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 29 extern "C" { 30 #endif 31 32 #include "./vpx_codec.h" 33 34 /*! Temporal Scalability: Maximum length of the sequence defining frame 35 * layer membership 36 */ 37 #define VPX_TS_MAX_PERIODICITY 16 38 39 /*! Temporal Scalability: Maximum number of coding layers */ 40 #define VPX_TS_MAX_LAYERS 5 41 42 /*!\deprecated Use #VPX_TS_MAX_PERIODICITY instead. */ 43 #define MAX_PERIODICITY VPX_TS_MAX_PERIODICITY 44 45 /*!\deprecated Use #VPX_TS_MAX_LAYERS instead. */ 46 #define MAX_LAYERS VPX_TS_MAX_LAYERS 47 48 /*! Spatial Scalability: Maximum number of coding layers */ 49 #define VPX_SS_MAX_LAYERS 5 50 51 /*! Spatial Scalability: Default number of coding layers */ 52 #define VPX_SS_DEFAULT_LAYERS 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 /*! Can support input images at greater than 8 bitdepth. 84 */ 85 #define VPX_CODEC_CAP_HIGHBITDEPTH 0x40000 86 87 /*! \brief Initialization-time Feature Enabling 88 * 89 * Certain codec features must be known at initialization time, to allow 90 * for proper memory allocation. 91 * 92 * The available flags are specified by VPX_CODEC_USE_* defines. 93 */ 94 #define VPX_CODEC_USE_PSNR 0x10000 /**< Calculate PSNR on each frame */ 95 #define VPX_CODEC_USE_OUTPUT_PARTITION 0x20000 /**< Make the encoder output one 96 partition at a time. */ 97 #define VPX_CODEC_USE_HIGHBITDEPTH 0x40000 /**< Use high bitdepth */ 98 99 100 /*!\brief Generic fixed size buffer structure 101 * 102 * This structure is able to hold a reference to any fixed size buffer. 103 */ 104 typedef struct vpx_fixed_buf { 105 void *buf; /**< Pointer to the data */ 106 size_t sz; /**< Length of the buffer, in chars */ 107 } vpx_fixed_buf_t; /**< alias for struct vpx_fixed_buf */ 108 109 110 /*!\brief Time Stamp Type 111 * 112 * An integer, which when multiplied by the stream's time base, provides 113 * the absolute time of a sample. 114 */ 115 typedef int64_t vpx_codec_pts_t; 116 117 118 /*!\brief Compressed Frame Flags 119 * 120 * This type represents a bitfield containing information about a compressed 121 * frame that may be useful to an application. The most significant 16 bits 122 * can be used by an algorithm to provide additional detail, for example to 123 * support frame types that are codec specific (MPEG-1 D-frames for example) 124 */ 125 typedef uint32_t vpx_codec_frame_flags_t; 126 #define VPX_FRAME_IS_KEY 0x1 /**< frame is the start of a GOP */ 127 #define VPX_FRAME_IS_DROPPABLE 0x2 /**< frame can be dropped without affecting 128 the stream (no future frame depends on 129 this one) */ 130 #define VPX_FRAME_IS_INVISIBLE 0x4 /**< frame should be decoded but will not 131 be shown */ 132 #define VPX_FRAME_IS_FRAGMENT 0x8 /**< this is a fragment of the encoded 133 frame */ 134 135 /*!\brief Error Resilient flags 136 * 137 * These flags define which error resilient features to enable in the 138 * encoder. The flags are specified through the 139 * vpx_codec_enc_cfg::g_error_resilient variable. 140 */ 141 typedef uint32_t vpx_codec_er_flags_t; 142 #define VPX_ERROR_RESILIENT_DEFAULT 0x1 /**< Improve resiliency against 143 losses of whole frames */ 144 #define VPX_ERROR_RESILIENT_PARTITIONS 0x2 /**< The frame partitions are 145 independently decodable by the 146 bool decoder, meaning that 147 partitions can be decoded even 148 though earlier partitions have 149 been lost. Note that intra 150 predicition is still done over 151 the partition boundary. */ 152 153 /*!\brief Encoder output packet variants 154 * 155 * This enumeration lists the different kinds of data packets that can be 156 * returned by calls to vpx_codec_get_cx_data(). Algorithms \ref MAY 157 * extend this list to provide additional functionality. 158 */ 159 enum vpx_codec_cx_pkt_kind { 160 VPX_CODEC_CX_FRAME_PKT, /**< Compressed video frame */ 161 VPX_CODEC_STATS_PKT, /**< Two-pass statistics for this frame */ 162 VPX_CODEC_FPMB_STATS_PKT, /**< first pass mb statistics for this frame */ 163 VPX_CODEC_PSNR_PKT, /**< PSNR statistics for this frame */ 164 #if CONFIG_SPATIAL_SVC 165 VPX_CODEC_SPATIAL_SVC_LAYER_SIZES, /**< Sizes for each layer in this frame*/ 166 VPX_CODEC_SPATIAL_SVC_LAYER_PSNR, /**< PSNR for each layer in this frame*/ 167 #endif 168 VPX_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions */ 169 }; 170 171 172 /*!\brief Encoder output packet 173 * 174 * This structure contains the different kinds of output data the encoder 175 * may produce while compressing a frame. 176 */ 177 typedef struct vpx_codec_cx_pkt { 178 enum vpx_codec_cx_pkt_kind kind; /**< packet variant */ 179 union { 180 struct { 181 void *buf; /**< compressed data buffer */ 182 size_t sz; /**< length of compressed data */ 183 vpx_codec_pts_t pts; /**< time stamp to show frame 184 (in timebase units) */ 185 unsigned long duration; /**< duration to show frame 186 (in timebase units) */ 187 vpx_codec_frame_flags_t flags; /**< flags for this frame */ 188 int partition_id; /**< the partition id 189 defines the decoding order 190 of the partitions. Only 191 applicable when "output partition" 192 mode is enabled. First partition 193 has id 0.*/ 194 195 } frame; /**< data for compressed frame packet */ 196 vpx_fixed_buf_t twopass_stats; /**< data for two-pass packet */ 197 vpx_fixed_buf_t firstpass_mb_stats; /**< first pass mb packet */ 198 struct vpx_psnr_pkt { 199 unsigned int samples[4]; /**< Number of samples, total/y/u/v */ 200 uint64_t sse[4]; /**< sum squared error, total/y/u/v */ 201 double psnr[4]; /**< PSNR, total/y/u/v */ 202 } psnr; /**< data for PSNR packet */ 203 vpx_fixed_buf_t raw; /**< data for arbitrary packets */ 204 #if CONFIG_SPATIAL_SVC 205 size_t layer_sizes[VPX_SS_MAX_LAYERS]; 206 struct vpx_psnr_pkt layer_psnr[VPX_SS_MAX_LAYERS]; 207 #endif 208 209 /* This packet size is fixed to allow codecs to extend this 210 * interface without having to manage storage for raw packets, 211 * i.e., if it's smaller than 128 bytes, you can store in the 212 * packet list directly. 213 */ 214 char pad[128 - sizeof(enum vpx_codec_cx_pkt_kind)]; /**< fixed sz */ 215 } data; /**< packet data */ 216 } vpx_codec_cx_pkt_t; /**< alias for struct vpx_codec_cx_pkt */ 217 218 219 /*!\brief Rational Number 220 * 221 * This structure holds a fractional value. 222 */ 223 typedef struct vpx_rational { 224 int num; /**< fraction numerator */ 225 int den; /**< fraction denominator */ 226 } vpx_rational_t; /**< alias for struct vpx_rational */ 227 228 229 /*!\brief Multi-pass Encoding Pass */ 230 enum vpx_enc_pass { 231 VPX_RC_ONE_PASS, /**< Single pass mode */ 232 VPX_RC_FIRST_PASS, /**< First pass of multi-pass mode */ 233 VPX_RC_LAST_PASS /**< Final pass of multi-pass mode */ 234 }; 235 236 237 /*!\brief Rate control mode */ 238 enum vpx_rc_mode { 239 VPX_VBR, /**< Variable Bit Rate (VBR) mode */ 240 VPX_CBR, /**< Constant Bit Rate (CBR) mode */ 241 VPX_CQ, /**< Constrained Quality (CQ) mode */ 242 VPX_Q, /**< Constant Quality (Q) mode */ 243 }; 244 245 246 /*!\brief Keyframe placement mode. 247 * 248 * This enumeration determines whether keyframes are placed automatically by 249 * the encoder or whether this behavior is disabled. Older releases of this 250 * SDK were implemented such that VPX_KF_FIXED meant keyframes were disabled. 251 * This name is confusing for this behavior, so the new symbols to be used 252 * are VPX_KF_AUTO and VPX_KF_DISABLED. 253 */ 254 enum vpx_kf_mode { 255 VPX_KF_FIXED, /**< deprecated, implies VPX_KF_DISABLED */ 256 VPX_KF_AUTO, /**< Encoder determines optimal placement automatically */ 257 VPX_KF_DISABLED = 0 /**< Encoder does not place keyframes. */ 258 }; 259 260 261 /*!\brief Encoded Frame Flags 262 * 263 * This type indicates a bitfield to be passed to vpx_codec_encode(), defining 264 * per-frame boolean values. By convention, bits common to all codecs will be 265 * named VPX_EFLAG_*, and bits specific to an algorithm will be named 266 * /algo/_eflag_*. The lower order 16 bits are reserved for common use. 267 */ 268 typedef long vpx_enc_frame_flags_t; 269 #define VPX_EFLAG_FORCE_KF (1<<0) /**< Force this frame to be a keyframe */ 270 271 272 /*!\brief Encoder configuration structure 273 * 274 * This structure contains the encoder settings that have common representations 275 * across all codecs. This doesn't imply that all codecs support all features, 276 * however. 277 */ 278 typedef struct vpx_codec_enc_cfg { 279 /* 280 * generic settings (g) 281 */ 282 283 /*!\brief Algorithm specific "usage" value 284 * 285 * Algorithms may define multiple values for usage, which may convey the 286 * intent of how the application intends to use the stream. If this value 287 * is non-zero, consult the documentation for the codec to determine its 288 * meaning. 289 */ 290 unsigned int g_usage; 291 292 293 /*!\brief Maximum number of threads to use 294 * 295 * For multi-threaded implementations, use no more than this number of 296 * threads. The codec may use fewer threads than allowed. The value 297 * 0 is equivalent to the value 1. 298 */ 299 unsigned int g_threads; 300 301 302 /*!\brief Bitstream profile to use 303 * 304 * Some codecs support a notion of multiple bitstream profiles. Typically 305 * this maps to a set of features that are turned on or off. Often the 306 * profile to use is determined by the features of the intended decoder. 307 * Consult the documentation for the codec to determine the valid values 308 * for this parameter, or set to zero for a sane default. 309 */ 310 unsigned int g_profile; /**< profile of bitstream to use */ 311 312 313 314 /*!\brief Width of the frame 315 * 316 * This value identifies the presentation resolution of the frame, 317 * in pixels. Note that the frames passed as input to the encoder must 318 * have this resolution. Frames will be presented by the decoder in this 319 * resolution, independent of any spatial resampling the encoder may do. 320 */ 321 unsigned int g_w; 322 323 324 /*!\brief Height of the frame 325 * 326 * This value identifies the presentation resolution of the frame, 327 * in pixels. Note that the frames passed as input to the encoder must 328 * have this resolution. Frames will be presented by the decoder in this 329 * resolution, independent of any spatial resampling the encoder may do. 330 */ 331 unsigned int g_h; 332 333 /*!\brief Bit-depth of the codec 334 * 335 * This value identifies the bit_depth of the codec, 336 * Only certain bit-depths are supported as identified in the 337 * vpx_bit_depth_t enum. 338 */ 339 vpx_bit_depth_t g_bit_depth; 340 341 /*!\brief Bit-depth of the input frames 342 * 343 * This value identifies the bit_depth of the input frames in bits. 344 * Note that the frames passed as input to the encoder must have 345 * this bit-depth. 346 */ 347 unsigned int g_input_bit_depth; 348 349 /*!\brief Stream timebase units 350 * 351 * Indicates the smallest interval of time, in seconds, used by the stream. 352 * For fixed frame rate material, or variable frame rate material where 353 * frames are timed at a multiple of a given clock (ex: video capture), 354 * the \ref RECOMMENDED method is to set the timebase to the reciprocal 355 * of the frame rate (ex: 1001/30000 for 29.970 Hz NTSC). This allows the 356 * pts to correspond to the frame number, which can be handy. For 357 * re-encoding video from containers with absolute time timestamps, the 358 * \ref RECOMMENDED method is to set the timebase to that of the parent 359 * container or multimedia framework (ex: 1/1000 for ms, as in FLV). 360 */ 361 struct vpx_rational g_timebase; 362 363 364 /*!\brief Enable error resilient modes. 365 * 366 * The error resilient bitfield indicates to the encoder which features 367 * it should enable to take measures for streaming over lossy or noisy 368 * links. 369 */ 370 vpx_codec_er_flags_t g_error_resilient; 371 372 373 /*!\brief Multi-pass Encoding Mode 374 * 375 * This value should be set to the current phase for multi-pass encoding. 376 * For single pass, set to #VPX_RC_ONE_PASS. 377 */ 378 enum vpx_enc_pass g_pass; 379 380 381 /*!\brief Allow lagged encoding 382 * 383 * If set, this value allows the encoder to consume a number of input 384 * frames before producing output frames. This allows the encoder to 385 * base decisions for the current frame on future frames. This does 386 * increase the latency of the encoding pipeline, so it is not appropriate 387 * in all situations (ex: realtime encoding). 388 * 389 * Note that this is a maximum value -- the encoder may produce frames 390 * sooner than the given limit. Set this value to 0 to disable this 391 * feature. 392 */ 393 unsigned int g_lag_in_frames; 394 395 396 /* 397 * rate control settings (rc) 398 */ 399 400 /*!\brief Temporal resampling configuration, if supported by the codec. 401 * 402 * Temporal resampling allows the codec to "drop" frames as a strategy to 403 * meet its target data rate. This can cause temporal discontinuities in 404 * the encoded video, which may appear as stuttering during playback. This 405 * trade-off is often acceptable, but for many applications is not. It can 406 * be disabled in these cases. 407 * 408 * Note that not all codecs support this feature. All vpx VPx codecs do. 409 * For other codecs, consult the documentation for that algorithm. 410 * 411 * This threshold is described as a percentage of the target data buffer. 412 * When the data buffer falls below this percentage of fullness, a 413 * dropped frame is indicated. Set the threshold to zero (0) to disable 414 * this feature. 415 */ 416 unsigned int rc_dropframe_thresh; 417 418 419 /*!\brief Enable/disable spatial resampling, if supported by the codec. 420 * 421 * Spatial resampling allows the codec to compress a lower resolution 422 * version of the frame, which is then upscaled by the encoder to the 423 * correct presentation resolution. This increases visual quality at 424 * low data rates, at the expense of CPU time on the encoder/decoder. 425 */ 426 unsigned int rc_resize_allowed; 427 428 /*!\brief Internal coded frame width. 429 * 430 * If spatial resampling is enabled this specifies the width of the 431 * encoded frame. 432 */ 433 unsigned int rc_scaled_width; 434 435 /*!\brief Internal coded frame height. 436 * 437 * If spatial resampling is enabled this specifies the height of the 438 * encoded frame. 439 */ 440 unsigned int rc_scaled_height; 441 442 /*!\brief Spatial resampling up watermark. 443 * 444 * This threshold is described as a percentage of the target data buffer. 445 * When the data buffer rises above this percentage of fullness, the 446 * encoder will step up to a higher resolution version of the frame. 447 */ 448 unsigned int rc_resize_up_thresh; 449 450 451 /*!\brief Spatial resampling down watermark. 452 * 453 * This threshold is described as a percentage of the target data buffer. 454 * When the data buffer falls below this percentage of fullness, the 455 * encoder will step down to a lower resolution version of the frame. 456 */ 457 unsigned int rc_resize_down_thresh; 458 459 460 /*!\brief Rate control algorithm to use. 461 * 462 * Indicates whether the end usage of this stream is to be streamed over 463 * a bandwidth constrained link, indicating that Constant Bit Rate (CBR) 464 * mode should be used, or whether it will be played back on a high 465 * bandwidth link, as from a local disk, where higher variations in 466 * bitrate are acceptable. 467 */ 468 enum vpx_rc_mode rc_end_usage; 469 470 471 /*!\brief Two-pass stats buffer. 472 * 473 * A buffer containing all of the stats packets produced in the first 474 * pass, concatenated. 475 */ 476 vpx_fixed_buf_t rc_twopass_stats_in; 477 478 /*!\brief first pass mb stats buffer. 479 * 480 * A buffer containing all of the first pass mb stats packets produced 481 * in the first pass, concatenated. 482 */ 483 vpx_fixed_buf_t rc_firstpass_mb_stats_in; 484 485 /*!\brief Target data rate 486 * 487 * Target bandwidth to use for this stream, in kilobits per second. 488 */ 489 unsigned int rc_target_bitrate; 490 491 492 /* 493 * quantizer settings 494 */ 495 496 497 /*!\brief Minimum (Best Quality) Quantizer 498 * 499 * The quantizer is the most direct control over the quality of the 500 * encoded image. The range of valid values for the quantizer is codec 501 * specific. Consult the documentation for the codec to determine the 502 * values to use. To determine the range programmatically, call 503 * vpx_codec_enc_config_default() with a usage value of 0. 504 */ 505 unsigned int rc_min_quantizer; 506 507 508 /*!\brief Maximum (Worst Quality) Quantizer 509 * 510 * The quantizer is the most direct control over the quality of the 511 * encoded image. The range of valid values for the quantizer is codec 512 * specific. Consult the documentation for the codec to determine the 513 * values to use. To determine the range programmatically, call 514 * vpx_codec_enc_config_default() with a usage value of 0. 515 */ 516 unsigned int rc_max_quantizer; 517 518 519 /* 520 * bitrate tolerance 521 */ 522 523 524 /*!\brief Rate control adaptation undershoot control 525 * 526 * This value, expressed as a percentage of the target bitrate, 527 * controls the maximum allowed adaptation speed of the codec. 528 * This factor controls the maximum amount of bits that can 529 * be subtracted from the target bitrate in order to compensate 530 * for prior overshoot. 531 * 532 * Valid values in the range 0-1000. 533 */ 534 unsigned int rc_undershoot_pct; 535 536 537 /*!\brief Rate control adaptation overshoot control 538 * 539 * This value, expressed as a percentage of the target bitrate, 540 * controls the maximum allowed adaptation speed of the codec. 541 * This factor controls the maximum amount of bits that can 542 * be added to the target bitrate in order to compensate for 543 * prior undershoot. 544 * 545 * Valid values in the range 0-1000. 546 */ 547 unsigned int rc_overshoot_pct; 548 549 550 /* 551 * decoder buffer model parameters 552 */ 553 554 555 /*!\brief Decoder Buffer Size 556 * 557 * This value indicates the amount of data that may be buffered by the 558 * decoding application. Note that this value is expressed in units of 559 * time (milliseconds). For example, a value of 5000 indicates that the 560 * client will buffer (at least) 5000ms worth of encoded data. Use the 561 * target bitrate (#rc_target_bitrate) to convert to bits/bytes, if 562 * necessary. 563 */ 564 unsigned int rc_buf_sz; 565 566 567 /*!\brief Decoder Buffer Initial Size 568 * 569 * This value indicates the amount of data that will be buffered by the 570 * decoding application prior to beginning playback. This value is 571 * expressed in units of time (milliseconds). Use the target bitrate 572 * (#rc_target_bitrate) to convert to bits/bytes, if necessary. 573 */ 574 unsigned int rc_buf_initial_sz; 575 576 577 /*!\brief Decoder Buffer Optimal Size 578 * 579 * This value indicates the amount of data that the encoder should try 580 * to maintain in the decoder's buffer. This value is expressed in units 581 * of time (milliseconds). Use the target bitrate (#rc_target_bitrate) 582 * to convert to bits/bytes, if necessary. 583 */ 584 unsigned int rc_buf_optimal_sz; 585 586 587 /* 588 * 2 pass rate control parameters 589 */ 590 591 592 /*!\brief Two-pass mode CBR/VBR bias 593 * 594 * Bias, expressed on a scale of 0 to 100, for determining target size 595 * for the current frame. The value 0 indicates the optimal CBR mode 596 * value should be used. The value 100 indicates the optimal VBR mode 597 * value should be used. Values in between indicate which way the 598 * encoder should "lean." 599 */ 600 unsigned int rc_2pass_vbr_bias_pct; /**< RC mode bias between CBR and VBR(0-100: 0->CBR, 100->VBR) */ 601 602 603 /*!\brief Two-pass mode per-GOP minimum bitrate 604 * 605 * This value, expressed as a percentage of the target bitrate, indicates 606 * the minimum bitrate to be used for a single GOP (aka "section") 607 */ 608 unsigned int rc_2pass_vbr_minsection_pct; 609 610 611 /*!\brief Two-pass mode per-GOP maximum bitrate 612 * 613 * This value, expressed as a percentage of the target bitrate, indicates 614 * the maximum bitrate to be used for a single GOP (aka "section") 615 */ 616 unsigned int rc_2pass_vbr_maxsection_pct; 617 618 619 /* 620 * keyframing settings (kf) 621 */ 622 623 /*!\brief Keyframe placement mode 624 * 625 * This value indicates whether the encoder should place keyframes at a 626 * fixed interval, or determine the optimal placement automatically 627 * (as governed by the #kf_min_dist and #kf_max_dist parameters) 628 */ 629 enum vpx_kf_mode kf_mode; 630 631 632 /*!\brief Keyframe minimum interval 633 * 634 * This value, expressed as a number of frames, prevents the encoder from 635 * placing a keyframe nearer than kf_min_dist to the previous keyframe. At 636 * least kf_min_dist frames non-keyframes will be coded before the next 637 * keyframe. Set kf_min_dist equal to kf_max_dist for a fixed interval. 638 */ 639 unsigned int kf_min_dist; 640 641 642 /*!\brief Keyframe maximum interval 643 * 644 * This value, expressed as a number of frames, forces the encoder to code 645 * a keyframe if one has not been coded in the last kf_max_dist frames. 646 * A value of 0 implies all frames will be keyframes. Set kf_min_dist 647 * equal to kf_max_dist for a fixed interval. 648 */ 649 unsigned int kf_max_dist; 650 651 /* 652 * Spatial scalability settings (ss) 653 */ 654 655 /*!\brief Number of spatial coding layers. 656 * 657 * This value specifies the number of spatial coding layers to be used. 658 */ 659 unsigned int ss_number_layers; 660 661 /*!\brief Enable auto alt reference flags for each spatial layer. 662 * 663 * These values specify if auto alt reference frame is enabled for each 664 * spatial layer. 665 */ 666 int ss_enable_auto_alt_ref[VPX_SS_MAX_LAYERS]; 667 668 /*!\brief Target bitrate for each spatial layer. 669 * 670 * These values specify the target coding bitrate to be used for each 671 * spatial layer. 672 */ 673 unsigned int ss_target_bitrate[VPX_SS_MAX_LAYERS]; 674 675 /*!\brief Number of temporal coding layers. 676 * 677 * This value specifies the number of temporal layers to be used. 678 */ 679 unsigned int ts_number_layers; 680 681 /*!\brief Target bitrate for each temporal layer. 682 * 683 * These values specify the target coding bitrate to be used for each 684 * temporal layer. 685 */ 686 unsigned int ts_target_bitrate[VPX_TS_MAX_LAYERS]; 687 688 /*!\brief Frame rate decimation factor for each temporal layer. 689 * 690 * These values specify the frame rate decimation factors to apply 691 * to each temporal layer. 692 */ 693 unsigned int ts_rate_decimator[VPX_TS_MAX_LAYERS]; 694 695 /*!\brief Length of the sequence defining frame temporal layer membership. 696 * 697 * This value specifies the length of the sequence that defines the 698 * membership of frames to temporal layers. For example, if the 699 * ts_periodicity = 8, then the frames are assigned to coding layers with a 700 * repeated sequence of length 8. 701 */ 702 unsigned int ts_periodicity; 703 704 /*!\brief Template defining the membership of frames to temporal layers. 705 * 706 * This array defines the membership of frames to temporal coding layers. 707 * For a 2-layer encoding that assigns even numbered frames to one temporal 708 * layer (0) and odd numbered frames to a second temporal layer (1) with 709 * ts_periodicity=8, then ts_layer_id = (0,1,0,1,0,1,0,1). 710 */ 711 unsigned int ts_layer_id[VPX_TS_MAX_PERIODICITY]; 712 } vpx_codec_enc_cfg_t; /**< alias for struct vpx_codec_enc_cfg */ 713 714 /*!\brief vp9 svc extra configure parameters 715 * 716 * This defines max/min quantizers and scale factors for each layer 717 * 718 */ 719 typedef struct vpx_svc_parameters { 720 int max_quantizers[VPX_SS_MAX_LAYERS]; 721 int min_quantizers[VPX_SS_MAX_LAYERS]; 722 int scaling_factor_num[VPX_SS_MAX_LAYERS]; 723 int scaling_factor_den[VPX_SS_MAX_LAYERS]; 724 } vpx_svc_extra_cfg_t; 725 726 727 /*!\brief Initialize an encoder instance 728 * 729 * Initializes a encoder context using the given interface. Applications 730 * should call the vpx_codec_enc_init convenience macro instead of this 731 * function directly, to ensure that the ABI version number parameter 732 * is properly initialized. 733 * 734 * If the library was configured with --disable-multithread, this call 735 * is not thread safe and should be guarded with a lock if being used 736 * in a multithreaded context. 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] flags Bitfield of VPX_CODEC_USE_* flags 742 * \param[in] ver ABI version number. Must be set to 743 * VPX_ENCODER_ABI_VERSION 744 * \retval #VPX_CODEC_OK 745 * The decoder algorithm initialized. 746 * \retval #VPX_CODEC_MEM_ERROR 747 * Memory allocation failed. 748 */ 749 vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t *ctx, 750 vpx_codec_iface_t *iface, 751 const vpx_codec_enc_cfg_t *cfg, 752 vpx_codec_flags_t flags, 753 int ver); 754 755 756 /*!\brief Convenience macro for vpx_codec_enc_init_ver() 757 * 758 * Ensures the ABI version parameter is properly set. 759 */ 760 #define vpx_codec_enc_init(ctx, iface, cfg, flags) \ 761 vpx_codec_enc_init_ver(ctx, iface, cfg, flags, VPX_ENCODER_ABI_VERSION) 762 763 764 /*!\brief Initialize multi-encoder instance 765 * 766 * Initializes multi-encoder context using the given interface. 767 * Applications should call the vpx_codec_enc_init_multi convenience macro 768 * instead of this function directly, to ensure that the ABI version number 769 * parameter is properly initialized. 770 * 771 * \param[in] ctx Pointer to this instance's context. 772 * \param[in] iface Pointer to the algorithm interface to use. 773 * \param[in] cfg Configuration to use, if known. May be NULL. 774 * \param[in] num_enc Total number of encoders. 775 * \param[in] flags Bitfield of VPX_CODEC_USE_* flags 776 * \param[in] dsf Pointer to down-sampling factors. 777 * \param[in] ver ABI version number. Must be set to 778 * VPX_ENCODER_ABI_VERSION 779 * \retval #VPX_CODEC_OK 780 * The decoder algorithm initialized. 781 * \retval #VPX_CODEC_MEM_ERROR 782 * Memory allocation failed. 783 */ 784 vpx_codec_err_t vpx_codec_enc_init_multi_ver(vpx_codec_ctx_t *ctx, 785 vpx_codec_iface_t *iface, 786 vpx_codec_enc_cfg_t *cfg, 787 int num_enc, 788 vpx_codec_flags_t flags, 789 vpx_rational_t *dsf, 790 int ver); 791 792 793 /*!\brief Convenience macro for vpx_codec_enc_init_multi_ver() 794 * 795 * Ensures the ABI version parameter is properly set. 796 */ 797 #define vpx_codec_enc_init_multi(ctx, iface, cfg, num_enc, flags, dsf) \ 798 vpx_codec_enc_init_multi_ver(ctx, iface, cfg, num_enc, flags, dsf, \ 799 VPX_ENCODER_ABI_VERSION) 800 801 802 /*!\brief Get a default configuration 803 * 804 * Initializes a encoder configuration structure with default values. Supports 805 * the notion of "usages" so that an algorithm may offer different default 806 * settings depending on the user's intended goal. This function \ref SHOULD 807 * be called by all applications to initialize the configuration structure 808 * before specializing the configuration with application specific values. 809 * 810 * \param[in] iface Pointer to the algorithm interface to use. 811 * \param[out] cfg Configuration buffer to populate 812 * \param[in] usage End usage. Set to 0 or use codec specific values. 813 * 814 * \retval #VPX_CODEC_OK 815 * The configuration was populated. 816 * \retval #VPX_CODEC_INCAPABLE 817 * Interface is not an encoder interface. 818 * \retval #VPX_CODEC_INVALID_PARAM 819 * A parameter was NULL, or the usage value was not recognized. 820 */ 821 vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface, 822 vpx_codec_enc_cfg_t *cfg, 823 unsigned int usage); 824 825 826 /*!\brief Set or change configuration 827 * 828 * Reconfigures an encoder instance according to the given configuration. 829 * 830 * \param[in] ctx Pointer to this instance's context 831 * \param[in] cfg Configuration buffer to use 832 * 833 * \retval #VPX_CODEC_OK 834 * The configuration was populated. 835 * \retval #VPX_CODEC_INCAPABLE 836 * Interface is not an encoder interface. 837 * \retval #VPX_CODEC_INVALID_PARAM 838 * A parameter was NULL, or the usage value was not recognized. 839 */ 840 vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx, 841 const vpx_codec_enc_cfg_t *cfg); 842 843 844 /*!\brief Get global stream headers 845 * 846 * Retrieves a stream level global header packet, if supported by the codec. 847 * 848 * \param[in] ctx Pointer to this instance's context 849 * 850 * \retval NULL 851 * Encoder does not support global header 852 * \retval Non-NULL 853 * Pointer to buffer containing global header packet 854 */ 855 vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx); 856 857 858 #define VPX_DL_REALTIME (1) /**< deadline parameter analogous to 859 * VPx REALTIME mode. */ 860 #define VPX_DL_GOOD_QUALITY (1000000) /**< deadline parameter analogous to 861 * VPx GOOD QUALITY mode. */ 862 #define VPX_DL_BEST_QUALITY (0) /**< deadline parameter analogous to 863 * VPx BEST QUALITY mode. */ 864 /*!\brief Encode a frame 865 * 866 * Encodes a video frame at the given "presentation time." The presentation 867 * time stamp (PTS) \ref MUST be strictly increasing. 868 * 869 * The encoder supports the notion of a soft real-time deadline. Given a 870 * non-zero value to the deadline parameter, the encoder will make a "best 871 * effort" guarantee to return before the given time slice expires. It is 872 * implicit that limiting the available time to encode will degrade the 873 * output quality. The encoder can be given an unlimited time to produce the 874 * best possible frame by specifying a deadline of '0'. This deadline 875 * supercedes the VPx notion of "best quality, good quality, realtime". 876 * Applications that wish to map these former settings to the new deadline 877 * based system can use the symbols #VPX_DL_REALTIME, #VPX_DL_GOOD_QUALITY, 878 * and #VPX_DL_BEST_QUALITY. 879 * 880 * When the last frame has been passed to the encoder, this function should 881 * continue to be called, with the img parameter set to NULL. This will 882 * signal the end-of-stream condition to the encoder and allow it to encode 883 * any held buffers. Encoding is complete when vpx_codec_encode() is called 884 * and vpx_codec_get_cx_data() returns no data. 885 * 886 * \param[in] ctx Pointer to this instance's context 887 * \param[in] img Image data to encode, NULL to flush. 888 * \param[in] pts Presentation time stamp, in timebase units. 889 * \param[in] duration Duration to show frame, in timebase units. 890 * \param[in] flags Flags to use for encoding this frame. 891 * \param[in] deadline Time to spend encoding, in microseconds. (0=infinite) 892 * 893 * \retval #VPX_CODEC_OK 894 * The configuration was populated. 895 * \retval #VPX_CODEC_INCAPABLE 896 * Interface is not an encoder interface. 897 * \retval #VPX_CODEC_INVALID_PARAM 898 * A parameter was NULL, the image format is unsupported, etc. 899 */ 900 vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx, 901 const vpx_image_t *img, 902 vpx_codec_pts_t pts, 903 unsigned long duration, 904 vpx_enc_frame_flags_t flags, 905 unsigned long deadline); 906 907 /*!\brief Set compressed data output buffer 908 * 909 * Sets the buffer that the codec should output the compressed data 910 * into. This call effectively sets the buffer pointer returned in the 911 * next VPX_CODEC_CX_FRAME_PKT packet. Subsequent packets will be 912 * appended into this buffer. The buffer is preserved across frames, 913 * so applications must periodically call this function after flushing 914 * the accumulated compressed data to disk or to the network to reset 915 * the pointer to the buffer's head. 916 * 917 * `pad_before` bytes will be skipped before writing the compressed 918 * data, and `pad_after` bytes will be appended to the packet. The size 919 * of the packet will be the sum of the size of the actual compressed 920 * data, pad_before, and pad_after. The padding bytes will be preserved 921 * (not overwritten). 922 * 923 * Note that calling this function does not guarantee that the returned 924 * compressed data will be placed into the specified buffer. In the 925 * event that the encoded data will not fit into the buffer provided, 926 * the returned packet \ref MAY point to an internal buffer, as it would 927 * if this call were never used. In this event, the output packet will 928 * NOT have any padding, and the application must free space and copy it 929 * to the proper place. This is of particular note in configurations 930 * that may output multiple packets for a single encoded frame (e.g., lagged 931 * encoding) or if the application does not reset the buffer periodically. 932 * 933 * Applications may restore the default behavior of the codec providing 934 * the compressed data buffer by calling this function with a NULL 935 * buffer. 936 * 937 * Applications \ref MUSTNOT call this function during iteration of 938 * vpx_codec_get_cx_data(). 939 * 940 * \param[in] ctx Pointer to this instance's context 941 * \param[in] buf Buffer to store compressed data into 942 * \param[in] pad_before Bytes to skip before writing compressed data 943 * \param[in] pad_after Bytes to skip after writing compressed data 944 * 945 * \retval #VPX_CODEC_OK 946 * The buffer was set successfully. 947 * \retval #VPX_CODEC_INVALID_PARAM 948 * A parameter was NULL, the image format is unsupported, etc. 949 */ 950 vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t *ctx, 951 const vpx_fixed_buf_t *buf, 952 unsigned int pad_before, 953 unsigned int pad_after); 954 955 956 /*!\brief Encoded data iterator 957 * 958 * Iterates over a list of data packets to be passed from the encoder to the 959 * application. The different kinds of packets available are enumerated in 960 * #vpx_codec_cx_pkt_kind. 961 * 962 * #VPX_CODEC_CX_FRAME_PKT packets should be passed to the application's 963 * muxer. Multiple compressed frames may be in the list. 964 * #VPX_CODEC_STATS_PKT packets should be appended to a global buffer. 965 * 966 * The application \ref MUST silently ignore any packet kinds that it does 967 * not recognize or support. 968 * 969 * The data buffers returned from this function are only guaranteed to be 970 * valid until the application makes another call to any vpx_codec_* function. 971 * 972 * \param[in] ctx Pointer to this instance's context 973 * \param[in,out] iter Iterator storage, initialized to NULL 974 * 975 * \return Returns a pointer to an output data packet (compressed frame data, 976 * two-pass statistics, etc.) or NULL to signal end-of-list. 977 * 978 */ 979 const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx, 980 vpx_codec_iter_t *iter); 981 982 983 /*!\brief Get Preview Frame 984 * 985 * Returns an image that can be used as a preview. Shows the image as it would 986 * exist at the decompressor. The application \ref MUST NOT write into this 987 * image buffer. 988 * 989 * \param[in] ctx Pointer to this instance's context 990 * 991 * \return Returns a pointer to a preview image, or NULL if no image is 992 * available. 993 * 994 */ 995 const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t *ctx); 996 997 998 /*!@} - end defgroup encoder*/ 999 #ifdef __cplusplus 1000 } 1001 #endif 1002 #endif // VPX_VPX_ENCODER_H_ 1003 1004
