[FFmpeg-trac] #2224(undetermined:new): HLS output using main or high profile has crazy initial timestamps
FFmpeg
trac at avcodec.org
Sat Feb 2 00:10:30 CET 2013
#2224: HLS output using main or high profile has crazy initial timestamps
-------------------------------------+-------------------------------------
Reporter: nealzebub | Owner:
Type: defect | Status: new
Priority: normal | Component:
Version: unspecified | undetermined
Keywords: libx264 | Resolution:
mpegts | Blocked By:
Blocking: | Reproduced by developer: 0
Analyzed by developer: 0 |
-------------------------------------+-------------------------------------
Comment (by nealzebub):
Replying to [comment:1 cehoyos]:
> Is there any specification that clarifies that timestamp wraps are not
allowed in HLS streams?
>
> And could you confirm that there are no problems with the validator if
you disable B-frames?
Spec. I'm not exactly sure what you mean by timestamp wrap, but I think
this is what you mean:
[http://tools.ietf.org/html/draft-pantos-http-live-
streaming-10#section-3.4.11]
It basically says that if the timestamps do not increment between
segments, that tag they mention needs to be stuck in the playlist before
the next segment to warn the player.
I just encoded main with no b-frame(same command with that one addition):
''./ffmpeg.20130129 -report -i ./source_media/bigbuckbunny_1500.mp4
-frames:v 480 -vsync -1 -c:v libx264 -crf 23 -profile:v main -bf 0 -g 48
-sc_threshold 0 -flags +cgop -c:a aac -strict -2 -b:a 112k -map 0:v:0 -map
0:a:0 -f ssegment -segment_time 10 -segment_list hls_main.m3u8
-segment_list_flags +live -segment_format mpegts "hlsAV_main_%04d.ts"''
When I do a comparison of the ffprobe -show_packets output from baseline
and main minus bframes, the only difference are position and size. That
means that the timestamps are identical. I suppose the ticket title
should change to "b-frames" from "main or high profile"!
--
Ticket URL: <https://ffmpeg.org/trac/ffmpeg/ticket/2224#comment:2>
FFmpeg <http://ffmpeg.org>
FFmpeg issue tracker
More information about the FFmpeg-trac
mailing list