3 * This file is part of FFmpeg.
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.
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.
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
23 * reference-counted frame API
26 #ifndef AVUTIL_FRAME_H
27 #define AVUTIL_FRAME_H
35 #include "samplefmt.h"
41 * @defgroup lavu_frame AVFrame
45 * AVFrame is an abstraction for reference-counted raw multimedia data.
48 enum AVFrameSideDataType {
50 * The data is the AVPanScan struct defined in libavcodec.
52 AV_FRAME_DATA_PANSCAN,
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.
60 * Stereoscopic 3d metadata.
61 * The data is the AVStereo3D struct defined in libavutil/stereo3d.h.
63 AV_FRAME_DATA_STEREO3D,
65 * The data is the AVMatrixEncoding enum defined in libavutil/channel_layout.h.
67 AV_FRAME_DATA_MATRIXENCODING,
69 * Metadata relevant to a downmix procedure.
70 * The data is the AVDownmixInfo struct defined in libavutil/downmix_info.h.
72 AV_FRAME_DATA_DOWNMIX_INFO,
74 * ReplayGain information in the form of the AVReplayGain struct.
76 AV_FRAME_DATA_REPLAYGAIN,
78 * This side data contains a 3x3 transformation matrix describing an affine
79 * transformation that needs to be applied to the frame for correct
82 * See libavutil/display.h for a detailed description of the data.
84 AV_FRAME_DATA_DISPLAYMATRIX,
86 * Active Format Description data consisting of a single byte as specified
87 * in ETSI TS 101 154 using AVActiveFormatDescription enum.
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.
96 AV_FRAME_DATA_MOTION_VECTORS,
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.
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)
108 AV_FRAME_DATA_SKIP_SAMPLES,
111 * This side data must be associated with an audio frame and corresponds to
112 * enum AVAudioServiceType defined in avcodec.h.
114 AV_FRAME_DATA_AUDIO_SERVICE_TYPE,
117 enum AVActiveFormatDescription {
122 AV_AFD_4_3_SP_14_9 = 13,
123 AV_AFD_16_9_SP_14_9 = 14,
129 * Structure to hold side data for an AVFrame.
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.
134 typedef struct AVFrameSideData {
135 enum AVFrameSideDataType type;
138 AVDictionary *metadata;
143 * This structure describes decoded (raw) audio or video data.
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().
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
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.
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.
171 typedef struct AVFrame {
172 #define AV_NUM_DATA_POINTERS 8
174 * pointer to the picture/channel planes.
175 * This might be different from the first allocated byte
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.
182 uint8_t *data[AV_NUM_DATA_POINTERS];
185 * For video, size in bytes of each picture line.
186 * For audio, size in bytes of each plane.
188 * For audio, only linesize[0] may be set. For planar audio, each channel
189 * plane must be the same size.
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.
196 * @note The linesize may be larger than the size of usable data -- there
197 * may be extra padding present for performance reasons.
199 int linesize[AV_NUM_DATA_POINTERS];
202 * pointers to the data planes/channels.
204 * For video, this should simply point to data[].
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.
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.
215 uint8_t **extended_data;
218 * width and height of the video frame
223 * number of audio samples (per channel) described by this frame
228 * format of the frame, -1 if unknown or unset
229 * Values correspond to enum AVPixelFormat for video frames,
230 * enum AVSampleFormat for audio)
235 * 1 -> keyframe, 0-> not
240 * Picture type of the frame.
242 enum AVPictureType pict_type;
245 * Sample aspect ratio for the video frame, 0/1 if unknown/unspecified.
247 AVRational sample_aspect_ratio;
250 * Presentation timestamp in time_base units (time when frame should be shown to user).
255 * PTS copied from the AVPacket that was decoded to produce this frame.
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.
267 * picture number in bitstream order
269 int coded_picture_number;
271 * picture number in display order
273 int display_picture_number;
276 * quality (between 1 (good) and FF_LAMBDA_MAX (bad))
281 * for some private data of the user
288 uint64_t error[AV_NUM_DATA_POINTERS];
291 * When decoding, this signals how much the picture must be delayed.
292 * extra_delay = repeat_pict / (2*fps)
297 * The content of the picture is interlaced.
299 int interlaced_frame;
302 * If the content is interlaced, is top field displayed first.
307 * Tell user application that palette has changed from previous frame.
309 int palette_has_changed;
312 * reordered opaque 64bit (generally an integer or a double precision float
313 * PTS but can be anything).
314 * The user sets AVCodecContext.reordered_opaque to represent the input at
316 * the decoder reorders values as needed and sets AVFrame.reordered_opaque
317 * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque
318 * @deprecated in favor of pkt_pts
320 int64_t reordered_opaque;
323 * Sample rate of the audio data.
328 * Channel layout of the audio data.
330 uint64_t channel_layout;
333 * AVBuffer references backing the data for this frame. If all elements of
334 * this array are NULL, then this frame is not reference counted. This array
335 * must be filled contiguously -- if buf[i] is non-NULL then buf[j] must
336 * also be non-NULL for all j < i.
338 * There may be at most one AVBuffer per data plane, so for video this array
339 * always contains all the references. For planar audio with more than
340 * AV_NUM_DATA_POINTERS channels, there may be more buffers than can fit in
341 * this array. Then the extra AVBufferRef pointers are stored in the
342 * extended_buf array.
344 AVBufferRef *buf[AV_NUM_DATA_POINTERS];
347 * For planar audio which requires more than AV_NUM_DATA_POINTERS
348 * AVBufferRef pointers, this array will hold all the references which
349 * cannot fit into AVFrame.buf.
351 * Note that this is different from AVFrame.extended_data, which always
352 * contains all the pointers. This array only contains the extra pointers,
353 * which cannot fit into AVFrame.buf.
355 * This array is always allocated using av_malloc() by whoever constructs
356 * the frame. It is freed in av_frame_unref().
358 AVBufferRef **extended_buf;
360 * Number of elements in extended_buf.
364 AVFrameSideData **side_data;
368 * @defgroup lavu_frame_flags AV_FRAME_FLAGS
369 * Flags describing additional frame properties.
375 * The frame data may be corrupted, e.g. due to decoding errors.
377 #define AV_FRAME_FLAG_CORRUPT (1 << 0)
383 * Frame flags, a combination of @ref lavu_frame_flags
388 * MPEG vs JPEG YUV range.
389 * It must be accessed using av_frame_get_color_range() and
390 * av_frame_set_color_range().
391 * - encoding: Set by user
392 * - decoding: Set by libavcodec
394 enum AVColorRange color_range;
396 enum AVColorPrimaries color_primaries;
398 enum AVColorTransferCharacteristic color_trc;
401 * YUV colorspace type.
402 * It must be accessed using av_frame_get_colorspace() and
403 * av_frame_set_colorspace().
404 * - encoding: Set by user
405 * - decoding: Set by libavcodec
407 enum AVColorSpace colorspace;
409 enum AVChromaLocation chroma_location;
412 * frame timestamp estimated using various heuristics, in stream time base
413 * Code outside libavutil should access this field using:
414 * av_frame_get_best_effort_timestamp(frame)
416 * - decoding: set by libavcodec, read by user.
418 int64_t best_effort_timestamp;
421 * reordered pos from the last AVPacket that has been input into the decoder
422 * Code outside libavutil should access this field using:
423 * av_frame_get_pkt_pos(frame)
425 * - decoding: Read by user.
430 * duration of the corresponding packet, expressed in
431 * AVStream->time_base units, 0 if unknown.
432 * Code outside libavutil should access this field using:
433 * av_frame_get_pkt_duration(frame)
435 * - decoding: Read by user.
437 int64_t pkt_duration;
441 * Code outside libavutil should access this field using:
442 * av_frame_get_metadata(frame)
443 * - encoding: Set by user.
444 * - decoding: Set by libavcodec.
446 AVDictionary *metadata;
449 * decode error flags of the frame, set to a combination of
450 * FF_DECODE_ERROR_xxx flags if the decoder produced a frame, but there
451 * were errors during the decoding.
452 * Code outside libavutil should access this field using:
453 * av_frame_get_decode_error_flags(frame)
455 * - decoding: set by libavcodec, read by user.
457 int decode_error_flags;
458 #define FF_DECODE_ERROR_INVALID_BITSTREAM 1
459 #define FF_DECODE_ERROR_MISSING_REFERENCE 2
462 * number of audio channels, only used for audio.
463 * Code outside libavutil should access this field using:
464 * av_frame_get_channels(frame)
466 * - decoding: Read by user.
471 * size of the corresponding packet containing the compressed
472 * frame. It must be accessed using av_frame_get_pkt_size() and
473 * av_frame_set_pkt_size().
474 * It is set to a negative value if unknown.
476 * - decoding: set by libavcodec, read by user.
483 * Not to be accessed directly from outside libavutil
486 int8_t *qscale_table;
489 * Not to be accessed directly from outside libavutil
498 * Not to be accessed directly from outside libavutil
500 AVBufferRef *qp_table_buf;
505 * Accessors for some AVFrame fields.
506 * The position of these field in the structure is not part of the ABI,
507 * they should not be accessed directly outside libavutil.
509 int64_t av_frame_get_best_effort_timestamp(const AVFrame *frame);
510 void av_frame_set_best_effort_timestamp(AVFrame *frame, int64_t val);
511 int64_t av_frame_get_pkt_duration (const AVFrame *frame);
512 void av_frame_set_pkt_duration (AVFrame *frame, int64_t val);
513 int64_t av_frame_get_pkt_pos (const AVFrame *frame);
514 void av_frame_set_pkt_pos (AVFrame *frame, int64_t val);
515 int64_t av_frame_get_channel_layout (const AVFrame *frame);
516 void av_frame_set_channel_layout (AVFrame *frame, int64_t val);
517 int av_frame_get_channels (const AVFrame *frame);
518 void av_frame_set_channels (AVFrame *frame, int val);
519 int av_frame_get_sample_rate (const AVFrame *frame);
520 void av_frame_set_sample_rate (AVFrame *frame, int val);
521 AVDictionary *av_frame_get_metadata (const AVFrame *frame);
522 void av_frame_set_metadata (AVFrame *frame, AVDictionary *val);
523 int av_frame_get_decode_error_flags (const AVFrame *frame);
524 void av_frame_set_decode_error_flags (AVFrame *frame, int val);
525 int av_frame_get_pkt_size(const AVFrame *frame);
526 void av_frame_set_pkt_size(AVFrame *frame, int val);
527 AVDictionary **avpriv_frame_get_metadatap(AVFrame *frame);
529 int8_t *av_frame_get_qp_table(AVFrame *f, int *stride, int *type);
530 int av_frame_set_qp_table(AVFrame *f, AVBufferRef *buf, int stride, int type);
532 enum AVColorSpace av_frame_get_colorspace(const AVFrame *frame);
533 void av_frame_set_colorspace(AVFrame *frame, enum AVColorSpace val);
534 enum AVColorRange av_frame_get_color_range(const AVFrame *frame);
535 void av_frame_set_color_range(AVFrame *frame, enum AVColorRange val);
538 * Get the name of a colorspace.
539 * @return a static string identifying the colorspace; can be NULL.
541 const char *av_get_colorspace_name(enum AVColorSpace val);
544 * Allocate an AVFrame and set its fields to default values. The resulting
545 * struct must be freed using av_frame_free().
547 * @return An AVFrame filled with default values or NULL on failure.
549 * @note this only allocates the AVFrame itself, not the data buffers. Those
550 * must be allocated through other means, e.g. with av_frame_get_buffer() or
553 AVFrame *av_frame_alloc(void);
556 * Free the frame and any dynamically allocated objects in it,
557 * e.g. extended_data. If the frame is reference counted, it will be
558 * unreferenced first.
560 * @param frame frame to be freed. The pointer will be set to NULL.
562 void av_frame_free(AVFrame **frame);
565 * Set up a new reference to the data described by the source frame.
567 * Copy frame properties from src to dst and create a new reference for each
568 * AVBufferRef from src.
570 * If src is not reference counted, new buffers are allocated and the data is
573 * @return 0 on success, a negative AVERROR on error
575 int av_frame_ref(AVFrame *dst, const AVFrame *src);
578 * Create a new frame that references the same data as src.
580 * This is a shortcut for av_frame_alloc()+av_frame_ref().
582 * @return newly created AVFrame on success, NULL on error.
584 AVFrame *av_frame_clone(const AVFrame *src);
587 * Unreference all the buffers referenced by frame and reset the frame fields.
589 void av_frame_unref(AVFrame *frame);
592 * Move everything contained in src to dst and reset src.
594 void av_frame_move_ref(AVFrame *dst, AVFrame *src);
597 * Allocate new buffer(s) for audio or video data.
599 * The following fields must be set on frame before calling this function:
600 * - format (pixel format for video, sample format for audio)
601 * - width and height for video
602 * - nb_samples and channel_layout for audio
604 * This function will fill AVFrame.data and AVFrame.buf arrays and, if
605 * necessary, allocate and fill AVFrame.extended_data and AVFrame.extended_buf.
606 * For planar formats, one buffer will be allocated for each plane.
608 * @param frame frame in which to store the new buffers.
609 * @param align required buffer size alignment
611 * @return 0 on success, a negative AVERROR on error.
613 int av_frame_get_buffer(AVFrame *frame, int align);
616 * Check if the frame data is writable.
618 * @return A positive value if the frame data is writable (which is true if and
619 * only if each of the underlying buffers has only one reference, namely the one
620 * stored in this frame). Return 0 otherwise.
622 * If 1 is returned the answer is valid until av_buffer_ref() is called on any
623 * of the underlying AVBufferRefs (e.g. through av_frame_ref() or directly).
625 * @see av_frame_make_writable(), av_buffer_is_writable()
627 int av_frame_is_writable(AVFrame *frame);
630 * Ensure that the frame data is writable, avoiding data copy if possible.
632 * Do nothing if the frame is writable, allocate new buffers and copy the data
635 * @return 0 on success, a negative AVERROR on error.
637 * @see av_frame_is_writable(), av_buffer_is_writable(),
638 * av_buffer_make_writable()
640 int av_frame_make_writable(AVFrame *frame);
643 * Copy the frame data from src to dst.
645 * This function does not allocate anything, dst must be already initialized and
646 * allocated with the same parameters as src.
648 * This function only copies the frame data (i.e. the contents of the data /
649 * extended data arrays), not any other properties.
651 * @return >= 0 on success, a negative AVERROR on error.
653 int av_frame_copy(AVFrame *dst, const AVFrame *src);
656 * Copy only "metadata" fields from src to dst.
658 * Metadata for the purpose of this function are those fields that do not affect
659 * the data layout in the buffers. E.g. pts, sample rate (for audio) or sample
660 * aspect ratio (for video), but not width/height or channel layout.
661 * Side data is also copied.
663 int av_frame_copy_props(AVFrame *dst, const AVFrame *src);
666 * Get the buffer reference a given data plane is stored in.
668 * @param plane index of the data plane of interest in frame->extended_data.
670 * @return the buffer reference that contains the plane or NULL if the input
671 * frame is not valid.
673 AVBufferRef *av_frame_get_plane_buffer(AVFrame *frame, int plane);
676 * Add a new side data to a frame.
678 * @param frame a frame to which the side data should be added
679 * @param type type of the added side data
680 * @param size size of the side data
682 * @return newly added side data on success, NULL on error
684 AVFrameSideData *av_frame_new_side_data(AVFrame *frame,
685 enum AVFrameSideDataType type,
689 * @return a pointer to the side data of a given type on success, NULL if there
690 * is no side data with such type in this frame.
692 AVFrameSideData *av_frame_get_side_data(const AVFrame *frame,
693 enum AVFrameSideDataType type);
696 * If side data of the supplied type exists in the frame, free it and remove it
699 void av_frame_remove_side_data(AVFrame *frame, enum AVFrameSideDataType type);
702 * @return a string identifying the side data type
704 const char *av_frame_side_data_name(enum AVFrameSideDataType type);
710 #endif /* AVUTIL_FRAME_H */