Page 1 of 1

WebP request

Posted: Fri Mar 04, 2011 9:00 pm
by MaxSt
I love that new version supports WebP, but it looks like the code is based on the older WebP decoder, which had some quality issues.

Here's the issue that I reported to WebP developers, and they already fixed it in the new version:

http://code.google.com/p/webp/issues/detail?id=14

Could this fix be included in XnView too? Thanks.

Re: WebP request

Posted: Thu Mar 10, 2011 12:45 pm
by xnview
Could you try XnView 1.98 beta 2

Re: WebP request

Posted: Thu Mar 10, 2011 6:38 pm
by MaxSt
Very good WebP quality in beta 2 !

Thanks!

Re: WebP request

Posted: Mon Mar 14, 2011 1:13 am
by Drahken
Is the lack of chroma subsampling options a characteristic of webp itself, or is it an issue with xnview's implementation?
If it's a characteristic of webp, then webp is totally useless. The quality is as bad as jpeg has the reputation for (due to subsampling), yet the filesize is actually LARGER than the original (but higher quality) JPG.

edit: It seems that I misjudged the original quality setting of the jpg that I started with. After using comparable settings with webp, the result was approximately 1/4 smaller than the comparable quality jpg (with the jpg resaved with the same crappy subsampling levels that webp is stuck with) & has the same subjective quality. Still, the lack of subsampling options limits webp to extremely low quality images.

Re: WebP request

Posted: Mon Mar 14, 2011 6:24 am
by MaxSt
Well, chroma subsampling is often implemented in lower resolution because our eyes have lower chroma resolution.

Re: WebP request

Posted: Mon Mar 14, 2011 6:59 am
by Drahken
That's the theory they spout, and there may even be some level of truth to it, but when it comes to computer images there's a very visible difference.

Look at the bright blue lines in the background of these 2 pics. The pics are from the same original image, saved in the same format, with the same quality level. The only difference is one has high quality subsampling (actually, no subsampling) and the other has maximum subsampling (the same level of subsampling webp seems locked into);

Image Image

Re: WebP request

Posted: Mon Mar 14, 2011 8:18 am
by MaxSt
You wouldn't want to use Jpeg on sharp grid lines either. It's not suitable for that kind of content.

And in other areas I can't see the difference, so it must be working.

Re: WebP request

Posted: Mon Mar 14, 2011 12:28 pm
by Drahken
1) Jpeg works just fine on gird lines, if you have the subsampling turned off. (Both the pics I posted here are in jpeg format, you can clearly see how well the lines turned out on the higher quality one.)

2) When you have a pic like the sample one I posted, no format is ideal (much of the pic is content that would make insanely bloated png files), but jpeg (with subsampling disabled) comes mighty close.

3) You can also see a distinct difference in the red areas if you open both in xnview (or any other viewer) & flip back and forth between them, or if you zoom in on the pics.

Re: WebP request

Posted: Tue Mar 15, 2011 11:14 am
by MaxSt
Well, both formats have areas where they are performed well.

So JPEG area is a little bigger than WebP, but that doesn't mean WebP isn't good or anything.