WebP - new image format by Google

[lovinglinux]

I'm leeching OMG Ubuntu today :)

http://www.omgubuntu.co.uk/2010/09/install-googles-webp-image-convertor-in-ubuntu/

http://code.google.com/speed/webp/
http://code.google.com/speed/webp/docs/using.html#viewing

[phrostbyte]

Where is the Firefox support for the image format? :)

[Dustin2128]

Where is the Firefox support for the image format? :)
oh don't worry. He'll develop an extension for it within 30 seconds of writing this that will be merged back into core firefox, thus giving him the confidence to apply for the job openings at mozilla.

[Starks]

Watch this get ignored like APNG.

[lovinglinux]

Where is the Firefox support for the image format? :)

No browser has support for this, not even Chrome.


oh don't worry. He'll develop an extension for it within 30 seconds of writing this that will be merged back into core firefox, thus giving him the confidence to apply for the job openings at mozilla.

:lolflag:

I must admit it crossed my mind developing an extension to act as a GUI for the converter :)

[Dustin2128]

No browser has support for this, not even Chrome.



:lolflag:

I must admit it crossed my mind developing an extension to act as a GUI for the converter :)
Do it. Make firefox the first, opera supported WebM (google format) before chrome, have us beat them both to it.

[lovinglinux]

Gallery http://code.google.com/speed/webp/gallery.html

[lovinglinux]

Do it. Make firefox the first, opera supported WebM (google format) before chrome, have us beat them both to it.

What I meant was a GUI for the converter, not the actual code that would allow Firefox to render the images. That is also out of my league :)

[chris200x9]

Gallery http://code.google.com/speed/webp/gallery.html

if no browser supports them yet how are we seeing them? Are they converting them and seeing the size then converting them to something lossless like tiff for the web?

[AllRadioisDead]

I believe they are being masked as PNG's.

[chris200x9]

I believe they are being masked as PNG's.

this brings me to another question, why is it vs jpeg wouldn't they do vs png since png is FOSS and most likely what it would replace?

[lovinglinux]

if no browser supports them yet how are we seeing them? Are they converting them and seeing the size then converting them to something lossless like tiff for the web?

Yes, they are converting back to PNG.

[Bachstelze]

*sigh*

If we didn't have enough "news" spam with Sporkman...

We can use the Internet too, we don't need you to deliver links to us.

[foxmulder881]

<post removed by foxmulder881>

[foxmulder881]

Great, just what I don't have the time to deal with in my profession. Another image file format to contend with.

[lovinglinux]

*sigh*

If we didn't have enough "news" spam with Sporkman...

We can use the Internet too, we don't need you to deliver links to us.

Why are you so negative? I only see posts from you being negative about everything. Relax and enjoy. If you don't like my threads, then don't read them. Besides, is not just about clipping the news, but about discussing them with fellow Ubuntu forum members.

[lovinglinux]

I have just created an extension to act as GUI for the converter. It isn't working tho and I'm tired, so I will try to make it work tomorrow.

[renkinjutsu]

Why can't things be like this more often? =[
http://ubuntuforums.org/attachment.php?attachmentid=170931&stc=1&d=1285901511

[formaldehyde_spoon]

Why are you so negative? I only see posts from you being negative about everything. Relax and enjoy. If you don't like my threads, then don't read them. Besides, is not just about clipping the news, but about discussing them with fellow Ubuntu forum members.

Agreed. We won't all find everything ourselves, and even if we did, this provides us with an opportunity to discuss it.


this brings me to another question, why is it vs jpeg wouldn't they do vs png since png is FOSS and most likely what it would replace?


FOSS or not is not the point.
Png is lossless, jpeg is lossy, most images on the web are jpeg.

So webp won't/can't replace png, and jpeg is the perfect target to aim for, which is what google are doing.

[laststop]

The most important part:


We plan to add support for a transparency layer, also known as alpha channel in a future update.

Alpha channel could be a deal breaker.

[EarthMind]

This is interesting. Faster load times + less bandwidth consumption. I'm going to try it out.

[lovinglinux]

The most important part:

Alpha channel could be a deal breaker.

I agree. That would be nice.

[Lucradia]

Watch this get ignored like APNG.

I agree.

[m4tic]

www.tgdaily.com/software-features/51815-microsoft-promises-to-speed-up-the-web-with-new-image-format

They keep saying microsoft created the format and google chrome, funniest and most stupid article ever.

[renkinjutsu]

www.tgdaily.com/software-features/51815-microsoft-promises-to-speed-up-the-web-with-new-image-format

They keep saying microsoft created the format and google chrome, funniest and most stupid article ever.

That was indeed a pretty big fail
It seems like he fixed up everything though...

Everything besides
WebP images can't be viewed until browsers support the format, but Microsoft says it's developing a patch for WebKit to provide native support for WebP in an upcoming release of Google Chrome.

[Ctrl-Alt-F1]

If we didn't have enough "news" spam with Sporkman...

We can use the Internet too, we don't need you to deliver links to us.
Totally agree. Although I don't think LovingLinux is guilty of the crime (or maybe I haven't been paying attention).

[NightwishFan]

Bachstelze

If we need a cynical opinion we will call you. :P

[Dustin2128]

if we need a cynical opinion we will call you. :p
+1

[foxmulder881]

I'd like to get into this in a little more technical detail because I find it rather interesting.
So what exactly are we gaining by this new format if it's simply being masked as a PNG image? If it does offer better compression and faster load times, aren't all those benefits simply ignored if the browser is decoding them as a PNG?

[renkinjutsu]

The images uploaded to the gallery are PNG's because no browsers (currently) can display the webp images. The images are only there for you to compare the quality of the format.

they converted jpeg -> webp, then webp -> png
since png is a lossless format, it should display the same quality as the webp image

[foxmulder881]

The images uploaded to the gallery are PNG's because no browsers (currently) can display the webp images. The images are only there for you to compare the quality of the format.

they converted jpeg -> webp, then webp -> png
since png is a lossless format, it should display the same quality as the webp image

But I'm wondering what the comparisons would be from jpeg > png and just skip the whole webp format? Is there really a difference in size?

[lovinglinux]

But I'm wondering what the comparisons would be from jpeg > png and just skip the whole webp format? Is there really a difference in size?

What they are doing is to use png just to show the image quality of the webp format in comparison with jpg. Since png is lossless, when they convert to png they can see the image quality as expected to be seen in webp when the browsers become capable of displaying the new format. Nevertheless, converting to png increases file size considerably.

[formaldehyde_spoon]

But I'm wondering what the comparisons would be from jpeg > png and just skip the whole webp format? Is there really a difference in size?

png is not part of this news. The only reason you're seeing png is because browsers can't display webp yet, but they want to show you what a webp image looks like - any image (eg webp, jpeg) when converted to a lossless format (png in this case) will look exactly the same as it did before the conversion.

Size-wise: png >* jpeg > webp
* usually - depends on what is shown in the image. jpeg is very good for photos. png is lossless but has compression.

jpeg is widely used online because in most cases it's smaller than lossless formats, with only a small loss in quality. webp is smaller again.

No such thing as a free lunch though: webp compression and decompression takes more processing than jpeg, although that's not a huge deal now that we have much more powerful computers than we did when jpeg came out.

[foxmulder881]

Nevertheless, converting to png increases file size considerably.

But that's what I'm getting at. What's the point then of introducing a new format when it offers nothing new that we don't have already?

[NightwishFan]

I do not really know, however I am guessing the advantages are it will be optimized in various ways for network transfer.

[lovinglinux]

But that's what I'm getting at. What's the point then of introducing a new format when it offers nothing new that we don't have already?

It does offer better compression than jpg, with the same image quality. So images will look the same, but will use less disk space and consume less bandwidth when transmitted over the web.

Forget about the png conversion. That is only being used because the browser can't display webp yet. It won't be part of the equation when the format is supported by browsers.

[formaldehyde_spoon]

But that's what I'm getting at. What's the point then of introducing a new format when it offers nothing new that we don't have already?

? Are my posts invisible to others? :p
webp is smaller. It's only been converted to png because the browser can't display webp. In the future they won't be converted to png.

[foxmulder881]

So essentially the gallery is showing the images off in a false sort of way. The developers are saying, "Hey, here's a great new format you should use. But we can't show you yet because no one has a browser to view it. Oh, but here's how they will look, but not look because this isn't actually it!"

[formaldehyde_spoon]

So essentially the gallery is showing the images off in a false sort of way. The developers are saying, &quot;Hey, here's a great new format you should use. But we can't show you yet because no one has a browser to view it. Oh, but here's how they will look, but not look because this isn't actually it!&quot;

This *is* what webp looks like. png is *lossless*, so it looks exactly like whatever was converted to it.

For example, you can open a jpeg in kolourpaint (or whatever) and save it as a png - it will still look exactly the same as it did when saved as a jpeg.

[lovinglinux]

So essentially the gallery is showing the images off in a false sort of way. The developers are saying, "Hey, here's a great new format you should use. But we can't show you yet because no one has a browser to view it. Oh, but here's how they will look, but not look because this isn't actually it!"

YES!

They will look exactly the same, because there isn't any quality loss when converting from webp to png. But since you can't view webp images yet, then they convert to png so you can have an idea of the image quality of the new format.

[formaldehyde_spoon]

YES!

...

NO! ;)
You just told foxmulder that this is correct: ''...but not look because this isn't actually it!''

[renkinjutsu]

So essentially the gallery is showing the images off in a false sort of way. The developers are saying, "Hey, here's a great new format you should use. But we can't show you yet because no one has a browser to view it. Oh, but here's how they will look, but not look because this isn't actually it!"

Have you read my post at all? I really put my time, effort and even my soul into that post, just for it to be ignored by you :(.

[lovinglinux]

NO! ;)
You just told foxmulder that this is correct: ''...but not look because this isn't actually it!''

It was close enough for me :)

[foxmulder881]

Have you read my post at all? I really put my time, effort and even my soul into that post, just for it to be ignored by you :(.

So don't put so much effort in next time if you're going to feel so unappreciated. This forums isn't a beauty contest.


For the record, I do realize what you're saying and what is being displayed here. You seem to be missing my point. All of you!
And for the record (2nd record), I do understand what a PNG image is, considering it is my job too and all. And considering I work with PNG images almost everyday. So please no more posts telling me PNG is a lossless format.

[formaldehyde_spoon]

It was close enough for me :)

Fair enough! :p
So don't put so much effort in next time if you're going to feel so unappreciated. This forums isn't a beauty contest.


For the record, I do realize what you're saying and what is being displayed here. You seem to be missing my point. All of you!
And for the record (2nd record), I do understand what a PNG image is, considering it is my job too and all. And considering I work with PNG images almost everyday. So please no more posts telling me PNG is a lossless format.

We're only trying to help with something that you say you don't understand.
And you do seem to ask questions that have already been answered earlier...

[NightwishFan]

Why bother to see the file if nothing could view it. I am sure if you wanted you could get your hands on one somehow.

[laststop]

Anyway, Goggle should have waited with the announcement until:

1. Alpha channel is implemented.
2. Converter/viewer software is available on all platforms (even if very simple)
3. Goggle browser supports it.

Then it would have made a much bigger splash.

[cammin]

For the comparison, they should have taken images from a lossless source, then converted it to both jpeg and webp. Not taken an already lossy image and done another lossy conversion to it.

[JDShu]

*sigh*

If we didn't have enough "news" spam with Sporkman...

We can use the Internet too, we don't need you to deliver links to us.

Isn't the point that we can discuss it here with other Linux users?

[formaldehyde_spoon]

For the comparison, they should have taken images from a lossless source, then converted it to both jpeg and webp. Not taken an already lossy image and done another lossy conversion to it.

That's exactly what they did...........

[spoons]

Bit misleading for them to compare it to the current ageing JPEG format and not any of the other competing newer formats. They may be better.

[formaldehyde_spoon]

Bit misleading for them to compare it to the current ageing JPEG format and not any of the other competing newer formats. They may be better.

This thread is full of misunderstanding.

How is it misleading? Do you actually mean ''unfair''?
It's not comparison for comparison's sake; jpeg is used all over the web, and the point is to replace it with something better(smaller), which is only really feasible because we all now have better hardware than we did when jpeg came out.

[foxmulder881]

Anyway, Goggle should have waited with the announcement

But why? What better way to generate publicity and interest among geeks other than to offer them something that they know can have, but not yet!

It's a bit like a stripteasing woman. You know what you can have, but not until the final details are uncovered.

[cammin]

That's exactly what they did...........

The source images are jpegs.

[NMFTM]

Bit misleading for them to compare it to the current ageing JPEG format and not any of the other competing newer formats. They may be better.
Newer formats aren't widely used like JPEG is though. If Webp weren't made by Google I'd say that it might go the way of some of those other formats. But Google is a big player and if they're supporting something it has a bigger chance for success.

[lovinglinux]

Newer formats aren't widely used like JPEG is though. If Webp weren't made by Google I'd say that it might go the way of some of those other formats. But Google is a big player and if they're supporting something it has a bigger chance for success.

I agree.

[bash]

The x264 developer(s) aren't particularly impressed with the new format, claiming that in it's current version it's actually worse than jpeg:

http://x264dev.multimedia.cx/archives/541

[lovinglinux]

The x264 developer(s) aren't particularly impressed with the new format, claiming that in it's current version it's actually worse than jpeg:

http://x264dev.multimedia.cx/archives/541

Interesting article.

[laststop]

But why? What better way to generate publicity and interest among geeks other than to offer them something that they know can have, but not yet!

It's a bit like a stripteasing woman. You know what you can have, but not until the final details are uncovered.

Haha. But seriously, people's attention span is too short these days.

[del_diablo]

The x264 developer(s) aren't particularly impressed with the new format, claiming that in it's current version it's actually worse than jpeg:

http://x264dev.multimedia.cx/archives/541

Haha, actually a good example of attempting FUD xD

[bash]

Haha, actually a good example of attempting FUD xD

How exactly is this "attempting FUD"? The author of that blog post is one (or the?) developer of x264, which is considered a superior encoder to the reference h.264 implementation. On the topic of vp8 he has a published a couple of comments generally respected, so I would say the author knows quite well about the topic and what he is talking about.

Please feel free to elaborate why exactly you find this to be "attempting FUD" and what is so discreditable about the blog post. Because currently your statement like this makes you appear like you have no idea what you are talking about.

[formaldehyde_spoon]

The source images are jpegs.

I don't know whether they re-compressed the jpeg as jpeg or not, but it doesn't matter, and it doesn't matter what the source is.
Jpegs files compress data. Images show the data from the file uncompressed.
The webp files have compressed the uncompressed data (from a displayed jpeg), and are smaller than jpeg.
The webp examples are compressing the image, not the jpeg file.

If the jpegs shown haven't been re-encoded as jpeg then that would certainly help explain why they look slightly better than the webp images.


How exactly is this &quot;attempting FUD&quot;? The author of that blog post is one (or the?) developer of x264, which is considered a superior encoder to the reference h.264 implementation. On the topic of vp8 he has a published a couple of comments generally respected, so I would say the author knows quite well about the topic and what he is talking about.

Please feel free to elaborate why exactly you find this to be &quot;attempting FUD&quot; and what is so discreditable about the blog post. Because currently your statement like this makes you appear like you have no idea what you are talking about.

I don't know that it's FUD either, but seeing as it's someone from x264 I would err on the side of caution and say that it probably is.
x264 devs are possibly the most biased, one-eyed commentators I've ever seen on any issue online.
They will blindly attack anyone or anything that threatens them on their little perch - eg webM/VP8, and because Webp is vp8 derived I think it's probably fair game as well as far as they're concerned.

[NCLI]

I don't know that it's FUD either, but seeing as it's someone from x264 I would err on the side of caution and say that it probably is.
x264 devs are possibly the most biased, one-eyed commentators I've ever seen on any issue online.
They will blindly attack anyone or anything that threatens them on their little perch - eg webM/VP8, and because Webp is vp8 derived I think it's probably fair game as well as far as they're concerned.
That's very true. I remember them trying to make an issue out of the fact that WebM doesn't have a very high-quality encoding option like x264, while failing to realize/ignoring that cinema quality video was never the point of WebM.

Let's wait for someone less biased to make a comparison, and with a more complete product.

[Npl]

err, the same guy was involved with creating the first independent VP8 decoder. Blindly attacking indeed... and would you actually claim that google, which paid the VP8 tech for good money is unbiased when it says WebP beats JPeg?

Their article announcing WebP is rather laughable, and lacking any kind of comparison... so you can get smaller filesizes with WebP while still looking decend, but how about comparing it to similar sized jpegs in the first place?
You know you can put out jpegs with any filesize you want.

[formaldehyde_spoon]

err, the same guy was involved with creating the first independent VP8 decoder. Blindly attacking indeed... and would you actually claim that google, which paid the VP8 tech for good money is unbiased when it says WebP beats JPeg?
So? How does that make even the tiniest bit of difference? Do you think he was part of that decoder because he's a fan of VP8?
I said nothing about google's bias, but they haven't launched any x264-style attacks on h264 anyway.

Their article announcing WebP is rather laughable, and lacking any kind of comparison... so you can get smaller filesizes with WebP while still looking decend, but how about comparing it to similar sized jpegs in the first place?
You know you can put out jpegs with any filesize you want.

The comparison is here: https://code.google.com/speed/webp/gallery.html
Kiss goodbye to your image quality when you encode small jpegs.
The examples have taken the image compressed int the jpeg files and recompressed them as webp instead - they're significantly smaller, with comparable quality.
You seem to have some strong feelings about webp/jpeg without any apparent technical reason.

[bash]

I don't know that it's FUD either, but seeing as it's someone from x264 I would err on the side of caution and say that it probably is.
x264 devs are possibly the most biased, one-eyed commentators I've ever seen on any issue online.
They will blindly attack anyone or anything that threatens them on their little perch - eg webM/VP8, and because Webp is vp8 derived I think it's probably fair game as well as far as they're concerned.

Ok that might kind of explain the FUD comment. I don't know about the personal views or biases of that x264 dev, but at least the technical reasoning seems sound. The reactions throughout the web seem to agree that at least the technical reasoning is ok and that he raises a few valid points.

The disagreement seems to be more on the ideological side with dismissing anything that he says is biased or he just wants to promote his tools.

Edit Clarified the blog reactions part with reactions throughout the web

[formaldehyde_spoon]

Ok that might kind of explain the FUD comment. I don't know about the personal views or biases of that x264 dev, but at least the technical reasoning seems sound. The blog reactions seem to agree that at least the technical reasoning is ok and that he raises a few valid points.

The disagreement seems to be more on the ideological side with dismissing anything that he says is biased or he just wants to promote his tools.

Blog reactions are to be taken with a grain of salt; you need to consider who makes up the majority of commenters: if someone read UF they might come to the conclusion that Ubuntu is far superior to all other OSs ;) (Not trying to say Ubuntu isn't great!)

That guy may well be raising great technical points, but unless I had the same level of knowledge of video and images he has to be able to say he's presenting an accurate picture I'd prefer to wait for comment by a more neutral expert.

All I can say is I've downloaded the images and can't see any difference between them with my naked eye.

[Npl]

So? How does that make even the tiniest bit of difference? Do you think he was part of that decoder because he's a fan of VP8?given that he aint got anything for doing this in his spare-time, I sure he hated every minute of it.

I said nothing about google's bias, but they haven't launched any x264-style attacks on h264 anyway.Likely cause technical sound comparisons wouldnt do them well.


The comparison is here: https://code.google.com/speed/webp/gallery.html
Kiss goodbye to your image quality when you encode small jpegs.
The examples have taken the image compressed int the jpeg files and recompressed them as webp instead - they're significantly smaller, with comparable quality.[\QUOTE]This was exactly the comparison I referenced. What stops you from recompressing them as jpegs, aslong you dont compare quality on equal footing its pointless.
[QUOTE=formaldehyde_spoon;9921643]You seem to have some strong feelings about webp/jpeg without any apparent technical reason.I have strong feelings against jpeg? :P
I have strong feelings against pushing unnecessary stuff for marketing reasons, so far there is no technical reason for WebP to exist at all.

[lovinglinux]

Ok that might kind of explain the FUD comment. I don't know about the personal views or biases of that x264 dev, but at least the technical reasoning seems sound. The blog reactions seem to agree that at least the technical reasoning is ok and that he raises a few valid points.

The disagreement seems to be more on the ideological side with dismissing anything that he says is biased or he just wants to promote his tools.

Although the article is interesting to read, I must agree I also think they are biased. Is not the first time I see them criticizing Google because of VP8.

[bash]

Blog reactions are to be taken with a grain of salt; you need to consider who makes up the majority of commenters: if someone read UF they might come to the conclusion that Ubuntu is far superior to all other OSs ;) (Not trying to say Ubuntu isn't great!)

With blog reactions I meant general reactions, like blog posts, articles or mails from other websites towards the raised criticism. Maybe I wasn't clear enough on that point. Things like this mail (http://lists.wikimedia.org/pipermail/wikitech-l/2010-May/047795.html) from a mailing list by Gregory Maxwell from the Xiph foundation (Note: the mail start of with the FUD but that is aimed at the previous mail in the mailing list not the x264 blog post). Or here a response from an Opera dev in his blog:

http://my.opera.com/haavard/blog/2010/05/20/webm-analysis

As said before the disagreements over the x264 devs blog posts seem more on a general ideological level between "VP8 is awesome, this guy has no idea and is just spreading FUD" and "His analysis is be all, end all; he is only one right, VP8 is clearly a complete failure".

Oh and just saw that Jobs used the x264 devs criticism as the explanation in his negative response to the question if Apple would use WebM [1]. Don't really know if that lends credibility to that guy or actually diminishes it.

[1] http://www.theregister.co.uk/2010/05/20/jobs_on_vp8/

[formaldehyde_spoon]

given that he aint got anything for doing this in his spare-time, I sure he hated every minute of it.
You didn't answer (either of) the question(s).
Likely cause technical sound comparisons wouldnt do them well.

Then how does this support your 'biased google' theory?
This was exactly the comparison I referenced. What stops you from recompressing them as jpegs, aslong you dont compare quality on equal footing its pointless.
The jpeg is the source file. The image from the jpeg is stored as a smaller wepb file. If you recompressed the jpeg as jpeg you'd end up with a lower quality picture than the webp.
The point is webp can encode with very little loss and excellent compression.

There are two competing variables to consider: size and quality.
They've done the comparison by checking the size of two images that apparently have similar quality.
Yes, they could instead have checked the quality of two equally sized images to reach the same conclusion, but quality is much harder to quantify than size.
I have strong feelings against jpeg? :P
I have strong feelings against pushing unnecessary stuff for marketing reasons, so far there is no technical reason for WebP to exist at all.Yes, how dare they! Marketing something unnecessary is outrageous!

Jpeg is 20 years old, optimized for much less powerful machines than we have now, and reducing data sent over our 'tubes' is always a good thing.

[formaldehyde_spoon]

With blog reactions I meant general reactions, like blog posts, articles or mails from other websites towards the raised criticism. Maybe I wasn't clear enough on that point. Things like this mail (http://lists.wikimedia.org/pipermail/wikitech-l/2010-May/047795.html) from a mailing list by Gregory Maxwell from the Xiph foundation (Note: the mail start of with the FUD but that is aimed at the previous mail in the mailing list not the x264 blog post). Or here a response from an Opera dev in his blog:

http://my.opera.com/haavard/blog/2010/05/20/webm-analysis

Oh just saw that Jobs used the x264 devs criticism as the explanation in his negative response to the question if Apple would use WebM [1]. Don't really know if that lends credibility to that guy or actually diminishes it.

[1] http://www.theregister.co.uk/2010/05/20/jobs_on_vp8/

Er, one of those links is a reaction to the (foaming mouth) x264 VP8 post, not the webp post (and it includes comments like ''Jason[an x264 dev] is talking out of his rear''), and the other recommends *against* taking the x264 dev's word as gospel.
I didn't bother reading Job's comment.

[Npl]

You didn't answer (either of) the question(s). Then how does this support your 'biased google theory'? The jpeg is the source file. The image from the jpeg is stored as a smaller wepb file. If you recompressed the jpeg as jpeg you'd end up with a lower quality picture than the webp.See, exactly that is an unproven hypothesis. Just pushing something out, and saying "its better" doesnt cut it. If you think the x264dev article is wrong then refute or let it be refuted with arguments and not because you think it aint so, unlike the google stuff the procedure how the images were archived is described and repeatable. And it allows a direct comparison between quality/filesize.


The point is webp can encode with very little loss and excellent compression.And so can Jpeg - in atleast one instance producing better results than WebP.


Yes, how dare they! Marketing something unnecessary is outrageous!The go buy one (http://it.slashdot.org/article.pl?sid=08/06/15/009216), must be true if Denon says so?


Jpeg is 20 years old, optimized for much less powerful machines than we have now, and reducing data sent over our 'tubes' is always a good thing.Yes, but WebP doesnt seem very fit to do so. Ignoring Quality, it doesnt even have several features that Jpg has, like beeing able to decode just a part of the picture, say you want a 300x200 thumbnail of a 14MP picture and Jpeg can do that withotu decoding the whole 14MP.
Jpeg might be 20 years old, but it was crafted with pictures in mind and it still does a pretty fine job. Something like Jpeg2000 would be a much better successor than WebP.

[formaldehyde_spoon]

See, exactly that is an unproven hypothesis. Just pushing something out, and saying &quot;its better&quot; doesnt cut it. If you think the x264dev article is wrong then refute or let it be refuted with arguments and not because you think it aint so, unlike the google stuff the procedure how the images were archived is described and repeatable. And it allows a direct comparison between quality/filesize. I guess my edit before was too slow:
There are two competing variables to consider: size and quality.
They've done the comparison by checking the size of two images that apparently have similar quality.
Yes, they could instead have checked the quality of two equally sized images to reach the same conclusion, but quality is much harder to quantify than size.

No-one is interested in replacing jpeg with something the same size! The whole point is to reduce the size of images transmitted, with minimal quality loss. Which is what webp does.


And so can Jpeg - in atleast one instance producing better results than WebP.But usually (much) worse - ie larger.
Yes, but WebP doesnt seem very fit to do so. Ignoring Quality, it doesnt even have several features that Jpg has, like beeing able to decode just a part of the picture, say you want a 300x200 thumbnail of a 14MP picture and Jpeg can do that withotu decoding the whole 14MP.
Jpeg might be 20 years old, but it was crafted with pictures in mind and it still does a pretty fine job. Something like Jpeg2000 would be a much better successor than WebP.Unlike you, I don't really care. Whether jpeg, webp or jpeg2000 is dominant this time next year makes no difference to me.
The point is webp can give us smaller images with little quality loss, google is pushing for it's adoption, and who knows - they may be successful. And if they are we'll reduce our data transmission, which is a good thing.

[Npl]

I guess my edit before was too slow:
There are two competing variables to consider: size and quality.
They've done the comparison by checking the size of two images that apparently have similar quality. Yes, they could instead have checked the quality of two equally sized images to reach the same conclusion, but quality is much harder to quantify than size.So how do you conclude you can compare size if you cant quantify quality?


No-one is interested in replacing jpeg with something the same size! The whole point is to reduce the size of images transmitted, with minimal quality loss. Which is what webp does.Actually I just downloaded the pictures that google themself used and tried recompressing the picture with jpg (via XNView) - a couple hundred bytes smaller than the webp file. Guess what I can easily spot differences between both the reencode/original and webp/original. and webp loses way more detail than the recompression with jpeg. I can upload the 3 files if you want,


But usually (much) worseBased on what evidence?


The point is webp can give us smaller images with little quality loss, google is pushing for it's adoption, and who knows - they may be successful. And if they are we'll reduce our data transmission, which is a good thing.unless theres a big gap in quality per bit, why bother.
The point is that the better quality is neither proven nor quantified now - all we got is googles word and their weird "test" which really is evidence against that IMHO.

Its something you can easily test for yourself, and unless you did whats the point in calling out bias or going so low and call "FUD" without even knowing anything?

[bash]

Er, one of those links is a reaction to the (foaming mouth) x264 VP8 post, not the webp post (and it includes comments like ''Jason[an x264 dev] is talking out of his rear''), and the other recommends *against* taking the x264 dev's word as gospel.
I didn't bother reading Job's comment.

This was about the guys general credibility about the technical assessment of WebP which is based on/uses VP8! If you read the two posted links closely you see none of them discards his technical analysis. The first one just argues that you have to put it into perspective:


Jason's comparison isn't unfair but you need to understand it for what it is— he's comparing a very raw, hardly out of development, set of
tools to his own project— which is the most sophisticated and mature
video encoder in existence.

He then goes on to question all of the x264 devs view on the patents which has nothing to with the technical analysis.

The Opera employee again doesn't discard the technical analysis but makes the point that the x264 devs is be all, end all on the question of the quality of VP8.

Which supports what I have been trying to say that most reactions here so far have either been a "This guy is completely wrong about his assessment of VP8 and the things based of it; he is completely biased and just talking cr*p" or "This guy is completely right". But not maybe a differentiated analysis between technical points he raises and more general ramblings and personal views he has - something along the lines of "Oh interesting read, he seems to know the technical details about different codecs, encoders and formats and raises a few valid points; although I wouldn't agree with all of his conclusions and non-technical points".

[Npl]

http://ubuntuforums.org/attachment.php?attachmentid=171251&stc=1&d=1286153329
WebP 122.260 bytes
http://ubuntuforums.org/attachment.php?attachmentid=171252&stc=1&d=1286153362
Jpeg 136.780 bytes
http://ubuntuforums.org/attachment.php?attachmentid=171253&stc=1&d=1286153362
Jpeg 117.172 bytes

Look for example at the solid blue lower left area of the painted city.
WebP smooths out detail considerably (noticeable everyhwere if you zoom in a bit). Recompressing with Jpeg (even smaller than the WebP file) retains the detail better.
So where exactly is the improvement over Jpeg?

[formaldehyde_spoon]

So how do you conclude you can compare size if you cant quantify quality?

You can compare quality, but it's hard to *quantify* it. You can't say ''quality is 15% better'' like you can with size, so it's not as useful to publish.
And of course the whole point is to reduce size.
Actually I just downloaded the pictures that google themself used and tried recompressing the picture with jpg (via XNView) - a couple hundred bytes smaller than the webp file. Guess what I can easily spot differences between both the reencode/original and webp/original. and webp loses way more detail than the recompression with jpeg. I can upload the 3 files if you want,

Good grief...quick tell google you've proved their publication wrong.
Did you also try recompressing again with webm?
The jpeg is the source, the webp image has compressed and stored the image from the jpeg, with little loss but excellent compression.

eg you have a 2x2 image. You can save it perfectly in a bmp of 4 bytes (say). Or you can save it in a 3 byte jpeg and lose some quality.
So the jpeg is now storing 4 pixels at 3 bytes. You can save those 4 pixels as webp in 2 bytes, perhaps with another small drop in quality.

The point is the webp image has almost the same quality as the jpeg even though it was produced from the jpeg not the bmp and it's smaller.
Do you not see that you have two images (almost) the same, but one is much smaller?
Based on what evidence?

I've already posted the link for you. 39% average.
unless theres a big gap in quality per bit, why bother.
The point is that the better quality is neither proven nor quantified now - all we got is googles word and their weird &quot;test&quot; which really is evidence against that IMHO.

Its something you can easily test for yourself, and unless you did whats the point in calling out bias or going so low and call &quot;FUD&quot; without even knowing anything?This thread isn't about the x264 post, it's about webp. The evidence is there for you to see, and I think you need to know more before you call it wrong.

[formaldehyde_spoon]

This was about the guys general credibility about the technical assessment of WebP which is based on/uses VP8! If you read the two posted links closely you see none of them discards his technical analysis. The first one just argues that you have to put it into perspective:



He then goes on to question all of the x264 devs view on the patents which has nothing to with the technical analysis.

The Opera employee again doesn't discard the technical analysis but makes the point that the x264 devs is be all, end all on the question of the quality of VP8.

Which supports what I have been trying to say that most reactions here so far have either been a &quot;This guy is completely wrong about his assessment of VP8 and the things based of it; he is completely biased and just talking cr*p&quot; or &quot;This guy is completely right&quot;. But not maybe a differentiated analysis between technical points he raises and more general ramblings and personal views he has - something along the lines of &quot;Oh interesting read, he seems to know the technical details about different codecs, encoders and formats and raises a few valid points; although I wouldn't agree with all of his conclusions and non-technical points&quot;.

I was only responding to you saying that comments were mostly positive. Your examples don't seem very positive.

[Npl]

You can compare quality, but it's hard to *quantify* it. You can't say ''quality is 15% better'' like you can with size, so it's not as useful to publish.size is useless if you cant guarantee equivalent quality, and why are you only allowed to dial filesize down using webp while keeping jpeg constant?
Its a really onesided comparison if you are allowed to find the "sweetspot" on one side and have an overblown file on the other.

Good grief...quick tell google you've proved their publication wrong.
Did you also try recompressing again with webm?why would I? google provides the webm compressed pictures. I recompressed the very same jpeg google used, seems like a fair comparison or not?


The point is the webp image has almost the same quality as the jpeg even though it was produced from the jpeg not the bmp and it's smaller.
Do you not see that you have two images (almost) the same, but one is much smaller? I've already posted the link for you. 39% average.Yeah and i could recompress it saving 40% with jpeg.. with little difference in quality. Now the question is, is the quality better or worse than webp. if wepb doesnt do a better job, why bother.


This thread isn't about the x264 post, it's about webp. The evidence is there for you to see, and I think you need to know more before you call it wrong.I try again (for the last time): I taken the webp-picture and the source-jpg directly from googles comparison.
I recompressed the original jpeg to something smaller, saving size while losing a bit of quality.
I posted all of those pictures a post above.

Do you see a huge loss in quality in the recompressed jpeg, or is quality almost the same and the filesize is less?
If you cant attest to the former, then there goes your evidence.

[formaldehyde_spoon]

Look for example at the solid blue lower left area of the painted city.
WebP smooths out detail considerably (noticeable everyhwere if you zoom in a bit). Recompressing with Jpeg (even smaller than the WebP file) retains the detail better.
So where exactly is the improvement over Jpeg?

Well done, you've just done what google and the rest of the web coudn't.../s

The jpeg has compressed a source to size X with quality loss Y.
Webp compression doesn't benefit from the fact that the image has already been encoded to jpeg, but still manages to compress much smaller, with minimal quality loss.

[formaldehyde_spoon]

size is useless if you cant guarantee equivalent quality, and why are you only allowed to dial filesize down using webp while keeping jpeg constant?
Its a really onesided comparison if you are allowed to find the &quot;sweetspot&quot; on one side and have an overblown file on the other.
why would I? google provides the webm compressed pictures. I recompressed the very same jpeg google used, seems like a fair comparison or not?

Yeah and i could recompress it saving 40% with jpeg.. with little difference in quality. Now the question is, is the quality better or worse than webp. if wepb doesnt do a better job, why bother.

I try again (for the last time): I taken the webp-picture and the source-jpg directly from googles comparison.
I recompressed the original jpeg to something smaller, saving size while losing a bit of quality.
I posted all of those pictures a post above.

Do you see a huge loss in quality in the recompressed jpeg, or is quality almost the same and the filesize is less?
If you cant attest to the former, then there goes your evidence.

Are you deliberately refusing to understand?
Ideally the webp would be allowed to start from the perfect source too, but it performs at the level you see even with the handicap of having to start with the jpeg.
Of course you can keep re-encoding images as jpeg, but quality deteriorates fast.

Yes, you can see jpeg artifacts appear all over the image.

[Dustin2128]

$150 says that internet explorer won't support it. Any takers? ;)

[formaldehyde_spoon]

$150 says that internet explorer won't support it. Any takers? ;)

I saw that before you edited it! :p :D Why the drop?

[bash]

I was only responding to you saying that comments were mostly positive. Your examples don't seem very positive.

Where did I said anything about positive? I merely said that "the reactions throughout the web seem to agree that at least the technical reasoning is ok". Which is exactly what the the comments do, they agree he raises a few valid points from a technical perspective, but don't share the same view on all of his conclusions.

But enough of this now.

[bash]

Edit Editing post fail. Instead of editing my previous post I added another one. Please delete

[Dustin2128]

I saw that before you edited it! :p :D Why the drop?
I don't have 300k :(

[formaldehyde_spoon]

Where did I said anything about positive? I merely said that &quot;the reactions throughout the web seem to agree that at least the technical reasoning is ok&quot;. Which is exactly what the the comments do, they agree he raises a few valid points from a technical perspective, but don't share the same view on all of his conclusions.

But enough of this now.

No sorry, you didn't actually say 'positive' ;)
I didn't really see any agreement on technical reasoning in those examples, just absence of disagreement.
But I've had enough of this too...:p

[foxmulder881]

Geez, I was only absent for less than 2 days and this thread has exploded! :P

[Frenziefrenz]

You can already have WebP in current browsers that support WebM.

http://extendopera.org/userjs/content/webp-support
http://antimatter15.github.com/weppy/demo.html

Enjoy. ;)

[ron999]

...

[Oxwivi]

Why can't things be like this more often? =[
http://ubuntuforums.org/attachment.php?attachmentid=170931&stc=1&d=1285901511
+1

Linux is more important than any Windows!

[ron999]

Hi
There's a beta version of XnView that will display WebP images.
It's for Windows and it seems to work OK with Linux and Wine.
Information here:-http://newsgroup.xnview.com/viewtopic.php?f=60&t=21031&p=87405

[alexan]

so, let me guess Google future plan.

WebM - Web Movie
WebP - Web Picture
WebOS - ---taken---
WebA - Web Audio
WebX - Web Software
WebV - Web Voip/Voice calls
Web? - ???Web pointless???

[renkinjutsu]

so, let me guess Google future plan.

WebM - Web Movie
WebP - Web Picture
WebOS - ---taken---
WebA - Web Audio
WebX - Web Software
WebV - Web Voip/Voice calls
Web? - ???Web pointless???

WebOS (WebX) - they already have ChromeOS (and they already think html5 / AJAX is their baby)
WebA - They already acknowledge Vorbis (and use it in webm)
WebV - They already have a really good voice implementation (I think they use On2 stuff.. not sure)

[alexan]

WebRight