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 enum AVActiveFormatDescription {
116 AV_AFD_4_3_SP_14_9 = 13,
117 AV_AFD_16_9_SP_14_9 = 14,
121 typedef struct AVFrameSideData {
122 enum AVFrameSideDataType type;
125 AVDictionary *metadata;
129 * This structure describes decoded (raw) audio or video data.
131 * AVFrame must be allocated using av_frame_alloc(). Note that this only
132 * allocates the AVFrame itself, the buffers for the data must be managed
133 * through other means (see below).
134 * AVFrame must be freed with av_frame_free().
136 * AVFrame is typically allocated once and then reused multiple times to hold
137 * different data (e.g. a single AVFrame to hold frames received from a
138 * decoder). In such a case, av_frame_unref() will free any references held by
139 * the frame and reset it to its original clean state before it
142 * The data described by an AVFrame is usually reference counted through the
143 * AVBuffer API. The underlying buffer references are stored in AVFrame.buf /
144 * AVFrame.extended_buf. An AVFrame is considered to be reference counted if at
145 * least one reference is set, i.e. if AVFrame.buf[0] != NULL. In such a case,
146 * every single data plane must be contained in one of the buffers in
147 * AVFrame.buf or AVFrame.extended_buf.
148 * There may be a single buffer for all the data, or one separate buffer for
149 * each plane, or anything in between.
151 * sizeof(AVFrame) is not a part of the public ABI, so new fields may be added
152 * to the end with a minor bump.
153 * Similarly fields that are marked as to be only accessed by
154 * av_opt_ptr() can be reordered. This allows 2 forks to add fields
155 * without breaking compatibility with each other.
157 typedef struct AVFrame {
158 #define AV_NUM_DATA_POINTERS 8
160 * pointer to the picture/channel planes.
161 * This might be different from the first allocated byte
163 * Some decoders access areas outside 0,0 - width,height, please
164 * see avcodec_align_dimensions2(). Some filters and swscale can read
165 * up to 16 bytes beyond the planes, if these filters are to be used,
166 * then 16 extra bytes must be allocated.
168 uint8_t *data[AV_NUM_DATA_POINTERS];
171 * For video, size in bytes of each picture line.
172 * For audio, size in bytes of each plane.
174 * For audio, only linesize[0] may be set. For planar audio, each channel
175 * plane must be the same size.
177 * For video the linesizes should be multiples of the CPUs alignment
178 * preference, this is 16 or 32 for modern desktop CPUs.
179 * Some code requires such alignment other code can be slower without
180 * correct alignment, for yet other it makes no difference.
182 * @note The linesize may be larger than the size of usable data -- there
183 * may be extra padding present for performance reasons.
185 int linesize[AV_NUM_DATA_POINTERS];
188 * pointers to the data planes/channels.
190 * For video, this should simply point to data[].
192 * For planar audio, each channel has a separate data pointer, and
193 * linesize[0] contains the size of each channel buffer.
194 * For packed audio, there is just one data pointer, and linesize[0]
195 * contains the total size of the buffer for all channels.
197 * Note: Both data and extended_data should always be set in a valid frame,
198 * but for planar audio with more channels that can fit in data,
199 * extended_data must be used in order to access all channels.
201 uint8_t **extended_data;
204 * width and height of the video frame
209 * number of audio samples (per channel) described by this frame
214 * format of the frame, -1 if unknown or unset
215 * Values correspond to enum AVPixelFormat for video frames,
216 * enum AVSampleFormat for audio)
221 * 1 -> keyframe, 0-> not
226 * Picture type of the frame.
228 enum AVPictureType pict_type;
230 #if FF_API_AVFRAME_LAVC
232 uint8_t *base[AV_NUM_DATA_POINTERS];
236 * Sample aspect ratio for the video frame, 0/1 if unknown/unspecified.
238 AVRational sample_aspect_ratio;
241 * Presentation timestamp in time_base units (time when frame should be shown to user).
246 * PTS copied from the AVPacket that was decoded to produce this frame.
251 * DTS copied from the AVPacket that triggered returning this frame. (if frame threading isn't used)
252 * This is also the Presentation time of this AVFrame calculated from
253 * only AVPacket.dts values without pts values.
258 * picture number in bitstream order
260 int coded_picture_number;
262 * picture number in display order
264 int display_picture_number;
267 * quality (between 1 (good) and FF_LAMBDA_MAX (bad))
271 #if FF_API_AVFRAME_LAVC
279 int8_t *qscale_table;
290 * mbskip_table[mb]>=1 if MB didn't change
291 * stride= mb_width = (width+15)>>4
294 uint8_t *mbskip_table;
297 * motion vector table
300 * int mv_sample_log2= 4 - motion_subsample_log2;
301 * int mb_width= (width+15)>>4;
302 * int mv_stride= (mb_width << mv_sample_log2) + 1;
303 * motion_val[direction][x + y*mv_stride][0->mv_x, 1->mv_y];
306 int16_t (*motion_val[2])[2];
309 * macroblock type table
310 * mb_type_base + mb_width + 2
322 * motion reference frame index
323 * the order in which these are stored can depend on the codec.
326 int8_t *ref_index[2];
330 * for some private data of the user
337 uint64_t error[AV_NUM_DATA_POINTERS];
339 #if FF_API_AVFRAME_LAVC
345 * When decoding, this signals how much the picture must be delayed.
346 * extra_delay = repeat_pict / (2*fps)
351 * The content of the picture is interlaced.
353 int interlaced_frame;
356 * If the content is interlaced, is top field displayed first.
361 * Tell user application that palette has changed from previous frame.
363 int palette_has_changed;
365 #if FF_API_AVFRAME_LAVC
373 struct AVPanScan *pan_scan;
377 * reordered opaque 64bit (generally an integer or a double precision float
378 * PTS but can be anything).
379 * The user sets AVCodecContext.reordered_opaque to represent the input at
381 * the decoder reorders values as needed and sets AVFrame.reordered_opaque
382 * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque
383 * @deprecated in favor of pkt_pts
385 int64_t reordered_opaque;
387 #if FF_API_AVFRAME_LAVC
389 * @deprecated this field is unused
391 attribute_deprecated void *hwaccel_picture_private;
394 struct AVCodecContext *owner;
399 * log2 of the size of the block which a single vector in motion_val represents:
400 * (4->16x16, 3->8x8, 2-> 4x4, 1-> 2x2)
402 uint8_t motion_subsample_log2;
406 * Sample rate of the audio data.
411 * Channel layout of the audio data.
413 uint64_t channel_layout;
416 * AVBuffer references backing the data for this frame. If all elements of
417 * this array are NULL, then this frame is not reference counted.
419 * There may be at most one AVBuffer per data plane, so for video this array
420 * always contains all the references. For planar audio with more than
421 * AV_NUM_DATA_POINTERS channels, there may be more buffers than can fit in
422 * this array. Then the extra AVBufferRef pointers are stored in the
423 * extended_buf array.
425 AVBufferRef *buf[AV_NUM_DATA_POINTERS];
428 * For planar audio which requires more than AV_NUM_DATA_POINTERS
429 * AVBufferRef pointers, this array will hold all the references which
430 * cannot fit into AVFrame.buf.
432 * Note that this is different from AVFrame.extended_data, which always
433 * contains all the pointers. This array only contains the extra pointers,
434 * which cannot fit into AVFrame.buf.
436 * This array is always allocated using av_malloc() by whoever constructs
437 * the frame. It is freed in av_frame_unref().
439 AVBufferRef **extended_buf;
441 * Number of elements in extended_buf.
445 AVFrameSideData **side_data;
449 * @defgroup lavu_frame_flags AV_FRAME_FLAGS
450 * Flags describing additional frame properties.
456 * The frame data may be corrupted, e.g. due to decoding errors.
458 #define AV_FRAME_FLAG_CORRUPT (1 << 0)
464 * Frame flags, a combination of @ref lavu_frame_flags
469 * MPEG vs JPEG YUV range.
470 * It must be accessed using av_frame_get_color_range() and
471 * av_frame_set_color_range().
472 * - encoding: Set by user
473 * - decoding: Set by libavcodec
475 enum AVColorRange color_range;
477 enum AVColorPrimaries color_primaries;
479 enum AVColorTransferCharacteristic color_trc;
482 * YUV colorspace type.
483 * It must be accessed using av_frame_get_colorspace() and
484 * av_frame_set_colorspace().
485 * - encoding: Set by user
486 * - decoding: Set by libavcodec
488 enum AVColorSpace colorspace;
490 enum AVChromaLocation chroma_location;
493 * frame timestamp estimated using various heuristics, in stream time base
494 * Code outside libavcodec should access this field using:
495 * av_frame_get_best_effort_timestamp(frame)
497 * - decoding: set by libavcodec, read by user.
499 int64_t best_effort_timestamp;
502 * reordered pos from the last AVPacket that has been input into the decoder
503 * Code outside libavcodec should access this field using:
504 * av_frame_get_pkt_pos(frame)
506 * - decoding: Read by user.
511 * duration of the corresponding packet, expressed in
512 * AVStream->time_base units, 0 if unknown.
513 * Code outside libavcodec should access this field using:
514 * av_frame_get_pkt_duration(frame)
516 * - decoding: Read by user.
518 int64_t pkt_duration;
522 * Code outside libavcodec should access this field using:
523 * av_frame_get_metadata(frame)
524 * - encoding: Set by user.
525 * - decoding: Set by libavcodec.
527 AVDictionary *metadata;
530 * decode error flags of the frame, set to a combination of
531 * FF_DECODE_ERROR_xxx flags if the decoder produced a frame, but there
532 * were errors during the decoding.
533 * Code outside libavcodec should access this field using:
534 * av_frame_get_decode_error_flags(frame)
536 * - decoding: set by libavcodec, read by user.
538 int decode_error_flags;
539 #define FF_DECODE_ERROR_INVALID_BITSTREAM 1
540 #define FF_DECODE_ERROR_MISSING_REFERENCE 2
543 * number of audio channels, only used for audio.
544 * Code outside libavcodec should access this field using:
545 * av_frame_get_channels(frame)
547 * - decoding: Read by user.
552 * size of the corresponding packet containing the compressed
553 * frame. It must be accessed using av_frame_get_pkt_size() and
554 * av_frame_set_pkt_size().
555 * It is set to a negative value if unknown.
557 * - decoding: set by libavcodec, read by user.
562 * Not to be accessed directly from outside libavutil
564 AVBufferRef *qp_table_buf;
568 * Accessors for some AVFrame fields.
569 * The position of these field in the structure is not part of the ABI,
570 * they should not be accessed directly outside libavcodec.
572 int64_t av_frame_get_best_effort_timestamp(const AVFrame *frame);
573 void av_frame_set_best_effort_timestamp(AVFrame *frame, int64_t val);
574 int64_t av_frame_get_pkt_duration (const AVFrame *frame);
575 void av_frame_set_pkt_duration (AVFrame *frame, int64_t val);
576 int64_t av_frame_get_pkt_pos (const AVFrame *frame);
577 void av_frame_set_pkt_pos (AVFrame *frame, int64_t val);
578 int64_t av_frame_get_channel_layout (const AVFrame *frame);
579 void av_frame_set_channel_layout (AVFrame *frame, int64_t val);
580 int av_frame_get_channels (const AVFrame *frame);
581 void av_frame_set_channels (AVFrame *frame, int val);
582 int av_frame_get_sample_rate (const AVFrame *frame);
583 void av_frame_set_sample_rate (AVFrame *frame, int val);
584 AVDictionary *av_frame_get_metadata (const AVFrame *frame);
585 void av_frame_set_metadata (AVFrame *frame, AVDictionary *val);
586 int av_frame_get_decode_error_flags (const AVFrame *frame);
587 void av_frame_set_decode_error_flags (AVFrame *frame, int val);
588 int av_frame_get_pkt_size(const AVFrame *frame);
589 void av_frame_set_pkt_size(AVFrame *frame, int val);
590 AVDictionary **avpriv_frame_get_metadatap(AVFrame *frame);
591 int8_t *av_frame_get_qp_table(AVFrame *f, int *stride, int *type);
592 int av_frame_set_qp_table(AVFrame *f, AVBufferRef *buf, int stride, int type);
593 enum AVColorSpace av_frame_get_colorspace(const AVFrame *frame);
594 void av_frame_set_colorspace(AVFrame *frame, enum AVColorSpace val);
595 enum AVColorRange av_frame_get_color_range(const AVFrame *frame);
596 void av_frame_set_color_range(AVFrame *frame, enum AVColorRange val);
599 * Get the name of a colorspace.
600 * @return a static string identifying the colorspace; can be NULL.
602 const char *av_get_colorspace_name(enum AVColorSpace val);
605 * Allocate an AVFrame and set its fields to default values. The resulting
606 * struct must be freed using av_frame_free().
608 * @return An AVFrame filled with default values or NULL on failure.
610 * @note this only allocates the AVFrame itself, not the data buffers. Those
611 * must be allocated through other means, e.g. with av_frame_get_buffer() or
614 AVFrame *av_frame_alloc(void);
617 * Free the frame and any dynamically allocated objects in it,
618 * e.g. extended_data. If the frame is reference counted, it will be
619 * unreferenced first.
621 * @param frame frame to be freed. The pointer will be set to NULL.
623 void av_frame_free(AVFrame **frame);
626 * Set up a new reference to the data described by the source frame.
628 * Copy frame properties from src to dst and create a new reference for each
629 * AVBufferRef from src.
631 * If src is not reference counted, new buffers are allocated and the data is
634 * @return 0 on success, a negative AVERROR on error
636 int av_frame_ref(AVFrame *dst, const AVFrame *src);
639 * Create a new frame that references the same data as src.
641 * This is a shortcut for av_frame_alloc()+av_frame_ref().
643 * @return newly created AVFrame on success, NULL on error.
645 AVFrame *av_frame_clone(const AVFrame *src);
648 * Unreference all the buffers referenced by frame and reset the frame fields.
650 void av_frame_unref(AVFrame *frame);
653 * Move everythnig contained in src to dst and reset src.
655 void av_frame_move_ref(AVFrame *dst, AVFrame *src);
658 * Allocate new buffer(s) for audio or video data.
660 * The following fields must be set on frame before calling this function:
661 * - format (pixel format for video, sample format for audio)
662 * - width and height for video
663 * - nb_samples and channel_layout for audio
665 * This function will fill AVFrame.data and AVFrame.buf arrays and, if
666 * necessary, allocate and fill AVFrame.extended_data and AVFrame.extended_buf.
667 * For planar formats, one buffer will be allocated for each plane.
669 * @param frame frame in which to store the new buffers.
670 * @param align required buffer size alignment
672 * @return 0 on success, a negative AVERROR on error.
674 int av_frame_get_buffer(AVFrame *frame, int align);
677 * Check if the frame data is writable.
679 * @return A positive value if the frame data is writable (which is true if and
680 * only if each of the underlying buffers has only one reference, namely the one
681 * stored in this frame). Return 0 otherwise.
683 * If 1 is returned the answer is valid until av_buffer_ref() is called on any
684 * of the underlying AVBufferRefs (e.g. through av_frame_ref() or directly).
686 * @see av_frame_make_writable(), av_buffer_is_writable()
688 int av_frame_is_writable(AVFrame *frame);
691 * Ensure that the frame data is writable, avoiding data copy if possible.
693 * Do nothing if the frame is writable, allocate new buffers and copy the data
696 * @return 0 on success, a negative AVERROR on error.
698 * @see av_frame_is_writable(), av_buffer_is_writable(),
699 * av_buffer_make_writable()
701 int av_frame_make_writable(AVFrame *frame);
704 * Copy the frame data from src to dst.
706 * This function does not allocate anything, dst must be already initialized and
707 * allocated with the same parameters as src.
709 * This function only copies the frame data (i.e. the contents of the data /
710 * extended data arrays), not any other properties.
712 * @return >= 0 on success, a negative AVERROR on error.
714 int av_frame_copy(AVFrame *dst, const AVFrame *src);
717 * Copy only "metadata" fields from src to dst.
719 * Metadata for the purpose of this function are those fields that do not affect
720 * the data layout in the buffers. E.g. pts, sample rate (for audio) or sample
721 * aspect ratio (for video), but not width/height or channel layout.
722 * Side data is also copied.
724 int av_frame_copy_props(AVFrame *dst, const AVFrame *src);
727 * Get the buffer reference a given data plane is stored in.
729 * @param plane index of the data plane of interest in frame->extended_data.
731 * @return the buffer reference that contains the plane or NULL if the input
732 * frame is not valid.
734 AVBufferRef *av_frame_get_plane_buffer(AVFrame *frame, int plane);
737 * Add a new side data to a frame.
739 * @param frame a frame to which the side data should be added
740 * @param type type of the added side data
741 * @param size size of the side data
743 * @return newly added side data on success, NULL on error
745 AVFrameSideData *av_frame_new_side_data(AVFrame *frame,
746 enum AVFrameSideDataType type,
750 * @return a pointer to the side data of a given type on success, NULL if there
751 * is no side data with such type in this frame.
753 AVFrameSideData *av_frame_get_side_data(const AVFrame *frame,
754 enum AVFrameSideDataType type);
757 * If side data of the supplied type exists in the frame, free it and remove it
760 void av_frame_remove_side_data(AVFrame *frame, enum AVFrameSideDataType type);
763 * @return a string identifying the side data type
765 const char *av_frame_side_data_name(enum AVFrameSideDataType type);
771 #endif /* AVUTIL_FRAME_H */