time. */ 95 96 97 /*!\brief Generic fixed size buffer structure 98 * 99 * This structure is able to hold a reference to any fixed size buffer. 100 */ 101 typedef struct vpx_fixed_buf { 102 void *buf; /**< Pointer to the data */ 103 size_t sz; /**< Length of the buffer, in chars */ 104 } vpx_fixed_buf_t; /**< alias for struct vpx_fixed_buf */ 105 106 107 /*!\brief Time Stamp Type 108 * 109 * An integer, which when multiplied by the stream's time base, provides 110 * the absolute time of a sample. 111 */ 112 typedef int64_t vpx_codec_pts_t; 113 114 115 /*!\brief Compressed Frame Flags 116 * 117 * This type represents a bitfield containing information about a compressed 118 * frame that may be useful to an application. The most significant 16 bits 119 * can be used by an algorithm to provide additional detail, for example to 120 * support frame types that are codec specific (MPEG-1 D-frames for example) 121 */ 122 typedef uint32_t vpx_codec_frame_flags_t; 123 #define VPX_FRAME_IS_KEY 0x1 /**< frame is the start of a GOP */ 124 #define VPX_FRAME_IS_DROPPABLE 0x2 /**< frame can be dropped without affecting 125 the stream (no future frame depends on 126 this one) */ 127 #define VPX_FRAME_IS_INVISIBLE 0x4 /**< frame should be decoded but will not 128 be shown */ 129 #define VPX_FRAME_IS_FRAGMENT 0x8 /**< this is a fragment of the encoded 130 frame */ 131 132 /*!\brief Error Resilient flags 133 * 134 * These flags define which error resilient features to enable in the 135 * encoder. The flags are specified through the 136 * vpx_codec_enc_cfg::g_error_resilient variable. 137 */ 138 typedef uint32_t vpx_codec_er_flags_t; 139 #define VPX_ERROR_RESILIENT_DEFAULT 0x1 /**< Improve resiliency against 140 losses of whole frames */ 141 #define VPX_ERROR_RESILIENT_PARTITIONS 0x2 /**< The frame partitions are 142 independently decodable by the 143 bool decoder, meaning that 144 partitions can be decoded even 145 though earlier partitions have 146 been lost. Note that intra 147 predicition is still done over 148 the partition boundary. */ 149 150 /*!\brief Encoder output packet variants 151 * 152 * This enumeration lists the different kinds of data packets that can be 153 * returned by calls to vpx_codec_get_cx_data(). Algorithms \ref MAY 154 * extend this list to provide additional functionality. 155 */ 156 enum vpx_codec_cx_pkt_kind { 157 VPX_CODEC_CX_FRAME_PKT, /**< Compressed video frame */ 158 VPX_CODEC_STATS_PKT, /**< Two-pass statistics for this frame */ 159 VPX_CODEC_PSNR_PKT, /**< PSNR statistics for this frame */ 160 VPX_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions */ 161 }; 162 163 164 /*!\brief Encoder output packet 165 * 166 * This structure contains the different kinds of output data the encoder 167 * may produce while compressing a frame. 168 */ 169 typedef struct vpx_codec_cx_pkt { 170 enum vpx_codec_cx_pkt_kind kind; /**< packet variant */ 171 union { 172 struct { 173 void *buf; /**< compressed data buffer */ 174 size_t sz; /**< length of compressed data */ 175 vpx_codec_pts_t pts; /**< time stamp to show frame 176 (in timebase units) */ 177 unsigned long duration; /**< duration to show frame 178 (in timebase units) */ 179 vpx_codec_frame_flags_t flags; /**< flags for this frame */ 180 int partition_id; /**< the partition id 181 defines the decoding order 182 of the partitions. Only 183 applicable when "output partition" 184 mode is enabled. First partition 185 has id 0.*/ 186 187 } frame; /**< data for compressed frame packet */ 188 struct vpx_fixed_buf twopass_stats; /**< data for two-pass packet */ 189 struct vpx_psnr_pkt { 190 unsigned int samples[4]; /**< Number of samples, total/y/u/v */ 191 uint64_t sse[4]; /**< sum squared error, total/y/u/v */ 192 double psnr[4]; /**< PSNR, total/y/u/v */ 193 } psnr; /**< data for PSNR packet */ 194 struct vpx_fixed_buf raw; /**< data for arbitrary packets */ 195 196 /* This packet size is fixed to allow codecs to extend this 197 * interface without having to manage storage for raw packets, 198 * i.e., if it's smaller than 128 bytes, you can store in the 199 * packet list directly. 200 */ 201 char pad[128 - sizeof(enum vpx_codec_cx_pkt_kind)]; /**< fixed sz */ 202 } data; /**< packet data */ 203 } vpx_codec_cx_pkt_t; /**< alias for struct vpx_codec_cx_pkt */ 204 205 206 /*!\brief Rational Number 207 * 208 * This structure holds a fractional value. 209 */ 210 typedef struct vpx_rational { 211 int num; /**< fraction numerator */ 212 int den; /**< fraction denominator */ 213 } vpx_rational_t; /**< alias for struct vpx_rational */ 214 215 216 /*!\brief Multi-pass Encoding Pass */ 217 enum vpx_enc_pass { 218 VPX_RC_ONE_PASS, /**< Single pass mode */ 219 VPX_RC_FIRST_PASS, /**< First pass of multi-pass mode */ 220 VPX_RC_LAST_PASS /**< Final pass of multi-pass mode */ 221 }; 222 223 224 /*!\brief Rate control mode */ 225 enum vpx_rc_mode { 226 VPX_VBR, /**< Variable Bit Rate (VBR) mode */ 227 VPX_CBR, /**< Constant Bit Rate (CBR) mode */ 228 VPX_CQ, /**< Constrained Quality (CQ) mode */ 229 VPX_Q, /**< Constant Quality (Q) mode */ 230 }; 231 232 233 /*!\brief Keyframe placement mode. 234 * 235 * This enumeration determines whether keyframes are placed automatically by 236 * the encoder or whether this behavior is disabled. Older releases of this 237 * SDK were implemented such that VPX_KF_FIXED meant keyframes were disabled. 238 * This name is confusing for this behavior, so the new symbols to be used 239 * are VPX_KF_AUTO and VPX_KF_DISABLED. 240 */ 241 enum vpx_kf_mode { 242 VPX_KF_FIXED, /**< deprecated, implies VPX_KF_DISABLED */ 243 VPX_KF_AUTO, /**< Encoder determines optimal placement automatically */ 244 VPX_KF_DISABLED = 0 /**< Encoder does not place keyframes. */ 245 }; 246 247 248 /*!\brief Encoded Frame Flags 249 * 250 * This type indicates a bitfield to be passed to vpx_codec_encode(), defining 251 * per-frame boolean values. By convention, bits common to all codecs will be 252 * named VPX_EFLAG_*, and bits specific to an algorithm will be named 253 * /algo/_eflag_*. The lower order 16 bits are reserved for common use. 254 */ 255 typedef long vpx_enc_frame_flags_t; 256 #define VPX_EFLAG_FORCE_KF (1<<0) /**< Force this frame to be a keyframe */ 257 258 259 /*!\brief Encoder configuration structure 260 * 261 * This structure contains the encoder settings that have common representations 262 * across all codecs. This doesn't imply that all codecs support all features, 263 * however. 264 */ 265 typedef struct vpx_codec_enc_cfg { 266 /* 267 * generic settings (g) 268 */ 269 270 /*!\brief Algorithm specific "usage" value 271 * 272 * Algorithms may define multiple values for usage, which may convey the 273 * intent of how the application intends to use the stream. If this value 274 * is non-zero, consult the documentation for the codec to determine its 275 * meaning. 276 */ 277 unsigned int g_usage; 278 279 280 /*!\brief Maximum number of threads to use 281 * 282 * For multi-threaded implementations, use no more than this number of 283 * threads. The codec may use fewer threads than allowed. The value 284 * 0 is equivalent to the value 1. 285 */ 286 unsigned int g_threads; 287 288 289 /*!\brief Bitstream profile to use 290 * 291 * Some codecs support a notion of multiple bitstream profiles. Typically 292 * this maps to a set of features that are turned on or off. Often the 293 * profile to use is determined by the features of the intended decoder. 294 * Consult the documentation for the codec to determine the valid values 295 * for this parameter, or set to zero for a sane default. 296 */ 297 unsigned int g_profile; /**< profile of bitstream to use */ 298 299 300 301 /*!\brief Width of the frame 302 * 303 * This value identifies the presentation resolution of the frame, 304 * in pixels. Note that the frames passed as input to the encoder must 305 * have this resolution. Frames will be presented by the decoder in this 306 * resolution, independent of any spatial resampling the encoder may do. 307 */ 308 unsigned int g_w; 309 310 311 /*!\brief Height of the frame 312 * 313 * This value identifies the presentation resolution of the frame, 314 * in pixels. Note that the frames passed as input to the encoder must 315 * have this resolution. Frames will be presented by the decoder in this 316 * resolution, independent of any spatial resampling the encoder may do. 317 */ 318 unsigned int g_h; 319 320 321 /*!\brief Stream timebase units 322 * 323 * Indicates the smallest interval of time, in seconds, used by the stream. 324 * For fixed frame rate material, or variable frame rate material where 325 * frames are timed at a multiple of a given clock (ex: video capture), 326 * the \ref RECOMMENDED method is to set the timebase to the reciprocal 327 * of the frame rate (ex: 1001/30000 for 29.970 Hz NTSC). This allows the 328 * pts to correspond to the frame number, which can be handy. For 329 * re-encoding video from containers with absolute time timestamps, the 330 * \ref RECOMMENDED method is to set the timebase to that of the parent 331 * container or multimedia framework (ex: 1/1000 for ms, as in FLV). 332 */ 333 struct vpx_rational g_timebase; 334 335 336 /*!\brief Enable error resilient modes. 337 * 338 * The error resilient bitfield indicates to the encoder which features 339 * it should enable to take measures for streaming over lossy or noisy 340 * links. 341 */ 342 vpx_codec_er_flags_t g_error_resilient; 343 344 345 /*!\brief Multi-pass Encoding Mode 346 * 347 * This value should be set to the current phase for multi-pass encoding. 348 * For single pass, set to #VPX_RC_ONE_PASS. 349 */ 350 enum vpx_enc_pass g_pass; 351 352 353 /*!\brief Allow lagged encoding 354 * 355 * If set, this value allows the encoder to consume a number of input 356 * frames before producing output frames. This allows the encoder to 357 * base decisions for the current frame on future frames. This does 358 * increase the latency of the encoding pipeline, so it is not appropriate 359 * in all situations (ex: realtime encoding). 360 * 361 * Note that this is a maximum value -- the encoder may produce frames 362 * sooner than the given limit. Set this value to 0 to disable this 363 * feature. 364 */ 365 unsigned int g_lag_in_frames; 366 367 368 /* 369 * rate control settings (rc) 370 */ 371 372 /*!\brief Temporal resampling configuration, if supported by the codec. 373 * 374 * Temporal resampling allows the codec to "drop" frames as a strategy to 375 * meet its target data rate. This can cause temporal discontinuities in 376 * the encoded video, which may appear as stuttering during playback. This 377 * trade-off is often acceptable, but for many applications is not. It can 378 * be disabled in these cases. 379 * 380 * Note that not all codecs support this feature. All vpx VPx codecs do. 381 * For other codecs, consult the documentation for that algorithm. 382 * 383 * This threshold is described as a percentage of the target data buffer. 384 * When the data buffer falls below this percentage of fullness, a 385 * dropped frame is indicated. Set the threshold to zero (0) to disable 386 * this feature. 387 */ 388 unsigned int rc_dropframe_thresh; 389 390 391 /*!\brief Enable/disable spatial resampling, if supported by the codec. 392 * 393 * Spatial resampling allows the codec to compress a lower resolution 394 * version of the frame, which is then upscaled by the encoder to the 395 * correct presentation resolution. This increases visual quality at 396 * low data rates, at the expense of CPU time on the encoder/decoder. 397 */ 398 unsigned int rc_resize_allowed; 399 400 401 /*!\brief Spatial resampling up watermark. 402 * 403 * This threshold is described as a percentage of the target data buffer. 404 * When the data buffer rises above this percentage of fullness, the 405 * encoder will step up to a higher resolution version of the frame. 406 */ 407 unsigned int rc_resize_up_thresh; 408 409 410 /*!\brief Spatial resampling down watermark. 411 * 412 * This threshold is described as a percentage of the target data buffer. 413 * When the data buffer falls below this percentage of fullness, the 414 * encoder will step down to a lower resolution version of the frame. 415 */ 416 unsigned int rc_resize_down_thresh; 417 418 419 /*!\brief Rate control algorithm to use. 420 * 421 * Indicates whether the end usage of this stream is to be streamed over 422 * a bandwidth constrained link, indicating that Constant Bit Rate (CBR) 423 * mode should be used, or whether it will be played back on a high 424 * bandwidth link, as from a local disk, where higher variations in 425 * bitrate are acceptable. 426 */ 427 enum vpx_rc_mode rc_end_usage; 428 429 430 /*!\brief Two-pass stats buffer. 431 * 432 * A buffer containing all of the stats packets produced in the first 433 * pass, concatenated. 434 */ 435 struct vpx_fixed_buf rc_twopass_stats_in; 436 437 438 /*!\brief Target data rate 439 * 440 * Target bandwidth to use for this stream, in kilobits per second. 441 */ 442 unsigned int rc_target_bitrate; 443 444 445 /* 446 * quantizer settings 447 */ 448 449 450 /*!\brief Minimum (Best Quality) Quantizer 451 * 452 * The quantizer is the most direct control over the quality of the 453 * encoded image. The range of valid values for the quantizer is codec 454 * specific. Consult the documentation for the codec to determine the 455 * values to use. To determine the range programmatically, call 456 * vpx_codec_enc_config_default() with a usage value of 0. 457 */ 458 unsigned int rc_min_quantizer; 459 460 461 /*!\brief Maximum (Worst Quality) Quantizer 462 * 463 * The quantizer is the most direct control over the quality of the 464 * encoded image. The range of valid values for the quantizer is codec 465 * specific. Consult the documentation for the codec to determine the 466 * values to use. To determine the range programmatically, call 467 * vpx_codec_enc_config_default() with a usage value of 0. 468 */ 469 unsigned int rc_max_quantizer; 470 471 472 /* 473 * bitrate tolerance 474 */ 475 476 477 /*!\brief Rate control adaptation undershoot control 478 * 479 * This value, expressed as a percentage of the target bitrate, 480 * controls the maximum allowed adaptation speed of the codec. 481 * This factor controls the maximum amount of bits that can 482 * be subtracted from the target bitrate in order to compensate 483 * for prior overshoot. 484 * 485 * Valid values in the range 0-1000. 486 */ 487 unsigned int rc_undershoot_pct; 488 489 490 /*!\brief Rate control adaptation overshoot control 491 * 492 * This value, expressed as a percentage of the target bitrate, 493 * controls the maximum allowed adaptation speed of the codec. 494 * This factor controls the maximum amount of bits that can 495 * be added to the target bitrate in order to compensate for 496 * prior undershoot. 497 * 498 * Valid values in the range 0-1000. 499 */ 500 unsigned int rc_overshoot_pct; 501 502 503 /* 504 * decoder buffer model parameters 505 */ 506 507 508 /*!\brief Decoder Buffer Size 509 * 510 * This value indicates the amount of data that may be buffered by the 511 * decoding application. Note that this value is expressed in units of 512 * time (milliseconds). For example, a value of 5000 indicates that the 513 * client will buffer (at least) 5000ms worth of encoded data. Use the 514 * target bitrate (#rc_target_bitrate) to convert to bits/bytes, if 515 * necessary. 516 */ 517 unsigned int rc_buf_sz; 518 519 520 /*!\brief Decoder Buffer Initial Size 521 * 522 * This value indicates the amount of data that will be buffered by the 523 * decoding application prior to beginning playback. This value is 524 * expressed in units of time (milliseconds). Use the target bitrate 525 * (#rc_target_bitrate) to convert to bits/bytes, if necessary. 526 */ 527 unsigned int rc_buf_initial_sz; 528 529 530 /*!\brief Decoder Buffer Optimal Size 531 * 532 * This value indicates the amount of data that the encoder should try 533 * to maintain in the decoder's buffer. This value is expressed in units 534 * of time (milliseconds). Use the target bitrate (#rc_target_bitrate) 535 * to convert to bits/bytes, if necessary. 536 */ 537 unsigned int rc_buf_optimal_sz; 538 539 540 /* 541 * 2 pass rate control parameters 542 */ 543 544 545 /*!\brief Two-pass mode CBR/VBR bias 546 * 547 * Bias, expressed on a scale of 0 to 100, for determining target size 548 * for the current frame. The value 0 indicates the optimal CBR mode 549 * value should be used. The value 100 indicates the optimal VBR mode 550 * value should be used. Values in between indicate which way the 551 * encoder should "lean." 552 */ 553 unsigned int rc_2pass_vbr_bias_pct; /**< RC mode bias between CBR and VBR(0-100: 0->CBR, 100->VBR) */ 554 555 556 /*!\brief Two-pass mode per-GOP minimum bitrate 557 * 558 * This value, expressed as a percentage of the target bitrate, indicates 559 * the minimum bitrate to be used for a single GOP (aka "section") 560 */ 561 unsigned int rc_2pass_vbr_minsection_pct; 562 563 564 /*!\brief Two-pass mode per-GOP maximum bitrate 565 * 566 * This value, expressed as a percentage of the target bitrate, indicates 567 * the maximum bitrate to be used for a single GOP (aka "section") 568 */ 569 unsigned int rc_2pass_vbr_maxsection_pct; 570 571 572 /* 573 * keyframing settings (kf) 574 */ 575 576 /*!\brief Keyframe placement mode 577 * 578 * This value indicates whether the encoder should place keyframes at a 579 * fixed interval, or determine the optimal placement automatically 580 * (as governed by the #kf_min_dist and #kf_max_dist parameters) 581 */ 582 enum vpx_kf_mode kf_mode; 583 584 585 /*!\brief Keyframe minimum interval 586 * 587 * This value, expressed as a number of frames, prevents the encoder from 588 * placing a keyframe nearer than kf_min_dist to the previous keyframe. At 589 * least kf_min_dist frames non-keyframes will be coded before the next 590 * keyframe. Set kf_min_dist equal to kf_max_dist for a fixed interval. 591 */ 592 unsigned int kf_min_dist; 593 594 595 /*!\brief Keyframe maximum interval 596 * 597 * This value, expressed as a number of frames, forces the encoder to code 598 * a keyframe if one has not been coded in the last kf_max_dist frames. 599 * A value of 0 implies all frames will be keyframes. Set kf_min_dist 600 * equal to kf_max_dist for a fixed interval. 601 */ 602 unsigned int kf_max_dist; 603 604 /* 605 * Spatial scalability settings (ss) 606 */ 607 608 /*!\brief Number of coding layers (spatial) 609 * 610 * This value specifies the number of coding layers to be used. 611 */ 612 unsigned int ss_number_layers; 613 614 /*!\brief Number of coding layers 615 * 616 * This value specifies the number of coding layers to be used. 617 */ 618 unsigned int ts_number_layers; 619 620 /*!\brief Target bitrate for each layer 621 * 622 * These values specify the target coding bitrate for each coding layer. 623 */ 624 unsigned int ts_target_bitrate[VPX_TS_MAX_LAYERS]; 625 626 /*!\brief Frame rate decimation factor for each layer 627 * 628 * These values specify the frame rate decimation factors to apply 629 * to each layer. 630 */ 631 unsigned int ts_rate_decimator[VPX_TS_MAX_LAYERS]; 632 633 /*!\brief Length of the sequence defining frame layer membership 634 * 635 * This value specifies the length of the sequence that defines the 636 * membership of frames to layers. For example, if ts_periodicity=8 then 637 * frames are assigned to coding layers with a repeated sequence of 638 * length 8. 639 */ 640 unsigned int ts_periodicity; 641 642 /*!\brief Template defining the membership of frames to coding layers 643 * 644 * This array defines the membership of frames to coding layers. For a 645 * 2-layer encoding that assigns even numbered frames to one layer (0) 646 * and odd numbered frames to a second layer (1) with ts_periodicity=8, 647 * then ts_layer_id = (0,1,0,1,0,1,0,1). 648 */ 649 unsigned int ts_layer_id[VPX_TS_MAX_PERIODICITY]; 650 } vpx_codec_enc_cfg_t; /**< alias for struct vpx_codec_enc_cfg */ 651 652 653 /*!\brief Initialize an encoder instance 654 * 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. 659 * 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. 663 * 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. 667 * 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. 678 */ 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, 683 int ver); 684 685 686 /*!\brief Convenience macro for vpx_codec_enc_init_ver() 687 * 688 * Ensures the ABI version parameter is properly set. 689 */ 690 #define vpx_codec_enc_init(ctx, iface, cfg, flags) \ 691 vpx_codec_enc_init_ver(ctx, iface, cfg, flags, VPX_ENCODER_ABI_VERSION) 692 693 694 /*!\brief Initialize multi-encoder instance 695 * 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. 700 * 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. 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] 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. 717 */ 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, 721 int num_enc, 722 vpx_codec_flags_t flags, 723 vpx_rational_t *dsf, 724 int ver); 725 726 727 /*!\brief Convenience macro for vpx_codec_enc_init_multi_ver() 728 * 729 * Ensures the ABI version parameter is properly set. 730 */ 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) 734 735 736 /*!\brief Get a default configuration 737 * 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. 743 * 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. 747 * 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. 754 */ 755 vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface, 756 vpx_codec_enc_cfg_t *cfg, 757 unsigned int usage); 758 759 760 /*!\brief Set or change configuration 761 * 762 * Reconfigures an encoder instance according to the given configuration. 763 * 764 * \param[in] ctx Pointer to this instance's context 765 * \param[in] cfg Configuration buffer to use 766 * 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. 773 */ 774 vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx, 775 const vpx_codec_enc_cfg_t *cfg); 776 777 778 /*!\brief Get global stream headers 779 * 780 * Retrieves a stream level global header packet, if supported by the codec. 781 * 782 * \param[in] ctx Pointer to this instance's context 783 * 784 * \retval NULL 785 * Encoder does not support global header 786 * \retval Non-NULL 787 * Pointer to buffer containing global header packet 788 */ 789 vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx); 790 791 792 #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 799 * 800 * Encodes a video frame at the given "presentation time." The presentation 801 * time stamp (PTS) \ref MUST be strictly increasing. 802 * 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. 813 * 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. 819 * 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) 826 * 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. 833 */ 834 vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx, 835 const vpx_image_t *img, 836 vpx_codec_pts_t pts, 837 unsigned long duration, 838 vpx_enc_frame_flags_t flags, 839 unsigned long deadline); 840 841 /*!\brief Set compressed data output buffer 842 * 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. 850 * 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 855 * (not overwritten). 856 * 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. 866 * 867 * Applications may restore the default behavior of the codec providing 868 * the compressed data buffer by calling this function with a NULL 869 * buffer. 870 * 871 * Applications \ref MUSTNOT call this function during iteration of 872 * vpx_codec_get_cx_data(). 873 * 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 878 * 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. 883 */ 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); 888 889 890 /*!\brief Encoded data iterator 891 * 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. 895 * 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. 899 * 900 * The application \ref MUST silently ignore any packet kinds that it does 901 * not recognize or support. 902 * 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. 905 * 906 * \param[in] ctx Pointer to this instance's context 907 * \param[in,out] iter Iterator storage, initialized to NULL 908 * 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. 911 * 912 */ 913 const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx, 914 vpx_codec_iter_t *iter); 915 916 917 /*!\brief Get Preview Frame 918 * 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 921 * image buffer. 922 * 923 * \param[in] ctx Pointer to this instance's context 924 * 925 * \return Returns a pointer to a preview image, or NULL if no image is 926 * available. 927 * 928 */ 929 const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t *ctx); 930 931 932 /*!@} - end defgroup encoder*/ 933 934 #endif 935 #ifdef __cplusplus 936 } 937 #endif 938
