FFmpeg笔记2——2.1结构体分析之AVFrame

AVFrame`结构体一般用于储存原始数据(即非压缩数据，例如对于视频来说YUV,RGB ,对于音频来说是PCM)，此外还包括一些相关的信息，比如说，解码的时候储存了宏块类型表，QP表，运动矢量表等数据，编码的时候也储存了相关的数据，因此在使用FFmpeg进行码流分析的时候，`AVFrame`是一个很重要的结构体 文件路径在`libavutil \frame.h

`AVFrame`结构体源码

typedef struct AVFrame {
#define AV_NUM_DATA_POINTERS 8
    /**
     * pointer to the picture/channel planes.
     * This might be different from the first allocated byte
     *
     * Some decoders access areas outside 0,0 - width,height, please
     * see avcodec_align_dimensions2(). Some filters and swscale can read
     * up to 16 bytes beyond the planes, if these filters are to be used,
     * then 16 extra bytes must be allocated.
     *
     * NOTE: Except for hwaccel formats, pointers not needed by the format
     * MUST be set to NULL.
     */
    uint8_t *data[AV_NUM_DATA_POINTERS];
​
    /**
     * For video, size in bytes of each picture line.
     * For audio, size in bytes of each plane.
     *
     * For audio, only linesize[0] may be set. For planar audio, each channel
     * plane must be the same size.
     *
     * For video the linesizes should be multiples of the CPUs alignment
     * preference, this is 16 or 32 for modern desktop CPUs.
     * Some code requires such alignment other code can be slower without
     * correct alignment, for yet other it makes no difference.
     *
     * @note The linesize may be larger than the size of usable data -- there
     * may be extra padding present for performance reasons.
     */
    int linesize[AV_NUM_DATA_POINTERS];
​
    /**
     * pointers to the data planes/channels.
     *
     * For video, this should simply point to data[].
     *
     * For planar audio, each channel has a separate data pointer, and
     * linesize[0] contains the size of each channel buffer.
     * For packed audio, there is just one data pointer, and linesize[0]
     * contains the total size of the buffer for all channels.
     *
     * Note: Both data and extended_data should always be set in a valid frame,
     * but for planar audio with more channels that can fit in data,
     * extended_data must be used in order to access all channels.
     */
    uint8_t **extended_data;
​
    /**
     * @name Video dimensions
     * Video frames only. The coded dimensions (in pixels) of the video frame,
     * i.e. the size of the rectangle that contains some well-defined values.
     *
     * @note The part of the frame intended for display/presentation is further
     * restricted by the @ref cropping "Cropping rectangle".
     * @{
     */
    int width, height;
    /**
     * @}
     */
​
    /**
     * number of audio samples (per channel) described by this frame
     */
    int nb_samples;
​
    /**
     * format of the frame, -1 if unknown or unset
     * Values correspond to enum AVPixelFormat for video frames,
     * enum AVSampleFormat for audio)
     */
    int format;
​
    /**
     * 1 -> keyframe, 0-> not
     */
    int key_frame;
​
    /**
     * Picture type of the frame.
     */
    enum AVPictureType pict_type;
​
    /**
     * Sample aspect ratio for the video frame, 0/1 if unknown/unspecified.
     */
    AVRational sample_aspect_ratio;
​
    /**
     * Presentation timestamp in time_base units (time when frame should be shown to user).
     */
    int64_t pts;
​
#if FF_API_PKT_PTS
    /**
     * PTS copied from the AVPacket that was decoded to produce this frame.
     * @deprecated use the pts field instead
     */
    attribute_deprecated
    int64_t pkt_pts;
#endif
​
    /**
     * DTS copied from the AVPacket that triggered returning this frame. (if frame threading isn't used)
     * This is also the Presentation time of this AVFrame calculated from
     * only AVPacket.dts values without pts values.
     */
    int64_t pkt_dts;
​
    /**
     * picture number in bitstream order
     */
    int coded_picture_number;
    /**
     * picture number in display order
     */
    int display_picture_number;
​
    /**
     * quality (between 1 (good) and FF_LAMBDA_MAX (bad))
     */
    int quality;
​
    /**
     * for some private data of the user
     */
    void *opaque;
​
#if FF_API_ERROR_FRAME
    /**
     * @deprecated unused
     */
    attribute_deprecated
    uint64_t error[AV_NUM_DATA_POINTERS];
#endif
​
    /**
     * When decoding, this signals how much the picture must be delayed.
     * extra_delay = repeat_pict / (2*fps)
     */
    int repeat_pict;
​
    /**
     * The content of the picture is interlaced.
     */
    int interlaced_frame;
​
    /**
     * If the content is interlaced, is top field displayed first.
     */
    int top_field_first;
​
    /**
     * Tell user application that palette has changed from previous frame.
     */
    int palette_has_changed;
​
    /**
     * reordered opaque 64 bits (generally an integer or a double precision float
     * PTS but can be anything).
     * The user sets AVCodecContext.reordered_opaque to represent the input at
     * that time,
     * the decoder reorders values as needed and sets AVFrame.reordered_opaque
     * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque
     * @deprecated in favor of pkt_pts
     */
    int64_t reordered_opaque;
​
    /**
     * Sample rate of the audio data.
     */
    int sample_rate;
​
    /**
     * Channel layout of the audio data.
     */
    uint64_t channel_layout;
​
    /**
     * AVBuffer references backing the data for this frame. If all elements of
     * this array are NULL, then this frame is not reference counted. This array
     * must be filled contiguously -- if buf[i] is non-NULL then buf[j] must
     * also be non-NULL for all j < i.
     *
     * There may be at most one AVBuffer per data plane, so for video this array
     * always contains all the references. For planar audio with more than
     * AV_NUM_DATA_POINTERS channels, there may be more buffers than can fit in
     * this array. Then the extra AVBufferRef pointers are stored in the
     * extended_buf array.
     */
    AVBufferRef *buf[AV_NUM_DATA_POINTERS];
​
    /**
     * For planar audio which requires more than AV_NUM_DATA_POINTERS
     * AVBufferRef pointers, this array will hold all the references which
     * cannot fit into AVFrame.buf.
     *
     * Note that this is different from AVFrame.extended_data, which always
     * contains all the pointers. This array only contains the extra pointers,
     * which cannot fit into AVFrame.buf.
     *
     * This array is always allocated using av_malloc() by whoever constructs
     * the frame. It is freed in av_frame_unref().
     */
    AVBufferRef **extended_buf;
    /**
     * Number of elements in extended_buf.
     */
    int        nb_extended_buf;
​
    AVFrameSideData **side_data;
    int            nb_side_data;
​
/**
 * @defgroup lavu_frame_flags AV_FRAME_FLAGS
 * @ingroup lavu_frame
 * Flags describing additional frame properties.
 *
 * @{
 */
​
/**
 * The frame data may be corrupted, e.g. due to decoding errors.
 */
#define AV_FRAME_FLAG_CORRUPT       (1 << 0)
/**
 * A flag to mark the frames which need to be decoded, but shouldn't be output.
 */
#define AV_FRAME_FLAG_DISCARD   (1 << 2)
/**
 * @}
 */
​
    /**
     * Frame flags, a combination of @ref lavu_frame_flags
     */
    int flags;
​
    /**
     * MPEG vs JPEG YUV range.
     * - encoding: Set by user
     * - decoding: Set by libavcodec
     */
    enum AVColorRange color_range;
​
    enum AVColorPrimaries color_primaries;
​
    enum AVColorTransferCharacteristic color_trc;
​
    /**
     * YUV colorspace type.
     * - encoding: Set by user
     * - decoding: Set by libavcodec
     */
    enum AVColorSpace colorspace;
​
    enum AVChromaLocation chroma_location;
​
    /**
     * frame timestamp estimated using various heuristics, in stream time base
     * - encoding: unused
     * - decoding: set by libavcodec, read by user.
     */
    int64_t best_effort_timestamp;
​
    /**
     * reordered pos from the last AVPacket that has been input into the decoder
     * - encoding: unused
     * - decoding: Read by user.
     */
    int64_t pkt_pos;
​
    /**
     * duration of the corresponding packet, expressed in
     * AVStream->time_base units, 0 if unknown.
     * - encoding: unused
     * - decoding: Read by user.
     */
    int64_t pkt_duration;
​
    /**
     * metadata.
     * - encoding: Set by user.
     * - decoding: Set by libavcodec.
     */
    AVDictionary *metadata;
​
    /**
     * decode error flags of the frame, set to a combination of
     * FF_DECODE_ERROR_xxx flags if the decoder produced a frame, but there
     * were errors during the decoding.
     * - encoding: unused
     * - decoding: set by libavcodec, read by user.
     */
    int decode_error_flags;
#define FF_DECODE_ERROR_INVALID_BITSTREAM   1
#define FF_DECODE_ERROR_MISSING_REFERENCE   2
​
    /**
     * number of audio channels, only used for audio.
     * - encoding: unused
     * - decoding: Read by user.
     */
    int channels;
​
    /**
     * size of the corresponding packet containing the compressed
     * frame.
     * It is set to a negative value if unknown.
     * - encoding: unused
     * - decoding: set by libavcodec, read by user.
     */
    int pkt_size;
​
#if FF_API_FRAME_QP
    /**
     * QP table
     */
    attribute_deprecated
    int8_t *qscale_table;
    /**
     * QP store stride
     */
    attribute_deprecated
    int qstride;
​
    attribute_deprecated
    int qscale_type;
​
    AVBufferRef *qp_table_buf;
#endif
    /**
     * For hwaccel-format frames, this should be a reference to the
     * AVHWFramesContext describing the frame.
     */
    AVBufferRef *hw_frames_ctx;
​
    /**
     * AVBufferRef for free use by the API user. FFmpeg will never check the
     * contents of the buffer ref. FFmpeg calls av_buffer_unref() on it when
     * the frame is unreferenced. av_frame_copy_props() calls create a new
     * reference with av_buffer_ref() for the target frame's opaque_ref field.
     *
     * This is unrelated to the opaque field, although it serves a similar
     * purpose.
     */
    AVBufferRef *opaque_ref;
​
    /**
     * @anchor cropping
     * @name Cropping
     * Video frames only. The number of pixels to discard from the the
     * top/bottom/left/right border of the frame to obtain the sub-rectangle of
     * the frame intended for presentation.
     * @{
     */
    size_t crop_top;
    size_t crop_bottom;
    size_t crop_left;
    size_t crop_right;
    /**
     * @}
     */
}AVFrame;
如下选几个主要变量做解释(这里主要考虑解码的情况)

uint8_t *data[AV_NUM_DATA_POINTERS]：解码后原始数据（对视频来说是YUV，RGB，对音频来说是PCM）
​
int linesize[AV_NUM_DATA_POINTERS]：data中“一行”数据的大小。注意：未必等于图像的宽，一般大于图像的宽。
​
int width, height：视频帧宽和高（1920x1080,1280x720...）
​
int nb_samples：音频的一个AVFrame中可能包含多个音频帧，在此标记包含了几个
​
int format：解码后原始数据类型（YUV420，YUV422，RGB24...）
​
int key_frame：是否是关键帧
​
enum AVPictureType pict_type：帧类型（I,B,P...）
​
AVRational sample_aspect_ratio：宽高比（16:9，4:3...）
​
int64_t pts：显示时间戳
​
int coded_picture_number：编码帧序号
​
int display_picture_number：显示帧序号
​
int8_t *qscale_table：QP表
​
uint8_t *mbskip_table：跳过宏块表
​
int16_t (*motion_val[2])[2]：运动矢量表
​
uint32_t *mb_type：宏块类型表
​
short *dct_coeff：DCT系数，这个没有提取过
​
int8_t *ref_index[2]：运动估计参考帧列表（貌似H.264这种比较新的标准才会涉及到多参考帧）
​
int interlaced_frame：是否是隔行扫描
​
uint8_t motion_subsample_log2：一个宏块中的运动矢量采样个数，取log的

【学习地址】：FFmpeg/WebRTC/RTMP/NDK/Android音视频流媒体高级开发

【文章福利】：免费领取更多音视频学习资料包、大厂面试题、技术视频和学习路线图，资料包括（C/C++，Linux，FFmpeg webRTC rtmp hls rtsp ffplay srs 等等）有需要的可以点击1079654574加群领取哦~

data[] data是个指针数组，每个元素都是uint8_t *的指针 对于packet格式的数据(例如RGB24),则会存到data[0]里面 对于planar格式的数据(例如YUV420P),则会分成data[0]，data[1]，data2

pict_type 帧类型（I,B,P...）

enum AVPictureType {
    AV_PICTURE_TYPE_NONE = 0, ///< Undefined
    AV_PICTURE_TYPE_I,     ///< Intra
    AV_PICTURE_TYPE_P,     ///< Predicted
    AV_PICTURE_TYPE_B,     ///< Bi-dir predicted
    AV_PICTURE_TYPE_S,     ///< S(GMC)-VOP MPEG-4
    AV_PICTURE_TYPE_SI,    ///< Switching Intra
    AV_PICTURE_TYPE_SP,    ///< Switching Predicted
    AV_PICTURE_TYPE_BI,    ///< BI type
};

原文链接：FFmpeg笔记2——2.1结构体分析之AVFrame - 简书