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 /*! Temporal+Spatial Scalability: Maximum number of coding layers */ 46 #define VPX_MAX_LAYERS 12 // 3 temporal + 4 spatial layers are allowed. 47 48 /*!\deprecated Use #VPX_MAX_LAYERS instead. */ 49 #define MAX_LAYERS VPX_MAX_LAYERS // 3 temporal + 4 spatial layers allowed. 50 51 /*! Spatial Scalability: Maximum number of coding layers */ 52 #define VPX_SS_MAX_LAYERS 5 53 54 /*! Spatial Scalability: Default number of coding layers */ 55 #define VPX_SS_DEFAULT_LAYERS 1 56 57 /*!\brief Current ABI version number 58 * 59 * \internal 60 * If this file is altered in any way that changes the ABI, this value 61 * must be bumped. Examples include, but are not limited to, changing 62 * types, removing or reassigning enums, adding/removing/rearranging 63 * fields to structures 64 */ 65 #define VPX_ENCODER_ABI_VERSION (5 + VPX_CODEC_ABI_VERSION) /**<\hideinitializer*/ 66 67 68 /*! \brief Encoder capabilities bitfield 69 * 70 * Each encoder advertises the capabilities it supports as part of its 71 * ::vpx_codec_iface_t interface structure. Capabilities are extra 72 * interfaces or functionality, and are not required to be supported 73 * by an encoder. 74 * 75 * The available flags are specified by VPX_CODEC_CAP_* defines. 76 */ 77 #define VPX_CODEC_CAP_PSNR 0x10000 /**< Can issue PSNR packets */ 78 79 /*! Can output one partition at a time. Each partition is returned in its 80 * own VPX_CODEC_CX_FRAME_PKT, with the FRAME_IS_FRAGMENT flag set for 81 * every partition but the last. In this mode all frames are always 82 * returned partition by partition. 83 */ 84 #define VPX_CODEC_CAP_OUTPUT_PARTITION 0x20000 85 86 /*! Can support input images at greater than 8 bitdepth. 87 */ 88 #define VPX_CODEC_CAP_HIGHBITDEPTH 0x40000 89 90 /*! \brief Initialization-time Feature Enabling 91 * 92 * Certain codec features must be known at initialization time, to allow 93 * for proper memory allocation. 94 * 95 * The available flags are specified by VPX_CODEC_USE_* defines. 96 */ 97 #define VPX_CODEC_USE_PSNR 0x10000 /**< Calculate PSNR on each frame */ 98 #define VPX_CODEC_USE_OUTPUT_PARTITION 0x20000 /**< Make the encoder output one 99 partition at a time. */ 100 #define VPX_CODEC_USE_HIGHBITDEPTH 0x40000 /**< Use high bitdepth */ 101 102 103 /*!\brief Generic fixed size buffer structure 104 * 105 * This structure is able to hold a reference to any fixed size buffer. 106 */ 107 typedef struct vpx_fixed_buf { 108 void *buf; /**< Pointer to the data */ 109 size_t sz; /**< Length of the buffer, in chars */ 110 } vpx_fixed_buf_t; /**< alias for struct vpx_fixed_buf */ 111 112 113 /*!\brief Time Stamp Type 114 * 115 * An integer, which when multiplied by the stream's time base, provides 116 * the absolute time of a sample. 117 */ 118 typedef int64_t vpx_codec_pts_t; 119 120 121 /*!\brief Compressed Frame Flags 122 * 123 * This type represents a bitfield containing information about a compressed 124 * frame that may be useful to an application. The most significant 16 bits 125 * can be used by an algorithm to provide additional detail, for example to 126 * support frame types that are codec specific (MPEG-1 D-frames for example) 127 */ 128 typedef uint32_t vpx_codec_frame_flags_t; 129 #define VPX_FRAME_IS_KEY 0x1 /**< frame is the start of a GOP */ 130 #define VPX_FRAME_IS_DROPPABLE 0x2 /**< frame can be dropped without affecting 131 the stream (no future frame depends on 132 this one) */ 133 #define VPX_FRAME_IS_INVISIBLE 0x4 /**< frame should be decoded but will not 134 be shown */ 135 #define VPX_FRAME_IS_FRAGMENT 0x8 /**< this is a fragment of the encoded 136 frame */ 137 138 /*!\brief Error Resilient flags 139 * 140 * These flags define which error resilient features to enable in the 141 * encoder. The flags are specified through the 142 * vpx_codec_enc_cfg::g_error_resilient variable. 143 */ 144 typedef uint32_t vpx_codec_er_flags_t; 145 #define VPX_ERROR_RESILIENT_DEFAULT 0x1 /**< Improve resiliency against 146 losses of whole frames */ 147 #define VPX_ERROR_RESILIENT_PARTITIONS 0x2 /**< The frame partitions are 148 independently decodable by the 149 bool decoder, meaning that 150 partitions can be decoded even 151 though earlier partitions have 152 been lost. Note that intra 153 predicition is still done over 154 the partition boundary. */ 155 156 /*!\brief Encoder output packet variants 157 * 158 * This enumeration lists the different kinds of data packets that can be 159 * returned by calls to vpx_codec_get_cx_data(). Algorithms \ref MAY 160 * extend this list to provide additional functionality. 161 */ 162 enum vpx_codec_cx_pkt_kind { 163 VPX_CODEC_CX_FRAME_PKT, /**< Compressed video frame */ 164 VPX_CODEC_STATS_PKT, /**< Two-pass statistics for this frame */ 165 VPX_CODEC_FPMB_STATS_PKT, /**< first pass mb statistics for this frame */ 166 VPX_CODEC_PSNR_PKT, /**< PSNR statistics for this frame */ 167 // Spatial SVC is still experimental and may be removed before the next ABI 168 // bump. 169 #if VPX_ENCODER_ABI_VERSION > (5 + VPX_CODEC_ABI_VERSION) 170 VPX_CODEC_SPATIAL_SVC_LAYER_SIZES, /**< Sizes for each layer in this frame*/ 171 VPX_CODEC_SPATIAL_SVC_LAYER_PSNR, /**< PSNR for each layer in this frame*/ 172 #endif 173 VPX_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions */ 174 }; 175 176 177 /*!\brief Encoder output packet 178 * 179 * This structure contains the different kinds of output data the encoder 180 * may produce while compressing a frame. 181 */ 182 typedef struct vpx_codec_cx_pkt { 183 enum vpx_codec_cx_pkt_kind kind; /**< packet variant */ 184 union { 185 struct { 186 void *buf; /**< compressed data buffer */ 187 size_t sz; /**< length of compressed data */ 188 vpx_codec_pts_t pts; /**< time stamp to show frame 189 (in timebase units) */ 190 unsigned long duration; /**< duration to show frame 191 (in timebase units) */ 192 vpx_codec_frame_flags_t flags; /**< flags for this frame */ 193 int partition_id; /**< the partition id 194 defines the decoding order 195 of the partitions. Only 196 applicable when "output partition" 197 mode is enabled. First partition 198 has id 0.*/ 199 200 } frame; /**< data for compressed frame packet */ 201 vpx_fixed_buf_t twopass_stats; /**< data for two-pass packet */ 202 vpx_fixed_buf_t firstpass_mb_stats; /**< first pass mb packet */ 203 struct vpx_psnr_pkt { 204 unsigned int samples[4]; /**< Number of samples, total/y/u/v */ 205 uint64_t sse[4]; /**< sum squared error, total/y/u/v */ 206 double psnr[4]; /**< PSNR, total/y/u/v */ 207 } psnr; /**< data for PSNR packet */ 208 vpx_fixed_buf_t raw; /**< data for arbitrary packets */ 209 // Spatial SVC is still experimental and may be removed before the next 210 // ABI bump. 211 #if VPX_ENCODER_ABI_VERSION > (5 + VPX_CODEC_ABI_VERSION) 212 size_t layer_sizes[VPX_SS_MAX_LAYERS]; 213 struct vpx_psnr_pkt layer_psnr[VPX_SS_MAX_LAYERS]; 214 #endif 215 216 /* This packet size is fixed to allow codecs to extend this 217 * interface without having to manage storage for raw packets, 218 * i.e., if it's smaller than 128 bytes, you can store in the 219 * packet list directly. 220 */ 221 char pad[128 - sizeof(enum vpx_codec_cx_pkt_kind)]; /**< fixed sz */ 222 } data; /**< packet data */ 223 } vpx_codec_cx_pkt_t; /**< alias for struct vpx_codec_cx_pkt */ 224 225 226 /*!\brief Encoder return output buffer callback 227 * 228 * This callback function, when registered, returns with packets when each 229 * spatial layer is encoded. 230 */ 231 // putting the definitions here for now. (agrange: find if there 232 // is a better place for this) 233 typedef void (* vpx_codec_enc_output_cx_pkt_cb_fn_t)(vpx_codec_cx_pkt_t *pkt, 234 void *user_data); 235 236 /*!\brief Callback function pointer / user data pair storage */ 237 typedef struct vpx_codec_enc_output_cx_cb_pair { 238 vpx_codec_enc_output_cx_pkt_cb_fn_t output_cx_pkt; /**< Callback function */ 239 void *user_priv; /**< Pointer to private data */ 240 } vpx_codec_priv_output_cx_pkt_cb_pair_t; 241 242 /*!\brief Rational Number 243 * 244 * This structure holds a fractional value. 245 */ 246 typedef struct vpx_rational { 247 int num; /**< fraction numerator */ 248 int den; /**< fraction denominator */ 249 } vpx_rational_t; /**< alias for struct vpx_rational */ 250 251 252 /*!\brief Multi-pass Encoding Pass */ 253 enum vpx_enc_pass { 254 VPX_RC_ONE_PASS, /**< Single pass mode */ 255 VPX_RC_FIRST_PASS, /**< First pass of multi-pass mode */ 256 VPX_RC_LAST_PASS /**< Final pass of multi-pass mode */ 257 }; 258 259 260 /*!\brief Rate control mode */ 261 enum vpx_rc_mode { 262 VPX_VBR, /**< Variable Bit Rate (VBR) mode */ 263 VPX_CBR, /**< Constant Bit Rate (CBR) mode */ 264 VPX_CQ, /**< Constrained Quality (CQ) mode */ 265 VPX_Q, /**< Constant Quality (Q) mode */ 266 }; 267 268 269 /*!\brief Keyframe placement mode. 270 * 271 * This enumeration determines whether keyframes are placed automatically by 272 * the encoder or whether this behavior is disabled. Older releases of this 273 * SDK were implemented such that VPX_KF_FIXED meant keyframes were disabled. 274 * This name is confusing for this behavior, so the new symbols to be used 275 * are VPX_KF_AUTO and VPX_KF_DISABLED. 276 */ 277 enum vpx_kf_mode { 278 VPX_KF_FIXED, /**< deprecated, implies VPX_KF_DISABLED */ 279 VPX_KF_AUTO, /**< Encoder determines optimal placement automatically */ 280 VPX_KF_DISABLED = 0 /**< Encoder does not place keyframes. */ 281 }; 282 283 284 /*!\brief Encoded Frame Flags 285 * 286 * This type indicates a bitfield to be passed to vpx_codec_encode(), defining 287 * per-frame boolean values. By convention, bits common to all codecs will be 288 * named VPX_EFLAG_*, and bits specific to an algorithm will be named 289 * /algo/_eflag_*. The lower order 16 bits are reserved for common use. 290 */ 291 typedef long vpx_enc_frame_flags_t; 292 #define VPX_EFLAG_FORCE_KF (1<<0) /**< Force this frame to be a keyframe */ 293 294 295 /*!\brief Encoder configuration structure 296 * 297 * This structure contains the encoder settings that have common representations 298 * across all codecs. This doesn't imply that all codecs support all features, 299 * however. 300 */ 301 typedef struct vpx_codec_enc_cfg { 302 /* 303 * generic settings (g) 304 */ 305 306 /*!\brief Algorithm specific "usage" value 307 * 308 * Algorithms may define multiple values for usage, which may convey the 309 * intent of how the application intends to use the stream. If this value 310 * is non-zero, consult the documentation for the codec to determine its 311 * meaning. 312 */ 313 unsigned int g_usage; 314 315 316 /*!\brief Maximum number of threads to use 317 * 318 * For multi-threaded implementations, use no more than this number of 319 * threads. The codec may use fewer threads than allowed. The value 320 * 0 is equivalent to the value 1. 321 */ 322 unsigned int g_threads; 323 324 325 /*!\brief Bitstream profile to use 326 * 327 * Some codecs support a notion of multiple bitstream profiles. Typically 328 * this maps to a set of features that are turned on or off. Often the 329 * profile to use is determined by the features of the intended decoder. 330 * Consult the documentation for the codec to determine the valid values 331 * for this parameter, or set to zero for a sane default. 332 */ 333 unsigned int g_profile; /**< profile of bitstream to use */ 334 335 336 337 /*!\brief Width of the frame 338 * 339 * This value identifies the presentation resolution of the frame, 340 * in pixels. Note that the frames passed as input to the encoder must 341 * have this resolution. Frames will be presented by the decoder in this 342 * resolution, independent of any spatial resampling the encoder may do. 343 */ 344 unsigned int g_w; 345 346 347 /*!\brief Height of the frame 348 * 349 * This value identifies the presentation resolution of the frame, 350 * in pixels. Note that the frames passed as input to the encoder must 351 * have this resolution. Frames will be presented by the decoder in this 352 * resolution, independent of any spatial resampling the encoder may do. 353 */ 354 unsigned int g_h; 355 356 /*!\brief Bit-depth of the codec 357 * 358 * This value identifies the bit_depth of the codec, 359 * Only certain bit-depths are supported as identified in the 360 * vpx_bit_depth_t enum. 361 */ 362 vpx_bit_depth_t g_bit_depth; 363 364 /*!\brief Bit-depth of the input frames 365 * 366 * This value identifies the bit_depth of the input frames in bits. 367 * Note that the frames passed as input to the encoder must have 368 * this bit-depth. 369 */ 370 unsigned int g_input_bit_depth; 371 372 /*!\brief Stream timebase units 373 * 374 * Indicates the smallest interval of time, in seconds, used by the stream. 375 * For fixed frame rate material, or variable frame rate material where 376 * frames are timed at a multiple of a given clock (ex: video capture), 377 * the \ref RECOMMENDED method is to set the timebase to the reciprocal 378 * of the frame rate (ex: 1001/30000 for 29.970 Hz NTSC). This allows the 379 * pts to correspond to the frame number, which can be handy. For 380 * re-encoding video from containers with absolute time timestamps, the 381 * \ref RECOMMENDED method is to set the timebase to that of the parent 382 * container or multimedia framework (ex: 1/1000 for ms, as in FLV). 383 */ 384 struct vpx_rational g_timebase; 385 386 387 /*!\brief Enable error resilient modes. 388 * 389 * The error resilient bitfield indicates to the encoder which features 390 * it should enable to take measures for streaming over lossy or noisy 391 * links. 392 */ 393 vpx_codec_er_flags_t g_error_resilient; 394 395 396 /*!\brief Multi-pass Encoding Mode 397 * 398 * This value should be set to the current phase for multi-pass encoding. 399 * For single pass, set to #VPX_RC_ONE_PASS. 400 */ 401 enum vpx_enc_pass g_pass; 402 403 404 /*!\brief Allow lagged encoding 405 * 406 * If set, this value allows the encoder to consume a number of input 407 * frames before producing output frames. This allows the encoder to 408 * base decisions for the current frame on future frames. This does 409 * increase the latency of the encoding pipeline, so it is not appropriate 410 * in all situations (ex: realtime encoding). 411 * 412 * Note that this is a maximum value -- the encoder may produce frames 413 * sooner than the given limit. Set this value to 0 to disable this 414 * feature. 415 */ 416 unsigned int g_lag_in_frames; 417 418 419 /* 420 * rate control settings (rc) 421 */ 422 423 /*!\brief Temporal resampling configuration, if supported by the codec. 424 * 425 * Temporal resampling allows the codec to "drop" frames as a strategy to 426 * meet its target data rate. This can cause temporal discontinuities in 427 * the encoded video, which may appear as stuttering during playback. This 428 * trade-off is often acceptable, but for many applications is not. It can 429 * be disabled in these cases. 430 * 431 * Note that not all codecs support this feature. All vpx VPx codecs do. 432 * For other codecs, consult the documentation for that algorithm. 433 * 434 * This threshold is described as a percentage of the target data buffer. 435 * When the data buffer falls below this percentage of fullness, a 436 * dropped frame is indicated. Set the threshold to zero (0) to disable 437 * this feature. 438 */ 439 unsigned int rc_dropframe_thresh; 440 441 442 /*!\brief Enable/disable spatial resampling, if supported by the codec. 443 * 444 * Spatial resampling allows the codec to compress a lower resolution 445 * version of the frame, which is then upscaled by the encoder to the 446 * correct presentation resolution. This increases visual quality at 447 * low data rates, at the expense of CPU time on the encoder/decoder. 448 */ 449 unsigned int rc_resize_allowed; 450 451 /*!\brief Internal coded frame width. 452 * 453 * If spatial resampling is enabled this specifies the width of the 454 * encoded frame. 455 */ 456 unsigned int rc_scaled_width; 457 458 /*!\brief Internal coded frame height. 459 * 460 * If spatial resampling is enabled this specifies the height of the 461 * encoded frame. 462 */ 463 unsigned int rc_scaled_height; 464 465 /*!\brief Spatial resampling up watermark. 466 * 467 * This threshold is described as a percentage of the target data buffer. 468 * When the data buffer rises above this percentage of fullness, the 469 * encoder will step up to a higher resolution version of the frame. 470 */ 471 unsigned int rc_resize_up_thresh; 472 473 474 /*!\brief Spatial resampling down watermark. 475 * 476 * This threshold is described as a percentage of the target data buffer. 477 * When the data buffer falls below this percentage of fullness, the 478 * encoder will step down to a lower resolution version of the frame. 479 */ 480 unsigned int rc_resize_down_thresh; 481 482 483 /*!\brief Rate control algorithm to use. 484 * 485 * Indicates whether the end usage of this stream is to be streamed over 486 * a bandwidth constrained link, indicating that Constant Bit Rate (CBR) 487 * mode should be used, or whether it will be played back on a high 488 * bandwidth link, as from a local disk, where higher variations in 489 * bitrate are acceptable. 490 */ 491 enum vpx_rc_mode rc_end_usage; 492 493 494 /*!\brief Two-pass stats buffer. 495 * 496 * A buffer containing all of the stats packets produced in the first 497 * pass, concatenated. 498 */ 499 vpx_fixed_buf_t rc_twopass_stats_in; 500 501 /*!\brief first pass mb stats buffer. 502 * 503 * A buffer containing all of the first pass mb stats packets produced 504 * in the first pass, concatenated. 505 */ 506 vpx_fixed_buf_t rc_firstpass_mb_stats_in; 507 508 /*!\brief Target data rate 509 * 510 * Target bandwidth to use for this stream, in kilobits per second. 511 */ 512 unsigned int rc_target_bitrate; 513 514 515 /* 516 * quantizer settings 517 */ 518 519 520 /*!\brief Minimum (Best Quality) Quantizer 521 * 522 * The quantizer is the most direct control over the quality of the 523 * encoded image. The range of valid values for the quantizer is codec 524 * specific. Consult the documentation for the codec to determine the 525 * values to use. To determine the range programmatically, call 526 * vpx_codec_enc_config_default() with a usage value of 0. 527 */ 528 unsigned int rc_min_quantizer; 529 530 531 /*!\brief Maximum (Worst Quality) Quantizer 532 * 533 * The quantizer is the most direct control over the quality of the 534 * encoded image. The range of valid values for the quantizer is codec 535 * specific. Consult the documentation for the codec to determine the 536 * values to use. To determine the range programmatically, call 537 * vpx_codec_enc_config_default() with a usage value of 0. 538 */ 539 unsigned int rc_max_quantizer; 540 541 542 /* 543 * bitrate tolerance 544 */ 545 546 547 /*!\brief Rate control adaptation undershoot control 548 * 549 * This value, expressed as a percentage of the target bitrate, 550 * controls the maximum allowed adaptation speed of the codec. 551 * This factor controls the maximum amount of bits that can 552 * be subtracted from the target bitrate in order to compensate 553 * for prior overshoot. 554 * 555 * Valid values in the range 0-1000. 556 */ 557 unsigned int rc_undershoot_pct; 558 559 560 /*!\brief Rate control adaptation overshoot control 561 * 562 * This value, expressed as a percentage of the target bitrate, 563 * controls the maximum allowed adaptation speed of the codec. 564 * This factor controls the maximum amount of bits that can 565 * be added to the target bitrate in order to compensate for 566 * prior undershoot. 567 * 568 * Valid values in the range 0-1000. 569 */ 570 unsigned int rc_overshoot_pct; 571 572 573 /* 574 * decoder buffer model parameters 575 */ 576 577 578 /*!\brief Decoder Buffer Size 579 * 580 * This value indicates the amount of data that may be buffered by the 581 * decoding application. Note that this value is expressed in units of 582 * time (milliseconds). For example, a value of 5000 indicates that the 583 * client will buffer (at least) 5000ms worth of encoded data. Use the 584 * target bitrate (#rc_target_bitrate) to convert to bits/bytes, if 585 * necessary. 586 */ 587 unsigned int rc_buf_sz; 588 589 590 /*!\brief Decoder Buffer Initial Size 591 * 592 * This value indicates the amount of data that will be buffered by the 593 * decoding application prior to beginning playback. This value is 594 * expressed in units of time (milliseconds). Use the target bitrate 595 * (#rc_target_bitrate) to convert to bits/bytes, if necessary. 596 */ 597 unsigned int rc_buf_initial_sz; 598 599 600 /*!\brief Decoder Buffer Optimal Size 601 * 602 * This value indicates the amount of data that the encoder should try 603 * to maintain in the decoder's buffer. This value is expressed in units 604 * of time (milliseconds). Use the target bitrate (#rc_target_bitrate) 605 * to convert to bits/bytes, if necessary. 606 */ 607 unsigned int rc_buf_optimal_sz; 608 609 610 /* 611 * 2 pass rate control parameters 612 */ 613 614 615 /*!\brief Two-pass mode CBR/VBR bias 616 * 617 * Bias, expressed on a scale of 0 to 100, for determining target size 618 * for the current frame. The value 0 indicates the optimal CBR mode 619 * value should be used. The value 100 indicates the optimal VBR mode 620 * value should be used. Values in between indicate which way the 621 * encoder should "lean." 622 */ 623 unsigned int rc_2pass_vbr_bias_pct; /**< RC mode bias between CBR and VBR(0-100: 0->CBR, 100->VBR) */ 624 625 626 /*!\brief Two-pass mode per-GOP minimum bitrate 627 * 628 * This value, expressed as a percentage of the target bitrate, indicates 629 * the minimum bitrate to be used for a single GOP (aka "section") 630 */ 631 unsigned int rc_2pass_vbr_minsection_pct; 632 633 634 /*!\brief Two-pass mode per-GOP maximum bitrate 635 * 636 * This value, expressed as a percentage of the target bitrate, indicates 637 * the maximum bitrate to be used for a single GOP (aka "section") 638 */ 639 unsigned int rc_2pass_vbr_maxsection_pct; 640 641 642 /* 643 * keyframing settings (kf) 644 */ 645 646 /*!\brief Keyframe placement mode 647 * 648 * This value indicates whether the encoder should place keyframes at a 649 * fixed interval, or determine the optimal placement automatically 650 * (as governed by the #kf_min_dist and #kf_max_dist parameters) 651 */ 652 enum vpx_kf_mode kf_mode; 653 654 655 /*!\brief Keyframe minimum interval 656 * 657 * This value, expressed as a number of frames, prevents the encoder from 658 * placing a keyframe nearer than kf_min_dist to the previous keyframe. At 659 * least kf_min_dist frames non-keyframes will be coded before the next 660 * keyframe. Set kf_min_dist equal to kf_max_dist for a fixed interval. 661 */ 662 unsigned int kf_min_dist; 663 664 665 /*!\brief Keyframe maximum interval 666 * 667 * This value, expressed as a number of frames, forces the encoder to code 668 * a keyframe if one has not been coded in the last kf_max_dist frames. 669 * A value of 0 implies all frames will be keyframes. Set kf_min_dist 670 * equal to kf_max_dist for a fixed interval. 671 */ 672 unsigned int kf_max_dist; 673 674 /* 675 * Spatial scalability settings (ss) 676 */ 677 678 /*!\brief Number of spatial coding layers. 679 * 680 * This value specifies the number of spatial coding layers to be used. 681 */ 682 unsigned int ss_number_layers; 683 684 /*!\brief Enable auto alt reference flags for each spatial layer. 685 * 686 * These values specify if auto alt reference frame is enabled for each 687 * spatial layer. 688 */ 689 int ss_enable_auto_alt_ref[VPX_SS_MAX_LAYERS]; 690 691 /*!\brief Target bitrate for each spatial layer. 692 * 693 * These values specify the target coding bitrate to be used for each 694 * spatial layer. 695 */ 696 unsigned int ss_target_bitrate[VPX_SS_MAX_LAYERS]; 697 698 /*!\brief Number of temporal coding layers. 699 * 700 * This value specifies the number of temporal layers to be used. 701 */ 702 unsigned int ts_number_layers; 703 704 /*!\brief Target bitrate for each temporal layer. 705 * 706 * These values specify the target coding bitrate to be used for each 707 * temporal layer. 708 */ 709 unsigned int ts_target_bitrate[VPX_TS_MAX_LAYERS]; 710 711 /*!\brief Frame rate decimation factor for each temporal layer. 712 * 713 * These values specify the frame rate decimation factors to apply 714 * to each temporal layer. 715 */ 716 unsigned int ts_rate_decimator[VPX_TS_MAX_LAYERS]; 717 718 /*!\brief Length of the sequence defining frame temporal layer membership. 719 * 720 * This value specifies the length of the sequence that defines the 721 * membership of frames to temporal layers. For example, if the 722 * ts_periodicity = 8, then the frames are assigned to coding layers with a 723 * repeated sequence of length 8. 724 */ 725 unsigned int ts_periodicity; 726 727 /*!\brief Template defining the membership of frames to temporal layers. 728 * 729 * This array defines the membership of frames to temporal coding layers. 730 * For a 2-layer encoding that assigns even numbered frames to one temporal 731 * layer (0) and odd numbered frames to a second temporal layer (1) with 732 * ts_periodicity=8, then ts_layer_id = (0,1,0,1,0,1,0,1). 733 */ 734 unsigned int ts_layer_id[VPX_TS_MAX_PERIODICITY]; 735 736 /*!\brief Target bitrate for each spatial/temporal layer. 737 * 738 * These values specify the target coding bitrate to be used for each 739 * spatial/temporal layer. 740 * 741 */ 742 unsigned int layer_target_bitrate[VPX_MAX_LAYERS]; 743 744 /*!\brief Temporal layering mode indicating which temporal layering scheme to use. 745 * 746 * The value (refer to VP9E_TEMPORAL_LAYERING_MODE) specifies the 747 * temporal layering mode to use. 748 * 749 */ 750 int temporal_layering_mode; 751 } vpx_codec_enc_cfg_t; /**< alias for struct vpx_codec_enc_cfg */ 752 753 /*!\brief vp9 svc extra configure parameters 754 * 755 * This defines max/min quantizers and scale factors for each layer 756 * 757 */ 758 typedef struct vpx_svc_parameters { 759 int max_quantizers[VPX_MAX_LAYERS]; /**< Max Q for each layer */ 760 int min_quantizers[VPX_MAX_LAYERS]; /**< Min Q for each layer */ 761 int scaling_factor_num[VPX_MAX_LAYERS]; /**< Scaling factor-numerator */ 762 int scaling_factor_den[VPX_MAX_LAYERS]; /**< Scaling factor-denominator */ 763 int temporal_layering_mode; /**< Temporal layering mode */ 764 } vpx_svc_extra_cfg_t; 765 766 767 /*!\brief Initialize an encoder instance 768 * 769 * Initializes a encoder context using the given interface. Applications 770 * should call the vpx_codec_enc_init convenience macro instead of this 771 * function directly, to ensure that the ABI version number parameter 772 * is properly initialized. 773 * 774 * If the library was configured with --disable-multithread, this call 775 * is not thread safe and should be guarded with a lock if being used 776 * in a multithreaded context. 777 * 778 * \param[in] ctx Pointer to this instance's context. 779 * \param[in] iface Pointer to the algorithm interface to use. 780 * \param[in] cfg Configuration to use, if known. May be NULL. 781 * \param[in] flags Bitfield of VPX_CODEC_USE_* flags 782 * \param[in] ver ABI version number. Must be set to 783 * VPX_ENCODER_ABI_VERSION 784 * \retval #VPX_CODEC_OK 785 * The decoder algorithm initialized. 786 * \retval #VPX_CODEC_MEM_ERROR 787 * Memory allocation failed. 788 */ 789 vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t *ctx, 790 vpx_codec_iface_t *iface, 791 const vpx_codec_enc_cfg_t *cfg, 792 vpx_codec_flags_t flags, 793 int ver); 794 795 796 /*!\brief Convenience macro for vpx_codec_enc_init_ver() 797 * 798 * Ensures the ABI version parameter is properly set. 799 */ 800 #define vpx_codec_enc_init(ctx, iface, cfg, flags) \ 801 vpx_codec_enc_init_ver(ctx, iface, cfg, flags, VPX_ENCODER_ABI_VERSION) 802 803 804 /*!\brief Initialize multi-encoder instance 805 * 806 * Initializes multi-encoder context using the given interface. 807 * Applications should call the vpx_codec_enc_init_multi convenience macro 808 * instead of this function directly, to ensure that the ABI version number 809 * parameter is properly initialized. 810 * 811 * \param[in] ctx Pointer to this instance's context. 812 * \param[in] iface Pointer to the algorithm interface to use. 813 * \param[in] cfg Configuration to use, if known. May be NULL. 814 * \param[in] num_enc Total number of encoders. 815 * \param[in] flags Bitfield of VPX_CODEC_USE_* flags 816 * \param[in] dsf Pointer to down-sampling factors. 817 * \param[in] ver ABI version number. Must be set to 818 * VPX_ENCODER_ABI_VERSION 819 * \retval #VPX_CODEC_OK 820 * The decoder algorithm initialized. 821 * \retval #VPX_CODEC_MEM_ERROR 822 * Memory allocation failed. 823 */ 824 vpx_codec_err_t vpx_codec_enc_init_multi_ver(vpx_codec_ctx_t *ctx, 825 vpx_codec_iface_t *iface, 826 vpx_codec_enc_cfg_t *cfg, 827 int num_enc, 828 vpx_codec_flags_t flags, 829 vpx_rational_t *dsf, 830 int ver); 831 832 833 /*!\brief Convenience macro for vpx_codec_enc_init_multi_ver() 834 * 835 * Ensures the ABI version parameter is properly set. 836 */ 837 #define vpx_codec_enc_init_multi(ctx, iface, cfg, num_enc, flags, dsf) \ 838 vpx_codec_enc_init_multi_ver(ctx, iface, cfg, num_enc, flags, dsf, \ 839 VPX_ENCODER_ABI_VERSION) 840 841 842 /*!\brief Get a default configuration 843 * 844 * Initializes a encoder configuration structure with default values. Supports 845 * the notion of "usages" so that an algorithm may offer different default 846 * settings depending on the user's intended goal. This function \ref SHOULD 847 * be called by all applications to initialize the configuration structure 848 * before specializing the configuration with application specific values. 849 * 850 * \param[in] iface Pointer to the algorithm interface to use. 851 * \param[out] cfg Configuration buffer to populate. 852 * \param[in] reserved Must set to 0 for VP8 and VP9. 853 * 854 * \retval #VPX_CODEC_OK 855 * The configuration was populated. 856 * \retval #VPX_CODEC_INCAPABLE 857 * Interface is not an encoder interface. 858 * \retval #VPX_CODEC_INVALID_PARAM 859 * A parameter was NULL, or the usage value was not recognized. 860 */ 861 vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface, 862 vpx_codec_enc_cfg_t *cfg, 863 unsigned int reserved); 864 865 866 /*!\brief Set or change configuration 867 * 868 * Reconfigures an encoder instance according to the given configuration. 869 * 870 * \param[in] ctx Pointer to this instance's context 871 * \param[in] cfg Configuration buffer to use 872 * 873 * \retval #VPX_CODEC_OK 874 * The configuration was populated. 875 * \retval #VPX_CODEC_INCAPABLE 876 * Interface is not an encoder interface. 877 * \retval #VPX_CODEC_INVALID_PARAM 878 * A parameter was NULL, or the usage value was not recognized. 879 */ 880 vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx, 881 const vpx_codec_enc_cfg_t *cfg); 882 883 884 /*!\brief Get global stream headers 885 * 886 * Retrieves a stream level global header packet, if supported by the codec. 887 * 888 * \param[in] ctx Pointer to this instance's context 889 * 890 * \retval NULL 891 * Encoder does not support global header 892 * \retval Non-NULL 893 * Pointer to buffer containing global header packet 894 */ 895 vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx); 896 897 898 #define VPX_DL_REALTIME (1) /**< deadline parameter analogous to 899 * VPx REALTIME mode. */ 900 #define VPX_DL_GOOD_QUALITY (1000000) /**< deadline parameter analogous to 901 * VPx GOOD QUALITY mode. */ 902 #define VPX_DL_BEST_QUALITY (0) /**< deadline parameter analogous to 903 * VPx BEST QUALITY mode. */ 904 /*!\brief Encode a frame 905 * 906 * Encodes a video frame at the given "presentation time." The presentation 907 * time stamp (PTS) \ref MUST be strictly increasing. 908 * 909 * The encoder supports the notion of a soft real-time deadline. Given a 910 * non-zero value to the deadline parameter, the encoder will make a "best 911 * effort" guarantee to return before the given time slice expires. It is 912 * implicit that limiting the available time to encode will degrade the 913 * output quality. The encoder can be given an unlimited time to produce the 914 * best possible frame by specifying a deadline of '0'. This deadline 915 * supercedes the VPx notion of "best quality, good quality, realtime". 916 * Applications that wish to map these former settings to the new deadline 917 * based system can use the symbols #VPX_DL_REALTIME, #VPX_DL_GOOD_QUALITY, 918 * and #VPX_DL_BEST_QUALITY. 919 * 920 * When the last frame has been passed to the encoder, this function should 921 * continue to be called, with the img parameter set to NULL. This will 922 * signal the end-of-stream condition to the encoder and allow it to encode 923 * any held buffers. Encoding is complete when vpx_codec_encode() is called 924 * and vpx_codec_get_cx_data() returns no data. 925 * 926 * \param[in] ctx Pointer to this instance's context 927 * \param[in] img Image data to encode, NULL to flush. 928 * \param[in] pts Presentation time stamp, in timebase units. 929 * \param[in] duration Duration to show frame, in timebase units. 930 * \param[in] flags Flags to use for encoding this frame. 931 * \param[in] deadline Time to spend encoding, in microseconds. (0=infinite) 932 * 933 * \retval #VPX_CODEC_OK 934 * The configuration was populated. 935 * \retval #VPX_CODEC_INCAPABLE 936 * Interface is not an encoder interface. 937 * \retval #VPX_CODEC_INVALID_PARAM 938 * A parameter was NULL, the image format is unsupported, etc. 939 */ 940 vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx, 941 const vpx_image_t *img, 942 vpx_codec_pts_t pts, 943 unsigned long duration, 944 vpx_enc_frame_flags_t flags, 945 unsigned long deadline); 946 947 /*!\brief Set compressed data output buffer 948 * 949 * Sets the buffer that the codec should output the compressed data 950 * into. This call effectively sets the buffer pointer returned in the 951 * next VPX_CODEC_CX_FRAME_PKT packet. Subsequent packets will be 952 * appended into this buffer. The buffer is preserved across frames, 953 * so applications must periodically call this function after flushing 954 * the accumulated compressed data to disk or to the network to reset 955 * the pointer to the buffer's head. 956 * 957 * `pad_before` bytes will be skipped before writing the compressed 958 * data, and `pad_after` bytes will be appended to the packet. The size 959 * of the packet will be the sum of the size of the actual compressed 960 * data, pad_before, and pad_after. The padding bytes will be preserved 961 * (not overwritten). 962 * 963 * Note that calling this function does not guarantee that the returned 964 * compressed data will be placed into the specified buffer. In the 965 * event that the encoded data will not fit into the buffer provided, 966 * the returned packet \ref MAY point to an internal buffer, as it would 967 * if this call were never used. In this event, the output packet will 968 * NOT have any padding, and the application must free space and copy it 969 * to the proper place. This is of particular note in configurations 970 * that may output multiple packets for a single encoded frame (e.g., lagged 971 * encoding) or if the application does not reset the buffer periodically. 972 * 973 * Applications may restore the default behavior of the codec providing 974 * the compressed data buffer by calling this function with a NULL 975 * buffer. 976 * 977 * Applications \ref MUSTNOT call this function during iteration of 978 * vpx_codec_get_cx_data(). 979 * 980 * \param[in] ctx Pointer to this instance's context 981 * \param[in] buf Buffer to store compressed data into 982 * \param[in] pad_before Bytes to skip before writing compressed data 983 * \param[in] pad_after Bytes to skip after writing compressed data 984 * 985 * \retval #VPX_CODEC_OK 986 * The buffer was set successfully. 987 * \retval #VPX_CODEC_INVALID_PARAM 988 * A parameter was NULL, the image format is unsupported, etc. 989 */ 990 vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t *ctx, 991 const vpx_fixed_buf_t *buf, 992 unsigned int pad_before, 993 unsigned int pad_after); 994 995 996 /*!\brief Encoded data iterator 997 * 998 * Iterates over a list of data packets to be passed from the encoder to the 999 * application. The different kinds of packets available are enumerated in 1000 * #vpx_codec_cx_pkt_kind. 1001 * 1002 * #VPX_CODEC_CX_FRAME_PKT packets should be passed to the application's 1003 * muxer. Multiple compressed frames may be in the list. 1004 * #VPX_CODEC_STATS_PKT packets should be appended to a global buffer. 1005 * 1006 * The application \ref MUST silently ignore any packet kinds that it does 1007 * not recognize or support. 1008 * 1009 * The data buffers returned from this function are only guaranteed to be 1010 * valid until the application makes another call to any vpx_codec_* function. 1011 * 1012 * \param[in] ctx Pointer to this instance's context 1013 * \param[in,out] iter Iterator storage, initialized to NULL 1014 * 1015 * \return Returns a pointer to an output data packet (compressed frame data, 1016 * two-pass statistics, etc.) or NULL to signal end-of-list. 1017 * 1018 */ 1019 const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx, 1020 vpx_codec_iter_t *iter); 1021 1022 1023 /*!\brief Get Preview Frame 1024 * 1025 * Returns an image that can be used as a preview. Shows the image as it would 1026 * exist at the decompressor. The application \ref MUST NOT write into this 1027 * image buffer. 1028 * 1029 * \param[in] ctx Pointer to this instance's context 1030 * 1031 * \return Returns a pointer to a preview image, or NULL if no image is 1032 * available. 1033 * 1034 */ 1035 const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t *ctx); 1036 1037 1038 /*!@} - end defgroup encoder*/ 1039 #ifdef __cplusplus 1040 } 1041 #endif 1042 #endif // VPX_VPX_ENCODER_H_ 1043 1044