[FFmpeg-devel] [PATCH] H.264/AVCHD interlaced fixes

[Ivan Schreter]

Michael Niedermayer wrote:
> On Sat, Jan 31, 2009 at 05:10:05PM +0100, Laurent Aimar wrote:
> [...]
>   
>> Example with a simple GOP structure with standard I/P, you can have
>> Type:               I P P P I
>> recovery_frame_cnt: 0 3 2 1 0
>>     
>
> IMHO more realistic example:
> Type:               P P P P P P P P P ...
> recovery_frame_cnt: 3 4 3 3 4 5 4 4 3 
>
> there never is one with 0 and there never is a I frame
> if there was we wouldnt need the SEI (obviously)
>   
Yes, this is indeed more realistic :-)
I haven't seen a sample with recovery_frame_cnt > 0, though. Do you have 
some?

Anyway, the files I have at hand all have I slices associated with SEI 
recovery point with frame count 0 and no SEI recovery point for all 
other slices. So effectively, this has same semantics as in MPEG-2 
(i.e., I-slice is then sure keyframe, just like I-frame in MPEG-2).

I-slice as such doesn't automatically mean key frame in H.264, though 
(due to potential long-term back-references in following slices). Only 
IDR picture is a sure keyframe (or I picture with SEI recovery point 
with frame cnt 0).

Regards,

Ivan