Hacker News new | comments | ask | show | jobs | submit login
WebP: A new image format for the Web (code.google.com)
62 points by pud on May 21, 2011 | hide | past | web | favorite | 28 comments

Here's a post from the Chromium blog [1], which has more details. It also states how Google is already using it in Gmail, Picasa and instant search previews.

Google is in a great position as they can write a spec, release a browser implementing the spec and update their sites (which account for a substantial portion of the web) in such a short time. In essence, there is no "chicken and the egg" problem of who goes first.

This also happened with SPDY [2], which Chrome already supports and most Google sites now use instead of HTTPS (for Chrome, at least).

I think we can expect Google to do everything they can to make the web faster.

[1] http://blog.chromium.org/2011/05/webp-in-chrome-picasa-gmail...

[2] http://www.chromium.org/spdy

I agree that Google wants to make the web as fast as possible and as good as possible because it's directly in their interest so they have incentive to do it. They are mostly a web company after all.

On the other hand Apple and Microsoft don't really have to want to make the Internet as fast or good as possible. They either do it to support a different agenda, like not needing Flash, or from a defensive point of view, like Microsoft with IE9. But when it comes to choose between what they really care about, they won't choose the web. That's why Apple hasn't made the embedded browser as good as the main one, and Microsoft would rather use more proprietary stuff to couple with their browser, like DirectX.

My point is it feels natural to these companies to use the web only to help their main objective indirectly, which is different than simply wanting to make the web better, while in Google's case, it feels natural to them to want to improve the web because that interests them directly. So when it comes to who really wants to improve the whole web the most, I'd trust Google.

It's interesting, intellectually, to compare this with how Microsoft have acted in the past when producing proprietary web extensions and pushing their own document formats.

I made a QuickLook plugin for it: https://github.com/dchest/webp-quicklook

I can understand Google's reason for making and promoting WebM, but is there really a need for another image format? Yes, it may be X% smaller than JPEG for the same image quality, but does that really matter? Compared with video, the bandwidth for serving images is practically nothing, and even with universal adoption by browser and operating system vendors it will still be years before you can rely on all your clients having WebP support.

I don't think there's any reason not to try to improve something even if its satisfactory. Image formats are no exception - I don't think JPEG is the best that can be done and I hope we have better options moving forward.

It doesn't matter. Get over it. Leave it alone. Let it go.

There is a need for better image formats for photography, for gigatextures, for voxels, for specialized cases, but not for general web use. This problem is solved. Move along.

ASCII has served us well, it's still used now, and even though it's broken in a lot of ways, UTF-8 addresses most of those to a degree that's satisfactory enough we don't need people inventing new character encoding systems.

Between GIF, PNG and JPEG you have what you need. Don't cry over a few wasted bytes or a few smudgy pixels.

I could say the same thing about H.264. Why should we bother making better video codecs for general movie viewing use? Let it go. We can just add more data layers to Blu-rays.

And it's not like JPEG or ASCII or UTF-8 or anything we use today is going to suddenly disappear if we add new browser support. Why not strive for excellence?

Tell that to Flickr, Wikimedia, Facebook and a large portion of the developing world that are stuck on slow and expensive mobile connections.

That was sarcasm, right?

Come on HN, he makes a justifiable point quite well; this doesn't deserve downvoting.

FWIW I disagree strongly with his sentiment.

Previous discussions: http://news.ycombinator.com/item?id=1744237 http://news.ycombinator.com/item?id=1746621 http://news.ycombinator.com/item?id=1745801 http://news.ycombinator.com/item?id=2507700

It may be X% smaller than JPEG for the same image quality, but does that really matter?

If the bandwidth cost saved is greater than the cost of adding WebP to Chrome, then it's a net win. Or if it makes some Google page load 10 ms faster, that would also pay for it.

years before you can rely on all your clients having WebP support.

That's what content negotiation is for.

The problem is that Chrome sends an accept header of / for images, meaning that mod_negotiation for instance can't determine what to send. The only real way is either by hard coded UA sniffing on the server (uggh!) or by detecting webp in JS (ugggh!) as far as I can tell.

It does matter when a lot of web browsing happens on mobile.


Majority of the mobile phones will only have access to 2G/3G speeds.

In a lot of countries, mobile data is metered. Saved bytes = saved money.

A cool option for mobile browsers, would be that an online service, could convert JPEG files into WebP. Then the mobile browser could be downloading the WebP file instead - that way saving money.

Opera (and others I think) does this with JPEG quality. They mirror the image files as far lower quality JPEG so as to improve speed for those on limited bandwidth. You can of course request the full files.

Correct. If you enable Turbo in Opera Mobile or desktop, all traffic is sent through a proxy server which converts images into WebP format. The quality is better than when we used jpegs and the file sizes are much smaller. (Disclaimer: I work for Opera)

If X was less than 10, you may have had a point. But a 39.8% decrease in size is substantial.

So, yes, it does really matter

PNG had similar issues with browser adoption. The huge thing about WebP for me is it's support for transparency (which it was planned to support last time I checked).

I think the lack of transparency right now is what prevents me from caring enough to use it. Compression along isn't enough of a win.

Alpha-channel support is in the trunk, marked as experimental, so I think it's coming soon.

Until they sort out the video tag and the audio tag they should keep away from messing up the img tag in the same way.

Introducing a new image format won't "mess up" anything. And even if they were doing a stellar job with the video and audio tags, it would have no bearing on WebP.

So does this mean that Chrome will drop support for JPEG, PNG and GIF? I mean, that was the case with the video tag.

Chrome only dropped support for H.264 because it would incur patent fees which many consider incompatible with the way web technology has developed previously.

They didn't, for example, drop support for Theora which like the image formats you name, is older and has less advanced compression technology but is still "free" in many senses that some consider important for interoperable web technologies.

How is WebP better than JPEG XR (HD Photo) which already has an ISO standard?

JPEG XR appears to be aimed at being a less resource intensive alternative to JPEG-2000 for camera manufacturers and semi-pro photographers. WebP seems to be setting itself up to be used to deliver images via the web.

Currently XR has a lossless mode, 4:4:4 color support, tiling and alpha which make it more featureful than WebP but those have all been announced for future versions with some features having experimental implementations already.

On the other hand XR, at least the currently available encoders, seems to perform very well under PSNR tests but fails compared with even JPEG in SSIM tests.


It also worth noting that being an ISO standard holds far less status ever since Microsoft bent the procedures beyond breaking point for OOXML.

I don't see an ISO standard as being less than it was. I think the big win for JPEG XR is that their are some pretty good people who built it. If camera manufactures adopt it, then I think it is going to be the default. I don't see any real advantage to WebP.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact