[Theora-dev] Encoding paramaters...

illiminable ogg
Sat Jul 3 03:40:53 PDT 2004

>----- Original Message -----
>From: "Maik Merten" <maikmerten at gmx.net>
>To: <theora-dev at xiph.org>
>Sent: Saturday, July 03, 2004 1:06 AM
>Subject: Re: [Theora-dev] Encoding paramaters...

>illiminable wrote:
>> The fields i don't understand are quick_p.... what does this do ? What
>> implications does it have on the other parameters.
>IIRC quick_p=0 enables a "high-quality-mode" - much slower, quality is
>slightly better (expensive motion-estimation, I assume).
>quick_p=1 is what VP3 called "Quick Compress".

Thanks for that... though i have another round of questions now...

Do the frame size parameters in theora_info have to match the values in
yuv_buffer ?

ie Is the idea that i can feed in a 800x600 video so the yuv values will be
these and the values in theora_info aer say 320x240, and it will do the
frame shrinking internally ?

Also i read somewhere or other about a multiple of 16 restriction on the
frame sizes... as i actually found out... if you don't follow this... this
line in encode.c

972: FragIndex = QuadMapToIndex1( cpi->pb.BlockMap, SB, MB, B );

returns -1 and when the following values are calculated... some pointer
arithmetic happens and you end up with an invalid address reference.

Anyway... what i'm wondering is... does this multiple of 16 restriction only
apply to the final frame sizes or also to the values in yuv_buffer... hence
you have to pad it out.


More information about the Theora-dev mailing list