| 1 | /* |
| 2 | * This file is part of FFmpeg. |
| 3 | * |
| 4 | * FFmpeg is free software; you can redistribute it and/or |
| 5 | * modify it under the terms of the GNU Lesser General Public |
| 6 | * License as published by the Free Software Foundation; either |
| 7 | * version 2.1 of the License, or (at your option) any later version. |
| 8 | * |
| 9 | * FFmpeg is distributed in the hope that it will be useful, |
| 10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 12 | * Lesser General Public License for more details. |
| 13 | * |
| 14 | * You should have received a copy of the GNU Lesser General Public |
| 15 | * License along with FFmpeg; if not, write to the Free Software |
| 16 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 17 | */ |
| 18 | |
| 19 | /** |
| 20 | * @file |
| 21 | * @ingroup lavu_frame |
| 22 | * reference-counted frame API |
| 23 | */ |
| 24 | |
| 25 | #ifndef AVUTIL_FRAME_H |
| 26 | #define AVUTIL_FRAME_H |
| 27 | |
| 28 | #include <stddef.h> |
| 29 | #include <stdint.h> |
| 30 | |
| 31 | #include "avutil.h" |
| 32 | #include "buffer.h" |
| 33 | #include "channel_layout.h" |
| 34 | #include "dict.h" |
| 35 | #include "rational.h" |
| 36 | #include "samplefmt.h" |
| 37 | #include "pixfmt.h" |
| 38 | #include "version.h" |
| 39 | |
| 40 | |
| 41 | /** |
| 42 | * @defgroup lavu_frame AVFrame |
| 43 | * @ingroup lavu_data |
| 44 | * |
| 45 | * @{ |
| 46 | * AVFrame is an abstraction for reference-counted raw multimedia data. |
| 47 | */ |
| 48 | |
| 49 | enum AVFrameSideDataType { |
| 50 | /** |
| 51 | * The data is the AVPanScan struct defined in libavcodec. |
| 52 | */ |
| 53 | AV_FRAME_DATA_PANSCAN, |
| 54 | /** |
| 55 | * ATSC A53 Part 4 Closed Captions. |
| 56 | * A53 CC bitstream is stored as uint8_t in AVFrameSideData.data. |
| 57 | * The number of bytes of CC data is AVFrameSideData.size. |
| 58 | */ |
| 59 | AV_FRAME_DATA_A53_CC, |
| 60 | /** |
| 61 | * Stereoscopic 3d metadata. |
| 62 | * The data is the AVStereo3D struct defined in libavutil/stereo3d.h. |
| 63 | */ |
| 64 | AV_FRAME_DATA_STEREO3D, |
| 65 | /** |
| 66 | * The data is the AVMatrixEncoding enum defined in libavutil/channel_layout.h. |
| 67 | */ |
| 68 | AV_FRAME_DATA_MATRIXENCODING, |
| 69 | /** |
| 70 | * Metadata relevant to a downmix procedure. |
| 71 | * The data is the AVDownmixInfo struct defined in libavutil/downmix_info.h. |
| 72 | */ |
| 73 | AV_FRAME_DATA_DOWNMIX_INFO, |
| 74 | /** |
| 75 | * ReplayGain information in the form of the AVReplayGain struct. |
| 76 | */ |
| 77 | AV_FRAME_DATA_REPLAYGAIN, |
| 78 | /** |
| 79 | * This side data contains a 3x3 transformation matrix describing an affine |
| 80 | * transformation that needs to be applied to the frame for correct |
| 81 | * presentation. |
| 82 | * |
| 83 | * See libavutil/display.h for a detailed description of the data. |
| 84 | */ |
| 85 | AV_FRAME_DATA_DISPLAYMATRIX, |
| 86 | /** |
| 87 | * Active Format Description data consisting of a single byte as specified |
| 88 | * in ETSI TS 101 154 using AVActiveFormatDescription enum. |
| 89 | */ |
| 90 | AV_FRAME_DATA_AFD, |
| 91 | /** |
| 92 | * Motion vectors exported by some codecs (on demand through the export_mvs |
| 93 | * flag set in the libavcodec AVCodecContext flags2 option). |
| 94 | * The data is the AVMotionVector struct defined in |
| 95 | * libavutil/motion_vector.h. |
| 96 | */ |
| 97 | AV_FRAME_DATA_MOTION_VECTORS, |
| 98 | /** |
| 99 | * Recommmends skipping the specified number of samples. This is exported |
| 100 | * only if the "skip_manual" AVOption is set in libavcodec. |
| 101 | * This has the same format as AV_PKT_DATA_SKIP_SAMPLES. |
| 102 | * @code |
| 103 | * u32le number of samples to skip from start of this packet |
| 104 | * u32le number of samples to skip from end of this packet |
| 105 | * u8 reason for start skip |
| 106 | * u8 reason for end skip (0=padding silence, 1=convergence) |
| 107 | * @endcode |
| 108 | */ |
| 109 | AV_FRAME_DATA_SKIP_SAMPLES, |
| 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 | * Mastering display metadata associated with a video frame. The payload is |
| 117 | * an AVMasteringDisplayMetadata type and contains information about the |
| 118 | * mastering display color volume. |
| 119 | */ |
| 120 | AV_FRAME_DATA_MASTERING_DISPLAY_METADATA, |
| 121 | /** |
| 122 | * The GOP timecode in 25 bit timecode format. Data format is 64-bit integer. |
| 123 | * This is set on the first frame of a GOP that has a temporal reference of 0. |
| 124 | */ |
| 125 | AV_FRAME_DATA_GOP_TIMECODE, |
| 126 | |
| 127 | /** |
| 128 | * The data represents the AVSphericalMapping structure defined in |
| 129 | * libavutil/spherical.h. |
| 130 | */ |
| 131 | AV_FRAME_DATA_SPHERICAL, |
| 132 | |
| 133 | /** |
| 134 | * Content light level (based on CTA-861.3). This payload contains data in |
| 135 | * the form of the AVContentLightMetadata struct. |
| 136 | */ |
| 137 | AV_FRAME_DATA_CONTENT_LIGHT_LEVEL, |
| 138 | |
| 139 | /** |
| 140 | * The data contains an ICC profile as an opaque octet buffer following the |
| 141 | * format described by ISO 15076-1 with an optional name defined in the |
| 142 | * metadata key entry "name". |
| 143 | */ |
| 144 | AV_FRAME_DATA_ICC_PROFILE, |
| 145 | |
| 146 | /** |
| 147 | * Timecode which conforms to SMPTE ST 12-1. The data is an array of 4 uint32_t |
| 148 | * where the first uint32_t describes how many (1-3) of the other timecodes are used. |
| 149 | * The timecode format is described in the documentation of av_timecode_get_smpte_from_framenum() |
| 150 | * function in libavutil/timecode.h. |
| 151 | */ |
| 152 | AV_FRAME_DATA_S12M_TIMECODE, |
| 153 | |
| 154 | /** |
| 155 | * HDR dynamic metadata associated with a video frame. The payload is |
| 156 | * an AVDynamicHDRPlus type and contains information for color |
| 157 | * volume transform - application 4 of SMPTE 2094-40:2016 standard. |
| 158 | */ |
| 159 | AV_FRAME_DATA_DYNAMIC_HDR_PLUS, |
| 160 | |
| 161 | /** |
| 162 | * Regions Of Interest, the data is an array of AVRegionOfInterest type, the number of |
| 163 | * array element is implied by AVFrameSideData.size / AVRegionOfInterest.self_size. |
| 164 | */ |
| 165 | AV_FRAME_DATA_REGIONS_OF_INTEREST, |
| 166 | |
| 167 | /** |
| 168 | * Encoding parameters for a video frame, as described by AVVideoEncParams. |
| 169 | */ |
| 170 | AV_FRAME_DATA_VIDEO_ENC_PARAMS, |
| 171 | |
| 172 | /** |
| 173 | * User data unregistered metadata associated with a video frame. |
| 174 | * This is the H.26[45] UDU SEI message, and shouldn't be used for any other purpose |
| 175 | * The data is stored as uint8_t in AVFrameSideData.data which is 16 bytes of |
| 176 | * uuid_iso_iec_11578 followed by AVFrameSideData.size - 16 bytes of user_data_payload_byte. |
| 177 | */ |
| 178 | AV_FRAME_DATA_SEI_UNREGISTERED, |
| 179 | |
| 180 | /** |
| 181 | * Film grain parameters for a frame, described by AVFilmGrainParams. |
| 182 | * Must be present for every frame which should have film grain applied. |
| 183 | */ |
| 184 | AV_FRAME_DATA_FILM_GRAIN_PARAMS, |
| 185 | |
| 186 | /** |
| 187 | * Bounding boxes for object detection and classification, |
| 188 | * as described by AVDetectionBBoxHeader. |
| 189 | */ |
| 190 | AV_FRAME_DATA_DETECTION_BBOXES, |
| 191 | |
| 192 | /** |
| 193 | * Dolby Vision RPU raw data, suitable for passing to x265 |
| 194 | * or other libraries. Array of uint8_t, with NAL emulation |
| 195 | * bytes intact. |
| 196 | */ |
| 197 | AV_FRAME_DATA_DOVI_RPU_BUFFER, |
| 198 | |
| 199 | /** |
| 200 | * Parsed Dolby Vision metadata, suitable for passing to a software |
| 201 | * implementation. The payload is the AVDOVIMetadata struct defined in |
| 202 | * libavutil/dovi_meta.h. |
| 203 | */ |
| 204 | AV_FRAME_DATA_DOVI_METADATA, |
| 205 | |
| 206 | /** |
| 207 | * HDR Vivid dynamic metadata associated with a video frame. The payload is |
| 208 | * an AVDynamicHDRVivid type and contains information for color |
| 209 | * volume transform - CUVA 005.1-2021. |
| 210 | */ |
| 211 | AV_FRAME_DATA_DYNAMIC_HDR_VIVID, |
| 212 | |
| 213 | /** |
| 214 | * Ambient viewing environment metadata, as defined by H.274. |
| 215 | */ |
| 216 | AV_FRAME_DATA_AMBIENT_VIEWING_ENVIRONMENT, |
| 217 | |
| 218 | /** |
| 219 | * Provide encoder-specific hinting information about changed/unchanged |
| 220 | * portions of a frame. It can be used to pass information about which |
| 221 | * macroblocks can be skipped because they didn't change from the |
| 222 | * corresponding ones in the previous frame. This could be useful for |
| 223 | * applications which know this information in advance to speed up |
| 224 | * encoding. |
| 225 | */ |
| 226 | AV_FRAME_DATA_VIDEO_HINT, |
| 227 | }; |
| 228 | |
| 229 | enum AVActiveFormatDescription { |
| 230 | AV_AFD_SAME = 8, |
| 231 | AV_AFD_4_3 = 9, |
| 232 | AV_AFD_16_9 = 10, |
| 233 | AV_AFD_14_9 = 11, |
| 234 | AV_AFD_4_3_SP_14_9 = 13, |
| 235 | AV_AFD_16_9_SP_14_9 = 14, |
| 236 | AV_AFD_SP_4_3 = 15, |
| 237 | }; |
| 238 | |
| 239 | |
| 240 | /** |
| 241 | * Structure to hold side data for an AVFrame. |
| 242 | * |
| 243 | * sizeof(AVFrameSideData) is not a part of the public ABI, so new fields may be added |
| 244 | * to the end with a minor bump. |
| 245 | */ |
| 246 | typedef struct AVFrameSideData { |
| 247 | enum AVFrameSideDataType type; |
| 248 | uint8_t *data; |
| 249 | size_t size; |
| 250 | AVDictionary *metadata; |
| 251 | AVBufferRef *buf; |
| 252 | } AVFrameSideData; |
| 253 | |
| 254 | /** |
| 255 | * Structure describing a single Region Of Interest. |
| 256 | * |
| 257 | * When multiple regions are defined in a single side-data block, they |
| 258 | * should be ordered from most to least important - some encoders are only |
| 259 | * capable of supporting a limited number of distinct regions, so will have |
| 260 | * to truncate the list. |
| 261 | * |
| 262 | * When overlapping regions are defined, the first region containing a given |
| 263 | * area of the frame applies. |
| 264 | */ |
| 265 | typedef struct AVRegionOfInterest { |
| 266 | /** |
| 267 | * Must be set to the size of this data structure (that is, |
| 268 | * sizeof(AVRegionOfInterest)). |
| 269 | */ |
| 270 | uint32_t self_size; |
| 271 | /** |
| 272 | * Distance in pixels from the top edge of the frame to the top and |
| 273 | * bottom edges and from the left edge of the frame to the left and |
| 274 | * right edges of the rectangle defining this region of interest. |
| 275 | * |
| 276 | * The constraints on a region are encoder dependent, so the region |
| 277 | * actually affected may be slightly larger for alignment or other |
| 278 | * reasons. |
| 279 | */ |
| 280 | int top; |
| 281 | int bottom; |
| 282 | int left; |
| 283 | int right; |
| 284 | /** |
| 285 | * Quantisation offset. |
| 286 | * |
| 287 | * Must be in the range -1 to +1. A value of zero indicates no quality |
| 288 | * change. A negative value asks for better quality (less quantisation), |
| 289 | * while a positive value asks for worse quality (greater quantisation). |
| 290 | * |
| 291 | * The range is calibrated so that the extreme values indicate the |
| 292 | * largest possible offset - if the rest of the frame is encoded with the |
| 293 | * worst possible quality, an offset of -1 indicates that this region |
| 294 | * should be encoded with the best possible quality anyway. Intermediate |
| 295 | * values are then interpolated in some codec-dependent way. |
| 296 | * |
| 297 | * For example, in 10-bit H.264 the quantisation parameter varies between |
| 298 | * -12 and 51. A typical qoffset value of -1/10 therefore indicates that |
| 299 | * this region should be encoded with a QP around one-tenth of the full |
| 300 | * range better than the rest of the frame. So, if most of the frame |
| 301 | * were to be encoded with a QP of around 30, this region would get a QP |
| 302 | * of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3). |
| 303 | * An extreme value of -1 would indicate that this region should be |
| 304 | * encoded with the best possible quality regardless of the treatment of |
| 305 | * the rest of the frame - that is, should be encoded at a QP of -12. |
| 306 | */ |
| 307 | AVRational qoffset; |
| 308 | } AVRegionOfInterest; |
| 309 | |
| 310 | /** |
| 311 | * This structure describes decoded (raw) audio or video data. |
| 312 | * |
| 313 | * AVFrame must be allocated using av_frame_alloc(). Note that this only |
| 314 | * allocates the AVFrame itself, the buffers for the data must be managed |
| 315 | * through other means (see below). |
| 316 | * AVFrame must be freed with av_frame_free(). |
| 317 | * |
| 318 | * AVFrame is typically allocated once and then reused multiple times to hold |
| 319 | * different data (e.g. a single AVFrame to hold frames received from a |
| 320 | * decoder). In such a case, av_frame_unref() will free any references held by |
| 321 | * the frame and reset it to its original clean state before it |
| 322 | * is reused again. |
| 323 | * |
| 324 | * The data described by an AVFrame is usually reference counted through the |
| 325 | * AVBuffer API. The underlying buffer references are stored in AVFrame.buf / |
| 326 | * AVFrame.extended_buf. An AVFrame is considered to be reference counted if at |
| 327 | * least one reference is set, i.e. if AVFrame.buf[0] != NULL. In such a case, |
| 328 | * every single data plane must be contained in one of the buffers in |
| 329 | * AVFrame.buf or AVFrame.extended_buf. |
| 330 | * There may be a single buffer for all the data, or one separate buffer for |
| 331 | * each plane, or anything in between. |
| 332 | * |
| 333 | * sizeof(AVFrame) is not a part of the public ABI, so new fields may be added |
| 334 | * to the end with a minor bump. |
| 335 | * |
| 336 | * Fields can be accessed through AVOptions, the name string used, matches the |
| 337 | * C structure field name for fields accessible through AVOptions. The AVClass |
| 338 | * for AVFrame can be obtained from avcodec_get_frame_class() |
| 339 | */ |
| 340 | typedef struct AVFrame { |
| 341 | #define AV_NUM_DATA_POINTERS 8 |
| 342 | /** |
| 343 | * pointer to the picture/channel planes. |
| 344 | * This might be different from the first allocated byte. For video, |
| 345 | * it could even point to the end of the image data. |
| 346 | * |
| 347 | * All pointers in data and extended_data must point into one of the |
| 348 | * AVBufferRef in buf or extended_buf. |
| 349 | * |
| 350 | * Some decoders access areas outside 0,0 - width,height, please |
| 351 | * see avcodec_align_dimensions2(). Some filters and swscale can read |
| 352 | * up to 16 bytes beyond the planes, if these filters are to be used, |
| 353 | * then 16 extra bytes must be allocated. |
| 354 | * |
| 355 | * NOTE: Pointers not needed by the format MUST be set to NULL. |
| 356 | * |
| 357 | * @attention In case of video, the data[] pointers can point to the |
| 358 | * end of image data in order to reverse line order, when used in |
| 359 | * combination with negative values in the linesize[] array. |
| 360 | */ |
| 361 | uint8_t *data[AV_NUM_DATA_POINTERS]; |
| 362 | |
| 363 | /** |
| 364 | * For video, a positive or negative value, which is typically indicating |
| 365 | * the size in bytes of each picture line, but it can also be: |
| 366 | * - the negative byte size of lines for vertical flipping |
| 367 | * (with data[n] pointing to the end of the data |
| 368 | * - a positive or negative multiple of the byte size as for accessing |
| 369 | * even and odd fields of a frame (possibly flipped) |
| 370 | * |
| 371 | * For audio, only linesize[0] may be set. For planar audio, each channel |
| 372 | * plane must be the same size. |
| 373 | * |
| 374 | * For video the linesizes should be multiples of the CPUs alignment |
| 375 | * preference, this is 16 or 32 for modern desktop CPUs. |
| 376 | * Some code requires such alignment other code can be slower without |
| 377 | * correct alignment, for yet other it makes no difference. |
| 378 | * |
| 379 | * @note The linesize may be larger than the size of usable data -- there |
| 380 | * may be extra padding present for performance reasons. |
| 381 | * |
| 382 | * @attention In case of video, line size values can be negative to achieve |
| 383 | * a vertically inverted iteration over image lines. |
| 384 | */ |
| 385 | int linesize[AV_NUM_DATA_POINTERS]; |
| 386 | |
| 387 | /** |
| 388 | * pointers to the data planes/channels. |
| 389 | * |
| 390 | * For video, this should simply point to data[]. |
| 391 | * |
| 392 | * For planar audio, each channel has a separate data pointer, and |
| 393 | * linesize[0] contains the size of each channel buffer. |
| 394 | * For packed audio, there is just one data pointer, and linesize[0] |
| 395 | * contains the total size of the buffer for all channels. |
| 396 | * |
| 397 | * Note: Both data and extended_data should always be set in a valid frame, |
| 398 | * but for planar audio with more channels that can fit in data, |
| 399 | * extended_data must be used in order to access all channels. |
| 400 | */ |
| 401 | uint8_t **extended_data; |
| 402 | |
| 403 | /** |
| 404 | * @name Video dimensions |
| 405 | * Video frames only. The coded dimensions (in pixels) of the video frame, |
| 406 | * i.e. the size of the rectangle that contains some well-defined values. |
| 407 | * |
| 408 | * @note The part of the frame intended for display/presentation is further |
| 409 | * restricted by the @ref cropping "Cropping rectangle". |
| 410 | * @{ |
| 411 | */ |
| 412 | int width, height; |
| 413 | /** |
| 414 | * @} |
| 415 | */ |
| 416 | |
| 417 | /** |
| 418 | * number of audio samples (per channel) described by this frame |
| 419 | */ |
| 420 | int nb_samples; |
| 421 | |
| 422 | /** |
| 423 | * format of the frame, -1 if unknown or unset |
| 424 | * Values correspond to enum AVPixelFormat for video frames, |
| 425 | * enum AVSampleFormat for audio) |
| 426 | */ |
| 427 | int format; |
| 428 | |
| 429 | #if FF_API_FRAME_KEY |
| 430 | /** |
| 431 | * 1 -> keyframe, 0-> not |
| 432 | * |
| 433 | * @deprecated Use AV_FRAME_FLAG_KEY instead |
| 434 | */ |
| 435 | attribute_deprecated |
| 436 | int key_frame; |
| 437 | #endif |
| 438 | |
| 439 | /** |
| 440 | * Picture type of the frame. |
| 441 | */ |
| 442 | enum AVPictureType pict_type; |
| 443 | |
| 444 | /** |
| 445 | * Sample aspect ratio for the video frame, 0/1 if unknown/unspecified. |
| 446 | */ |
| 447 | AVRational sample_aspect_ratio; |
| 448 | |
| 449 | /** |
| 450 | * Presentation timestamp in time_base units (time when frame should be shown to user). |
| 451 | */ |
| 452 | int64_t pts; |
| 453 | |
| 454 | /** |
| 455 | * DTS copied from the AVPacket that triggered returning this frame. (if frame threading isn't used) |
| 456 | * This is also the Presentation time of this AVFrame calculated from |
| 457 | * only AVPacket.dts values without pts values. |
| 458 | */ |
| 459 | int64_t pkt_dts; |
| 460 | |
| 461 | /** |
| 462 | * Time base for the timestamps in this frame. |
| 463 | * In the future, this field may be set on frames output by decoders or |
| 464 | * filters, but its value will be by default ignored on input to encoders |
| 465 | * or filters. |
| 466 | */ |
| 467 | AVRational time_base; |
| 468 | |
| 469 | #if FF_API_FRAME_PICTURE_NUMBER |
| 470 | /** |
| 471 | * picture number in bitstream order |
| 472 | */ |
| 473 | attribute_deprecated |
| 474 | int coded_picture_number; |
| 475 | /** |
| 476 | * picture number in display order |
| 477 | */ |
| 478 | attribute_deprecated |
| 479 | int display_picture_number; |
| 480 | #endif |
| 481 | |
| 482 | /** |
| 483 | * quality (between 1 (good) and FF_LAMBDA_MAX (bad)) |
| 484 | */ |
| 485 | int quality; |
| 486 | |
| 487 | /** |
| 488 | * Frame owner's private data. |
| 489 | * |
| 490 | * This field may be set by the code that allocates/owns the frame data. |
| 491 | * It is then not touched by any library functions, except: |
| 492 | * - it is copied to other references by av_frame_copy_props() (and hence by |
| 493 | * av_frame_ref()); |
| 494 | * - it is set to NULL when the frame is cleared by av_frame_unref() |
| 495 | * - on the caller's explicit request. E.g. libavcodec encoders/decoders |
| 496 | * will copy this field to/from @ref AVPacket "AVPackets" if the caller sets |
| 497 | * @ref AV_CODEC_FLAG_COPY_OPAQUE. |
| 498 | * |
| 499 | * @see opaque_ref the reference-counted analogue |
| 500 | */ |
| 501 | void *opaque; |
| 502 | |
| 503 | /** |
| 504 | * Number of fields in this frame which should be repeated, i.e. the total |
| 505 | * duration of this frame should be repeat_pict + 2 normal field durations. |
| 506 | * |
| 507 | * For interlaced frames this field may be set to 1, which signals that this |
| 508 | * frame should be presented as 3 fields: beginning with the first field (as |
| 509 | * determined by AV_FRAME_FLAG_TOP_FIELD_FIRST being set or not), followed |
| 510 | * by the second field, and then the first field again. |
| 511 | * |
| 512 | * For progressive frames this field may be set to a multiple of 2, which |
| 513 | * signals that this frame's duration should be (repeat_pict + 2) / 2 |
| 514 | * normal frame durations. |
| 515 | * |
| 516 | * @note This field is computed from MPEG2 repeat_first_field flag and its |
| 517 | * associated flags, H.264 pic_struct from picture timing SEI, and |
| 518 | * their analogues in other codecs. Typically it should only be used when |
| 519 | * higher-layer timing information is not available. |
| 520 | */ |
| 521 | int repeat_pict; |
| 522 | |
| 523 | #if FF_API_INTERLACED_FRAME |
| 524 | /** |
| 525 | * The content of the picture is interlaced. |
| 526 | * |
| 527 | * @deprecated Use AV_FRAME_FLAG_INTERLACED instead |
| 528 | */ |
| 529 | attribute_deprecated |
| 530 | int interlaced_frame; |
| 531 | |
| 532 | /** |
| 533 | * If the content is interlaced, is top field displayed first. |
| 534 | * |
| 535 | * @deprecated Use AV_FRAME_FLAG_TOP_FIELD_FIRST instead |
| 536 | */ |
| 537 | attribute_deprecated |
| 538 | int top_field_first; |
| 539 | #endif |
| 540 | |
| 541 | #if FF_API_PALETTE_HAS_CHANGED |
| 542 | /** |
| 543 | * Tell user application that palette has changed from previous frame. |
| 544 | */ |
| 545 | attribute_deprecated |
| 546 | int palette_has_changed; |
| 547 | #endif |
| 548 | |
| 549 | #if FF_API_REORDERED_OPAQUE |
| 550 | /** |
| 551 | * reordered opaque 64 bits (generally an integer or a double precision float |
| 552 | * PTS but can be anything). |
| 553 | * The user sets AVCodecContext.reordered_opaque to represent the input at |
| 554 | * that time, |
| 555 | * the decoder reorders values as needed and sets AVFrame.reordered_opaque |
| 556 | * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque |
| 557 | * |
| 558 | * @deprecated Use AV_CODEC_FLAG_COPY_OPAQUE instead |
| 559 | */ |
| 560 | attribute_deprecated |
| 561 | int64_t reordered_opaque; |
| 562 | #endif |
| 563 | |
| 564 | /** |
| 565 | * Sample rate of the audio data. |
| 566 | */ |
| 567 | int sample_rate; |
| 568 | |
| 569 | #if FF_API_OLD_CHANNEL_LAYOUT |
| 570 | /** |
| 571 | * Channel layout of the audio data. |
| 572 | * @deprecated use ch_layout instead |
| 573 | */ |
| 574 | attribute_deprecated |
| 575 | uint64_t channel_layout; |
| 576 | #endif |
| 577 | |
| 578 | /** |
| 579 | * AVBuffer references backing the data for this frame. All the pointers in |
| 580 | * data and extended_data must point inside one of the buffers in buf or |
| 581 | * extended_buf. This array must be filled contiguously -- if buf[i] is |
| 582 | * non-NULL then buf[j] must also be non-NULL for all j < i. |
| 583 | * |
| 584 | * There may be at most one AVBuffer per data plane, so for video this array |
| 585 | * always contains all the references. For planar audio with more than |
| 586 | * AV_NUM_DATA_POINTERS channels, there may be more buffers than can fit in |
| 587 | * this array. Then the extra AVBufferRef pointers are stored in the |
| 588 | * extended_buf array. |
| 589 | */ |
| 590 | AVBufferRef *buf[AV_NUM_DATA_POINTERS]; |
| 591 | |
| 592 | /** |
| 593 | * For planar audio which requires more than AV_NUM_DATA_POINTERS |
| 594 | * AVBufferRef pointers, this array will hold all the references which |
| 595 | * cannot fit into AVFrame.buf. |
| 596 | * |
| 597 | * Note that this is different from AVFrame.extended_data, which always |
| 598 | * contains all the pointers. This array only contains the extra pointers, |
| 599 | * which cannot fit into AVFrame.buf. |
| 600 | * |
| 601 | * This array is always allocated using av_malloc() by whoever constructs |
| 602 | * the frame. It is freed in av_frame_unref(). |
| 603 | */ |
| 604 | AVBufferRef **extended_buf; |
| 605 | /** |
| 606 | * Number of elements in extended_buf. |
| 607 | */ |
| 608 | int nb_extended_buf; |
| 609 | |
| 610 | AVFrameSideData **side_data; |
| 611 | int nb_side_data; |
| 612 | |
| 613 | /** |
| 614 | * @defgroup lavu_frame_flags AV_FRAME_FLAGS |
| 615 | * @ingroup lavu_frame |
| 616 | * Flags describing additional frame properties. |
| 617 | * |
| 618 | * @{ |
| 619 | */ |
| 620 | |
| 621 | /** |
| 622 | * The frame data may be corrupted, e.g. due to decoding errors. |
| 623 | */ |
| 624 | #define AV_FRAME_FLAG_CORRUPT (1 << 0) |
| 625 | /** |
| 626 | * A flag to mark frames that are keyframes. |
| 627 | */ |
| 628 | #define AV_FRAME_FLAG_KEY (1 << 1) |
| 629 | /** |
| 630 | * A flag to mark the frames which need to be decoded, but shouldn't be output. |
| 631 | */ |
| 632 | #define AV_FRAME_FLAG_DISCARD (1 << 2) |
| 633 | /** |
| 634 | * A flag to mark frames whose content is interlaced. |
| 635 | */ |
| 636 | #define AV_FRAME_FLAG_INTERLACED (1 << 3) |
| 637 | /** |
| 638 | * A flag to mark frames where the top field is displayed first if the content |
| 639 | * is interlaced. |
| 640 | */ |
| 641 | #define AV_FRAME_FLAG_TOP_FIELD_FIRST (1 << 4) |
| 642 | /** |
| 643 | * @} |
| 644 | */ |
| 645 | |
| 646 | /** |
| 647 | * Frame flags, a combination of @ref lavu_frame_flags |
| 648 | */ |
| 649 | int flags; |
| 650 | |
| 651 | /** |
| 652 | * MPEG vs JPEG YUV range. |
| 653 | * - encoding: Set by user |
| 654 | * - decoding: Set by libavcodec |
| 655 | */ |
| 656 | enum AVColorRange color_range; |
| 657 | |
| 658 | enum AVColorPrimaries color_primaries; |
| 659 | |
| 660 | enum AVColorTransferCharacteristic color_trc; |
| 661 | |
| 662 | /** |
| 663 | * YUV colorspace type. |
| 664 | * - encoding: Set by user |
| 665 | * - decoding: Set by libavcodec |
| 666 | */ |
| 667 | enum AVColorSpace colorspace; |
| 668 | |
| 669 | enum AVChromaLocation chroma_location; |
| 670 | |
| 671 | /** |
| 672 | * frame timestamp estimated using various heuristics, in stream time base |
| 673 | * - encoding: unused |
| 674 | * - decoding: set by libavcodec, read by user. |
| 675 | */ |
| 676 | int64_t best_effort_timestamp; |
| 677 | |
| 678 | #if FF_API_FRAME_PKT |
| 679 | /** |
| 680 | * reordered pos from the last AVPacket that has been input into the decoder |
| 681 | * - encoding: unused |
| 682 | * - decoding: Read by user. |
| 683 | * @deprecated use AV_CODEC_FLAG_COPY_OPAQUE to pass through arbitrary user |
| 684 | * data from packets to frames |
| 685 | */ |
| 686 | attribute_deprecated |
| 687 | int64_t pkt_pos; |
| 688 | #endif |
| 689 | |
| 690 | #if FF_API_PKT_DURATION |
| 691 | /** |
| 692 | * duration of the corresponding packet, expressed in |
| 693 | * AVStream->time_base units, 0 if unknown. |
| 694 | * - encoding: unused |
| 695 | * - decoding: Read by user. |
| 696 | * |
| 697 | * @deprecated use duration instead |
| 698 | */ |
| 699 | attribute_deprecated |
| 700 | int64_t pkt_duration; |
| 701 | #endif |
| 702 | |
| 703 | /** |
| 704 | * metadata. |
| 705 | * - encoding: Set by user. |
| 706 | * - decoding: Set by libavcodec. |
| 707 | */ |
| 708 | AVDictionary *metadata; |
| 709 | |
| 710 | /** |
| 711 | * decode error flags of the frame, set to a combination of |
| 712 | * FF_DECODE_ERROR_xxx flags if the decoder produced a frame, but there |
| 713 | * were errors during the decoding. |
| 714 | * - encoding: unused |
| 715 | * - decoding: set by libavcodec, read by user. |
| 716 | */ |
| 717 | int decode_error_flags; |
| 718 | #define FF_DECODE_ERROR_INVALID_BITSTREAM 1 |
| 719 | #define FF_DECODE_ERROR_MISSING_REFERENCE 2 |
| 720 | #define FF_DECODE_ERROR_CONCEALMENT_ACTIVE 4 |
| 721 | #define FF_DECODE_ERROR_DECODE_SLICES 8 |
| 722 | |
| 723 | #if FF_API_OLD_CHANNEL_LAYOUT |
| 724 | /** |
| 725 | * number of audio channels, only used for audio. |
| 726 | * - encoding: unused |
| 727 | * - decoding: Read by user. |
| 728 | * @deprecated use ch_layout instead |
| 729 | */ |
| 730 | attribute_deprecated |
| 731 | int channels; |
| 732 | #endif |
| 733 | |
| 734 | #if FF_API_FRAME_PKT |
| 735 | /** |
| 736 | * size of the corresponding packet containing the compressed |
| 737 | * frame. |
| 738 | * It is set to a negative value if unknown. |
| 739 | * - encoding: unused |
| 740 | * - decoding: set by libavcodec, read by user. |
| 741 | * @deprecated use AV_CODEC_FLAG_COPY_OPAQUE to pass through arbitrary user |
| 742 | * data from packets to frames |
| 743 | */ |
| 744 | attribute_deprecated |
| 745 | int pkt_size; |
| 746 | #endif |
| 747 | |
| 748 | /** |
| 749 | * For hwaccel-format frames, this should be a reference to the |
| 750 | * AVHWFramesContext describing the frame. |
| 751 | */ |
| 752 | AVBufferRef *hw_frames_ctx; |
| 753 | |
| 754 | /** |
| 755 | * Frame owner's private data. |
| 756 | * |
| 757 | * This field may be set by the code that allocates/owns the frame data. |
| 758 | * It is then not touched by any library functions, except: |
| 759 | * - a new reference to the underlying buffer is propagated by |
| 760 | * av_frame_copy_props() (and hence by av_frame_ref()); |
| 761 | * - it is unreferenced in av_frame_unref(); |
| 762 | * - on the caller's explicit request. E.g. libavcodec encoders/decoders |
| 763 | * will propagate a new reference to/from @ref AVPacket "AVPackets" if the |
| 764 | * caller sets @ref AV_CODEC_FLAG_COPY_OPAQUE. |
| 765 | * |
| 766 | * @see opaque the plain pointer analogue |
| 767 | */ |
| 768 | AVBufferRef *opaque_ref; |
| 769 | |
| 770 | /** |
| 771 | * @anchor cropping |
| 772 | * @name Cropping |
| 773 | * Video frames only. The number of pixels to discard from the the |
| 774 | * top/bottom/left/right border of the frame to obtain the sub-rectangle of |
| 775 | * the frame intended for presentation. |
| 776 | * @{ |
| 777 | */ |
| 778 | size_t crop_top; |
| 779 | size_t crop_bottom; |
| 780 | size_t crop_left; |
| 781 | size_t crop_right; |
| 782 | /** |
| 783 | * @} |
| 784 | */ |
| 785 | |
| 786 | /** |
| 787 | * AVBufferRef for internal use by a single libav* library. |
| 788 | * Must not be used to transfer data between libraries. |
| 789 | * Has to be NULL when ownership of the frame leaves the respective library. |
| 790 | * |
| 791 | * Code outside the FFmpeg libs should never check or change the contents of the buffer ref. |
| 792 | * |
| 793 | * FFmpeg calls av_buffer_unref() on it when the frame is unreferenced. |
| 794 | * av_frame_copy_props() calls create a new reference with av_buffer_ref() |
| 795 | * for the target frame's private_ref field. |
| 796 | */ |
| 797 | AVBufferRef *private_ref; |
| 798 | |
| 799 | /** |
| 800 | * Channel layout of the audio data. |
| 801 | */ |
| 802 | AVChannelLayout ch_layout; |
| 803 | |
| 804 | /** |
| 805 | * Duration of the frame, in the same units as pts. 0 if unknown. |
| 806 | */ |
| 807 | int64_t duration; |
| 808 | } AVFrame; |
| 809 | |
| 810 | |
| 811 | /** |
| 812 | * Allocate an AVFrame and set its fields to default values. The resulting |
| 813 | * struct must be freed using av_frame_free(). |
| 814 | * |
| 815 | * @return An AVFrame filled with default values or NULL on failure. |
| 816 | * |
| 817 | * @note this only allocates the AVFrame itself, not the data buffers. Those |
| 818 | * must be allocated through other means, e.g. with av_frame_get_buffer() or |
| 819 | * manually. |
| 820 | */ |
| 821 | AVFrame *av_frame_alloc(void); |
| 822 | |
| 823 | /** |
| 824 | * Free the frame and any dynamically allocated objects in it, |
| 825 | * e.g. extended_data. If the frame is reference counted, it will be |
| 826 | * unreferenced first. |
| 827 | * |
| 828 | * @param frame frame to be freed. The pointer will be set to NULL. |
| 829 | */ |
| 830 | void av_frame_free(AVFrame **frame); |
| 831 | |
| 832 | /** |
| 833 | * Set up a new reference to the data described by the source frame. |
| 834 | * |
| 835 | * Copy frame properties from src to dst and create a new reference for each |
| 836 | * AVBufferRef from src. |
| 837 | * |
| 838 | * If src is not reference counted, new buffers are allocated and the data is |
| 839 | * copied. |
| 840 | * |
| 841 | * @warning: dst MUST have been either unreferenced with av_frame_unref(dst), |
| 842 | * or newly allocated with av_frame_alloc() before calling this |
| 843 | * function, or undefined behavior will occur. |
| 844 | * |
| 845 | * @return 0 on success, a negative AVERROR on error |
| 846 | */ |
| 847 | int av_frame_ref(AVFrame *dst, const AVFrame *src); |
| 848 | |
| 849 | /** |
| 850 | * Ensure the destination frame refers to the same data described by the source |
| 851 | * frame, either by creating a new reference for each AVBufferRef from src if |
| 852 | * they differ from those in dst, by allocating new buffers and copying data if |
| 853 | * src is not reference counted, or by unrefencing it if src is empty. |
| 854 | * |
| 855 | * Frame properties on dst will be replaced by those from src. |
| 856 | * |
| 857 | * @return 0 on success, a negative AVERROR on error. On error, dst is |
| 858 | * unreferenced. |
| 859 | */ |
| 860 | int av_frame_replace(AVFrame *dst, const AVFrame *src); |
| 861 | |
| 862 | /** |
| 863 | * Create a new frame that references the same data as src. |
| 864 | * |
| 865 | * This is a shortcut for av_frame_alloc()+av_frame_ref(). |
| 866 | * |
| 867 | * @return newly created AVFrame on success, NULL on error. |
| 868 | */ |
| 869 | AVFrame *av_frame_clone(const AVFrame *src); |
| 870 | |
| 871 | /** |
| 872 | * Unreference all the buffers referenced by frame and reset the frame fields. |
| 873 | */ |
| 874 | void av_frame_unref(AVFrame *frame); |
| 875 | |
| 876 | /** |
| 877 | * Move everything contained in src to dst and reset src. |
| 878 | * |
| 879 | * @warning: dst is not unreferenced, but directly overwritten without reading |
| 880 | * or deallocating its contents. Call av_frame_unref(dst) manually |
| 881 | * before calling this function to ensure that no memory is leaked. |
| 882 | */ |
| 883 | void av_frame_move_ref(AVFrame *dst, AVFrame *src); |
| 884 | |
| 885 | /** |
| 886 | * Allocate new buffer(s) for audio or video data. |
| 887 | * |
| 888 | * The following fields must be set on frame before calling this function: |
| 889 | * - format (pixel format for video, sample format for audio) |
| 890 | * - width and height for video |
| 891 | * - nb_samples and ch_layout for audio |
| 892 | * |
| 893 | * This function will fill AVFrame.data and AVFrame.buf arrays and, if |
| 894 | * necessary, allocate and fill AVFrame.extended_data and AVFrame.extended_buf. |
| 895 | * For planar formats, one buffer will be allocated for each plane. |
| 896 | * |
| 897 | * @warning: if frame already has been allocated, calling this function will |
| 898 | * leak memory. In addition, undefined behavior can occur in certain |
| 899 | * cases. |
| 900 | * |
| 901 | * @param frame frame in which to store the new buffers. |
| 902 | * @param align Required buffer size alignment. If equal to 0, alignment will be |
| 903 | * chosen automatically for the current CPU. It is highly |
| 904 | * recommended to pass 0 here unless you know what you are doing. |
| 905 | * |
| 906 | * @return 0 on success, a negative AVERROR on error. |
| 907 | */ |
| 908 | int av_frame_get_buffer(AVFrame *frame, int align); |
| 909 | |
| 910 | /** |
| 911 | * Check if the frame data is writable. |
| 912 | * |
| 913 | * @return A positive value if the frame data is writable (which is true if and |
| 914 | * only if each of the underlying buffers has only one reference, namely the one |
| 915 | * stored in this frame). Return 0 otherwise. |
| 916 | * |
| 917 | * If 1 is returned the answer is valid until av_buffer_ref() is called on any |
| 918 | * of the underlying AVBufferRefs (e.g. through av_frame_ref() or directly). |
| 919 | * |
| 920 | * @see av_frame_make_writable(), av_buffer_is_writable() |
| 921 | */ |
| 922 | int av_frame_is_writable(AVFrame *frame); |
| 923 | |
| 924 | /** |
| 925 | * Ensure that the frame data is writable, avoiding data copy if possible. |
| 926 | * |
| 927 | * Do nothing if the frame is writable, allocate new buffers and copy the data |
| 928 | * if it is not. Non-refcounted frames behave as non-writable, i.e. a copy |
| 929 | * is always made. |
| 930 | * |
| 931 | * @return 0 on success, a negative AVERROR on error. |
| 932 | * |
| 933 | * @see av_frame_is_writable(), av_buffer_is_writable(), |
| 934 | * av_buffer_make_writable() |
| 935 | */ |
| 936 | int av_frame_make_writable(AVFrame *frame); |
| 937 | |
| 938 | /** |
| 939 | * Copy the frame data from src to dst. |
| 940 | * |
| 941 | * This function does not allocate anything, dst must be already initialized and |
| 942 | * allocated with the same parameters as src. |
| 943 | * |
| 944 | * This function only copies the frame data (i.e. the contents of the data / |
| 945 | * extended data arrays), not any other properties. |
| 946 | * |
| 947 | * @return >= 0 on success, a negative AVERROR on error. |
| 948 | */ |
| 949 | int av_frame_copy(AVFrame *dst, const AVFrame *src); |
| 950 | |
| 951 | /** |
| 952 | * Copy only "metadata" fields from src to dst. |
| 953 | * |
| 954 | * Metadata for the purpose of this function are those fields that do not affect |
| 955 | * the data layout in the buffers. E.g. pts, sample rate (for audio) or sample |
| 956 | * aspect ratio (for video), but not width/height or channel layout. |
| 957 | * Side data is also copied. |
| 958 | */ |
| 959 | int av_frame_copy_props(AVFrame *dst, const AVFrame *src); |
| 960 | |
| 961 | /** |
| 962 | * Get the buffer reference a given data plane is stored in. |
| 963 | * |
| 964 | * @param frame the frame to get the plane's buffer from |
| 965 | * @param plane index of the data plane of interest in frame->extended_data. |
| 966 | * |
| 967 | * @return the buffer reference that contains the plane or NULL if the input |
| 968 | * frame is not valid. |
| 969 | */ |
| 970 | AVBufferRef *av_frame_get_plane_buffer(const AVFrame *frame, int plane); |
| 971 | |
| 972 | /** |
| 973 | * Add a new side data to a frame. |
| 974 | * |
| 975 | * @param frame a frame to which the side data should be added |
| 976 | * @param type type of the added side data |
| 977 | * @param size size of the side data |
| 978 | * |
| 979 | * @return newly added side data on success, NULL on error |
| 980 | */ |
| 981 | AVFrameSideData *av_frame_new_side_data(AVFrame *frame, |
| 982 | enum AVFrameSideDataType type, |
| 983 | size_t size); |
| 984 | |
| 985 | /** |
| 986 | * Add a new side data to a frame from an existing AVBufferRef |
| 987 | * |
| 988 | * @param frame a frame to which the side data should be added |
| 989 | * @param type the type of the added side data |
| 990 | * @param buf an AVBufferRef to add as side data. The ownership of |
| 991 | * the reference is transferred to the frame. |
| 992 | * |
| 993 | * @return newly added side data on success, NULL on error. On failure |
| 994 | * the frame is unchanged and the AVBufferRef remains owned by |
| 995 | * the caller. |
| 996 | */ |
| 997 | AVFrameSideData *av_frame_new_side_data_from_buf(AVFrame *frame, |
| 998 | enum AVFrameSideDataType type, |
| 999 | AVBufferRef *buf); |
| 1000 | |
| 1001 | /** |
| 1002 | * @return a pointer to the side data of a given type on success, NULL if there |
| 1003 | * is no side data with such type in this frame. |
| 1004 | */ |
| 1005 | AVFrameSideData *av_frame_get_side_data(const AVFrame *frame, |
| 1006 | enum AVFrameSideDataType type); |
| 1007 | |
| 1008 | /** |
| 1009 | * Remove and free all side data instances of the given type. |
| 1010 | */ |
| 1011 | void av_frame_remove_side_data(AVFrame *frame, enum AVFrameSideDataType type); |
| 1012 | |
| 1013 | |
| 1014 | /** |
| 1015 | * Flags for frame cropping. |
| 1016 | */ |
| 1017 | enum { |
| 1018 | /** |
| 1019 | * Apply the maximum possible cropping, even if it requires setting the |
| 1020 | * AVFrame.data[] entries to unaligned pointers. Passing unaligned data |
| 1021 | * to FFmpeg API is generally not allowed, and causes undefined behavior |
| 1022 | * (such as crashes). You can pass unaligned data only to FFmpeg APIs that |
| 1023 | * are explicitly documented to accept it. Use this flag only if you |
| 1024 | * absolutely know what you are doing. |
| 1025 | */ |
| 1026 | AV_FRAME_CROP_UNALIGNED = 1 << 0, |
| 1027 | }; |
| 1028 | |
| 1029 | /** |
| 1030 | * Crop the given video AVFrame according to its crop_left/crop_top/crop_right/ |
| 1031 | * crop_bottom fields. If cropping is successful, the function will adjust the |
| 1032 | * data pointers and the width/height fields, and set the crop fields to 0. |
| 1033 | * |
| 1034 | * In all cases, the cropping boundaries will be rounded to the inherent |
| 1035 | * alignment of the pixel format. In some cases, such as for opaque hwaccel |
| 1036 | * formats, the left/top cropping is ignored. The crop fields are set to 0 even |
| 1037 | * if the cropping was rounded or ignored. |
| 1038 | * |
| 1039 | * @param frame the frame which should be cropped |
| 1040 | * @param flags Some combination of AV_FRAME_CROP_* flags, or 0. |
| 1041 | * |
| 1042 | * @return >= 0 on success, a negative AVERROR on error. If the cropping fields |
| 1043 | * were invalid, AVERROR(ERANGE) is returned, and nothing is changed. |
| 1044 | */ |
| 1045 | int av_frame_apply_cropping(AVFrame *frame, int flags); |
| 1046 | |
| 1047 | /** |
| 1048 | * @return a string identifying the side data type |
| 1049 | */ |
| 1050 | const char *av_frame_side_data_name(enum AVFrameSideDataType type); |
| 1051 | |
| 1052 | /** |
| 1053 | * @} |
| 1054 | */ |
| 1055 | |
| 1056 | #endif /* AVUTIL_FRAME_H */ |
| 1057 | |