[theora-dev] What sort of math i required?

[Holger Waechtler]

Timothy B. Terriberry wrote:
>> The resulting flow fields should be well compressible again with 
>> DCT/wavelet techniques. Has this been tried, does anybody knows 
>> research results about this?
> 
> 
> IIRC, this was not only tried several years ago, but patented by MS (the 
> wavelet approach). This may not be the best approach, however, because 
> these are lossy techniques applied to a lossless coding problem. You 
> can't just throw away bits of your motion vectors like you can your 
> residual data, because even a small change can introduce a large amount 
> of error (consider a checkerboard or striped pattern).

For both the DCT and he wavelet transform lossless integer 
implementations are known - since the motion fields are usually pretty 
smooth and have low gradients they should perform well and the resulting 
coefficients should be tiny.

btw, the wavelet transforms used in the old experimental w3d code are 
all lossless. Quantization is done explicitly in the entropy coder (you 
remember - the weak part of the w3d code;).

Do you have some paper references or the patent number handy?

Holger

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'theora-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.