Re: [video-codec] NHW: a wavelet image codec

[Raphael Canut <[e-mail]>]

Hi Timothy,

Many thanks for your answer.

2015-08-12 20:23 GMT+02:00 Timothy B. Terriberry <[e-mail]>:

>
> Thanks for the contribution. There are several ways to go from here.
>
> One thing that would be interesting to know would be the performance
> compared to Daala and Thor. I would suggest working with Thomas Daede <
> [e-mail]> to make that happen.
>
> That's a good idea, but keep in mind that the NHW codec is optimized for
neatness (and sharpness), so results for PSNR and SSIM are not good (there
are more "errors").For now, the best way I have found to evaluate the NHW
codec is to visually evaluate it.... It is also now optimized for
mid/normal compression, there are no high compression settings... It's
still experimental!



> Another thing would be to try to integrate pieces of the other submissions
> into your codec or pieces of your codec into the other submissions, to see
> if you can demonstrate improvements. Some examples of how we have been
> doing this for Daala and Thor are here:
>
> https://review.xiph.org/874/
> https://github.com/cisco/thor/pull/8
>
> We can try first with the entropy coding for the wavelet coefficients.I
would be really interested in trying Daala's arithmetic coder.


One potentially interesting area to experiment with might be be the
> prototype screencasting experiments in Daala presented at IETF 93:
> https://tools.ietf.org/html/draft-valin-netvc-l1tw-01
>
> Those are using wavelets, so they might benefit from some of what you are
> doing.
>

I have read the draft, in my codec I am using 5/3 (and 9/7) wavelets.I have
read that you are using tree coding like SPIHT and EZW.I would be
interested in understanding more and trying this technique.

Many thanks.
Cheers,
Raphael