1 /* 2 * 3 * This file is part of FFmpeg. 4 * 5 * FFmpeg is free software; you can redistribute it and/or 6 * modify it under the terms of the GNU Lesser General Public 7 * License as published by the Free Software Foundation; either 8 * version 2.1 of the License, or (at your option) any later version. 9 * 10 * FFmpeg is distributed in the hope that it will be useful, 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 13 * Lesser General Public License for more details. 14 * 15 * You should have received a copy of the GNU Lesser General Public 16 * License along with FFmpeg; if not, write to the Free Software 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA 18 */ 19 20 /** 21 * @file 22 * @ingroup lavu_frame 23 * reference-counted frame API 24 */ 25 26 #ifndef AVUTIL_FRAME_H 27 #define AVUTIL_FRAME_H 28 29 #include <stdint.h> 30 31 #include "avutil.h" 32 #include "buffer.h" 33 #include "dict.h" 34 #include "rational.h" 35 #include "samplefmt.h" 36 #include "pixfmt.h" 37 #include "version.h" 38 39 40 /** 41 * @defgroup lavu_frame AVFrame 42 * @ingroup lavu_data 43 * 44 * @{ 45 * AVFrame is an abstraction for reference-counted raw multimedia data. 46 */ 47 48 enum AVFrameSideDataType { 49 /** 50 * The data is the AVPanScan struct defined in libavcodec. 51 */ 52 AV_FRAME_DATA_PANSCAN, 53 /** 54 * ATSC A53 Part 4 Closed Captions. 55 * A53 CC bitstream is stored as uint8_t in AVFrameSideData.data. 56 * The number of bytes of CC data is AVFrameSideData.size. 57 */ 58 AV_FRAME_DATA_A53_CC, 59 /** 60 * Stereoscopic 3d metadata. 61 * The data is the AVStereo3D struct defined in libavutil/stereo3d.h. 62 */ 63 AV_FRAME_DATA_STEREO3D, 64 /** 65 * The data is the AVMatrixEncoding enum defined in libavutil/channel_layout.h. 66 */ 67 AV_FRAME_DATA_MATRIXENCODING, 68 /** 69 * Metadata relevant to a downmix procedure. 70 * The data is the AVDownmixInfo struct defined in libavutil/downmix_info.h. 71 */ 72 AV_FRAME_DATA_DOWNMIX_INFO, 73 /** 74 * ReplayGain information in the form of the AVReplayGain struct. 75 */ 76 AV_FRAME_DATA_REPLAYGAIN, 77 /** 78 * This side data contains a 3x3 transformation matrix describing an affine 79 * transformation that needs to be applied to the frame for correct 80 * presentation. 81 * 82 * See libavutil/display.h for a detailed description of the data. 83 */ 84 AV_FRAME_DATA_DISPLAYMATRIX, 85 /** 86 * Active Format Description data consisting of a single byte as specified 87 * in ETSI TS 101 154 using AVActiveFormatDescription enum. 88 */ 89 AV_FRAME_DATA_AFD, 90 /** 91 * Motion vectors exported by some codecs (on demand through the export_mvs 92 * flag set in the libavcodec AVCodecContext flags2 option). 93 * The data is the AVMotionVector struct defined in 94 * libavutil/motion_vector.h. 95 */ 96 AV_FRAME_DATA_MOTION_VECTORS, 97 /** 98 * Recommmends skipping the specified number of samples. This is exported 99 * only if the "skip_manual" AVOption is set in libavcodec. 100 * This has the same format as AV_PKT_DATA_SKIP_SAMPLES. 101 * @code 102 * u32le number of samples to skip from start of this packet 103 * u32le number of samples to skip from end of this packet 104 * u8 reason for start skip 105 * u8 reason for end skip (0=padding silence, 1=convergence) 106 * @endcode 107 */ 108 AV_FRAME_DATA_SKIP_SAMPLES, 109 110 /** 111 * This side data must be associated with an audio frame and corresponds to 112 * enum AVAudioServiceType defined in avcodec.h. 113 */ 114 AV_FRAME_DATA_AUDIO_SERVICE_TYPE, 115 }; 116 117 enum AVActiveFormatDescription { 118 AV_AFD_SAME = 8, 119 AV_AFD_4_3 = 9, 120 AV_AFD_16_9 = 10, 121 AV_AFD_14_9 = 11, 122 AV_AFD_4_3_SP_14_9 = 13, 123 AV_AFD_16_9_SP_14_9 = 14, 124 AV_AFD_SP_4_3 = 15, 125 }; 126 127 128 /** 129 * Structure to hold side data for an AVFrame. 130 * 131 * sizeof(AVFrameSideData) is not a part of the public ABI, so new fields may be added 132 * to the end with a minor bump. 133 */ 134 typedef struct AVFrameSideData { 135 enum AVFrameSideDataType type; 136 uint8_t *data; 137 int size; 138 AVDictionary *metadata; 139 AVBufferRef *buf; 140 } AVFrameSideData; 141 142 /** 143 * This structure describes decoded (raw) audio or video data. 144 * 145 * AVFrame must be allocated using av_frame_alloc(). Note that this only 146 * allocates the AVFrame itself, the buffers for the data must be managed 147 * through other means (see below). 148 * AVFrame must be freed with av_frame_free(). 149 * 150 * AVFrame is typically allocated once and then reused multiple times to hold 151 * different data (e.g. a single AVFrame to hold frames received from a 152 * decoder). In such a case, av_frame_unref() will free any references held by 153 * the frame and reset it to its original clean state before it 154 * is reused again. 155 * 156 * The data described by an AVFrame is usually reference counted through the 157 * AVBuffer API. The underlying buffer references are stored in AVFrame.buf / 158 * AVFrame.extended_buf. An AVFrame is considered to be reference counted if at 159 * least one reference is set, i.e. if AVFrame.buf[0] != NULL. In such a case, 160 * every single data plane must be contained in one of the buffers in 161 * AVFrame.buf or AVFrame.extended_buf. 162 * There may be a single buffer for all the data, or one separate buffer for 163 * each plane, or anything in between. 164 * 165 * sizeof(AVFrame) is not a part of the public ABI, so new fields may be added 166 * to the end with a minor bump. 167 * Similarly fields that are marked as to be only accessed by 168 * av_opt_ptr() can be reordered. This allows 2 forks to add fields 169 * without breaking compatibility with each other. 170 */ 171 typedef struct AVFrame { 172 #define AV_NUM_DATA_POINTERS 8 173 /** 174 * pointer to the picture/channel planes. 175 * This might be different from the first allocated byte 176 * 177 * Some decoders access areas outside 0,0 - width,height, please 178 * see avcodec_align_dimensions2(). Some filters and swscale can read 179 * up to 16 bytes beyond the planes, if these filters are to be used, 180 * then 16 extra bytes must be allocated. 181 */ 182 uint8_t *data[AV_NUM_DATA_POINTERS]; 183 184 /** 185 * For video, size in bytes of each picture line. 186 * For audio, size in bytes of each plane. 187 * 188 * For audio, only linesize[0] may be set. For planar audio, each channel 189 * plane must be the same size. 190 * 191 * For video the linesizes should be multiples of the CPUs alignment 192 * preference, this is 16 or 32 for modern desktop CPUs. 193 * Some code requires such alignment other code can be slower without 194 * correct alignment, for yet other it makes no difference. 195 * 196 * @note The linesize may be larger than the size of usable data -- there 197 * may be extra padding present for performance reasons. 198 */ 199 int linesize[AV_NUM_DATA_POINTERS]; 200 201 /** 202 * pointers to the data planes/channels. 203 * 204 * For video, this should simply point to data[]. 205 * 206 * For planar audio, each channel has a separate data pointer, and 207 * linesize[0] contains the size of each channel buffer. 208 * For packed audio, there is just one data pointer, and linesize[0] 209 * contains the total size of the buffer for all channels. 210 * 211 * Note: Both data and extended_data should always be set in a valid frame, 212 * but for planar audio with more channels that can fit in data, 213 * extended_data must be used in order to access all channels. 214 */ 215 uint8_t **extended_data; 216 217 /** 218 * width and height of the video frame 219 */ 220 int width, height; 221 222 /** 223 * number of audio samples (per channel) described by this frame 224 */ 225 int nb_samples; 226 227 /** 228 * format of the frame, -1 if unknown or unset 229 * Values correspond to enum AVPixelFormat for video frames, 230 * enum AVSampleFormat for audio) 231 */ 232 int format; 233 234 /** 235 * 1 -> keyframe, 0-> not 236 */ 237 int key_frame; 238 239 /** 240 * Picture type of the frame. 241 */ 242 enum AVPictureType pict_type; 243 244 /** 245 * Sample aspect ratio for the video frame, 0/1 if unknown/unspecified. 246 */ 247 AVRational sample_aspect_ratio; 248 249 /** 250 * Presentation timestamp in time_base units (time when frame should be shown to user). 251 */ 252 int64_t pts; 253 254 /** 255 * PTS copied from the AVPacket that was decoded to produce this frame. 256 */ 257 int64_t pkt_pts; 258 259 /** 260 * DTS copied from the AVPacket that triggered returning this frame. (if frame threading isn't used) 261 * This is also the Presentation time of this AVFrame calculated from 262 * only AVPacket.dts values without pts values. 263 */ 264 int64_t pkt_dts; 265 266 /** 267 * picture number in bitstream order 268 */ 269 int coded_picture_number; 270 /** 271 * picture number in display order 272 */ 273 int display_picture_number; 274 275 /** 276 * quality (between 1 (good) and FF_LAMBDA_MAX (bad)) 277 */ 278 int quality; 279 280 /** 281 * for some private data of the user 282 */ 283 void *opaque; 284 285 #if FF_API_ERROR_FRAME 286 /** 287 * @deprecated unused 288 */ 289 attribute_deprecated 290 uint64_t error[AV_NUM_DATA_POINTERS]; 291 #endif 292 293 /** 294 * When decoding, this signals how much the picture must be delayed. 295 * extra_delay = repeat_pict / (2*fps) 296 */ 297 int repeat_pict; 298 299 /** 300 * The content of the picture is interlaced. 301 */ 302 int interlaced_frame; 303 304 /** 305 * If the content is interlaced, is top field displayed first. 306 */ 307 int top_field_first; 308 309 /** 310 * Tell user application that palette has changed from previous frame. 311 */ 312 int palette_has_changed; 313 314 /** 315 * reordered opaque 64bit (generally an integer or a double precision float 316 * PTS but can be anything). 317 * The user sets AVCodecContext.reordered_opaque to represent the input at 318 * that time, 319 * the decoder reorders values as needed and sets AVFrame.reordered_opaque 320 * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque 321 * @deprecated in favor of pkt_pts 322 */ 323 int64_t reordered_opaque; 324 325 /** 326 * Sample rate of the audio data. 327 */ 328 int sample_rate; 329 330 /** 331 * Channel layout of the audio data. 332 */ 333 uint64_t channel_layout; 334 335 /** 336 * AVBuffer references backing the data for this frame. If all elements of 337 * this array are NULL, then this frame is not reference counted. This array 338 * must be filled contiguously -- if buf[i] is non-NULL then buf[j] must 339 * also be non-NULL for all j < i. 340 * 341 * There may be at most one AVBuffer per data plane, so for video this array 342 * always contains all the references. For planar audio with more than 343 * AV_NUM_DATA_POINTERS channels, there may be more buffers than can fit in 344 * this array. Then the extra AVBufferRef pointers are stored in the 345 * extended_buf array. 346 */ 347 AVBufferRef *buf[AV_NUM_DATA_POINTERS]; 348 349 /** 350 * For planar audio which requires more than AV_NUM_DATA_POINTERS 351 * AVBufferRef pointers, this array will hold all the references which 352 * cannot fit into AVFrame.buf. 353 * 354 * Note that this is different from AVFrame.extended_data, which always 355 * contains all the pointers. This array only contains the extra pointers, 356 * which cannot fit into AVFrame.buf. 357 * 358 * This array is always allocated using av_malloc() by whoever constructs 359 * the frame. It is freed in av_frame_unref(). 360 */ 361 AVBufferRef **extended_buf; 362 /** 363 * Number of elements in extended_buf. 364 */ 365 int nb_extended_buf; 366 367 AVFrameSideData **side_data; 368 int nb_side_data; 369 370 /** 371 * @defgroup lavu_frame_flags AV_FRAME_FLAGS 372 * Flags describing additional frame properties. 373 * 374 * @{ 375 */ 376 377 /** 378 * The frame data may be corrupted, e.g. due to decoding errors. 379 */ 380 #define AV_FRAME_FLAG_CORRUPT (1 << 0) 381 /** 382 * @} 383 */ 384 385 /** 386 * Frame flags, a combination of @ref lavu_frame_flags 387 */ 388 int flags; 389 390 /** 391 * MPEG vs JPEG YUV range. 392 * It must be accessed using av_frame_get_color_range() and 393 * av_frame_set_color_range(). 394 * - encoding: Set by user 395 * - decoding: Set by libavcodec 396 */ 397 enum AVColorRange color_range; 398 399 enum AVColorPrimaries color_primaries; 400 401 enum AVColorTransferCharacteristic color_trc; 402 403 /** 404 * YUV colorspace type. 405 * It must be accessed using av_frame_get_colorspace() and 406 * av_frame_set_colorspace(). 407 * - encoding: Set by user 408 * - decoding: Set by libavcodec 409 */ 410 enum AVColorSpace colorspace; 411 412 enum AVChromaLocation chroma_location; 413 414 /** 415 * frame timestamp estimated using various heuristics, in stream time base 416 * Code outside libavutil should access this field using: 417 * av_frame_get_best_effort_timestamp(frame) 418 * - encoding: unused 419 * - decoding: set by libavcodec, read by user. 420 */ 421 int64_t best_effort_timestamp; 422 423 /** 424 * reordered pos from the last AVPacket that has been input into the decoder 425 * Code outside libavutil should access this field using: 426 * av_frame_get_pkt_pos(frame) 427 * - encoding: unused 428 * - decoding: Read by user. 429 */ 430 int64_t pkt_pos; 431 432 /** 433 * duration of the corresponding packet, expressed in 434 * AVStream->time_base units, 0 if unknown. 435 * Code outside libavutil should access this field using: 436 * av_frame_get_pkt_duration(frame) 437 * - encoding: unused 438 * - decoding: Read by user. 439 */ 440 int64_t pkt_duration; 441 442 /** 443 * metadata. 444 * Code outside libavutil should access this field using: 445 * av_frame_get_metadata(frame) 446 * - encoding: Set by user. 447 * - decoding: Set by libavcodec. 448 */ 449 AVDictionary *metadata; 450 451 /** 452 * decode error flags of the frame, set to a combination of 453 * FF_DECODE_ERROR_xxx flags if the decoder produced a frame, but there 454 * were errors during the decoding. 455 * Code outside libavutil should access this field using: 456 * av_frame_get_decode_error_flags(frame) 457 * - encoding: unused 458 * - decoding: set by libavcodec, read by user. 459 */ 460 int decode_error_flags; 461 #define FF_DECODE_ERROR_INVALID_BITSTREAM 1 462 #define FF_DECODE_ERROR_MISSING_REFERENCE 2 463 464 /** 465 * number of audio channels, only used for audio. 466 * Code outside libavutil should access this field using: 467 * av_frame_get_channels(frame) 468 * - encoding: unused 469 * - decoding: Read by user. 470 */ 471 int channels; 472 473 /** 474 * size of the corresponding packet containing the compressed 475 * frame. It must be accessed using av_frame_get_pkt_size() and 476 * av_frame_set_pkt_size(). 477 * It is set to a negative value if unknown. 478 * - encoding: unused 479 * - decoding: set by libavcodec, read by user. 480 */ 481 int pkt_size; 482 483 #if FF_API_FRAME_QP 484 /** 485 * QP table 486 * Not to be accessed directly from outside libavutil 487 */ 488 attribute_deprecated 489 int8_t *qscale_table; 490 /** 491 * QP store stride 492 * Not to be accessed directly from outside libavutil 493 */ 494 attribute_deprecated 495 int qstride; 496 497 attribute_deprecated 498 int qscale_type; 499 500 /** 501 * Not to be accessed directly from outside libavutil 502 */ 503 AVBufferRef *qp_table_buf; 504 #endif 505 } AVFrame; 506 507 /** 508 * Accessors for some AVFrame fields. 509 * The position of these field in the structure is not part of the ABI, 510 * they should not be accessed directly outside libavutil. 511 */ 512 int64_t av_frame_get_best_effort_timestamp(const AVFrame *frame); 513 void av_frame_set_best_effort_timestamp(AVFrame *frame, int64_t val); 514 int64_t av_frame_get_pkt_duration (const AVFrame *frame); 515 void av_frame_set_pkt_duration (AVFrame *frame, int64_t val); 516 int64_t av_frame_get_pkt_pos (const AVFrame *frame); 517 void av_frame_set_pkt_pos (AVFrame *frame, int64_t val); 518 int64_t av_frame_get_channel_layout (const AVFrame *frame); 519 void av_frame_set_channel_layout (AVFrame *frame, int64_t val); 520 int av_frame_get_channels (const AVFrame *frame); 521 void av_frame_set_channels (AVFrame *frame, int val); 522 int av_frame_get_sample_rate (const AVFrame *frame); 523 void av_frame_set_sample_rate (AVFrame *frame, int val); 524 AVDictionary *av_frame_get_metadata (const AVFrame *frame); 525 void av_frame_set_metadata (AVFrame *frame, AVDictionary *val); 526 int av_frame_get_decode_error_flags (const AVFrame *frame); 527 void av_frame_set_decode_error_flags (AVFrame *frame, int val); 528 int av_frame_get_pkt_size(const AVFrame *frame); 529 void av_frame_set_pkt_size(AVFrame *frame, int val); 530 AVDictionary **avpriv_frame_get_metadatap(AVFrame *frame); 531 #if FF_API_FRAME_QP 532 int8_t *av_frame_get_qp_table(AVFrame *f, int *stride, int *type); 533 int av_frame_set_qp_table(AVFrame *f, AVBufferRef *buf, int stride, int type); 534 #endif 535 enum AVColorSpace av_frame_get_colorspace(const AVFrame *frame); 536 void av_frame_set_colorspace(AVFrame *frame, enum AVColorSpace val); 537 enum AVColorRange av_frame_get_color_range(const AVFrame *frame); 538 void av_frame_set_color_range(AVFrame *frame, enum AVColorRange val); 539 540 /** 541 * Get the name of a colorspace. 542 * @return a static string identifying the colorspace; can be NULL. 543 */ 544 const char *av_get_colorspace_name(enum AVColorSpace val); 545 546 /** 547 * Allocate an AVFrame and set its fields to default values. The resulting 548 * struct must be freed using av_frame_free(). 549 * 550 * @return An AVFrame filled with default values or NULL on failure. 551 * 552 * @note this only allocates the AVFrame itself, not the data buffers. Those 553 * must be allocated through other means, e.g. with av_frame_get_buffer() or 554 * manually. 555 */ 556 AVFrame *av_frame_alloc(void); 557 558 /** 559 * Free the frame and any dynamically allocated objects in it, 560 * e.g. extended_data. If the frame is reference counted, it will be 561 * unreferenced first. 562 * 563 * @param frame frame to be freed. The pointer will be set to NULL. 564 */ 565 void av_frame_free(AVFrame **frame); 566 567 /** 568 * Set up a new reference to the data described by the source frame. 569 * 570 * Copy frame properties from src to dst and create a new reference for each 571 * AVBufferRef from src. 572 * 573 * If src is not reference counted, new buffers are allocated and the data is 574 * copied. 575 * 576 * @return 0 on success, a negative AVERROR on error 577 */ 578 int av_frame_ref(AVFrame *dst, const AVFrame *src); 579 580 /** 581 * Create a new frame that references the same data as src. 582 * 583 * This is a shortcut for av_frame_alloc()+av_frame_ref(). 584 * 585 * @return newly created AVFrame on success, NULL on error. 586 */ 587 AVFrame *av_frame_clone(const AVFrame *src); 588 589 /** 590 * Unreference all the buffers referenced by frame and reset the frame fields. 591 */ 592 void av_frame_unref(AVFrame *frame); 593 594 /** 595 * Move everything contained in src to dst and reset src. 596 */ 597 void av_frame_move_ref(AVFrame *dst, AVFrame *src); 598 599 /** 600 * Allocate new buffer(s) for audio or video data. 601 * 602 * The following fields must be set on frame before calling this function: 603 * - format (pixel format for video, sample format for audio) 604 * - width and height for video 605 * - nb_samples and channel_layout for audio 606 * 607 * This function will fill AVFrame.data and AVFrame.buf arrays and, if 608 * necessary, allocate and fill AVFrame.extended_data and AVFrame.extended_buf. 609 * For planar formats, one buffer will be allocated for each plane. 610 * 611 * @param frame frame in which to store the new buffers. 612 * @param align required buffer size alignment 613 * 614 * @return 0 on success, a negative AVERROR on error. 615 */ 616 int av_frame_get_buffer(AVFrame *frame, int align); 617 618 /** 619 * Check if the frame data is writable. 620 * 621 * @return A positive value if the frame data is writable (which is true if and 622 * only if each of the underlying buffers has only one reference, namely the one 623 * stored in this frame). Return 0 otherwise. 624 * 625 * If 1 is returned the answer is valid until av_buffer_ref() is called on any 626 * of the underlying AVBufferRefs (e.g. through av_frame_ref() or directly). 627 * 628 * @see av_frame_make_writable(), av_buffer_is_writable() 629 */ 630 int av_frame_is_writable(AVFrame *frame); 631 632 /** 633 * Ensure that the frame data is writable, avoiding data copy if possible. 634 * 635 * Do nothing if the frame is writable, allocate new buffers and copy the data 636 * if it is not. 637 * 638 * @return 0 on success, a negative AVERROR on error. 639 * 640 * @see av_frame_is_writable(), av_buffer_is_writable(), 641 * av_buffer_make_writable() 642 */ 643 int av_frame_make_writable(AVFrame *frame); 644 645 /** 646 * Copy the frame data from src to dst. 647 * 648 * This function does not allocate anything, dst must be already initialized and 649 * allocated with the same parameters as src. 650 * 651 * This function only copies the frame data (i.e. the contents of the data / 652 * extended data arrays), not any other properties. 653 * 654 * @return >= 0 on success, a negative AVERROR on error. 655 */ 656 int av_frame_copy(AVFrame *dst, const AVFrame *src); 657 658 /** 659 * Copy only "metadata" fields from src to dst. 660 * 661 * Metadata for the purpose of this function are those fields that do not affect 662 * the data layout in the buffers. E.g. pts, sample rate (for audio) or sample 663 * aspect ratio (for video), but not width/height or channel layout. 664 * Side data is also copied. 665 */ 666 int av_frame_copy_props(AVFrame *dst, const AVFrame *src); 667 668 /** 669 * Get the buffer reference a given data plane is stored in. 670 * 671 * @param plane index of the data plane of interest in frame->extended_data. 672 * 673 * @return the buffer reference that contains the plane or NULL if the input 674 * frame is not valid. 675 */ 676 AVBufferRef *av_frame_get_plane_buffer(AVFrame *frame, int plane); 677 678 /** 679 * Add a new side data to a frame. 680 * 681 * @param frame a frame to which the side data should be added 682 * @param type type of the added side data 683 * @param size size of the side data 684 * 685 * @return newly added side data on success, NULL on error 686 */ 687 AVFrameSideData *av_frame_new_side_data(AVFrame *frame, 688 enum AVFrameSideDataType type, 689 int size); 690 691 /** 692 * @return a pointer to the side data of a given type on success, NULL if there 693 * is no side data with such type in this frame. 694 */ 695 AVFrameSideData *av_frame_get_side_data(const AVFrame *frame, 696 enum AVFrameSideDataType type); 697 698 /** 699 * If side data of the supplied type exists in the frame, free it and remove it 700 * from the frame. 701 */ 702 void av_frame_remove_side_data(AVFrame *frame, enum AVFrameSideDataType type); 703 704 /** 705 * @return a string identifying the side data type 706 */ 707 const char *av_frame_side_data_name(enum AVFrameSideDataType type); 708 709 /** 710 * @} 711 */ 712 713 #endif /* AVUTIL_FRAME_H */ 714