[x264-devel] Re: bit allocation algorithms among frames

[Loren Merritt]

On Thu, 13 Jul 2006, Tuukka Toivonen wrote:

> Hi, stumbled some time ago into this paper:
> http://foulard.ee.cornell.edu/publications/sermadevi_dcc2005.pdf
> "When is Bit Allocation for Predictive Video Coding Easy?"
>
> On page 7, it describes a "Non-backtracking Coordinate SD Algorithm"
> (SD=steepest descent) which appears to be very simple to implement
> (albeit a little bit slow). If someone is looking some experiments to do
> with x264, this would be a nice project. I'd very much like to see the
> results (especially how QP varies between frames) but don't have time for
> this project.

Interesting idea, but it requires random access in the encoder, which 
makes it O(movie length * # of QPs * GOP size). I'll have to think 
about ways to convert it to a standard multipass framework.
Also, there must be some way to start with an approximation of a good bit 
allocation, rather than decrementing from qp=51. Of course I could just 
pick some overestimate of QP, but I'd like to refine an estimate at the 
same bitrate.

--Loren Merritt

-- 
This is the x264-devel mailing-list
To unsubscribe, go to: http://developers.videolan.org/lists.html