WebP 1.0 Image Format Released

[FishPls]

But is it actually being used? 'Cause I've never ever ran across an image with the WebP format. Almost all of the images I come across are either JPG, JPEG or PNG and the ones that aren't are either GIF, TIFF or SVG.

Nope, because only Chromium supports WebP's. :/

Why use an image format if only one browser supports it?

[Royi]

What about JPEG-XR?
I like it is based on integer Math hence no quantization error on decoding and encoding.

[squash]

Funny how everyone predicts doom and gloom for webp. It wasn't that long ago the same predictions were made for png, ogg, vorbis... But people used them. They produced a good result, and took the cost of a commercial encoder license off the cost of producing their product. I'm serving hundreds of thousands of webp images per day (thanks mod_pagespeed!) to browsers that can use them, which lowers my bottom line.

[chithanh]

Web sites which serve images and text (vast majority) are not paying royalties for distributing images. It will also never have full web browser support since browsers like Firefox won't support it. This is why the only uptake of HEIF is from companies who is already paying the HEVC royalty cap. Again it has ZERO chance of replacing jpeg as the de facto standard lossy image format.

If you are a website that serves videos to a non-trivial number of mobile users, then there is probably no way around HEVC.
Firefox also resisted implementation of AVC/H.264 and whatnot, but not for long.

Mobile devices has hardware accelerated jpeg decoding, which since it's simpler format is more battery efficient. The reason to use a new format is not to save battery, it is to get better quality per bit, but since websites won't pay royalties to distribute images, HEIF is never going to see any uptake on the web.

As if that were mutually exclusive.

Serving HEIC to mobile users will be feasible and save on bandwidth. Serving them AV1 based images is not feasible due to battery life reasons for at least another couple of years.

An AV1 based image format however, very well may, since it can be supported by all browsers and all websites given how it's royalty free, and also having full hardware accelerated support. But in all honesty I'm not sure even that is enough to topple jpeg as the 'standard' lossy format.

Existing images will remain JPEG of course.

[Grinch]

If you are a website that serves videos to a non-trivial number of mobile users, then there is probably no way around HEVC.

Of course there is, ever heard of h264 ? Better yet, ever heard of Youtube, it's a site that serves videos to a non-trivial number of mobile users, it does not use HEVC, it uses h264 and vp9.

HEVC currently has the 4k/UHD niche, but that's a lunch AV1 is sure to eat.

Firefox also resisted implementation of AVC/H.264 and whatnot, but not for long.

They implemented h264 when it became free (as in cost) for them to do so, courtesy of Cisco.

Serving HEIC to mobile users will be feasible and save on bandwidth.

Saving bandwidth is nice and all, but my wager is that a new image format won't result in overall smaller images in bytes on the web, but that you will get much better quality images for those same bytes. The big bandwidth eater on the web is video, not images.

[chithanh]

Of course there is, ever heard of h264 ? Better yet, ever heard of Youtube, it's a site that serves videos to a non-trivial number of mobile users, it does not use HEVC, it uses h264 and vp9.

HEVC currently has the 4k/UHD niche, but that's a lunch AV1 is sure to eat.

YouTube does feel enormous pressure to limit their bandwidth use. YouTube is of course backed by Google's deep pockets, so strategic decisions may outweigh mid-term economic factors. It may be that they are trying to hold out until AV1. But waiting until 2021 for AV1 hardware decoders was too long for Amazon and Netflix, and I don't think it is certain that YouTube will stay away from HEVC.

They implemented h264 when it became free (as in cost) for them to do so, courtesy of Cisco.

The Cisco OpenH264 plugin is used for video conferencing (WebRTC). It is not used to decode videos embedded in websites. Those are passed to the operating system multimedia framework.

Saving bandwidth is nice and all, but my wager is that a new image format won't result in overall smaller images in bytes on the web, but that you will get much better quality images for those same bytes. The big bandwidth eater on the web is video, not images.

As I wrote, this is not mutually exclusive.
When the terrestrial TV broadcaster here switched from H.262 (MPEG-2) to H.265 (HEVC), they both increased the number of channels per transponder and increased the resolution from SD (576p50) to HD (1080p50).
Similar I expect for video and images on websites. Part of the bandwidth savings will be re-invested in increased quality and part will lower the operating cost.

[Grinch]

and I don't think it is certain that YouTube will stay away from HEVC.

Nothing is certain, but it's extremely likely given that they have done so up until now, when AV1 is practically finalized and thus will start rolling out support in web browsers and of course Youtube.

Also I don't think it will take three years for AV1 hardware support to show, particularly given that development on AV1 has been done with full continous input from the hardware companies who are to implement it.

It is not used to decode videos embedded in websites. Those are passed to the operating system multimedia framework.

AFAIK it's used by Firefox for all kinds of h264 video on Linux when Gstreamer is not installed.

When the terrestrial TV broadcaster here switched from H.262 (MPEG-2) to H.265 (HEVC)

Where's this ?

Similar I expect for video and images on websites.

The amount of bandwidth to save in web images is abysmal when compared to video, which is also why there's been so relatively little interest in adopting a new lossy image format to replace jpeg on the web. As we've seen it's not a big issue, which leads me to the conclusion that the current bandwidth use of images is not an actual problem.

However if it that was to increase markedly as a result of images needing to be larger/more detailed due to the increasing overall resolution on screens, then that would become a problem.

So as I stated before, I believe the actual size in bytes of images on the web will remain pretty much the same with the advent of a new format with superior compression, but you will get much better detail/resolution for those same bytes.

[chithanh]

Also I don't think it will take three years for AV1 hardware support to show, particularly given that development on AV1 has been done with full continous input from the hardware companies who are to implement it.

AV1 bitstream has only been finalized a month ago. Only now it can be implemented in hardware designs. Three years from conception to hardware in consumer hands is more or less a lower limit (unless you are really pushing it).

AFAIK it's used by Firefox for all kinds of h264 video on Linux when Gstreamer is not installed.

https://wiki.mozilla.org/Media/openh264
Also AFAIK gstreamer got nuked from Firefox and ffmpeg is now used directly.

Where's this ?

Germany, transition started in June 2016.

[Grinch]

Three years from conception to hardware in consumer hands is more or less a lower limit (unless you are really pushing it).

Well, we'll have to wait and see, this is however not a design which just lands in the manufacturers lap as of bitstream finalization, the companies in question (Intel, Broadcom, NVidia, AMD etc) have been part of the bitstream design process step by step.

Also AFAIK gstreamer got nuked from Firefox and ffmpeg is now used directly.

I stand corrected, thanks for the update.