It's just another google scam to track users using images. What a low-life company. Google is trying to be one of those giant companies that dictates every aspects of our live, just like those movies, like in Lorax (off the top of my head).

It's an open format. And the 30 percent smaller images is a valid reason for the format.

ChangBot, your payment from M$oft an ¢rapple should be in your account and available as we type...

It's just another google scam to track users using images. What a low-life company. Google is trying to be one of those giant companies that dictates every aspects of our live, just like those movies, like in Lorax (off the top of my head).

Note the irony: citing the movie instead of the book. Who's a giant company - Theodor Geisel, or Universal Pictures?

Unfortunately it won't challenge PNG, because the people who (ab)use PNG do so because it's the default format in many software and they're lazy, or because they heard it's lossless so it must be better for everything everywhere.

Those who actually need PNG won't use WebP because they don't do the same thing. The rest will use PNG regardless because they're ignorant or stupid.

Besides, they did a rather sneaky trick:

Clamp to 8 bits per component: convert input.png -depth 8 output.png

They recompressed the PNG files to assume 8 bits per component and transparency, which makes them 32 bit images, which blows up the filesize if the image in question happens to be optimized with a limited color palette.

I did a little experiment with a simple X-Y plot diagram, restricted the palette to 16 colors and saved it. Resulting file size was 17 kB. The same file upconverted to 8 bits per channel plus transparency became 37 kB.

PNG is actually not very good for full-color pictures. It's a possibility, but PNG shouldn't be used like that over the internet because it results in very large files. Any picture that requires full-color support most likely has the kind of content that is ill suited for PNG's compression algorithm.

Ps. because of that, and because people misuse PNG a lot, a random selection of PNG files on the web will contain a large number of files where the content is e.g. a photograph or a screenshot of a game, which shouldn't even be done with PNG.

With that sort of material, it's easy to prove the supremacy of a format.

For example, this image should not be in PNG format: http://www.person...png8.png

Eikka: Interesting. What format should your test photo be in then? Are you saying that Jpeg is superior? Is there a format that is uniformly better no matter the content (a compressed format, that is)?

What format should your test photo be in then?

Quite possibly in JPEG format. If WebP is better, then that. I haven't seen the independent reviews though.

For a picture that contains few colors, which can do with a less than 256 color palette, PNG is better. For anything else, you probably shouldn't be using PNG. It was a mistake to add full-color support in PNG in the first place because it has such horrible performance (up to 5x larger files vs. JPEG at neglible difference).

My understanding of PNG is that is very similar to GIF but doesn't use LZW compression since it was patented by Unisys/Compuserve.

LZW compresses repeating patterns reasonably well, so graphics containing solid colors and reduced palettes are compressed reasonably well. Photographs have less regularity and hence don't compress well with LZW or similar.

It's just another google scam to track users using images. What a low-life company.

I doubt it. My wild guess is that ... Google wants more compressed images out there on the web because it directly effects it's own bandwidth and storage. Even saving a few percent would be huge, 39% is out of this world.

What I'd like is a variable quality setting in your browser so that you could speed up browsing by trading off image quality unless you specifically want to see images more clearly.

My guess is that Google won't have to do much to win over web developers because, as the article states, smaller file sizes means faster page loads, and anything that can improve the performance of a developer's product is something that the developer will have no reservations about implementing.

What's the technology here? Wavelets?

WebP is losing more details of picture than the JPEG and it even doesn't support 4:2:2 and 4:4:4 chroma subsamplings of JPEG or alpha transparency mask of PNG. Actually it's not so difficult to achieve 40% saving of space with more aggressive compression of JPEG and you'll not recognize the difference from WebP anyway. Google should develop a good encoder first before its promotion - the opposite way will not work.
BTW Why most of comments in PO discussions are started with twaddlers, who have absolutely no idea about subject?

Jpeg is superior in terms of image quality.

Is that quality needed for web page browsing?

No.

Jpeg is superior in terms of image quality. Is that quality needed for web page browsing? No.

Then you can adjust the JPEG quality during compression in such a way, you'll get picture smaller by 40% - and you're not required to implement another format. Google's problem solved.

WebP is losing more details of picture than the JPEG and it even doesn't support 4:2:2 and 4:4:4 chroma subsamplings of JPEG or alpha transparency mask of PNG

Mind you, that article is three years out of date. They've added those things.

Is that quality needed for web page browsing?

Well, yes, seeing how it's supposed to be used for replacing PNG for the graphics on the facade of a website, where you'd want small files but you can't tolerate JPEG's compression artifacts.

Whenever JPEG is used for images that are downloaded a lot, the limiting factor becomes how bad you can make it look before your customers or users complain about it.