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