Hacker News Reader

Comment by monsterix | Hacker News Reader

Wonder what happened to the WebP format from Google? Last when I tried using it, there were no nice image converters that would convert my PNGs into WebP properly.

Any live websites using it?

View on HN

Mozilla is still against it. Makes me kinda angry, actually. They killed MNG/JNG, because it added like 100kB to the installer and because we should just use Flash instead (seriously). It's now a decade later and we still don't have a lossy RGBA format.

If quantized PNG8s aren't good enough, you can only use gigantic PNG32 files, which are over 5 times larger than JNG or WebP files. I've seen websites which were over 5MB thanks to that limitation. It's ridiculous.

I wish they had at least kept JNG. It's a fairly simple JPG/PNG mashup format. If you already use JPG and PNG libraries, supporting JNG won't add much weight.

Actually this is a non-issue. You could always implement comma separated image fallbacks[1]:

background-image: image("wavy.svg", 'wavy.png' , "wavy.gif"); and so on...

It seems that Google lowered emphasis on the project, or is still working on it behind-the-scenes. I'd much prefer to have 'less of images' moving over the web (without cutting back on visuals, you know) and save precious bandwidth.

[1] http://dev.w3.org/csswg/css3-images/#image-fallbacks

And then the fact that Google Chrome has over 30% market share. Makes sense to promote WebP.

That only works for background images. Plus, the UA will send a request for the webp image even if it can't read the format. It needs the mime type to figure that out.

Ah yes. Then probably paste image raw data in the second fallback image? Though I do agree with you that something which benefits the web should be adopted across the board.

We do have lossy RGBA that works in all browsers since IE7 (and degrades well in IE4+):

http://pngquant.org http://tinypng.org

And while it's not as good as WebP, it's 20% of the filesize you get from Photoshop.

http://pngmini.com/vs-webp/

256 colors (well, 256 RGBA tuples) aren't always enough. Just look at the kind of images used by those heavy parallax websites.

Larger photo-based images will always look terrible, no matter which kind of dithering is used.

Parallax websites are actually easy, because images are divided into many sprites/layers, so there's many palettes to work with.

I've got a pretty large collection of PNGs to test quantisation on and 68% of them end up converted in great quality, 89% are within acceptable range. It's about double of what you can get with Fireworks.

And for images that can't absolutely fit in 256-colors there's a hack for lossy true-color PNGs that can reduce size by 20-40%: https://github.com/pornel/mediancut-posterizer

WebP seemed fairly pointless; its advantages over existing formats were slim, it had some disadvantages, and lacked any compelling new functionality (e.g. good HDR support). Basically it was pretty "meh."

Given the huge effort required to get any sort of widespread adoption of a new format—and without widespread adoption there's little point unless you've got some unusual features (which WebP didn't)—I'm not sure there was ever much chance WebP would go anywhere.

I'm generally no fan of Microsoft, but MS's new image format, JPEG XR†, seems vastly superior to WebP in almost every way... and it actually does have compelling new features.

† http://en.wikipedia.org/wiki/JPEG_XR

>JPEG XR†, seems vastly superior to WebP in almost every way

No animations duh.

Is that really a problem? I would much prefer animated gifs never existed.

http://www.reddit.com/r/Cinemagraphs/

Boy, huge loss there... ><

Yes, it would be a big loss.

> Any live websites using it?

Even google isn't using it on thier websites. That says a lot.

There you are:

http://blog.chromium.org/2013/02/using-webp-to-improve-speed...

An official update from Google.

Reply on news.ycombinator.com