[theora] Wavelet Image Codec - advice

[Raphael Canut]

Hello,
Just contact you quickly again as my last post and questions were not maybe
very precise, and maybe rather vague...
To precise, really hope that my question is relevant on this channel (and
that I am not disturbing you), my approach would be to have a little more
neatness (or sharpness) to the detriment, of course, of precision (at the
same compression ratio).
For example (very quickly), for me Google WebP is a codec that has a very
(very) impressive precision, but on the other hand it maybe tends to
decrease a little neatness (always for me at my very humble level, I am not
an expert).My codec doesn't have this same (impressive) level of precision,
but would have maybe a little more neatness... Would you think this could
be also an "acceptable" approach? Or do you think that precision is the
definitely first "criteria of selection" for an image codec, far above
neatness for example? And that I have to add more precision to my codec,
even if it is to decrease neatness...(My algorithm which is not the same,
would be however coded this way)...
I thank you very much for any advice and comment.Also maybe this thread is
meaningless without source code.I am ok to give the current source code
quickly if it would add interest to my work, but it would be maybe better
to finalize the codec first (and adapt it to any size, clean up the code
and add quality settings,...)? I will also ask for it to Monty Montgomery
as he also helped me to improve my codec these last 3 months (thank you
very much again).If the source code would be required, also do not hesitate
to let me know, but maybe first would you think this work and its approach
could be of interest?
If you would have time, any comment and advice on the codec and demo page
in general, would be really very welcome and helpful.
Many thanks,
Raphael

Le 14 mars 2012 14:11, Raphael Canut <nhwcodec at gmail.com> a écrit :

> Hello,
>
> Is is a great pleasure to post on a Xiph IRC Channel.I contacted you to
> have your advice and comment about wavelets for image/video compression, as
> I have made a wavelet-based image compression codec.
>
> I know that currently the best image/video compression codecs use the DCT
> block-based transform, like JPEG, Ogg Theora and the recent Google WebP,
> for example, but what would you think of wavelet transform?
>
> Here is the link for my wavelet-based image compression codec:
> http://nhwcodec.blogspot.com/
>
> I am still trying to improve my codec, and so any comments and advice are
> very welcome.I had few, but great discussions with Monty Montgomery and he
> also encouraged me to post and present my work on the channel.
>
> Very quickly, the main features would be a fast decoder and maybe a good
> neatness.You can find few results on the demo page.You can also download
> the win32 binaries to test it. -I am planning to give the source code and
> algorithm detail rather quickly, but currently the codec is absolutely not
> finalized, I am changing algorithm every month currently(...), would
> also have to adapt the codec to any size of image, code 30-40 quality
> settings,... the current version is just a very first (but functional) one.-
>
> Do not hesitate to leave any comment and advice on this work, if you would
> have time, would be really so much appreciated, and/or in general what do
> you think of wavelets for image compression?
>
> Many thanks!
> Best regards,
> Raphael
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.xiph.org/pipermail/theora/attachments/20120326/85e02d24/attachment.htm