[theora] NHW Project - very high compression settings

[Raphael Canut]

Hello,

I think I have nearly validated the -l8 "very high compression" setting, I
have started to work on -l9 setting, and I have just tested these ones
against x265 (HEVC), and on rather good quality images, they are very
competitive, that's not a joke!!! I have tested -l8 quality setting on 30
rather good quality images against x265, and on 20-22 images I prefer the
results of the NHW Project because it has more neatness, which I find more
pleasant.

Really you can test it, it is 4-5 days that I have published the -l8
quality setting and I just had 2 visits on my demo page,
http://nhwcodec.blogspot.com/ , that's very few... )-: But really I am not
kidding, you can test it by yourself and make your own opinion.I would be
really very interested in any opinion on the NHW Project at high
compression.

Also, yes that's right, I think we can save 2KB in average per .nhw
compressed file.I know how to save 1Ko on the wavelet DC image compression,
I have explained it on the Theora forum in February 2015, iirc.To give an
update, we can do smarter DC image splitting into different size blocks and
give each block an appropriate coding mode (I have defined 4 modes for now,
but just 1 mode is applied to the whole DC image for now, we can add bypass
and 3 other modes)... I have also other ideas for the wavelet coefficients
and residuals compression...

Really if the Xiph Theora community could test the -l8 "very high
compression" setting, it would be really great!!! Because if we could
validate very high compression, I would release a new version, and the NHW
Project would be nearly ready!!! (then stays to adapt to any size of
image...)

Many thanks!
Cheers,
Raphael
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.xiph.org/pipermail/theora/attachments/20180822/85990680/attachment.html>