Hello, Just a very quick message to let you know that I have added a higher quality setting (-h2, +10Ko) and I have added now residual coding on the 512x256 wavelet image for the higher settings.Precision is a little better.-I also still apply the post-processing function on the 512x256 wavelet image for now.- Here is the link for the binaries (as I can not link them in my blog for now): https://docs.google.com/file/d/0B9F8eU1TkGfiVDdFZkpWd3I2ZDQ/edit?usp=sharing For the lower quality settings, will have to improve the -l3 setting first, raise the quantization to something like 0.82 for example for -l3 and -l4 settings, and better select/remove the wavelet coefficients. Many thanks, Raphael -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.xiph.org/pipermail/theora/attachments/20130703/8ff1d8fc/attachment.htm
Hi, Sorry, there was a (slight) error with 512x256 residual coding in the decoder.Now that's ok. source code: https://github.com/rcanut/nhwcodec binaries: https://docs.google.com/file/d/0B9F8eU1TkGfiS1ZJRHJ2N1dsbEE/edit?usp=sharing Raphael 2013/7/3 Raphael Canut <nhwcodec at gmail.com>> Hello, > > Just a very quick message to let you know that I have added a higher > quality setting (-h2, +10Ko) and I have added now residual coding on the > 512x256 wavelet image for the higher settings.Precision is a little > better.-I also still apply the post-processing function on the 512x256 > wavelet image for now.- > > Here is the link for the binaries (as I can not link them in my blog for > now): > https://docs.google.com/file/d/0B9F8eU1TkGfiVDdFZkpWd3I2ZDQ/edit?usp=sharing > > For the lower quality settings, will have to improve the -l3 setting > first, raise the quantization to something like 0.82 for example for -l3 > and -l4 settings, and better select/remove the wavelet coefficients. > > Many thanks, > Raphael > > > > >-------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.xiph.org/pipermail/theora/attachments/20130703/a6efc173/attachment.htm
Hello, I have just added a higher quality setting for the NHW codec (-h3, +4/5Ko), with extended residual coding on 512x256 wavelet image and more higher wavelet coefficients.This setting has more precision. I have also tried for the -l3 lower quality setting to raise the quantization from 0.77 to 0.79, and then remove 1Ko with a fast denoising (exp, actually I use data computed during pre-processing function in the encoder, and so this simple denoising is at nearly no extra computational cost).The -l3 setting would have then a little more precision, but don't know if it is better (would still have to better select the wavelet coefficients...).Any comment is welcome. I have updated my demo page: http://nhwcodec.blogspot.com/ . Many thanks, Raphael 2013/7/3 Raphael Canut <nhwcodec at gmail.com>> Hello, > > Just a very quick message to let you know that I have added a higher > quality setting (-h2, +10Ko) and I have added now residual coding on the > 512x256 wavelet image for the higher settings.Precision is a little > better.-I also still apply the post-processing function on the 512x256 > wavelet image for now.- > > Here is the link for the binaries (as I can not link them in my blog for > now): > https://docs.google.com/file/d/0B9F8eU1TkGfiVDdFZkpWd3I2ZDQ/edit?usp=sharing > > For the lower quality settings, will have to improve the -l3 setting > first, raise the quantization to something like 0.82 for example for -l3 > and -l4 settings, and better select/remove the wavelet coefficients. > > Many thanks, > Raphael > > > > >-------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.xiph.org/pipermail/theora/attachments/20130711/b30c7186/attachment.htm