Hi Luca,

thanks that I now know about old stuff that I can wipe out of my mind :-)

Luca Abeni wrote:

[...]
>> But here I 
>> get either the entry point from the first sequence header when I point 
>> to the beginning of the packet's data member or I get anywhere in the 
>> buffer's data when I look at packet.data[packet.pos] ... there is no 
>> valid stream start code at that position.
>>     
>
> I think pkt.data (and not pkt.data[pkt.pos]) should contain the frame... 
> If you already tried looking at it and it does not contain correct data, 
> then I do not know what to say. Maybe someone else (who knows 
> libavformat's internals better) will have more information about this.
> Anyway, what's the problem you are seeing with pkt.data?
>
>                               Luca
> _______________________________________________
>   
Indeed pkt.data points to "00 00 01 b3" after the first time call. So I 
thought there is no point to really get the picture header by libav 
interface. In fact on the second call pkt.data points to a valid picture 
header "00 00 01 00". After that I first made a mistake caused by 
reading the 2nd DWORD after the picture start code instead of reading 
the 1st one. I've fixed that and now I really can read the picture flags 
requested ... but only with my own interface (some C++ support classes 
capable of basic MPEG header interpretation, sure not covering all of them).

If it is possible I would like to avoid to grab on MPEG code with two 
different interfaces. So is there a way to generate picture information 
data (e.g. filling an AVFrame) right from the buffer address at the 
picture header by means of using libav calls? - surely again without 
decoding the frame data?

Konran

_______________________________________________
libav-user mailing list
[email protected]
https://lists.mplayerhq.hu/mailman/listinfo/libav-user

Reply via email to