[theora] NHW Project - post-processing functions

Raphael Canut nhwcodec at gmail.com
Thu Dec 28 17:34:40 UTC 2017


I am slowly working on the NHW codec.

I have re-tested the post-processing function in the NHW decoder that was
supposed to remove aliasing but that doesn't work at all, and after several
tests I still find this post-processing useful.I find that it gives some
good results, rebalances the image (but it is very slight).I know that this
kind of post-processing must be taken very carefully, but after several
tests, I have decided to validate it.

Else I know that Monty, Odin, Franz urged me to make technology demos for
the NHW Project, I have heard you but I currently lack that graphical
design and web space to make it well.But I'll see what I can do next
year.Any help is very welcome!

For the near future, I would like to create a post-processing function that
removes aliasing on the reconstructed Y comp of the NHW decoder.Starts to
become imperative.Don't know if it is the right approach, but for now I am
interested in trying to design such a post-processing, (I will maybe reuse
the Laplacian kernel) but I will have to detect blurred edges, parts.If you
have such a aliasing detector, please do not hesitate to contact me, and
any contribution from the community is also welcome!

Again, if you care about the NHW Project and you think there are things
that really don't go in the right direction, please, please do not hesitate
to reach out! Would be much appreciated!

Many thanks!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.xiph.org/pipermail/theora/attachments/20171228/eb861b7a/attachment.html>

More information about the theora mailing list