[theora] NHW Project - some results

[Raphael Canut]

Hello,

Just a quick message to let you know that I get back to NHW Project
currently.

I downloaded 20 images from the Internet with rather good quality (mainly
faces), I compressed them at high compression -l7 setting with the NHW
Project, and on 19 images out of 20, I visually prefer the results of NHW
compared to x265 (HEVC)!!! -I can make available these 20x3=60 512x512
24bit bitmap images for those who want, just let me know.-

I did not select these 20 images, it was the 20 first rather good quality
images that Google Images gave me (I entered the name of an actress...).

So I am quite satisfied with these results, and I think now that the NHW
Project is (visually) better than x265 (HEVC) for rather good quality
images!!!

So I would like to finish the codec, even if the NHW Project will not find
an application in the Industry. -I think the NHW Project has good
advantages, it is furthermore very fast and royalty-free, but as told me
Dave Johnson in his previous post, the giants of the Industry and of the
Internet are too attached to their respective works, codecs... even if for
me, and it is only my opinion, they are too slow for mobile devices and for
me it seems that we are in a phase to save energy...-

For very high compression, for now I would like to apply first a
pre-processing on the image that would remove the details and then compress
the resulting processed image at -l7 quality setting.Would some of you know
a processing that removes details inside the contours and that totally
preserves the contours, edges? Would be so much helpful!

Then will mainly lack, after this task, to adapt the NHW Project to any
size of image.But this time, as for now I am working on the NHW Project on
my spare time, I would like to find a company that will sponsorize me to
adapt the codec to any image size.Do you think it is realistic, possible?

Again, any comment, help is very welcome!!!

Many thanks!
Cheers,
Raphael
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.xiph.org/pipermail/theora/attachments/20180714/80a90f87/attachment.html>