Guest post by Google has released a new set of tools for the WebP Image format. The release includes a completely rewritten encoder with improved image quality. Also available is a Windows Imaging Component providing WebP import to any application that supports WIC such as Microsoft Office 2010. There is a side-by-side comparison between the new encoder, the old encoder and JPEG here.
Isn’t yet another web aimed picture format. What we do need is replacement for TIFF. Why? Because TIFF is only picture format that is generally specified in archiving standards. Another format is PDF 1.5 which isn’t very good either. I been waiting proper XML-based image format since Microsoft released XPS or atleast something that has proper custom metadata format. Problem is that XPS isn’t aimed for long time archive purpose at least Microsoft isn’t doing much to push it in archiving standards.
Well, I wouldn’t pretend to know ‘what the world needs’ when it comes to image formats, but I don’t think webp will make a splash unless the improvements over Jpeg become dramatic. From my own (now outdated) tests webp is better when using low quality, with higher quality there was very little difference.
You know, PNG-MNG-JNG covers almost everything about raster images, and SVG for vector images. The rest, PDF-PSD-whatever is gross and useful only on specific cases (proprietary format)
Kochise
PostScript is a powerful interpreted programming language, and it’s a really good way of producing quality printed material.
It’s too bad that people think that it’s unreadable because they look at generated code and say “look! it’s garbage!” well, you probably couldn’t read the intermediate code spat out by most compilers that have concern for neither speed nor readability, but rather accuracy.
Also, PostScript files aren’t proprietary. Very easy to open one up in any raseriser (including a capable printer) or even a text editor (but again, generated code isn’t easily parsed by humans).
It’s too bad that Adobe’s PDF reader was free, and their implementations of PS were not. Then again, big companies have always wanted a stranglehold on anything they produce, so they would have used PDF anyway (not that it’s at all hard to bypass PDF locks).
*sigh*
Money spoils technology again.
Edited 2011-02-27 00:37 UTC
I hope you don’t mean PDF is a good format ?
It’s fileformat is possible more complicated then those from Microsoft Office.
It is not even possible to make a working/valid parser, because the format is to arbitrary:
http://www.youtube.com/watch?v=54XYqsf4JEY
If you are on Windows and have Acrobat installed and you understood the video. I suggest disabling the acrobat-plugin in your browser if you haven’t already.
Edited 2011-02-27 02:04 UTC
PDF used to be a decent document format for its time, and Adobe provides free patent licenses for implementations which makes it infinitely superior to PS which is patent-encumbered. Some of the recent additions to it for tagging and accessibility are good.
Problem is people wouldn’t pay for Adobe products if they only released the archiving PDF standard. Adding Flash and other crap means that they can have features over other readers and editors.
But make no mistake, if PS was better, that it wasn’t, it is massively slower and otherwise unsuitable for document exchange, they would have done the same to it, and PS is a full programming language by itself so it would be even worse.
Why pay any attention to nonsense software patents that have gone unenforced for _decades_, and are thus unenforceable?
They have a responsibility to defend their IP if they want to retain ownership, and GhostScript has been around for many years. I’m willing to bet it has a larger installed base than Adobe’s implementation.
Also, links for any evidence of any of this, and why anyone should care?
“Why pay any attention to nonsense software patents that have gone unenforced for _decades_, and are thus unenforceable? ”
Patents just don’t work that way. If I am a patent holder, I can wait till even a year before the patent expires to begin enforcing it. There is no requirement to actively protect it unlike say a trademark.
…but I don’t see the slightest bit of validity of any software patent.
“…but I don’t see the slightest bit of validity of any software patent.”
What you see or not see is entirely irrelevant. It is the question of what the courts would see in a patent lawsuit or what the organizations involved in such technology would deal with their asset or how other organizations would respond when faced with a patent risk would do that is relevant.
It’s nigh impossible to sue everyone who’s using ghostscript.
Java is at a much greater risk, and it’s… awful.
Seriously, it’s foolish to claim patents on the implementation of a programming language.
“It’s nigh impossible to sue everyone who’s using ghostscript. ”
You don’t have to sue everyone. Just some high profile users.
Please don’t skim my posts then reply to them.
Read them, or don’t reply.
How do you find the image comparisons? I really want the webp format to be good, but i can’t say i am blown away or even all to impressed by the comparisons.
With the same quality setting i find the webp blurrier (still about half or 3/5 of the file size)
When i match the file sizes, i don’t know which one i prefer, for example
http://webscaws.x10.mx/webp/RAW_CANON_S5IS.CRW/jpeg/jpeg/RAW_CANON_…
http://webscaws.x10.mx/webp/RAW_CANON_S5IS.CRW/libwebp-0.1/png/RAW_…
In the original you can clearly see that some grasses are “striped” vertically. In the jpeg image i see more artifacts, but i still see more of the stripes in the grasses, while the webp one only shows green-without-detail. ( i also suggest watching example 0 with the clouds)
I suspect it might be a matter of personal preference (“can you try to look… blockier?”), just as some people prefer the artifacts of 128/192kbit mp3 to cd-quality flac.
Edit:
But holy smokes, check out the low qualityies. Webp beats jpeg hands down.
Edited 2011-02-26 18:19 UTC
I believe that is the real point of webp
I agree but I think you can see some improvement between the older webpconv encoder and the new encoder in terms of sharpness (less blurriness).
The new encoder also provides settings that can be used to specify different filter strengths. All the examples on the websca.ws page are encoded with the default settings.
JPEG XR is a far more advanced format and it has the adventages of an open standard. Unfortunately, at this time there are is no open source decoder partially because of the unfavorable license terms of the reference decoder. It would be very nice to see the broader adoption of this standard but it would require a good quality FLOSS encoder and a well designed fast decoding library (like libjpeg).
Hmm, looks new webp is pretty good improvement over previous (old) one. It is also sharper and smaller than jpeg. But it looks new version also introduced slightly more block artifacts, but they are still much more smaller than in jpeg.
Any way, i still think webp idea is bad idea. It should be compared first with improved jpeg (bigger DCT blocks, state-of-the-art encoders/decoders for jpeg, which makes really big difference, and include deblocking loop in decoder to remove this artifacts). I also think webp should be compared with wavelet based formats, most importantly PGF, which is fast, supports also alpha channel, progressive download, multiple color spaces and color depths and is free and open. I think if we want new image format it should be PGF, not webp. And will not be for just introducing webp becuase it is better than jpeg. It is better, but we can get MUCH more better, and if we want to go throught the pain of implementing this in all webbrowser, and tools, please wait, think deaply, and introduce just one format, not one, and next year, next one, and next year, next one, because somene forgot to support some obvious futures.
Edited 2011-02-27 14:32 UTC
since I have taken the time to work on image distortion by jpeg I would like to see
1. Distortion variance versus pixel intensity
2. Spatial correlation of noise
3. Inter channel correlation
4. Quality versus percentage of original file size
for both webp and jpeg.
Or even better I can start evaluating on my own