[Theora] Ffdshow-20042003 Theora support fixed.

illiminable ogg at illiminable.com
Mon Oct 11 04:52:28 PDT 2004

----- Original Message ----- 
From: "Neil Morris" <neil at adm1.ph.man.ac.uk>
To: <theora at xiph.org>
Sent: Monday, October 11, 2004 7:25 PM
Subject: Re: [Theora] Ffdshow-20042003 Theora support fixed.

> I've never delved into ffdshow source so I couldn't tell if this was the
> reason but I'm currently wading through the Ogg/Vorbis/Theora 
> documentation
> and trying to get to grips with "granule pos" and I would guess that IS
> the reason<g>.

The easiest way to understand granule pos... is that it's an arbitrary unit 
of time. In everything but theora... every granule has the same duration... 
so you could say there is a granule rate, such that x granules occur every 
second. (in the audio codecs, this is generally 1 granule = 1 sample)

And like time... it can't go backwards.

In theora things get messy... each granule no longer represents the same 
amount of time, there is no direct way to convert time to granule pos, only 
from granule pos to time. Between keyframes each granule represents 1 video 
frame... but as you hit each key frame, weird magic happens and the granule 
pos jumps up. You'll have to look at the specs for the exact calculations... 
i can't remember it.

> I'm new to A/V programing and Theora/Ogg but I'm interested in trying my
> hand with something to familiarize myself with Xiph formats, so I have
> been toying with the idea attempting to writing a program to convert the
> ffdshow avi theora files to ogg as a learning exercise, and as my 1st
> programing language is Fortran it should make the code very 
> "interesting"<BG>.

It would certainly be an exercise... but i think you'd find a lot of the 
effort taken up trying to do things that are pertty unrelated to the xiph 


More information about the Theora mailing list