[FFmpeg-devel] [PATCH] H.264 I-frames with SEI are also key frames

Jason Garrett-Glaser darkshikari
Fri Mar 6 11:28:39 CET 2009

On Fri, Mar 6, 2009 at 2:26 AM, Jason Garrett-Glaser
<darkshikari at gmail.com> wrote:
> On Fri, Mar 6, 2009 at 2:01 AM, madshi <dear at madshi.net> wrote:
>> Jason Garrett-Glaser schrieb:
>>>> Recommendation: Default, or 10x whatever your framerate is.
>>>> So based on this my new suggestion would be:
>>>> (1) If we're at runtime 11 seconds or later and
>>>> (2) there was no recovery point SEI found yet and
>>>> (3) there was one IDR frame at the start of the stream and
>>>> (4) there was no further IDR frame found yet, ...
>>>> ... in this situation treat all future I-frames as key frames, until
>>>> a recovery point SEI or a new IDR frame is found.
>>> This is potentially reasonable logic for the general case, but your
>>> justification for that logic--of adapting the decoder based on
>>> x264--is a phenomenally stupid idea, IMO.
>>> Dark Shikari
>> Well, it's easy to criticize suggestions.
> I didn't say there was a problem with the suggestion; I said there was
> a problem with the justification of that suggestion.
> The idea is reasonable IMO.
> Dark Shikari

Actually--wait a minute here.  Any frame--period--that can be seeked
to--must do an *effective* reference list refresh.  This can be done
through MMCO or, in the simplest case, having a stream with just 1
reference frame and then having an I-frame.  Wouldn't the proper
solution be to "detect" seekable frames--even those that aren't
signaled as such--and allow seeking to them?

Dark Shikari

More information about the ffmpeg-devel mailing list