2 * This file is part of FFmpeg.
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.
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.
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
22 * reference-counted frame API
25 #ifndef AVUTIL_FRAME_H
26 #define AVUTIL_FRAME_H
34 #include "samplefmt.h"
40 * @defgroup lavu_frame AVFrame
44 * AVFrame is an abstraction for reference-counted raw multimedia data.
47 enum AVFrameSideDataType {
49 * The data is the AVPanScan struct defined in libavcodec.
51 AV_FRAME_DATA_PANSCAN,
53 * ATSC A53 Part 4 Closed Captions.
54 * A53 CC bitstream is stored as uint8_t in AVFrameSideData.data.
55 * The number of bytes of CC data is AVFrameSideData.size.
59 * Stereoscopic 3d metadata.
60 * The data is the AVStereo3D struct defined in libavutil/stereo3d.h.
62 AV_FRAME_DATA_STEREO3D,
64 * The data is the AVMatrixEncoding enum defined in libavutil/channel_layout.h.
66 AV_FRAME_DATA_MATRIXENCODING,
68 * Metadata relevant to a downmix procedure.
69 * The data is the AVDownmixInfo struct defined in libavutil/downmix_info.h.
71 AV_FRAME_DATA_DOWNMIX_INFO,
73 * ReplayGain information in the form of the AVReplayGain struct.
75 AV_FRAME_DATA_REPLAYGAIN,
77 * This side data contains a 3x3 transformation matrix describing an affine
78 * transformation that needs to be applied to the frame for correct
81 * See libavutil/display.h for a detailed description of the data.
83 AV_FRAME_DATA_DISPLAYMATRIX,
85 * Active Format Description data consisting of a single byte as specified
86 * in ETSI TS 101 154 using AVActiveFormatDescription enum.
90 * Motion vectors exported by some codecs (on demand through the export_mvs
91 * flag set in the libavcodec AVCodecContext flags2 option).
92 * The data is the AVMotionVector struct defined in
93 * libavutil/motion_vector.h.
95 AV_FRAME_DATA_MOTION_VECTORS,
97 * Recommmends skipping the specified number of samples. This is exported
98 * only if the "skip_manual" AVOption is set in libavcodec.
99 * This has the same format as AV_PKT_DATA_SKIP_SAMPLES.
101 * u32le number of samples to skip from start of this packet
102 * u32le number of samples to skip from end of this packet
103 * u8 reason for start skip
104 * u8 reason for end skip (0=padding silence, 1=convergence)
107 AV_FRAME_DATA_SKIP_SAMPLES,
109 * This side data must be associated with an audio frame and corresponds to
110 * enum AVAudioServiceType defined in avcodec.h.
112 AV_FRAME_DATA_AUDIO_SERVICE_TYPE,
114 * Mastering display metadata associated with a video frame. The payload is
115 * an AVMasteringDisplayMetadata type and contains information about the
116 * mastering display color volume.
118 AV_FRAME_DATA_MASTERING_DISPLAY_METADATA,
120 * The GOP timecode in 25 bit timecode format. Data format is 64-bit integer.
121 * This is set on the first frame of a GOP that has a temporal reference of 0.
123 AV_FRAME_DATA_GOP_TIMECODE
126 enum AVActiveFormatDescription {
131 AV_AFD_4_3_SP_14_9 = 13,
132 AV_AFD_16_9_SP_14_9 = 14,
138 * Structure to hold side data for an AVFrame.
140 * sizeof(AVFrameSideData) is not a part of the public ABI, so new fields may be added
141 * to the end with a minor bump.
143 typedef struct AVFrameSideData {
144 enum AVFrameSideDataType type;
147 AVDictionary *metadata;
152 * This structure describes decoded (raw) audio or video data.
154 * AVFrame must be allocated using av_frame_alloc(). Note that this only
155 * allocates the AVFrame itself, the buffers for the data must be managed
156 * through other means (see below).
157 * AVFrame must be freed with av_frame_free().
159 * AVFrame is typically allocated once and then reused multiple times to hold
160 * different data (e.g. a single AVFrame to hold frames received from a
161 * decoder). In such a case, av_frame_unref() will free any references held by
162 * the frame and reset it to its original clean state before it
165 * The data described by an AVFrame is usually reference counted through the
166 * AVBuffer API. The underlying buffer references are stored in AVFrame.buf /
167 * AVFrame.extended_buf. An AVFrame is considered to be reference counted if at
168 * least one reference is set, i.e. if AVFrame.buf[0] != NULL. In such a case,
169 * every single data plane must be contained in one of the buffers in
170 * AVFrame.buf or AVFrame.extended_buf.
171 * There may be a single buffer for all the data, or one separate buffer for
172 * each plane, or anything in between.
174 * sizeof(AVFrame) is not a part of the public ABI, so new fields may be added
175 * to the end with a minor bump.
176 * Similarly fields that are marked as to be only accessed by
177 * av_opt_ptr() can be reordered. This allows 2 forks to add fields
178 * without breaking compatibility with each other.
180 typedef struct AVFrame {
181 #define AV_NUM_DATA_POINTERS 8
183 * pointer to the picture/channel planes.
184 * This might be different from the first allocated byte
186 * Some decoders access areas outside 0,0 - width,height, please
187 * see avcodec_align_dimensions2(). Some filters and swscale can read
188 * up to 16 bytes beyond the planes, if these filters are to be used,
189 * then 16 extra bytes must be allocated.
191 * NOTE: Except for hwaccel formats, pointers not needed by the format
192 * MUST be set to NULL.
194 uint8_t *data[AV_NUM_DATA_POINTERS];
197 * For video, size in bytes of each picture line.
198 * For audio, size in bytes of each plane.
200 * For audio, only linesize[0] may be set. For planar audio, each channel
201 * plane must be the same size.
203 * For video the linesizes should be multiples of the CPUs alignment
204 * preference, this is 16 or 32 for modern desktop CPUs.
205 * Some code requires such alignment other code can be slower without
206 * correct alignment, for yet other it makes no difference.
208 * @note The linesize may be larger than the size of usable data -- there
209 * may be extra padding present for performance reasons.
211 int linesize[AV_NUM_DATA_POINTERS];
214 * pointers to the data planes/channels.
216 * For video, this should simply point to data[].
218 * For planar audio, each channel has a separate data pointer, and
219 * linesize[0] contains the size of each channel buffer.
220 * For packed audio, there is just one data pointer, and linesize[0]
221 * contains the total size of the buffer for all channels.
223 * Note: Both data and extended_data should always be set in a valid frame,
224 * but for planar audio with more channels that can fit in data,
225 * extended_data must be used in order to access all channels.
227 uint8_t **extended_data;
230 * width and height of the video frame
235 * number of audio samples (per channel) described by this frame
240 * format of the frame, -1 if unknown or unset
241 * Values correspond to enum AVPixelFormat for video frames,
242 * enum AVSampleFormat for audio)
247 * 1 -> keyframe, 0-> not
252 * Picture type of the frame.
254 enum AVPictureType pict_type;
257 * Sample aspect ratio for the video frame, 0/1 if unknown/unspecified.
259 AVRational sample_aspect_ratio;
262 * Presentation timestamp in time_base units (time when frame should be shown to user).
267 * PTS copied from the AVPacket that was decoded to produce this frame.
272 * DTS copied from the AVPacket that triggered returning this frame. (if frame threading isn't used)
273 * This is also the Presentation time of this AVFrame calculated from
274 * only AVPacket.dts values without pts values.
279 * picture number in bitstream order
281 int coded_picture_number;
283 * picture number in display order
285 int display_picture_number;
288 * quality (between 1 (good) and FF_LAMBDA_MAX (bad))
293 * for some private data of the user
297 #if FF_API_ERROR_FRAME
302 uint64_t error[AV_NUM_DATA_POINTERS];
306 * When decoding, this signals how much the picture must be delayed.
307 * extra_delay = repeat_pict / (2*fps)
312 * The content of the picture is interlaced.
314 int interlaced_frame;
317 * If the content is interlaced, is top field displayed first.
322 * Tell user application that palette has changed from previous frame.
324 int palette_has_changed;
327 * reordered opaque 64bit (generally an integer or a double precision float
328 * PTS but can be anything).
329 * The user sets AVCodecContext.reordered_opaque to represent the input at
331 * the decoder reorders values as needed and sets AVFrame.reordered_opaque
332 * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque
333 * @deprecated in favor of pkt_pts
335 int64_t reordered_opaque;
338 * Sample rate of the audio data.
343 * Channel layout of the audio data.
345 uint64_t channel_layout;
348 * AVBuffer references backing the data for this frame. If all elements of
349 * this array are NULL, then this frame is not reference counted. This array
350 * must be filled contiguously -- if buf[i] is non-NULL then buf[j] must
351 * also be non-NULL for all j < i.
353 * There may be at most one AVBuffer per data plane, so for video this array
354 * always contains all the references. For planar audio with more than
355 * AV_NUM_DATA_POINTERS channels, there may be more buffers than can fit in
356 * this array. Then the extra AVBufferRef pointers are stored in the
357 * extended_buf array.
359 AVBufferRef *buf[AV_NUM_DATA_POINTERS];
362 * For planar audio which requires more than AV_NUM_DATA_POINTERS
363 * AVBufferRef pointers, this array will hold all the references which
364 * cannot fit into AVFrame.buf.
366 * Note that this is different from AVFrame.extended_data, which always
367 * contains all the pointers. This array only contains the extra pointers,
368 * which cannot fit into AVFrame.buf.
370 * This array is always allocated using av_malloc() by whoever constructs
371 * the frame. It is freed in av_frame_unref().
373 AVBufferRef **extended_buf;
375 * Number of elements in extended_buf.
379 AVFrameSideData **side_data;
383 * @defgroup lavu_frame_flags AV_FRAME_FLAGS
384 * Flags describing additional frame properties.
390 * The frame data may be corrupted, e.g. due to decoding errors.
392 #define AV_FRAME_FLAG_CORRUPT (1 << 0)
398 * Frame flags, a combination of @ref lavu_frame_flags
403 * MPEG vs JPEG YUV range.
404 * It must be accessed using av_frame_get_color_range() and
405 * av_frame_set_color_range().
406 * - encoding: Set by user
407 * - decoding: Set by libavcodec
409 enum AVColorRange color_range;
411 enum AVColorPrimaries color_primaries;
413 enum AVColorTransferCharacteristic color_trc;
416 * YUV colorspace type.
417 * It must be accessed using av_frame_get_colorspace() and
418 * av_frame_set_colorspace().
419 * - encoding: Set by user
420 * - decoding: Set by libavcodec
422 enum AVColorSpace colorspace;
424 enum AVChromaLocation chroma_location;
427 * For hwaccel-format frames, this should be a reference to the
428 * AVHWFramesContext describing the frame.
430 AVBufferRef *hw_frames_ctx;
433 * frame timestamp estimated using various heuristics, in stream time base
434 * Code outside libavutil should access this field using:
435 * av_frame_get_best_effort_timestamp(frame)
437 * - decoding: set by libavcodec, read by user.
439 int64_t best_effort_timestamp;
442 * reordered pos from the last AVPacket that has been input into the decoder
443 * Code outside libavutil should access this field using:
444 * av_frame_get_pkt_pos(frame)
446 * - decoding: Read by user.
451 * duration of the corresponding packet, expressed in
452 * AVStream->time_base units, 0 if unknown.
453 * Code outside libavutil should access this field using:
454 * av_frame_get_pkt_duration(frame)
456 * - decoding: Read by user.
458 int64_t pkt_duration;
462 * Code outside libavutil should access this field using:
463 * av_frame_get_metadata(frame)
464 * - encoding: Set by user.
465 * - decoding: Set by libavcodec.
467 AVDictionary *metadata;
470 * decode error flags of the frame, set to a combination of
471 * FF_DECODE_ERROR_xxx flags if the decoder produced a frame, but there
472 * were errors during the decoding.
473 * Code outside libavutil should access this field using:
474 * av_frame_get_decode_error_flags(frame)
476 * - decoding: set by libavcodec, read by user.
478 int decode_error_flags;
479 #define FF_DECODE_ERROR_INVALID_BITSTREAM 1
480 #define FF_DECODE_ERROR_MISSING_REFERENCE 2
483 * number of audio channels, only used for audio.
484 * Code outside libavutil should access this field using:
485 * av_frame_get_channels(frame)
487 * - decoding: Read by user.
492 * size of the corresponding packet containing the compressed
493 * frame. It must be accessed using av_frame_get_pkt_size() and
494 * av_frame_set_pkt_size().
495 * It is set to a negative value if unknown.
497 * - decoding: set by libavcodec, read by user.
504 * Not to be accessed directly from outside libavutil
507 int8_t *qscale_table;
510 * Not to be accessed directly from outside libavutil
519 * Not to be accessed directly from outside libavutil
521 AVBufferRef *qp_table_buf;
526 * Accessors for some AVFrame fields.
527 * The position of these field in the structure is not part of the ABI,
528 * they should not be accessed directly outside libavutil.
530 int64_t av_frame_get_best_effort_timestamp(const AVFrame *frame);
531 void av_frame_set_best_effort_timestamp(AVFrame *frame, int64_t val);
532 int64_t av_frame_get_pkt_duration (const AVFrame *frame);
533 void av_frame_set_pkt_duration (AVFrame *frame, int64_t val);
534 int64_t av_frame_get_pkt_pos (const AVFrame *frame);
535 void av_frame_set_pkt_pos (AVFrame *frame, int64_t val);
536 int64_t av_frame_get_channel_layout (const AVFrame *frame);
537 void av_frame_set_channel_layout (AVFrame *frame, int64_t val);
538 int av_frame_get_channels (const AVFrame *frame);
539 void av_frame_set_channels (AVFrame *frame, int val);
540 int av_frame_get_sample_rate (const AVFrame *frame);
541 void av_frame_set_sample_rate (AVFrame *frame, int val);
542 AVDictionary *av_frame_get_metadata (const AVFrame *frame);
543 void av_frame_set_metadata (AVFrame *frame, AVDictionary *val);
544 int av_frame_get_decode_error_flags (const AVFrame *frame);
545 void av_frame_set_decode_error_flags (AVFrame *frame, int val);
546 int av_frame_get_pkt_size(const AVFrame *frame);
547 void av_frame_set_pkt_size(AVFrame *frame, int val);
548 AVDictionary **avpriv_frame_get_metadatap(AVFrame *frame);
550 int8_t *av_frame_get_qp_table(AVFrame *f, int *stride, int *type);
551 int av_frame_set_qp_table(AVFrame *f, AVBufferRef *buf, int stride, int type);
553 enum AVColorSpace av_frame_get_colorspace(const AVFrame *frame);
554 void av_frame_set_colorspace(AVFrame *frame, enum AVColorSpace val);
555 enum AVColorRange av_frame_get_color_range(const AVFrame *frame);
556 void av_frame_set_color_range(AVFrame *frame, enum AVColorRange val);
559 * Get the name of a colorspace.
560 * @return a static string identifying the colorspace; can be NULL.
562 const char *av_get_colorspace_name(enum AVColorSpace val);
565 * Allocate an AVFrame and set its fields to default values. The resulting
566 * struct must be freed using av_frame_free().
568 * @return An AVFrame filled with default values or NULL on failure.
570 * @note this only allocates the AVFrame itself, not the data buffers. Those
571 * must be allocated through other means, e.g. with av_frame_get_buffer() or
574 AVFrame *av_frame_alloc(void);
577 * Free the frame and any dynamically allocated objects in it,
578 * e.g. extended_data. If the frame is reference counted, it will be
579 * unreferenced first.
581 * @param frame frame to be freed. The pointer will be set to NULL.
583 void av_frame_free(AVFrame **frame);
586 * Set up a new reference to the data described by the source frame.
588 * Copy frame properties from src to dst and create a new reference for each
589 * AVBufferRef from src.
591 * If src is not reference counted, new buffers are allocated and the data is
594 * @warning: dst MUST have been either unreferenced with av_frame_unref(dst),
595 * or newly allocated with av_frame_alloc() before calling this
596 * function, or undefined behavior will occur.
598 * @return 0 on success, a negative AVERROR on error
600 int av_frame_ref(AVFrame *dst, const AVFrame *src);
603 * Create a new frame that references the same data as src.
605 * This is a shortcut for av_frame_alloc()+av_frame_ref().
607 * @return newly created AVFrame on success, NULL on error.
609 AVFrame *av_frame_clone(const AVFrame *src);
612 * Unreference all the buffers referenced by frame and reset the frame fields.
614 void av_frame_unref(AVFrame *frame);
617 * Move everything contained in src to dst and reset src.
619 * @warning: dst is not unreferenced, but directly overwritten without reading
620 * or deallocating its contents. Call av_frame_unref(dst) manually
621 * before calling this function to ensure that no memory is leaked.
623 void av_frame_move_ref(AVFrame *dst, AVFrame *src);
626 * Allocate new buffer(s) for audio or video data.
628 * The following fields must be set on frame before calling this function:
629 * - format (pixel format for video, sample format for audio)
630 * - width and height for video
631 * - nb_samples and channel_layout for audio
633 * This function will fill AVFrame.data and AVFrame.buf arrays and, if
634 * necessary, allocate and fill AVFrame.extended_data and AVFrame.extended_buf.
635 * For planar formats, one buffer will be allocated for each plane.
637 * @warning: if frame already has been allocated, calling this function will
638 * leak memory. In addition, undefined behavior can occur in certain
641 * @param frame frame in which to store the new buffers.
642 * @param align required buffer size alignment
644 * @return 0 on success, a negative AVERROR on error.
646 int av_frame_get_buffer(AVFrame *frame, int align);
649 * Check if the frame data is writable.
651 * @return A positive value if the frame data is writable (which is true if and
652 * only if each of the underlying buffers has only one reference, namely the one
653 * stored in this frame). Return 0 otherwise.
655 * If 1 is returned the answer is valid until av_buffer_ref() is called on any
656 * of the underlying AVBufferRefs (e.g. through av_frame_ref() or directly).
658 * @see av_frame_make_writable(), av_buffer_is_writable()
660 int av_frame_is_writable(AVFrame *frame);
663 * Ensure that the frame data is writable, avoiding data copy if possible.
665 * Do nothing if the frame is writable, allocate new buffers and copy the data
668 * @return 0 on success, a negative AVERROR on error.
670 * @see av_frame_is_writable(), av_buffer_is_writable(),
671 * av_buffer_make_writable()
673 int av_frame_make_writable(AVFrame *frame);
676 * Copy the frame data from src to dst.
678 * This function does not allocate anything, dst must be already initialized and
679 * allocated with the same parameters as src.
681 * This function only copies the frame data (i.e. the contents of the data /
682 * extended data arrays), not any other properties.
684 * @return >= 0 on success, a negative AVERROR on error.
686 int av_frame_copy(AVFrame *dst, const AVFrame *src);
689 * Copy only "metadata" fields from src to dst.
691 * Metadata for the purpose of this function are those fields that do not affect
692 * the data layout in the buffers. E.g. pts, sample rate (for audio) or sample
693 * aspect ratio (for video), but not width/height or channel layout.
694 * Side data is also copied.
696 int av_frame_copy_props(AVFrame *dst, const AVFrame *src);
699 * Get the buffer reference a given data plane is stored in.
701 * @param plane index of the data plane of interest in frame->extended_data.
703 * @return the buffer reference that contains the plane or NULL if the input
704 * frame is not valid.
706 AVBufferRef *av_frame_get_plane_buffer(AVFrame *frame, int plane);
709 * Add a new side data to a frame.
711 * @param frame a frame to which the side data should be added
712 * @param type type of the added side data
713 * @param size size of the side data
715 * @return newly added side data on success, NULL on error
717 AVFrameSideData *av_frame_new_side_data(AVFrame *frame,
718 enum AVFrameSideDataType type,
722 * @return a pointer to the side data of a given type on success, NULL if there
723 * is no side data with such type in this frame.
725 AVFrameSideData *av_frame_get_side_data(const AVFrame *frame,
726 enum AVFrameSideDataType type);
729 * If side data of the supplied type exists in the frame, free it and remove it
732 void av_frame_remove_side_data(AVFrame *frame, enum AVFrameSideDataType type);
735 * @return a string identifying the side data type
737 const char *av_frame_side_data_name(enum AVFrameSideDataType type);
743 #endif /* AVUTIL_FRAME_H */