vpx_encoder.h
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_VPX_ENCODER_H_ 11 #define VPX_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 /*! Temporal+Spatial Scalability: Maximum number of coding layers */ 43 #define VPX_MAX_LAYERS 12 // 3 temporal + 4 spatial layers are allowed. 44 45 /*! Spatial Scalability: Maximum number of coding layers */ 46 #define VPX_SS_MAX_LAYERS 5 47 48 /*! Spatial Scalability: Default number of coding layers */ 49 #define VPX_SS_DEFAULT_LAYERS 1 50 51 /*!\brief Current ABI version number 52 * 53 * \internal 54 * If this file is altered in any way that changes the ABI, this value 55 * must be bumped. Examples include, but are not limited to, changing 56 * types, removing or reassigning enums, adding/removing/rearranging 57 * fields to structures 58 */ 59 #define VPX_ENCODER_ABI_VERSION \ 60 (14 + VPX_CODEC_ABI_VERSION) /**<\hideinitializer*/ 61 62 /*! \brief Encoder capabilities bitfield 63 * 64 * Each encoder advertises the capabilities it supports as part of its 65 * ::vpx_codec_iface_t interface structure. Capabilities are extra 66 * interfaces or functionality, and are not required to be supported 67 * by an encoder. 68 * 69 * The available flags are specified by VPX_CODEC_CAP_* defines. 70 */ 71 #define VPX_CODEC_CAP_PSNR 0x10000 /**< Can issue PSNR packets */ 72 73 /*! Can output one partition at a time. Each partition is returned in its 74 * own VPX_CODEC_CX_FRAME_PKT, with the FRAME_IS_FRAGMENT flag set for 75 * every partition but the last. In this mode all frames are always 76 * returned partition by partition. 77 */ 78 #define VPX_CODEC_CAP_OUTPUT_PARTITION 0x20000 79 80 /*! \brief Initialization-time Feature Enabling 81 * 82 * Certain codec features must be known at initialization time, to allow 83 * for proper memory allocation. 84 * 85 * The available flags are specified by VPX_CODEC_USE_* defines. 86 */ 87 #define VPX_CODEC_USE_PSNR 0x10000 /**< Calculate PSNR on each frame */ 88 /*!\brief Make the encoder output one partition at a time. */ 89 #define VPX_CODEC_USE_OUTPUT_PARTITION 0x20000 90 #define VPX_CODEC_USE_HIGHBITDEPTH 0x40000 /**< Use high bitdepth */ 91 92 /*!\brief Generic fixed size buffer structure 93 * 94 * This structure is able to hold a reference to any fixed size buffer. 95 */ 96 typedef struct vpx_fixed_buf { 97 void *buf; /**< Pointer to the data */ 98 size_t sz; /**< Length of the buffer, in chars */ 99 } vpx_fixed_buf_t; /**< alias for struct vpx_fixed_buf */ 100 101 /*!\brief Time Stamp Type 102 * 103 * An integer, which when multiplied by the stream's time base, provides 104 * the absolute time of a sample. 105 */ 106 typedef int64_t vpx_codec_pts_t; 107 108 /*!\brief Compressed Frame Flags 109 * 110 * This type represents a bitfield containing information about a compressed 111 * frame that may be useful to an application. The most significant 16 bits 112 * can be used by an algorithm to provide additional detail, for example to 113 * support frame types that are codec specific (MPEG-1 D-frames for example) 114 */ 115 typedef uint32_t vpx_codec_frame_flags_t; 116 #define VPX_FRAME_IS_KEY 0x1 /**< frame is the start of a GOP */ 117 /*!\brief frame can be dropped without affecting the stream (no future frame 118 * depends on this one) */ 119 #define VPX_FRAME_IS_DROPPABLE 0x2 120 /*!\brief frame should be decoded but will not be shown */ 121 #define VPX_FRAME_IS_INVISIBLE 0x4 122 /*!\brief this is a fragment of the encoded frame */ 123 #define VPX_FRAME_IS_FRAGMENT 0x8 124 125 /*!\brief Error Resilient flags 126 * 127 * These flags define which error resilient features to enable in the 128 * encoder. The flags are specified through the 129 * vpx_codec_enc_cfg::g_error_resilient variable. 130 */ 131 typedef uint32_t vpx_codec_er_flags_t; 132 /*!\brief Improve resiliency against losses of whole frames */ 133 #define VPX_ERROR_RESILIENT_DEFAULT 0x1 134 /*!\brief The frame partitions are independently decodable by the bool decoder, 135 * meaning that partitions can be decoded even though earlier partitions have 136 * been lost. Note that intra prediction is still done over the partition 137 * boundary. */ 138 #define VPX_ERROR_RESILIENT_PARTITIONS 0x2 139 140 /*!\brief Encoder output packet variants 141 * 142 * This enumeration lists the different kinds of data packets that can be 143 * returned by calls to vpx_codec_get_cx_data(). Algorithms \ref MAY 144 * extend this list to provide additional functionality. 145 */ 146 enum vpx_codec_cx_pkt_kind { 147 VPX_CODEC_CX_FRAME_PKT, /**< Compressed video frame */ 148 VPX_CODEC_STATS_PKT, /**< Two-pass statistics for this frame */ 149 VPX_CODEC_FPMB_STATS_PKT, /**< first pass mb statistics for this frame */ 150 VPX_CODEC_PSNR_PKT, /**< PSNR statistics for this frame */ 151 VPX_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions */ 152 }; 153 154 /*!\brief Encoder output packet 155 * 156 * This structure contains the different kinds of output data the encoder 157 * may produce while compressing a frame. 158 */ 159 typedef struct vpx_codec_cx_pkt { 160 enum vpx_codec_cx_pkt_kind kind; /**< packet variant */ 161 union { 162 struct { 163 void *buf; /**< compressed data buffer */ 164 size_t sz; /**< length of compressed data */ 165 /*!\brief time stamp to show frame (in timebase units) */ 166 vpx_codec_pts_t pts; 167 /*!\brief duration to show frame (in timebase units) */ 168 unsigned long duration; 169 vpx_codec_frame_flags_t flags; /**< flags for this frame */ 170 /*!\brief the partition id defines the decoding order of the partitions. 171 * Only applicable when "output partition" mode is enabled. First 172 * partition has id 0.*/ 173 int partition_id; 174 /*!\brief Width and height of frames in this packet. VP8 will only use the 175 * first one.*/ 176 unsigned int width[VPX_SS_MAX_LAYERS]; /**< frame width */ 177 unsigned int height[VPX_SS_MAX_LAYERS]; /**< frame height */ 178 /*!\brief Flag to indicate if spatial layer frame in this packet is 179 * encoded or dropped. VP8 will always be set to 1.*/ 180 uint8_t spatial_layer_encoded[VPX_SS_MAX_LAYERS]; 181 } frame; /**< data for compressed frame packet */ 182 vpx_fixed_buf_t twopass_stats; /**< data for two-pass packet */ 183 vpx_fixed_buf_t firstpass_mb_stats; /**< first pass mb packet */ 184 struct vpx_psnr_pkt { 185 unsigned int samples[4]; /**< Number of samples, total/y/u/v */ 186 uint64_t sse[4]; /**< sum squared error, total/y/u/v */ 187 double psnr[4]; /**< PSNR, total/y/u/v */ 188 } psnr; /**< data for PSNR packet */ 189 vpx_fixed_buf_t raw; /**< data for arbitrary packets */ 190 191 /* This packet size is fixed to allow codecs to extend this 192 * interface without having to manage storage for raw packets, 193 * i.e., if it's smaller than 128 bytes, you can store in the 194 * packet list directly. 195 */ 196 char pad[128 - sizeof(enum vpx_codec_cx_pkt_kind)]; /**< fixed sz */ 197 } data; /**< packet data */ 198 } vpx_codec_cx_pkt_t; /**< alias for struct vpx_codec_cx_pkt */ 199 200 /*!\brief Encoder return output buffer callback 201 * 202 * This callback function, when registered, returns with packets when each 203 * spatial layer is encoded. 204 */ 205 typedef void (*vpx_codec_enc_output_cx_pkt_cb_fn_t)(vpx_codec_cx_pkt_t *pkt, 206 void *user_data); 207 208 /*!\brief Callback function pointer / user data pair storage */ 209 typedef struct vpx_codec_enc_output_cx_cb_pair { 210 vpx_codec_enc_output_cx_pkt_cb_fn_t output_cx_pkt; /**< Callback function */ 211 void *user_priv; /**< Pointer to private data */ 212 } vpx_codec_priv_output_cx_pkt_cb_pair_t; 213 214 /*!\brief Rational Number 215 * 216 * This structure holds a fractional value. 217 */ 218 typedef struct vpx_rational { 219 int num; /**< fraction numerator */ 220 int den; /**< fraction denominator */ 221 } vpx_rational_t; /**< alias for struct vpx_rational */ 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 /*!\brief Rate control mode */ 231 enum vpx_rc_mode { 232 VPX_VBR, /**< Variable Bit Rate (VBR) mode */ 233 VPX_CBR, /**< Constant Bit Rate (CBR) mode */ 234 VPX_CQ, /**< Constrained Quality (CQ) mode */ 235 VPX_Q, /**< Constant Quality (Q) mode */ 236 }; 237 238 /*!\brief Keyframe placement mode. 239 * 240 * This enumeration determines whether keyframes are placed automatically by 241 * the encoder or whether this behavior is disabled. Older releases of this 242 * SDK were implemented such that VPX_KF_FIXED meant keyframes were disabled. 243 * This name is confusing for this behavior, so the new symbols to be used 244 * are VPX_KF_AUTO and VPX_KF_DISABLED. 245 */ 246 enum vpx_kf_mode { 247 VPX_KF_FIXED, /**< deprecated, implies VPX_KF_DISABLED */ 248 VPX_KF_AUTO, /**< Encoder determines optimal placement automatically */ 249 VPX_KF_DISABLED = 0 /**< Encoder does not place keyframes. */ 250 }; 251 252 /*!\brief Encoded Frame Flags 253 * 254 * This type indicates a bitfield to be passed to vpx_codec_encode(), defining 255 * per-frame boolean values. By convention, bits common to all codecs will be 256 * named VPX_EFLAG_*, and bits specific to an algorithm will be named 257 * /algo/_eflag_*. The lower order 16 bits are reserved for common use. 258 */ 259 typedef long vpx_enc_frame_flags_t; 260 #define VPX_EFLAG_FORCE_KF (1 << 0) /**< Force this frame to be a keyframe */ 261 262 /*!\brief Encoder configuration structure 263 * 264 * This structure contains the encoder settings that have common representations 265 * across all codecs. This doesn't imply that all codecs support all features, 266 * however. 267 */ 268 typedef struct vpx_codec_enc_cfg { 269 /* 270 * generic settings (g) 271 */ 272 273 /*!\brief Deprecated: Algorithm specific "usage" value 274 * 275 * This value must be zero. 276 */ 277 unsigned int g_usage; 278 279 /*!\brief Maximum number of threads to use 280 * 281 * For multi-threaded implementations, use no more than this number of 282 * threads. The codec may use fewer threads than allowed. The value 283 * 0 is equivalent to the value 1. 284 */ 285 unsigned int g_threads; 286 287 /*!\brief Bitstream profile to use 288 * 289 * Some codecs support a notion of multiple bitstream profiles. Typically 290 * this maps to a set of features that are turned on or off. Often the 291 * profile to use is determined by the features of the intended decoder. 292 * Consult the documentation for the codec to determine the valid values 293 * for this parameter, or set to zero for a sane default. 294 */ 295 unsigned int g_profile; /**< profile of bitstream to use */ 296 297 /*!\brief Width of the frame 298 * 299 * This value identifies the presentation resolution of the frame, 300 * in pixels. Note that the frames passed as input to the encoder must 301 * have this resolution. Frames will be presented by the decoder in this 302 * resolution, independent of any spatial resampling the encoder may do. 303 */ 304 unsigned int g_w; 305 306 /*!\brief Height of the frame 307 * 308 * This value identifies the presentation resolution of the frame, 309 * in pixels. Note that the frames passed as input to the encoder must 310 * have this resolution. Frames will be presented by the decoder in this 311 * resolution, independent of any spatial resampling the encoder may do. 312 */ 313 unsigned int g_h; 314 315 /*!\brief Bit-depth of the codec 316 * 317 * This value identifies the bit_depth of the codec, 318 * Only certain bit-depths are supported as identified in the 319 * vpx_bit_depth_t enum. 320 */ 321 vpx_bit_depth_t g_bit_depth; 322 323 /*!\brief Bit-depth of the input frames 324 * 325 * This value identifies the bit_depth of the input frames in bits. 326 * Note that the frames passed as input to the encoder must have 327 * this bit-depth. 328 */ 329 unsigned int g_input_bit_depth; 330 331 /*!\brief Stream timebase units 332 * 333 * Indicates the smallest interval of time, in seconds, used by the stream. 334 * For fixed frame rate material, or variable frame rate material where 335 * frames are timed at a multiple of a given clock (ex: video capture), 336 * the \ref RECOMMENDED method is to set the timebase to the reciprocal 337 * of the frame rate (ex: 1001/30000 for 29.970 Hz NTSC). This allows the 338 * pts to correspond to the frame number, which can be handy. For 339 * re-encoding video from containers with absolute time timestamps, the 340 * \ref RECOMMENDED method is to set the timebase to that of the parent 341 * container or multimedia framework (ex: 1/1000 for ms, as in FLV). 342 */ 343 struct vpx_rational g_timebase; 344 345 /*!\brief Enable error resilient modes. 346 * 347 * The error resilient bitfield indicates to the encoder which features 348 * it should enable to take measures for streaming over lossy or noisy 349 * links. 350 */ 351 vpx_codec_er_flags_t g_error_resilient; 352 353 /*!\brief Multi-pass Encoding Mode 354 * 355 * This value should be set to the current phase for multi-pass encoding. 356 * For single pass, set to #VPX_RC_ONE_PASS. 357 */ 358 enum vpx_enc_pass g_pass; 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 * rate control settings (rc) 376 */ 377 378 /*!\brief Temporal resampling configuration, if supported by the codec. 379 * 380 * Temporal resampling allows the codec to "drop" frames as a strategy to 381 * meet its target data rate. This can cause temporal discontinuities in 382 * the encoded video, which may appear as stuttering during playback. This 383 * trade-off is often acceptable, but for many applications is not. It can 384 * be disabled in these cases. 385 * 386 * This threshold is described as a percentage of the target data buffer. 387 * When the data buffer falls below this percentage of fullness, a 388 * dropped frame is indicated. Set the threshold to zero (0) to disable 389 * this feature. 390 */ 391 unsigned int rc_dropframe_thresh; 392 393 /*!\brief Enable/disable spatial resampling, if supported by the codec. 394 * 395 * Spatial resampling allows the codec to compress a lower resolution 396 * version of the frame, which is then upscaled by the encoder to the 397 * correct presentation resolution. This increases visual quality at 398 * low data rates, at the expense of CPU time on the encoder/decoder. 399 */ 400 unsigned int rc_resize_allowed; 401 402 /*!\brief Internal coded frame width. 403 * 404 * If spatial resampling is enabled this specifies the width of the 405 * encoded frame. 406 */ 407 unsigned int rc_scaled_width; 408 409 /*!\brief Internal coded frame height. 410 * 411 * If spatial resampling is enabled this specifies the height of the 412 * encoded frame. 413 */ 414 unsigned int rc_scaled_height; 415 416 /*!\brief Spatial resampling up watermark. 417 * 418 * This threshold is described as a percentage of the target data buffer. 419 * When the data buffer rises above this percentage of fullness, the 420 * encoder will step up to a higher resolution version of the frame. 421 */ 422 unsigned int rc_resize_up_thresh; 423 424 /*!\brief Spatial resampling down watermark. 425 * 426 * This threshold is described as a percentage of the target data buffer. 427 * When the data buffer falls below this percentage of fullness, the 428 * encoder will step down to a lower resolution version of the frame. 429 */ 430 unsigned int rc_resize_down_thresh; 431 432 /*!\brief Rate control algorithm to use. 433 * 434 * Indicates whether the end usage of this stream is to be streamed over 435 * a bandwidth constrained link, indicating that Constant Bit Rate (CBR) 436 * mode should be used, or whether it will be played back on a high 437 * bandwidth link, as from a local disk, where higher variations in 438 * bitrate are acceptable. 439 */ 440 enum vpx_rc_mode rc_end_usage; 441 442 /*!\brief Two-pass stats buffer. 443 * 444 * A buffer containing all of the stats packets produced in the first 445 * pass, concatenated. 446 */ 447 vpx_fixed_buf_t rc_twopass_stats_in; 448 449 /*!\brief first pass mb stats buffer. 450 * 451 * A buffer containing all of the first pass mb stats packets produced 452 * in the first pass, concatenated. 453 */ 454 vpx_fixed_buf_t rc_firstpass_mb_stats_in; 455 456 /*!\brief Target data rate 457 * 458 * Target bandwidth to use for this stream, in kilobits per second. 459 */ 460 unsigned int rc_target_bitrate; 461 462 /* 463 * quantizer settings 464 */ 465 466 /*!\brief Minimum (Best Quality) Quantizer 467 * 468 * The quantizer is the most direct control over the quality of the 469 * encoded image. The range of valid values for the quantizer is codec 470 * specific. Consult the documentation for the codec to determine the 471 * values to use. 472 */ 473 unsigned int rc_min_quantizer; 474 475 /*!\brief Maximum (Worst Quality) Quantizer 476 * 477 * The quantizer is the most direct control over the quality of the 478 * encoded image. The range of valid values for the quantizer is codec 479 * specific. Consult the documentation for the codec to determine the 480 * values to use. 481 */ 482 unsigned int rc_max_quantizer; 483 484 /* 485 * bitrate tolerance 486 */ 487 488 /*!\brief Rate control adaptation undershoot control 489 * 490 * VP8: Expressed as a percentage of the target bitrate, 491 * controls the maximum allowed adaptation speed of the codec. 492 * This factor controls the maximum amount of bits that can 493 * be subtracted from the target bitrate in order to compensate 494 * for prior overshoot. 495 * VP9: Expressed as a percentage of the target bitrate, a threshold 496 * undershoot level (current rate vs target) beyond which more aggressive 497 * corrective measures are taken. 498 * * 499 * Valid values in the range VP8:0-1000 VP9: 0-100. 500 */ 501 unsigned int rc_undershoot_pct; 502 503 /*!\brief Rate control adaptation overshoot control 504 * 505 * VP8: 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 added to the target bitrate in order to compensate for 509 * prior undershoot. 510 * VP9: Expressed as a percentage of the target bitrate, a threshold 511 * overshoot level (current rate vs target) beyond which more aggressive 512 * corrective measures are taken. 513 * 514 * Valid values in the range VP8:0-1000 VP9: 0-100. 515 */ 516 unsigned int rc_overshoot_pct; 517 518 /* 519 * decoder buffer model parameters 520 */ 521 522 /*!\brief Decoder Buffer Size 523 * 524 * This value indicates the amount of data that may be buffered by the 525 * decoding application. Note that this value is expressed in units of 526 * time (milliseconds). For example, a value of 5000 indicates that the 527 * client will buffer (at least) 5000ms worth of encoded data. Use the 528 * target bitrate (#rc_target_bitrate) to convert to bits/bytes, if 529 * necessary. 530 */ 531 unsigned int rc_buf_sz; 532 533 /*!\brief Decoder Buffer Initial Size 534 * 535 * This value indicates the amount of data that will be buffered by the 536 * decoding application prior to beginning playback. This value is 537 * expressed in units of time (milliseconds). Use the target bitrate 538 * (#rc_target_bitrate) to convert to bits/bytes, if necessary. 539 */ 540 unsigned int rc_buf_initial_sz; 541 542 /*!\brief Decoder Buffer Optimal Size 543 * 544 * This value indicates the amount of data that the encoder should try 545 * to maintain in the decoder's buffer. This value is expressed in units 546 * of time (milliseconds). Use the target bitrate (#rc_target_bitrate) 547 * to convert to bits/bytes, if necessary. 548 */ 549 unsigned int rc_buf_optimal_sz; 550 551 /* 552 * 2 pass rate control parameters 553 */ 554 555 /*!\brief Two-pass mode CBR/VBR bias 556 * 557 * Bias, expressed on a scale of 0 to 100, for determining target size 558 * for the current frame. The value 0 indicates the optimal CBR mode 559 * value should be used. The value 100 indicates the optimal VBR mode 560 * value should be used. Values in between indicate which way the 561 * encoder should "lean." 562 */ 563 unsigned int rc_2pass_vbr_bias_pct; 564 565 /*!\brief Two-pass mode per-GOP minimum bitrate 566 * 567 * This value, expressed as a percentage of the target bitrate, indicates 568 * the minimum bitrate to be used for a single GOP (aka "section") 569 */ 570 unsigned int rc_2pass_vbr_minsection_pct; 571 572 /*!\brief Two-pass mode per-GOP maximum bitrate 573 * 574 * This value, expressed as a percentage of the target bitrate, indicates 575 * the maximum bitrate to be used for a single GOP (aka "section") 576 */ 577 unsigned int rc_2pass_vbr_maxsection_pct; 578 579 /*!\brief Two-pass corpus vbr mode complexity control 580 * Used only in VP9: A value representing the corpus midpoint complexity 581 * for corpus vbr mode. This value defaults to 0 which disables corpus vbr 582 * mode in favour of normal vbr mode. 583 */ 584 unsigned int rc_2pass_vbr_corpus_complexity; 585 586 /* 587 * keyframing settings (kf) 588 */ 589 590 /*!\brief Keyframe placement mode 591 * 592 * This value indicates whether the encoder should place keyframes at a 593 * fixed interval, or determine the optimal placement automatically 594 * (as governed by the #kf_min_dist and #kf_max_dist parameters) 595 */ 596 enum vpx_kf_mode kf_mode; 597 598 /*!\brief Keyframe minimum interval 599 * 600 * This value, expressed as a number of frames, prevents the encoder from 601 * placing a keyframe nearer than kf_min_dist to the previous keyframe. At 602 * least kf_min_dist frames non-keyframes will be coded before the next 603 * keyframe. Set kf_min_dist equal to kf_max_dist for a fixed interval. 604 */ 605 unsigned int kf_min_dist; 606 607 /*!\brief Keyframe maximum interval 608 * 609 * This value, expressed as a number of frames, forces the encoder to code 610 * a keyframe if one has not been coded in the last kf_max_dist frames. 611 * A value of 0 implies all frames will be keyframes. Set kf_min_dist 612 * equal to kf_max_dist for a fixed interval. 613 */ 614 unsigned int kf_max_dist; 615 616 /* 617 * Spatial scalability settings (ss) 618 */ 619 620 /*!\brief Number of spatial coding layers. 621 * 622 * This value specifies the number of spatial coding layers to be used. 623 */ 624 unsigned int ss_number_layers; 625 626 /*!\brief Enable auto alt reference flags for each spatial layer. 627 * 628 * These values specify if auto alt reference frame is enabled for each 629 * spatial layer. 630 */ 631 int ss_enable_auto_alt_ref[VPX_SS_MAX_LAYERS]; 632 633 /*!\brief Target bitrate for each spatial layer. 634 * 635 * These values specify the target coding bitrate to be used for each 636 * spatial layer. 637 */ 638 unsigned int ss_target_bitrate[VPX_SS_MAX_LAYERS]; 639 640 /*!\brief Number of temporal coding layers. 641 * 642 * This value specifies the number of temporal layers to be used. 643 */ 644 unsigned int ts_number_layers; 645 646 /*!\brief Target bitrate for each temporal layer. 647 * 648 * These values specify the target coding bitrate to be used for each 649 * temporal layer. 650 */ 651 unsigned int ts_target_bitrate[VPX_TS_MAX_LAYERS]; 652 653 /*!\brief Frame rate decimation factor for each temporal layer. 654 * 655 * These values specify the frame rate decimation factors to apply 656 * to each temporal layer. 657 */ 658 unsigned int ts_rate_decimator[VPX_TS_MAX_LAYERS]; 659 660 /*!\brief Length of the sequence defining frame temporal layer membership. 661 * 662 * This value specifies the length of the sequence that defines the 663 * membership of frames to temporal layers. For example, if the 664 * ts_periodicity = 8, then the frames are assigned to coding layers with a 665 * repeated sequence of length 8. 666 */ 667 unsigned int ts_periodicity; 668 669 /*!\brief Template defining the membership of frames to temporal layers. 670 * 671 * This array defines the membership of frames to temporal coding layers. 672 * For a 2-layer encoding that assigns even numbered frames to one temporal 673 * layer (0) and odd numbered frames to a second temporal layer (1) with 674 * ts_periodicity=8, then ts_layer_id = (0,1,0,1,0,1,0,1). 675 */ 676 unsigned int ts_layer_id[VPX_TS_MAX_PERIODICITY]; 677 678 /*!\brief Target bitrate for each spatial/temporal layer. 679 * 680 * These values specify the target coding bitrate to be used for each 681 * spatial/temporal layer. 682 * 683 */ 684 unsigned int layer_target_bitrate[VPX_MAX_LAYERS]; 685 686 /*!\brief Temporal layering mode indicating which temporal layering scheme to 687 * use. 688 * 689 * The value (refer to VP9E_TEMPORAL_LAYERING_MODE) specifies the 690 * temporal layering mode to use. 691 * 692 */ 693 int temporal_layering_mode; 694 } vpx_codec_enc_cfg_t; /**< alias for struct vpx_codec_enc_cfg */ 695 696 /*!\brief vp9 svc extra configure parameters 697 * 698 * This defines max/min quantizers and scale factors for each layer 699 * 700 */ 701 typedef struct vpx_svc_parameters { 702 int max_quantizers[VPX_MAX_LAYERS]; /**< Max Q for each layer */ 703 int min_quantizers[VPX_MAX_LAYERS]; /**< Min Q for each layer */ 704 int scaling_factor_num[VPX_MAX_LAYERS]; /**< Scaling factor-numerator */ 705 int scaling_factor_den[VPX_MAX_LAYERS]; /**< Scaling factor-denominator */ 706 int speed_per_layer[VPX_MAX_LAYERS]; /**< Speed setting for each sl */ 707 int temporal_layering_mode; /**< Temporal layering mode */ 708 } vpx_svc_extra_cfg_t; 709 710 /*!\brief Initialize an encoder instance 711 * 712 * Initializes a encoder context using the given interface. Applications 713 * should call the vpx_codec_enc_init convenience macro instead of this 714 * function directly, to ensure that the ABI version number parameter 715 * is properly initialized. 716 * 717 * If the library was configured with --disable-multithread, this call 718 * is not thread safe and should be guarded with a lock if being used 719 * in a multithreaded context. 720 * 721 * \param[in] ctx Pointer to this instance's context. 722 * \param[in] iface Pointer to the algorithm interface to use. 723 * \param[in] cfg Configuration to use, if known. May be NULL. 724 * \param[in] flags Bitfield of VPX_CODEC_USE_* flags 725 * \param[in] ver ABI version number. Must be set to 726 * VPX_ENCODER_ABI_VERSION 727 * \retval #VPX_CODEC_OK 728 * The decoder algorithm initialized. 729 * \retval #VPX_CODEC_MEM_ERROR 730 * Memory allocation failed. 731 */ 732 vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t *ctx, 733 vpx_codec_iface_t *iface, 734 const vpx_codec_enc_cfg_t *cfg, 735 vpx_codec_flags_t flags, int ver); 736 737 /*!\brief Convenience macro for vpx_codec_enc_init_ver() 738 * 739 * Ensures the ABI version parameter is properly set. 740 */ 741 #define vpx_codec_enc_init(ctx, iface, cfg, flags) \ 742 vpx_codec_enc_init_ver(ctx, iface, cfg, flags, VPX_ENCODER_ABI_VERSION) 743 744 /*!\brief Initialize multi-encoder instance 745 * 746 * Initializes multi-encoder context using the given interface. 747 * Applications should call the vpx_codec_enc_init_multi convenience macro 748 * instead of this function directly, to ensure that the ABI version number 749 * parameter is properly initialized. 750 * 751 * \param[in] ctx Pointer to this instance's context. 752 * \param[in] iface Pointer to the algorithm interface to use. 753 * \param[in] cfg Configuration to use, if known. May be NULL. 754 * \param[in] num_enc Total number of encoders. 755 * \param[in] flags Bitfield of VPX_CODEC_USE_* flags 756 * \param[in] dsf Pointer to down-sampling factors. 757 * \param[in] ver ABI version number. Must be set to 758 * VPX_ENCODER_ABI_VERSION 759 * \retval #VPX_CODEC_OK 760 * The decoder algorithm initialized. 761 * \retval #VPX_CODEC_MEM_ERROR 762 * Memory allocation failed. 763 */ 764 vpx_codec_err_t vpx_codec_enc_init_multi_ver( 765 vpx_codec_ctx_t *ctx, vpx_codec_iface_t *iface, vpx_codec_enc_cfg_t *cfg, 766 int num_enc, vpx_codec_flags_t flags, vpx_rational_t *dsf, int ver); 767 768 /*!\brief Convenience macro for vpx_codec_enc_init_multi_ver() 769 * 770 * Ensures the ABI version parameter is properly set. 771 */ 772 #define vpx_codec_enc_init_multi(ctx, iface, cfg, num_enc, flags, dsf) \ 773 vpx_codec_enc_init_multi_ver(ctx, iface, cfg, num_enc, flags, dsf, \ 774 VPX_ENCODER_ABI_VERSION) 775 776 /*!\brief Get a default configuration 777 * 778 * Initializes a encoder configuration structure with default values. Supports 779 * the notion of "usages" so that an algorithm may offer different default 780 * settings depending on the user's intended goal. This function \ref SHOULD 781 * be called by all applications to initialize the configuration structure 782 * before specializing the configuration with application specific values. 783 * 784 * \param[in] iface Pointer to the algorithm interface to use. 785 * \param[out] cfg Configuration buffer to populate. 786 * \param[in] usage Must be set to 0. 787 * 788 * \retval #VPX_CODEC_OK 789 * The configuration was populated. 790 * \retval #VPX_CODEC_INCAPABLE 791 * Interface is not an encoder interface. 792 * \retval #VPX_CODEC_INVALID_PARAM 793 * A parameter was NULL, or the usage value was not recognized. 794 */ 795 vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface, 796 vpx_codec_enc_cfg_t *cfg, 797 unsigned int usage); 798 799 /*!\brief Set or change configuration 800 * 801 * Reconfigures an encoder instance according to the given configuration. 802 * 803 * \param[in] ctx Pointer to this instance's context 804 * \param[in] cfg Configuration buffer to use 805 * 806 * \retval #VPX_CODEC_OK 807 * The configuration was populated. 808 * \retval #VPX_CODEC_INCAPABLE 809 * Interface is not an encoder interface. 810 * \retval #VPX_CODEC_INVALID_PARAM 811 * A parameter was NULL, or the usage value was not recognized. 812 */ 813 vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx, 814 const vpx_codec_enc_cfg_t *cfg); 815 816 /*!\brief Get global stream headers 817 * 818 * Retrieves a stream level global header packet, if supported by the codec. 819 * 820 * \param[in] ctx Pointer to this instance's context 821 * 822 * \retval NULL 823 * Encoder does not support global header 824 * \retval Non-NULL 825 * Pointer to buffer containing global header packet 826 */ 827 vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx); 828 829 /*!\brief deadline parameter analogous to VPx REALTIME mode. */ 830 #define VPX_DL_REALTIME (1) 831 /*!\brief deadline parameter analogous to VPx GOOD QUALITY mode. */ 832 #define VPX_DL_GOOD_QUALITY (1000000) 833 /*!\brief deadline parameter analogous to VPx BEST QUALITY mode. */ 834 #define VPX_DL_BEST_QUALITY (0) 835 /*!\brief Encode a frame 836 * 837 * Encodes a video frame at the given "presentation time." The presentation 838 * time stamp (PTS) \ref MUST be strictly increasing. 839 * 840 * The encoder supports the notion of a soft real-time deadline. Given a 841 * non-zero value to the deadline parameter, the encoder will make a "best 842 * effort" guarantee to return before the given time slice expires. It is 843 * implicit that limiting the available time to encode will degrade the 844 * output quality. The encoder can be given an unlimited time to produce the 845 * best possible frame by specifying a deadline of '0'. This deadline 846 * supersedes the VPx notion of "best quality, good quality, realtime". 847 * Applications that wish to map these former settings to the new deadline 848 * based system can use the symbols #VPX_DL_REALTIME, #VPX_DL_GOOD_QUALITY, 849 * and #VPX_DL_BEST_QUALITY. 850 * 851 * When the last frame has been passed to the encoder, this function should 852 * continue to be called, with the img parameter set to NULL. This will 853 * signal the end-of-stream condition to the encoder and allow it to encode 854 * any held buffers. Encoding is complete when vpx_codec_encode() is called 855 * and vpx_codec_get_cx_data() returns no data. 856 * 857 * \param[in] ctx Pointer to this instance's context 858 * \param[in] img Image data to encode, NULL to flush. 859 * \param[in] pts Presentation time stamp, in timebase units. 860 * \param[in] duration Duration to show frame, in timebase units. 861 * \param[in] flags Flags to use for encoding this frame. 862 * \param[in] deadline Time to spend encoding, in microseconds. (0=infinite) 863 * 864 * \retval #VPX_CODEC_OK 865 * The configuration was populated. 866 * \retval #VPX_CODEC_INCAPABLE 867 * Interface is not an encoder interface. 868 * \retval #VPX_CODEC_INVALID_PARAM 869 * A parameter was NULL, the image format is unsupported, etc. 870 */ 871 vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx, const vpx_image_t *img, 872 vpx_codec_pts_t pts, unsigned long duration, 873 vpx_enc_frame_flags_t flags, 874 unsigned long deadline); 875 876 /*!\brief Set compressed data output buffer 877 * 878 * Sets the buffer that the codec should output the compressed data 879 * into. This call effectively sets the buffer pointer returned in the 880 * next VPX_CODEC_CX_FRAME_PKT packet. Subsequent packets will be 881 * appended into this buffer. The buffer is preserved across frames, 882 * so applications must periodically call this function after flushing 883 * the accumulated compressed data to disk or to the network to reset 884 * the pointer to the buffer's head. 885 * 886 * `pad_before` bytes will be skipped before writing the compressed 887 * data, and `pad_after` bytes will be appended to the packet. The size 888 * of the packet will be the sum of the size of the actual compressed 889 * data, pad_before, and pad_after. The padding bytes will be preserved 890 * (not overwritten). 891 * 892 * Note that calling this function does not guarantee that the returned 893 * compressed data will be placed into the specified buffer. In the 894 * event that the encoded data will not fit into the buffer provided, 895 * the returned packet \ref MAY point to an internal buffer, as it would 896 * if this call were never used. In this event, the output packet will 897 * NOT have any padding, and the application must free space and copy it 898 * to the proper place. This is of particular note in configurations 899 * that may output multiple packets for a single encoded frame (e.g., lagged 900 * encoding) or if the application does not reset the buffer periodically. 901 * 902 * Applications may restore the default behavior of the codec providing 903 * the compressed data buffer by calling this function with a NULL 904 * buffer. 905 * 906 * Applications \ref MUSTNOT call this function during iteration of 907 * vpx_codec_get_cx_data(). 908 * 909 * \param[in] ctx Pointer to this instance's context 910 * \param[in] buf Buffer to store compressed data into 911 * \param[in] pad_before Bytes to skip before writing compressed data 912 * \param[in] pad_after Bytes to skip after writing compressed data 913 * 914 * \retval #VPX_CODEC_OK 915 * The buffer was set successfully. 916 * \retval #VPX_CODEC_INVALID_PARAM 917 * A parameter was NULL, the image format is unsupported, etc. 918 */ 919 vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t *ctx, 920 const vpx_fixed_buf_t *buf, 921 unsigned int pad_before, 922 unsigned int pad_after); 923 924 /*!\brief Encoded data iterator 925 * 926 * Iterates over a list of data packets to be passed from the encoder to the 927 * application. The different kinds of packets available are enumerated in 928 * #vpx_codec_cx_pkt_kind. 929 * 930 * #VPX_CODEC_CX_FRAME_PKT packets should be passed to the application's 931 * muxer. Multiple compressed frames may be in the list. 932 * #VPX_CODEC_STATS_PKT packets should be appended to a global buffer. 933 * 934 * The application \ref MUST silently ignore any packet kinds that it does 935 * not recognize or support. 936 * 937 * The data buffers returned from this function are only guaranteed to be 938 * valid until the application makes another call to any vpx_codec_* function. 939 * 940 * \param[in] ctx Pointer to this instance's context 941 * \param[in,out] iter Iterator storage, initialized to NULL 942 * 943 * \return Returns a pointer to an output data packet (compressed frame data, 944 * two-pass statistics, etc.) or NULL to signal end-of-list. 945 * 946 */ 947 const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx, 948 vpx_codec_iter_t *iter); 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 /*!@} - end defgroup encoder*/ 965 #ifdef __cplusplus 966 } 967 #endif 968 #endif // VPX_VPX_VPX_ENCODER_H_