I basically only use webp for lossless compression where it beats PNG hands down in everything I've tried.
Overall I think it will be extremely difficult for any new lossy image format to make inroads as jpeg is simply 'good enough' and supported EVERYWHERE.
Unlike what others said I watched presentations, videos etc and WebP is a lot like systemD or the Opus audio codec (and don't give me the apples/oranges shitty excuse) - it fits pretty much every usecase, is often (a lot) better and is pretty much free as in beer and porn and solves most problems like no other alternative does. When taken all of it's pros/cons together, just like with Opus/systemd (or even to a degree wayland) - it is the clear winner, though needless to mention - not perfect.
The biggest issue with WebP is that it's a lot slower at encoding and somewhat slower at decoding. Other than that it: yields better quality than jpg, uses less space, animates like gif, has true alpha support, better meta support, has lossless mode but unlike png creates smaller lossless files, also unlike png its API is not crappy (like that of libpng) etc etc. Don't be lazy or morons, look it up on youtube for all the pros and cons.
I was curious about this "animated WebP" and how was it any different from plain webM video. Turns out it's a similar hack to webp as apng was to png, so completely different from webm.
Without digging up that old horse, I saw little value in "show the first frame" of APNG - if you want an animation, the first frame is practically useless. Even worse, it may give the user the impression it's supposed to be a static image, while a "missing support" box lets the user know it's supposed to be something else.
I applaud you for the phrase "free as in porn"Unlike what others said I watched presentations, videos etc and WebP is a lot like systemD or the Opus audio codec (and don't give me the apples/oranges shitty excuse) - it fits pretty much every usecase, is often (a lot) better and is pretty much free as in beer and porn and solves most problems like no other alternative does. When taken all of it's pros/cons together, just like with Opus/systemd (or even to a degree wayland) - it is the clear winner, though needless to mention - not perfect.
Still, flawed comparison: Opus is indeed better in every way, Wayland and systemd are not.
I've also been tracking the Daala codec, every once in a while I'd compile it and try out the encoder. It's certainly doing very good for a codec from scratch that has only been in development for about a year: 2012-02-22 is the first commit but it said it was to import someone's Daala stuff, so it's probably around a year.
But wait, this is about animated webp right? I should probably perform my own tests because the only one I saw, that had gif vs webp vs apng but using a gif source. It's one of the stupidest things to do in a test, even if it's file size. GIF is a lossy format and then they convert to a lossless format (Webp is both lossy and lossless)
PNG's strength comes from being able to produce a small file, similar to how GIF can do it, where you basically tell it that the PNG can only have these colours selected. It didn't need all that extra information and it's still lossless if done right. The other strength is that PNG is basically the lossless image standard right now. Webp's strength comes from Google and everything that Google does. They make sure it is fast and small and gets the job done, like everything else. If I remember right it's based off TIFF or at least uses a TIFF wrapper.
Actually, for filesize comparison, such "gif vs webp vs apng" test makes perfect sense, because it ensures all 3 animations are identical. It's like comparing ZIP and RAR, you have to use identical set of files. Otherwise it's not fair.
Even in their announcement on the Chromium Blog, google devs used the same approach, they took a rotating cube GIF and converted it (very lossy) into WebP using gif2webp tool. If you want to perform your own tests, you will probably use gif2webp as well.