1 /* 2 * Copyright (c) 2016, Alliance for Open Media. All rights reserved 3 * 4 * This source code is subject to the terms of the BSD 2 Clause License and 5 * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License 6 * was not distributed with this source code in the LICENSE file, you can 7 * obtain it at www.aomedia.org/license/software. If the Alliance for Open 8 * Media Patent License 1.0 was not distributed with this source code in the 9 * PATENTS file, you can obtain it at www.aomedia.org/license/patent. 10 */ 11 #ifndef AOM_AOM_AOM_ENCODER_H_ 12 #define AOM_AOM_AOM_ENCODER_H_ 13 14 /*!\defgroup encoder Encoder Algorithm Interface 15 * \ingroup codec 16 * This abstraction allows applications using this encoder to easily support 17 * multiple video formats with minimal code duplication. This section describes 18 * the interface common to all encoders. 19 * @{ 20 */ 21 22 /*!\file 23 * \brief Describes the encoder algorithm interface to applications. 24 * 25 * This file describes the interface between an application and a 26 * video encoder algorithm. 27 * 28 */ 29 #ifdef __cplusplus 30 extern "C" { 31 #endif 32 33 #include "aom/aom_codec.h" 34 35 /*!\brief Current ABI version number 36 * 37 * \internal 38 * If this file is altered in any way that changes the ABI, this value 39 * must be bumped. Examples include, but are not limited to, changing 40 * types, removing or reassigning enums, adding/removing/rearranging 41 * fields to structures 42 */ 43 #define AOM_ENCODER_ABI_VERSION \ 44 (5 + AOM_CODEC_ABI_VERSION) /**<\hideinitializer*/ 45 46 /*! \brief Encoder capabilities bitfield 47 * 48 * Each encoder advertises the capabilities it supports as part of its 49 * ::aom_codec_iface_t interface structure. Capabilities are extra 50 * interfaces or functionality, and are not required to be supported 51 * by an encoder. 52 * 53 * The available flags are specified by AOM_CODEC_CAP_* defines. 54 */ 55 #define AOM_CODEC_CAP_PSNR 0x10000 /**< Can issue PSNR packets */ 56 57 /*! Can support input images at greater than 8 bitdepth. 58 */ 59 #define AOM_CODEC_CAP_HIGHBITDEPTH 0x40000 60 61 /*! \brief Initialization-time Feature Enabling 62 * 63 * Certain codec features must be known at initialization time, to allow 64 * for proper memory allocation. 65 * 66 * The available flags are specified by AOM_CODEC_USE_* defines. 67 */ 68 #define AOM_CODEC_USE_PSNR 0x10000 /**< Calculate PSNR on each frame */ 69 /*!\brief Make the encoder output one partition at a time. */ 70 #define AOM_CODEC_USE_HIGHBITDEPTH 0x40000 /**< Use high bitdepth */ 71 72 /*!\brief Generic fixed size buffer structure 73 * 74 * This structure is able to hold a reference to any fixed size buffer. 75 */ 76 typedef struct aom_fixed_buf { 77 void *buf; /**< Pointer to the data */ 78 size_t sz; /**< Length of the buffer, in chars */ 79 } aom_fixed_buf_t; /**< alias for struct aom_fixed_buf */ 80 81 /*!\brief Time Stamp Type 82 * 83 * An integer, which when multiplied by the stream's time base, provides 84 * the absolute time of a sample. 85 */ 86 typedef int64_t aom_codec_pts_t; 87 88 /*!\brief Compressed Frame Flags 89 * 90 * This type represents a bitfield containing information about a compressed 91 * frame that may be useful to an application. The most significant 16 bits 92 * can be used by an algorithm to provide additional detail, for example to 93 * support frame types that are codec specific (MPEG-1 D-frames for example) 94 */ 95 typedef uint32_t aom_codec_frame_flags_t; 96 #define AOM_FRAME_IS_KEY 0x1 /**< frame is the start of a GOP */ 97 /*!\brief frame can be dropped without affecting the stream (no future frame 98 * depends on this one) */ 99 #define AOM_FRAME_IS_DROPPABLE 0x2 100 /*!\brief this is an INTRA_ONLY frame */ 101 #define AOM_FRAME_IS_INTRAONLY 0x10 102 /*!\brief this is an S-frame */ 103 #define AOM_FRAME_IS_SWITCH 0x20 104 /*!\brief this is an error-resilient frame */ 105 #define AOM_FRAME_IS_ERROR_RESILIENT 0x40 106 /*!\brief this is a key-frame dependent recovery-point frame */ 107 #define AOM_FRAME_IS_DELAYED_RANDOM_ACCESS_POINT 0x80 108 109 /*!\brief Error Resilient flags 110 * 111 * These flags define which error resilient features to enable in the 112 * encoder. The flags are specified through the 113 * aom_codec_enc_cfg::g_error_resilient variable. 114 */ 115 typedef uint32_t aom_codec_er_flags_t; 116 /*!\brief Improve resiliency against losses of whole frames */ 117 #define AOM_ERROR_RESILIENT_DEFAULT 0x1 118 119 /*!\brief Encoder output packet variants 120 * 121 * This enumeration lists the different kinds of data packets that can be 122 * returned by calls to aom_codec_get_cx_data(). Algorithms \ref MAY 123 * extend this list to provide additional functionality. 124 */ 125 enum aom_codec_cx_pkt_kind { 126 AOM_CODEC_CX_FRAME_PKT, /**< Compressed video frame */ 127 AOM_CODEC_STATS_PKT, /**< Two-pass statistics for this frame */ 128 AOM_CODEC_FPMB_STATS_PKT, /**< first pass mb statistics for this frame */ 129 AOM_CODEC_PSNR_PKT, /**< PSNR statistics for this frame */ 130 AOM_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions */ 131 }; 132 133 /*!\brief Encoder output packet 134 * 135 * This structure contains the different kinds of output data the encoder 136 * may produce while compressing a frame. 137 */ 138 typedef struct aom_codec_cx_pkt { 139 enum aom_codec_cx_pkt_kind kind; /**< packet variant */ 140 union { 141 struct { 142 void *buf; /**< compressed data buffer */ 143 size_t sz; /**< length of compressed data */ 144 /*!\brief time stamp to show frame (in timebase units) */ 145 aom_codec_pts_t pts; 146 /*!\brief duration to show frame (in timebase units) */ 147 unsigned long duration; 148 aom_codec_frame_flags_t flags; /**< flags for this frame */ 149 /*!\brief the partition id defines the decoding order of the partitions. 150 * Only applicable when "output partition" mode is enabled. First 151 * partition has id 0.*/ 152 int partition_id; 153 /*!\brief size of the visible frame in this packet */ 154 size_t vis_frame_size; 155 } frame; /**< data for compressed frame packet */ 156 aom_fixed_buf_t twopass_stats; /**< data for two-pass packet */ 157 aom_fixed_buf_t firstpass_mb_stats; /**< first pass mb packet */ 158 struct aom_psnr_pkt { 159 unsigned int samples[4]; /**< Number of samples, total/y/u/v */ 160 uint64_t sse[4]; /**< sum squared error, total/y/u/v */ 161 double psnr[4]; /**< PSNR, total/y/u/v */ 162 } psnr; /**< data for PSNR packet */ 163 aom_fixed_buf_t raw; /**< data for arbitrary packets */ 164 165 /* This packet size is fixed to allow codecs to extend this 166 * interface without having to manage storage for raw packets, 167 * i.e., if it's smaller than 128 bytes, you can store in the 168 * packet list directly. 169 */ 170 char pad[128 - sizeof(enum aom_codec_cx_pkt_kind)]; /**< fixed sz */ 171 } data; /**< packet data */ 172 } aom_codec_cx_pkt_t; /**< alias for struct aom_codec_cx_pkt */ 173 174 /*!\brief Rational Number 175 * 176 * This structure holds a fractional value. 177 */ 178 typedef struct aom_rational { 179 int num; /**< fraction numerator */ 180 int den; /**< fraction denominator */ 181 } aom_rational_t; /**< alias for struct aom_rational */ 182 183 /*!\brief Multi-pass Encoding Pass */ 184 enum aom_enc_pass { 185 AOM_RC_ONE_PASS, /**< Single pass mode */ 186 AOM_RC_FIRST_PASS, /**< First pass of multi-pass mode */ 187 AOM_RC_LAST_PASS /**< Final pass of multi-pass mode */ 188 }; 189 190 /*!\brief Rate control mode */ 191 enum aom_rc_mode { 192 AOM_VBR, /**< Variable Bit Rate (VBR) mode */ 193 AOM_CBR, /**< Constant Bit Rate (CBR) mode */ 194 AOM_CQ, /**< Constrained Quality (CQ) mode */ 195 AOM_Q, /**< Constant Quality (Q) mode */ 196 }; 197 198 /*!\brief Keyframe placement mode. 199 * 200 * This enumeration determines whether keyframes are placed automatically by 201 * the encoder or whether this behavior is disabled. Older releases of this 202 * SDK were implemented such that AOM_KF_FIXED meant keyframes were disabled. 203 * This name is confusing for this behavior, so the new symbols to be used 204 * are AOM_KF_AUTO and AOM_KF_DISABLED. 205 */ 206 enum aom_kf_mode { 207 AOM_KF_FIXED, /**< deprecated, implies AOM_KF_DISABLED */ 208 AOM_KF_AUTO, /**< Encoder determines optimal placement automatically */ 209 AOM_KF_DISABLED = 0 /**< Encoder does not place keyframes. */ 210 }; 211 212 /*!\brief Encoded Frame Flags 213 * 214 * This type indicates a bitfield to be passed to aom_codec_encode(), defining 215 * per-frame boolean values. By convention, bits common to all codecs will be 216 * named AOM_EFLAG_*, and bits specific to an algorithm will be named 217 * /algo/_eflag_*. The lower order 16 bits are reserved for common use. 218 */ 219 typedef long aom_enc_frame_flags_t; 220 #define AOM_EFLAG_FORCE_KF (1 << 0) /**< Force this frame to be a keyframe */ 221 222 /*!\brief Encoder configuration structure 223 * 224 * This structure contains the encoder settings that have common representations 225 * across all codecs. This doesn't imply that all codecs support all features, 226 * however. 227 */ 228 typedef struct aom_codec_enc_cfg { 229 /* 230 * generic settings (g) 231 */ 232 233 /*!\brief Algorithm specific "usage" value 234 * 235 * Algorithms may define multiple values for usage, which may convey the 236 * intent of how the application intends to use the stream. If this value 237 * is non-zero, consult the documentation for the codec to determine its 238 * meaning. 239 */ 240 unsigned int g_usage; 241 242 /*!\brief Maximum number of threads to use 243 * 244 * For multi-threaded implementations, use no more than this number of 245 * threads. The codec may use fewer threads than allowed. The value 246 * 0 is equivalent to the value 1. 247 */ 248 unsigned int g_threads; 249 250 /*!\brief Bitstream profile to use 251 * 252 * Some codecs support a notion of multiple bitstream profiles. Typically 253 * this maps to a set of features that are turned on or off. Often the 254 * profile to use is determined by the features of the intended decoder. 255 * Consult the documentation for the codec to determine the valid values 256 * for this parameter, or set to zero for a sane default. 257 */ 258 unsigned int g_profile; /**< profile of bitstream to use */ 259 260 /*!\brief Width of the frame 261 * 262 * This value identifies the presentation resolution of the frame, 263 * in pixels. Note that the frames passed as input to the encoder must 264 * have this resolution. Frames will be presented by the decoder in this 265 * resolution, independent of any spatial resampling the encoder may do. 266 */ 267 unsigned int g_w; 268 269 /*!\brief Height of the frame 270 * 271 * This value identifies the presentation resolution of the frame, 272 * in pixels. Note that the frames passed as input to the encoder must 273 * have this resolution. Frames will be presented by the decoder in this 274 * resolution, independent of any spatial resampling the encoder may do. 275 */ 276 unsigned int g_h; 277 278 /*!\brief Max number of frames to encode 279 * 280 */ 281 unsigned int g_limit; 282 283 /*!\brief Forced maximum width of the frame 284 * 285 * If this value is non-zero then it is used to force the maximum frame 286 * width written in write_sequence_header(). 287 */ 288 unsigned int g_forced_max_frame_width; 289 290 /*!\brief Forced maximum height of the frame 291 * 292 * If this value is non-zero then it is used to force the maximum frame 293 * height written in write_sequence_header(). 294 */ 295 unsigned int g_forced_max_frame_height; 296 297 /*!\brief Bit-depth of the codec 298 * 299 * This value identifies the bit_depth of the codec, 300 * Only certain bit-depths are supported as identified in the 301 * aom_bit_depth_t enum. 302 */ 303 aom_bit_depth_t g_bit_depth; 304 305 /*!\brief Bit-depth of the input frames 306 * 307 * This value identifies the bit_depth of the input frames in bits. 308 * Note that the frames passed as input to the encoder must have 309 * this bit-depth. 310 */ 311 unsigned int g_input_bit_depth; 312 313 /*!\brief Stream timebase units 314 * 315 * Indicates the smallest interval of time, in seconds, used by the stream. 316 * For fixed frame rate material, or variable frame rate material where 317 * frames are timed at a multiple of a given clock (ex: video capture), 318 * the \ref RECOMMENDED method is to set the timebase to the reciprocal 319 * of the frame rate (ex: 1001/30000 for 29.970 Hz NTSC). This allows the 320 * pts to correspond to the frame number, which can be handy. For 321 * re-encoding video from containers with absolute time timestamps, the 322 * \ref RECOMMENDED method is to set the timebase to that of the parent 323 * container or multimedia framework (ex: 1/1000 for ms, as in FLV). 324 */ 325 struct aom_rational g_timebase; 326 327 /*!\brief Enable error resilient modes. 328 * 329 * The error resilient bitfield indicates to the encoder which features 330 * it should enable to take measures for streaming over lossy or noisy 331 * links. 332 */ 333 aom_codec_er_flags_t g_error_resilient; 334 335 /*!\brief Multi-pass Encoding Mode 336 * 337 * This value should be set to the current phase for multi-pass encoding. 338 * For single pass, set to #AOM_RC_ONE_PASS. 339 */ 340 enum aom_enc_pass g_pass; 341 342 /*!\brief Allow lagged encoding 343 * 344 * If set, this value allows the encoder to consume a number of input 345 * frames before producing output frames. This allows the encoder to 346 * base decisions for the current frame on future frames. This does 347 * increase the latency of the encoding pipeline, so it is not appropriate 348 * in all situations (ex: realtime encoding). 349 * 350 * Note that this is a maximum value -- the encoder may produce frames 351 * sooner than the given limit. Set this value to 0 to disable this 352 * feature. 353 */ 354 unsigned int g_lag_in_frames; 355 356 /* 357 * rate control settings (rc) 358 */ 359 360 /*!\brief Temporal resampling configuration, if supported by the codec. 361 * 362 * Temporal resampling allows the codec to "drop" frames as a strategy to 363 * meet its target data rate. This can cause temporal discontinuities in 364 * the encoded video, which may appear as stuttering during playback. This 365 * trade-off is often acceptable, but for many applications is not. It can 366 * be disabled in these cases. 367 * 368 * Note that not all codecs support this feature. All aom AVx codecs do. 369 * For other codecs, consult the documentation for that algorithm. 370 * 371 * This threshold is described as a percentage of the target data buffer. 372 * When the data buffer falls below this percentage of fullness, a 373 * dropped frame is indicated. Set the threshold to zero (0) to disable 374 * this feature. 375 */ 376 unsigned int rc_dropframe_thresh; 377 378 /*!\brief Mode for spatial resampling, if supported by the codec. 379 * 380 * Spatial resampling allows the codec to compress a lower resolution 381 * version of the frame, which is then upscaled by the decoder to the 382 * correct presentation resolution. This increases visual quality at 383 * low data rates, at the expense of CPU time on the encoder/decoder. 384 */ 385 unsigned int rc_resize_mode; 386 387 /*!\brief Frame resize denominator. 388 * 389 * The denominator for resize to use, assuming 8 as the numerator. 390 * 391 * Valid denominators are 8 - 16 for now. 392 */ 393 unsigned int rc_resize_denominator; 394 395 /*!\brief Keyframe resize denominator. 396 * 397 * The denominator for resize to use, assuming 8 as the numerator. 398 * 399 * Valid denominators are 8 - 16 for now. 400 */ 401 unsigned int rc_resize_kf_denominator; 402 403 /*!\brief Frame super-resolution scaling mode. 404 * 405 * Similar to spatial resampling, frame super-resolution integrates 406 * upscaling after the encode/decode process. Taking control of upscaling and 407 * using restoration filters should allow it to outperform normal resizing. 408 * 409 * Valid values are 0 to 4 as defined in enum SUPERRES_MODE. 410 */ 411 unsigned int rc_superres_mode; 412 413 /*!\brief Frame super-resolution denominator. 414 * 415 * The denominator for superres to use. If fixed it will only change if the 416 * cumulative scale change over resizing and superres is greater than 1/2; 417 * this forces superres to reduce scaling. 418 * 419 * Valid denominators are 8 to 16. 420 * 421 * Used only by SUPERRES_FIXED. 422 */ 423 unsigned int rc_superres_denominator; 424 425 /*!\brief Keyframe super-resolution denominator. 426 * 427 * The denominator for superres to use. If fixed it will only change if the 428 * cumulative scale change over resizing and superres is greater than 1/2; 429 * this forces superres to reduce scaling. 430 * 431 * Valid denominators are 8 - 16 for now. 432 */ 433 unsigned int rc_superres_kf_denominator; 434 435 /*!\brief Frame super-resolution q threshold. 436 * 437 * The q level threshold after which superres is used. 438 * Valid values are 1 to 63. 439 * 440 * Used only by SUPERRES_QTHRESH 441 */ 442 unsigned int rc_superres_qthresh; 443 444 /*!\brief Keyframe super-resolution q threshold. 445 * 446 * The q level threshold after which superres is used for key frames. 447 * Valid values are 1 to 63. 448 * 449 * Used only by SUPERRES_QTHRESH 450 */ 451 unsigned int rc_superres_kf_qthresh; 452 453 /*!\brief Rate control algorithm to use. 454 * 455 * Indicates whether the end usage of this stream is to be streamed over 456 * a bandwidth constrained link, indicating that Constant Bit Rate (CBR) 457 * mode should be used, or whether it will be played back on a high 458 * bandwidth link, as from a local disk, where higher variations in 459 * bitrate are acceptable. 460 */ 461 enum aom_rc_mode rc_end_usage; 462 463 /*!\brief Two-pass stats buffer. 464 * 465 * A buffer containing all of the stats packets produced in the first 466 * pass, concatenated. 467 */ 468 aom_fixed_buf_t rc_twopass_stats_in; 469 470 /*!\brief first pass mb stats buffer. 471 * 472 * A buffer containing all of the first pass mb stats packets produced 473 * in the first pass, concatenated. 474 */ 475 aom_fixed_buf_t rc_firstpass_mb_stats_in; 476 477 /*!\brief Target data rate 478 * 479 * Target bandwidth to use for this stream, in kilobits per second. 480 */ 481 unsigned int rc_target_bitrate; 482 483 /* 484 * quantizer settings 485 */ 486 487 /*!\brief Minimum (Best Quality) Quantizer 488 * 489 * The quantizer is the most direct control over the quality of the 490 * encoded image. The range of valid values for the quantizer is codec 491 * specific. Consult the documentation for the codec to determine the 492 * values to use. To determine the range programmatically, call 493 * aom_codec_enc_config_default() with a usage value of 0. 494 */ 495 unsigned int rc_min_quantizer; 496 497 /*!\brief Maximum (Worst Quality) Quantizer 498 * 499 * The quantizer is the most direct control over the quality of the 500 * encoded image. The range of valid values for the quantizer is codec 501 * specific. Consult the documentation for the codec to determine the 502 * values to use. To determine the range programmatically, call 503 * aom_codec_enc_config_default() with a usage value of 0. 504 */ 505 unsigned int rc_max_quantizer; 506 507 /* 508 * bitrate tolerance 509 */ 510 511 /*!\brief Rate control adaptation undershoot control 512 * 513 * This value, expressed as a percentage of the target bitrate, 514 * controls the maximum allowed adaptation speed of the codec. 515 * This factor controls the maximum amount of bits that can 516 * be subtracted from the target bitrate in order to compensate 517 * for prior overshoot. 518 * 519 * Valid values in the range 0-1000. 520 */ 521 unsigned int rc_undershoot_pct; 522 523 /*!\brief Rate control adaptation overshoot control 524 * 525 * This value, expressed as a percentage of the target bitrate, 526 * controls the maximum allowed adaptation speed of the codec. 527 * This factor controls the maximum amount of bits that can 528 * be added to the target bitrate in order to compensate for 529 * prior undershoot. 530 * 531 * Valid values in the range 0-1000. 532 */ 533 unsigned int rc_overshoot_pct; 534 535 /* 536 * decoder buffer model parameters 537 */ 538 539 /*!\brief Decoder Buffer Size 540 * 541 * This value indicates the amount of data that may be buffered by the 542 * decoding application. Note that this value is expressed in units of 543 * time (milliseconds). For example, a value of 5000 indicates that the 544 * client will buffer (at least) 5000ms worth of encoded data. Use the 545 * target bitrate (#rc_target_bitrate) to convert to bits/bytes, if 546 * necessary. 547 */ 548 unsigned int rc_buf_sz; 549 550 /*!\brief Decoder Buffer Initial Size 551 * 552 * This value indicates the amount of data that will be buffered by the 553 * decoding application prior to beginning playback. This value is 554 * expressed in units of time (milliseconds). Use the target bitrate 555 * (#rc_target_bitrate) to convert to bits/bytes, if necessary. 556 */ 557 unsigned int rc_buf_initial_sz; 558 559 /*!\brief Decoder Buffer Optimal Size 560 * 561 * This value indicates the amount of data that the encoder should try 562 * to maintain in the decoder's buffer. This value is expressed in units 563 * of time (milliseconds). Use the target bitrate (#rc_target_bitrate) 564 * to convert to bits/bytes, if necessary. 565 */ 566 unsigned int rc_buf_optimal_sz; 567 568 /* 569 * 2 pass rate control parameters 570 */ 571 572 /*!\brief Two-pass mode CBR/VBR bias 573 * 574 * Bias, expressed on a scale of 0 to 100, for determining target size 575 * for the current frame. The value 0 indicates the optimal CBR mode 576 * value should be used. The value 100 indicates the optimal VBR mode 577 * value should be used. Values in between indicate which way the 578 * encoder should "lean." 579 */ 580 unsigned int rc_2pass_vbr_bias_pct; 581 582 /*!\brief Two-pass mode per-GOP minimum bitrate 583 * 584 * This value, expressed as a percentage of the target bitrate, indicates 585 * the minimum bitrate to be used for a single GOP (aka "section") 586 */ 587 unsigned int rc_2pass_vbr_minsection_pct; 588 589 /*!\brief Two-pass mode per-GOP maximum bitrate 590 * 591 * This value, expressed as a percentage of the target bitrate, indicates 592 * the maximum bitrate to be used for a single GOP (aka "section") 593 */ 594 unsigned int rc_2pass_vbr_maxsection_pct; 595 596 /* 597 * keyframing settings (kf) 598 */ 599 600 /*!\brief Option to enable forward reference key frame 601 * 602 */ 603 int fwd_kf_enabled; 604 605 /*!\brief Keyframe placement mode 606 * 607 * This value indicates whether the encoder should place keyframes at a 608 * fixed interval, or determine the optimal placement automatically 609 * (as governed by the #kf_min_dist and #kf_max_dist parameters) 610 */ 611 enum aom_kf_mode kf_mode; 612 613 /*!\brief Keyframe minimum interval 614 * 615 * This value, expressed as a number of frames, prevents the encoder from 616 * placing a keyframe nearer than kf_min_dist to the previous keyframe. At 617 * least kf_min_dist frames non-keyframes will be coded before the next 618 * keyframe. Set kf_min_dist equal to kf_max_dist for a fixed interval. 619 */ 620 unsigned int kf_min_dist; 621 622 /*!\brief Keyframe maximum interval 623 * 624 * This value, expressed as a number of frames, forces the encoder to code 625 * a keyframe if one has not been coded in the last kf_max_dist frames. 626 * A value of 0 implies all frames will be keyframes. Set kf_min_dist 627 * equal to kf_max_dist for a fixed interval. 628 */ 629 unsigned int kf_max_dist; 630 631 /*!\brief sframe interval 632 * 633 * This value, expressed as a number of frames, forces the encoder to code 634 * an S-Frame every sframe_dist frames. 635 */ 636 unsigned int sframe_dist; 637 638 /*!\brief sframe insertion mode 639 * 640 * This value must be set to 1 or 2, and tells the encoder how to insert 641 * S-Frames. It will only have an effect if sframe_dist != 0. 642 * 643 * If altref is enabled: 644 * - if sframe_mode == 1, the considered frame will be made into an 645 * S-Frame only if it is an altref frame 646 * - if sframe_mode == 2, the next altref frame will be made into an 647 * S-Frame. 648 * 649 * Otherwise: the considered frame will be made into an S-Frame. 650 */ 651 unsigned int sframe_mode; 652 653 /*!\brief Tile coding mode 654 * 655 * This value indicates the tile coding mode. 656 * A value of 0 implies a normal non-large-scale tile coding. A value of 1 657 * implies a large-scale tile coding. 658 */ 659 unsigned int large_scale_tile; 660 661 /*!\brief Monochrome mode 662 * 663 * If this is nonzero, the encoder will generate a monochrome stream 664 * with no chroma planes. 665 */ 666 unsigned int monochrome; 667 668 /*!\brief full_still_picture_hdr 669 * 670 * If this is nonzero, the encoder will generate a full header even for 671 * still picture encoding. if zero, a reduced header is used for still 672 * picture. This flag has no effect when a regular video with more than 673 * a single frame is encoded. 674 */ 675 unsigned int full_still_picture_hdr; 676 677 /*!\brief Bitstream syntax mode 678 * 679 * This value indicates the bitstream syntax mode. 680 * A value of 0 indicates bitstream is saved as Section 5 bitstream. A value 681 * of 1 indicates the bitstream is saved in Annex-B format 682 */ 683 unsigned int save_as_annexb; 684 685 /*!\brief Number of explicit tile widths specified 686 * 687 * This value indicates the number of tile widths specified 688 * A value of 0 implies no tile widths are specified. 689 * Tile widths are given in the array tile_widths[] 690 */ 691 int tile_width_count; 692 693 /*!\brief Number of explicit tile heights specified 694 * 695 * This value indicates the number of tile heights specified 696 * A value of 0 implies no tile heights are specified. 697 * Tile heights are given in the array tile_heights[] 698 */ 699 int tile_height_count; 700 701 /*!\brief Maximum number of tile widths in tile widths array 702 * 703 * This define gives the maximum number of elements in the tile_widths array. 704 */ 705 #define MAX_TILE_WIDTHS 64 // maximum tile width array length 706 707 /*!\brief Array of specified tile widths 708 * 709 * This array specifies tile widths (and may be empty) 710 * The number of widths specified is given by tile_width_count 711 */ 712 int tile_widths[MAX_TILE_WIDTHS]; 713 714 /*!\brief Maximum number of tile heights in tile heights array. 715 * 716 * This define gives the maximum number of elements in the tile_heights array. 717 */ 718 #define MAX_TILE_HEIGHTS 64 // maximum tile height array length 719 720 /*!\brief Array of specified tile heights 721 * 722 * This array specifies tile heights (and may be empty) 723 * The number of heights specified is given by tile_height_count 724 */ 725 int tile_heights[MAX_TILE_HEIGHTS]; 726 727 /*!\brief Options defined per config file 728 * 729 */ 730 cfg_options_t cfg; 731 } aom_codec_enc_cfg_t; /**< alias for struct aom_codec_enc_cfg */ 732 733 /*!\brief Initialize an encoder instance 734 * 735 * Initializes a encoder context using the given interface. Applications 736 * should call the aom_codec_enc_init convenience macro instead of this 737 * function directly, to ensure that the ABI version number parameter 738 * is properly initialized. 739 * 740 * If the library was configured with --disable-multithread, this call 741 * is not thread safe and should be guarded with a lock if being used 742 * in a multithreaded context. 743 * 744 * \param[in] ctx Pointer to this instance's context. 745 * \param[in] iface Pointer to the algorithm interface to use. 746 * \param[in] cfg Configuration to use, if known. 747 * \param[in] flags Bitfield of AOM_CODEC_USE_* flags 748 * \param[in] ver ABI version number. Must be set to 749 * AOM_ENCODER_ABI_VERSION 750 * \retval #AOM_CODEC_OK 751 * The decoder algorithm initialized. 752 * \retval #AOM_CODEC_MEM_ERROR 753 * Memory allocation failed. 754 */ 755 aom_codec_err_t aom_codec_enc_init_ver(aom_codec_ctx_t *ctx, 756 aom_codec_iface_t *iface, 757 const aom_codec_enc_cfg_t *cfg, 758 aom_codec_flags_t flags, int ver); 759 760 /*!\brief Convenience macro for aom_codec_enc_init_ver() 761 * 762 * Ensures the ABI version parameter is properly set. 763 */ 764 #define aom_codec_enc_init(ctx, iface, cfg, flags) \ 765 aom_codec_enc_init_ver(ctx, iface, cfg, flags, AOM_ENCODER_ABI_VERSION) 766 767 /*!\brief Initialize multi-encoder instance 768 * 769 * Initializes multi-encoder context using the given interface. 770 * Applications should call the aom_codec_enc_init_multi convenience macro 771 * instead of this function directly, to ensure that the ABI version number 772 * parameter is properly initialized. 773 * 774 * \param[in] ctx Pointer to this instance's context. 775 * \param[in] iface Pointer to the algorithm interface to use. 776 * \param[in] cfg Configuration to use, if known. 777 * \param[in] num_enc Total number of encoders. 778 * \param[in] flags Bitfield of AOM_CODEC_USE_* flags 779 * \param[in] dsf Pointer to down-sampling factors. 780 * \param[in] ver ABI version number. Must be set to 781 * AOM_ENCODER_ABI_VERSION 782 * \retval #AOM_CODEC_OK 783 * The decoder algorithm initialized. 784 * \retval #AOM_CODEC_MEM_ERROR 785 * Memory allocation failed. 786 */ 787 aom_codec_err_t aom_codec_enc_init_multi_ver( 788 aom_codec_ctx_t *ctx, aom_codec_iface_t *iface, aom_codec_enc_cfg_t *cfg, 789 int num_enc, aom_codec_flags_t flags, aom_rational_t *dsf, int ver); 790 791 /*!\brief Convenience macro for aom_codec_enc_init_multi_ver() 792 * 793 * Ensures the ABI version parameter is properly set. 794 */ 795 #define aom_codec_enc_init_multi(ctx, iface, cfg, num_enc, flags, dsf) \ 796 aom_codec_enc_init_multi_ver(ctx, iface, cfg, num_enc, flags, dsf, \ 797 AOM_ENCODER_ABI_VERSION) 798 799 /*!\brief Get a default configuration 800 * 801 * Initializes a encoder configuration structure with default values. Supports 802 * the notion of "usages" so that an algorithm may offer different default 803 * settings depending on the user's intended goal. This function \ref SHOULD 804 * be called by all applications to initialize the configuration structure 805 * before specializing the configuration with application specific values. 806 * 807 * \param[in] iface Pointer to the algorithm interface to use. 808 * \param[out] cfg Configuration buffer to populate. 809 * \param[in] reserved Must set to 0. 810 * 811 * \retval #AOM_CODEC_OK 812 * The configuration was populated. 813 * \retval #AOM_CODEC_INCAPABLE 814 * Interface is not an encoder interface. 815 * \retval #AOM_CODEC_INVALID_PARAM 816 * A parameter was NULL, or the usage value was not recognized. 817 */ 818 aom_codec_err_t aom_codec_enc_config_default(aom_codec_iface_t *iface, 819 aom_codec_enc_cfg_t *cfg, 820 unsigned int reserved); 821 822 /*!\brief Set or change configuration 823 * 824 * Reconfigures an encoder instance according to the given configuration. 825 * 826 * \param[in] ctx Pointer to this instance's context 827 * \param[in] cfg Configuration buffer to use 828 * 829 * \retval #AOM_CODEC_OK 830 * The configuration was populated. 831 * \retval #AOM_CODEC_INCAPABLE 832 * Interface is not an encoder interface. 833 * \retval #AOM_CODEC_INVALID_PARAM 834 * A parameter was NULL, or the usage value was not recognized. 835 */ 836 aom_codec_err_t aom_codec_enc_config_set(aom_codec_ctx_t *ctx, 837 const aom_codec_enc_cfg_t *cfg); 838 839 /*!\brief Get global stream headers 840 * 841 * Retrieves a stream level global header packet, if supported by the codec. 842 * Calls to this function should be deferred until all configuration information 843 * has been passed to libaom. Otherwise the global header data may be 844 * invalidated by additional configuration changes. 845 * 846 * The AV1 implementation of this function returns an OBU. The OBU returned is 847 * in Low Overhead Bitstream Format. Specifically, the obu_has_size_field bit is 848 * set, and the buffer contains the obu_size field for the returned OBU. 849 * 850 * \param[in] ctx Pointer to this instance's context 851 * 852 * \retval NULL 853 * Encoder does not support global header, or an error occurred while 854 * generating the global header. 855 * 856 * \retval Non-NULL 857 * Pointer to buffer containing global header packet. The caller owns the 858 * memory associated with this buffer, and must free the 'buf' member of the 859 * aom_fixed_buf_t as well as the aom_fixed_buf_t pointer. Memory returned 860 * must be freed via call to free(). 861 */ 862 aom_fixed_buf_t *aom_codec_get_global_headers(aom_codec_ctx_t *ctx); 863 864 /*!\brief usage parameter analogous to AV1 GOOD QUALITY mode. */ 865 #define AOM_USAGE_GOOD_QUALITY (0) 866 /*!\brief usage parameter analogous to AV1 REALTIME mode. */ 867 #define AOM_USAGE_REALTIME (1) 868 869 /*!\brief Encode a frame 870 * 871 * Encodes a video frame at the given "presentation time." The presentation 872 * time stamp (PTS) \ref MUST be strictly increasing. 873 * 874 * When the last frame has been passed to the encoder, this function should 875 * continue to be called, with the img parameter set to NULL. This will 876 * signal the end-of-stream condition to the encoder and allow it to encode 877 * any held buffers. Encoding is complete when aom_codec_encode() is called 878 * and aom_codec_get_cx_data() returns no data. 879 * 880 * \param[in] ctx Pointer to this instance's context 881 * \param[in] img Image data to encode, NULL to flush. 882 * \param[in] pts Presentation time stamp, in timebase units. 883 * \param[in] duration Duration to show frame, in timebase units. 884 * \param[in] flags Flags to use for encoding this frame. 885 * 886 * \retval #AOM_CODEC_OK 887 * The configuration was populated. 888 * \retval #AOM_CODEC_INCAPABLE 889 * Interface is not an encoder interface. 890 * \retval #AOM_CODEC_INVALID_PARAM 891 * A parameter was NULL, the image format is unsupported, etc. 892 */ 893 aom_codec_err_t aom_codec_encode(aom_codec_ctx_t *ctx, const aom_image_t *img, 894 aom_codec_pts_t pts, unsigned long duration, 895 aom_enc_frame_flags_t flags); 896 897 /*!\brief Set compressed data output buffer 898 * 899 * Sets the buffer that the codec should output the compressed data 900 * into. This call effectively sets the buffer pointer returned in the 901 * next AOM_CODEC_CX_FRAME_PKT packet. Subsequent packets will be 902 * appended into this buffer. The buffer is preserved across frames, 903 * so applications must periodically call this function after flushing 904 * the accumulated compressed data to disk or to the network to reset 905 * the pointer to the buffer's head. 906 * 907 * `pad_before` bytes will be skipped before writing the compressed 908 * data, and `pad_after` bytes will be appended to the packet. The size 909 * of the packet will be the sum of the size of the actual compressed 910 * data, pad_before, and pad_after. The padding bytes will be preserved 911 * (not overwritten). 912 * 913 * Note that calling this function does not guarantee that the returned 914 * compressed data will be placed into the specified buffer. In the 915 * event that the encoded data will not fit into the buffer provided, 916 * the returned packet \ref MAY point to an internal buffer, as it would 917 * if this call were never used. In this event, the output packet will 918 * NOT have any padding, and the application must free space and copy it 919 * to the proper place. This is of particular note in configurations 920 * that may output multiple packets for a single encoded frame (e.g., lagged 921 * encoding) or if the application does not reset the buffer periodically. 922 * 923 * Applications may restore the default behavior of the codec providing 924 * the compressed data buffer by calling this function with a NULL 925 * buffer. 926 * 927 * Applications \ref MUSTNOT call this function during iteration of 928 * aom_codec_get_cx_data(). 929 * 930 * \param[in] ctx Pointer to this instance's context 931 * \param[in] buf Buffer to store compressed data into 932 * \param[in] pad_before Bytes to skip before writing compressed data 933 * \param[in] pad_after Bytes to skip after writing compressed data 934 * 935 * \retval #AOM_CODEC_OK 936 * The buffer was set successfully. 937 * \retval #AOM_CODEC_INVALID_PARAM 938 * A parameter was NULL, the image format is unsupported, etc. 939 */ 940 aom_codec_err_t aom_codec_set_cx_data_buf(aom_codec_ctx_t *ctx, 941 const aom_fixed_buf_t *buf, 942 unsigned int pad_before, 943 unsigned int pad_after); 944 945 /*!\brief Encoded data iterator 946 * 947 * Iterates over a list of data packets to be passed from the encoder to the 948 * application. The different kinds of packets available are enumerated in 949 * #aom_codec_cx_pkt_kind. 950 * 951 * #AOM_CODEC_CX_FRAME_PKT packets should be passed to the application's 952 * muxer. Multiple compressed frames may be in the list. 953 * #AOM_CODEC_STATS_PKT packets should be appended to a global buffer. 954 * 955 * The application \ref MUST silently ignore any packet kinds that it does 956 * not recognize or support. 957 * 958 * The data buffers returned from this function are only guaranteed to be 959 * valid until the application makes another call to any aom_codec_* function. 960 * 961 * \param[in] ctx Pointer to this instance's context 962 * \param[in,out] iter Iterator storage, initialized to NULL 963 * 964 * \return Returns a pointer to an output data packet (compressed frame data, 965 * two-pass statistics, etc.) or NULL to signal end-of-list. 966 * 967 */ 968 const aom_codec_cx_pkt_t *aom_codec_get_cx_data(aom_codec_ctx_t *ctx, 969 aom_codec_iter_t *iter); 970 971 /*!\brief Get Preview Frame 972 * 973 * Returns an image that can be used as a preview. Shows the image as it would 974 * exist at the decompressor. The application \ref MUST NOT write into this 975 * image buffer. 976 * 977 * \param[in] ctx Pointer to this instance's context 978 * 979 * \return Returns a pointer to a preview image, or NULL if no image is 980 * available. 981 * 982 */ 983 const aom_image_t *aom_codec_get_preview_frame(aom_codec_ctx_t *ctx); 984 985 /*!@} - end defgroup encoder*/ 986 #ifdef __cplusplus 987 } 988 #endif 989 #endif // AOM_AOM_AOM_ENCODER_H_ 990