ffmpeg time_base

83 阅读 0 评论 55 点赞

我是靠谱客的博主开放秋天，最近开发中收集的这篇文章主要介绍ffmpeg time_base，觉得挺不错的，现在分享给大家，希望可以做个参考。

概述

ffmpeg存在多个时间基准(time_base)，对应不同的阶段(结构体)，每个time_base具体的值不一样，ffmpeg提供函数在各个time_base中进行切换。搞清楚各个time_base的来源，对于阅读ffmpeg的代码很重要。

一、time_base

1、AVStream(libavformat/avformat.h)

typedef struct AVStream {
/**
* This is the fundamental unit of time (in seconds) in terms
* of which frame timestamps are represented.
*
* decoding: set by libavformat
* encoding: May be set by the caller before avformat_write_header() to
*
provide a hint to the muxer about the desired timebase. In
*
avformat_write_header(), the muxer will overwrite this field
*
with the timebase that will actually be used for the timestamps
*
written into the file (which may or may not be related to the
*
user-provided one, depending on the format).
*/
AVRational time_base;
/**
* Decoding: pts of the first frame of the stream in presentation order, in stream time base.
* Only set this if you are absolutely 100% sure that the value you set
* it to really is the pts of the first frame.
* This may be undefined (AV_NOPTS_VALUE).
* @note The ASF header does NOT contain a correct start_time the ASF
* demuxer must NOT set this.
*/
int64_t start_time;
/**
* Decoding: duration of the stream, in stream time base.
* If a source file does not specify a duration, but does specify
* a bitrate, this value will be estimated from bitrate and file size.
*/
int64_t duration;

从上面的信息可以看到，AVStream->time_base单位为秒。

那AVStream->time_base具体的值是多少呢？下面以mpegts_demuxer为例：

static int mpegts_set_stream_info(AVStream *st, PESContext *pes,
uint32_t stream_type, uint32_t prog_reg_desc)
{
avpriv_set_pts_info(st, 33, 1, 90000);
void avpriv_set_pts_info(AVStream *s, int pts_wrap_bits,
unsigned int pts_num, unsigned int pts_den)
{
AVRational new_tb;
if (av_reduce(&new_tb.num, &new_tb.den, pts_num, pts_den, INT_MAX)) {
if (new_tb.num != pts_num)
av_log(NULL, AV_LOG_DEBUG,
"st:%d removing common factor %d from timebasen",
s->index, pts_num / new_tb.num);
} else
av_log(NULL, AV_LOG_WARNING,
"st:%d has too large timebase, reducingn", s->index);
if (new_tb.num <= 0 || new_tb.den <= 0) {
av_log(NULL, AV_LOG_ERROR,
"Ignoring attempt to set invalid timebase %d/%d for st:%dn",
new_tb.num, new_tb.den,
s->index);
return;
}
s->time_base
= new_tb;
av_codec_set_pkt_timebase(s->codec, new_tb);
s->pts_wrap_bits = pts_wrap_bits;
}

通过avpriv_set_pts_info(st, 33, 1, 90000)函数，设置AVStream->time_base为1/90000。为什么是90000？因为mpeg的pts、dts都是以90kHz来采样的，所以采样间隔为1/90000秒。

2、AVCodecContext

typedef struct AVCodecContext {
/**
* This is the fundamental unit of time (in seconds) in terms
* of which frame timestamps are represented. For fixed-fps content,
* timebase should be 1/framerate and timestamp increments should be
* identically 1.
* - encoding: MUST be set by user.
* - decoding: Set by libavcodec.
*/
AVRational time_base;

从上面的信息可以看到，AVCodecContext->time_base单位同样为秒，不过精度没有AVStream->time_base高，大小为1/framerate。

下面以ffmpeg转码工具为例：

static int transcode_init(void)
{
if (enc_ctx->codec_type == AVMEDIA_TYPE_VIDEO) {
if (ost->filter && !ost->frame_rate.num)
ost->frame_rate = av_buffersink_get_frame_rate(ost->filter->filter);
if (ist && !ost->frame_rate.num)
ost->frame_rate = ist->framerate;
if (ist && !ost->frame_rate.num)
ost->frame_rate = ist->st->r_frame_rate;
if (ist && !ost->frame_rate.num) {
ost->frame_rate = (AVRational){25, 1};
av_log(NULL, AV_LOG_WARNING,
"No information "
"about the input framerate is available. Falling "
"back to a default value of 25fps for output stream #%d:%d. Use the -r option "
"if you want a different framerate.n",
ost->file_index, ost->index);
}
//ost->frame_rate = ist->st->avg_frame_rate.num ? ist->st->avg_frame_rate : (AVRational){25, 1};
if (ost->enc && ost->enc->supported_framerates && !ost->force_fps) {
int idx = av_find_nearest_q_idx(ost->frame_rate, ost->enc->supported_framerates);
ost->frame_rate = ost->enc->supported_framerates[idx];
}
if (enc_ctx->codec_id == AV_CODEC_ID_MPEG4) {
av_reduce(&ost->frame_rate.num, &ost->frame_rate.den,
ost->frame_rate.num, ost->frame_rate.den, 65535);
}
}
switch (enc_ctx->codec_type) {
case AVMEDIA_TYPE_VIDEO:
enc_ctx->time_base = av_inv_q(ost->frame_rate);
if (ost->filter && !(enc_ctx->time_base.num && enc_ctx->time_base.den))
enc_ctx->time_base = ost->filter->filter->inputs[0]->time_base;
if (av_q2d(enc_ctx->time_base) < 0.001 && video_sync_method != VSYNC_PASSTHROUGH && (video_sync_method == VSYNC_CFR || video_sync_method == VSYNC_VSCFR || (video_sync_method == VSYNC_AUTO && !(oc->oformat->flags & AVFMT_VARIABLE_FPS)))){
av_log(oc, AV_LOG_WARNING, "Frame rate very high for a muxer not efficiently supporting it.n"
"Please consider specifying a lower framerate, a different muxer or -vsync 2n");
}

首先获取ost->frame_rate，然后计算enc_ctx->time_base = 1/ost->frame_rate。

总结：
AVStream->time_base比AVCodecContext->time_base精度要高(数值要小)，比如AVStream->time_base为1/90000，而AVCodecContext->time_base为1/30(假设frame_rate为30)；同样的pts和dts，以AVStream->time_base为单位，数值要比以AVCodecContext->time_base为单位要大。

二、pts、dts

那各个结构下，pts和dts使用哪个time_base来表示呢？

1、AVPacket

typedef struct AVPacket {
/**
* Presentation timestamp in AVStream->time_base units; the time at which
* the decompressed packet will be presented to the user.
* Can be AV_NOPTS_VALUE if it is not stored in the file.
* pts MUST be larger or equal to dts as presentation cannot happen before
* decompression, unless one wants to view hex dumps. Some formats misuse
* the terms dts and pts/cts to mean something different. Such timestamps
* must be converted to true pts/dts before they are stored in AVPacket.
*/
int64_t pts;
/**
* Decompression timestamp in AVStream->time_base units; the time at which
* the packet is decompressed.
* Can be AV_NOPTS_VALUE if it is not stored in the file.
*/
int64_t dts;

从上面可以看到，AVPacket下的pts和dts以AVStream->time_base为单位(数值比较大)。这也很容易理解，根据mpeg的协议，压缩后或解压前的数据，pts和dts是90kHz时钟的采样值，时间间隔就是AVStream->time_base。
2、AVFrame

typedef struct AVFrame {
/**
* Presentation timestamp in time_base units (time when frame should be shown to user).
*/
int64_t pts;
/**
* PTS copied from the AVPacket that was decoded to produce this frame.
*/
int64_t pkt_pts;
/**
* 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;

注意：
AVFrame里面的pkt_pts和pkt_dts是拷贝自AVPacket，同样以AVStream->time_base为单位；而pts是为输出(显示)准备的，以AVCodecContex->time_base为单位)。//FIXME
3、InputStream

typedef struct InputStream {
int file_index;
AVStream *st;
AVCodecContext *dec_ctx;
int64_t
start;
/* time when read started */
/* predicted dts of the next packet read for this stream or (when there are
* several frames in a packet) of the next frame in current packet (in AV_TIME_BASE units) */
int64_t
next_dts;
int64_t
dts;
///< dts of the last packet read for this stream (in AV_TIME_BASE units)
int64_t
next_pts;
///< synthetic pts for the next decode frame (in AV_TIME_BASE units)
int64_t
pts;
///< current pts of the decoded frame
(in AV_TIME_BASE units)

从上面可以看到，InputStream下的pts和dts以AV_TIME_BASE为单位(微秒)，至于为什么要转化为微妙，可能是为了避免使用浮点数。
4、OutputStream

typedef struct OutputStream {
int file_index;
/* file index */
int index;
/* stream index in the output file */
int source_index;
/* InputStream index */
AVStream *st;
/* stream in the output file */
int encoding_needed;
/* true if encoding needed for this stream */
int frame_number;
/* input pts and corresponding output pts
for A/V sync */
struct InputStream *sync_ist; /* input stream to sync against */
int64_t sync_opts;
/* output frame counter, could be changed to some true timestamp */ // FIXME look at frame_number
/* pts of the first frame encoded for this stream, used for limiting
* recording time */
int64_t first_pts;
/* dts of the last packet sent to the muxer */
int64_t last_mux_dts;
AVBitStreamFilterContext *bitstream_filters;
AVCodecContext *enc_ctx;
AVCodec *enc;
int64_t max_frames;
AVFrame *filtered_frame;

OutputStream涉及音视频同步，结构和InputStream不同，暂时只作记录，不分析。

三、各个time_base之间转换

ffmpeg提供av_rescale_q函数用于time_base之间转换，av_rescale_q(a,b,c)作用相当于执行a*b/c，通过设置b,c的值，可以很方便的实现time_base之间转换。
例如：
1、InputStream(AV_TIME_BASE)到AVPacket(AVStream->time_base)

static int decode_video(InputStream *ist, AVPacket *pkt, int *got_output)
{
pkt->dts
= av_rescale_q(ist->dts, AV_TIME_BASE_Q, ist->st->time_base);

2、AVPacket(AVStream->time_base)到InputStream(AV_TIME_BASE)

static int process_input_packet(InputStream *ist, const AVPacket *pkt)
{
if (pkt->dts != AV_NOPTS_VALUE) {
ist->next_dts = ist->dts = av_rescale_q(pkt->dts, ist->st->time_base, AV_TIME_BASE_Q);

四、后记：

AVFrame->pts和AVPacket->pts、AVPacket->dts的值，在解码/编码后，会经历短暂的time_base不匹配的情况：

1、解码后

static int decode_video(InputStream *ist, AVPacket *pkt, int *got_output)
{
decoded_frame = ist->decoded_frame;
pkt->dts
= av_rescale_q(ist->dts, AV_TIME_BASE_Q, ist->st->time_base);
update_benchmark(NULL);
ret = avcodec_decode_video2(ist->dec_ctx,
decoded_frame, got_output, pkt);
best_effort_timestamp= av_frame_get_best_effort_timestamp(decoded_frame);
if(best_effort_timestamp != AV_NOPTS_VALUE)
ist->next_pts = ist->pts = av_rescale_q(decoded_frame->pts = best_effort_timestamp, ist->st->time_base, AV_TIME_BASE_Q);

解码后，decoded_frame->pts的值使用AVStream->time_base为单位，后在AVFilter里面转换成以AVCodecContext->time_base为单位。 //FIXME

2、编码后

static void do_video_out(AVFormatContext *s,
OutputStream *ost,
AVFrame *in_picture)
{
ret = avcodec_encode_video2(enc, &pkt, in_picture, &got_packet);
if (got_packet) {
if (debug_ts) {
av_log(NULL, AV_LOG_INFO, "encoder -> type:video "
"pkt_pts:%s pkt_pts_time:%s pkt_dts:%s pkt_dts_time:%sn",
av_ts2str(pkt.pts), av_ts2timestr(pkt.pts, &enc->time_base),
av_ts2str(pkt.dts), av_ts2timestr(pkt.dts, &enc->time_base));
}
if (pkt.pts == AV_NOPTS_VALUE && !(enc->codec->capabilities & CODEC_CAP_DELAY))
pkt.pts = ost->sync_opts;
av_packet_rescale_ts(&pkt, enc->time_base, ost->st->time_base);
if (debug_ts) {
av_log(NULL, AV_LOG_INFO, "encoder -> type:video "
"pkt_pts:%s pkt_pts_time:%s pkt_dts:%s pkt_dts_time:%sn",
av_ts2str(pkt.pts), av_ts2timestr(pkt.pts, &ost->st->time_base),
av_ts2str(pkt.dts), av_ts2timestr(pkt.dts, &ost->st->time_base));
}
frame_size = pkt.size;
write_frame(s, &pkt, ost);
/* if two pass, output log */
if (ost->logfile && enc->stats_out) {
fprintf(ost->logfile, "%s", enc->stats_out);
}
}