Slashdot Mirror
JPEG Patent Challenged
ChocLinux writes "The Public Patent Foundation has filed a request at the US Patent Office to revoke Compression Labs' data compression patent, which it is reportedly using to harrass anyone that implements the JPEG format. 'CLI's aggressive assertion of the '672 patent is causing substantial public harm by threatening this international standard on which the public relies,' says Pubpat in its filing."
then beware of geeks bearing GIFs.
Who wasted time chasing this while nearly putting themselves out of business. How about focusing on some real products???
Get this patent overturned. It's extremely important to get these ridiculous technology stifling unoriginal patents overturned.
Where the hell is EFF on this? Pubpatents is getting my money this year and I recommend you guys donate there as well if you are into donating to tech freedom.
... was the prior art in JPEG format?
How can a post be modded "overrated" or "underrated" when it hasn't been rated yet?
Is it possible that if JPEG patents were enforced we would actually persue the use a PNG (a significantly better format). As a web developer, if I could rely on people being able to see all the different derivations of PNG, life and design would be much easier.
zork% mv *.asp
283 files eaten by a grue
Compression Lab Inc:
"All your pr0n are belongs to us"
This is covered in details over at Groklaw
Help fight continental drift.
Not always. Sometimes the file size can go from a few hundred KB to two or three megs, which is more than a bit, especially if you're putting the images online where they will be accessed by people on dial-up. Also, for some things like photographs, lossless compression isn't always necessary. So while PNG would make a good fall-back option if JPG wasn't available, it isn't close to a perfect replacement.
Gee, where have I seen that before?
As a small added bonus, PNG was the first W3C Recommendation (in 1996) (see the REC). It came well before their HTML 4.0 (1998). But I guess Slashdotters already knows such things... ;)
You can hold down the "B" button for continuous firing.
Actually, IE does not render PNGs with transparency properly. The current version in common use (6.0) renders transparency in PNGs as a gray background.
"Think about how stupid the average person is, then realize that half of 'em are stupider than that!" - George Carlin.
Um...did you even read the wikipedia article you referenced?
From your post:
Unlike JPEGs, PNGs can be lossless
And from the article:
PNG (Portable Network Graphics) is a lossless bitmap image format.
By saying PNGs can be lossless, you imply they can be lossy as well...which is not what they were designed for.
From your post:
Sure, they tend to be a bit larger than JPEGs, but I figure the gain in quality is often worth it.
And from the article:
Using PNG instead of a high quality JPEG for such images would result in a large increase in filesize (often 5-10 times) with negligible gain in quality.
And finally, from the article:
PNG was not intended to replace the other popular web image format JPEG.
PNG is intended as a replacement for GIF, not JPG.
Hope this clears things up.
____
~ |rip/\/\aster /\/\onkey
Here's what. What happens when someone claims rights on PNG too after everyone starts using it? What we have happening here is the patent board is unable to understand prior art and granting patents on obvious and pre-existing technology.
The issue is not this particular patent, but the issue is blocking other corporations and lawyer clusters from trying to gain broadly worded patents that incorporate pre existing technology from obscure sources so they can make money.
If, and I say IF the patent is valid from a 'no prior art' and is not intuitively obvious, then the complaint is rather suspect.
... also asks the US Patent Office to take notice of Compression Labs' "aggressive assertion" of its patent, which it says is causing substantial public harm."
"The Public Patent Foundation
If the patent is valid and the public has used it regardless, then they are within their rights (legal and quite likely moral) to defend it.
"People who do stupid things with hazardous materials often die." -- Jim Davidson on alt.folklore.urban
The problem seems to arise when a patent is sold. It seems like the patent rules should be changed so that patent infringement can only apply when the original party filing the patent has been harmed. It makes sense to protect the inventor - that's what patents are for. The problem seems to arise when the party who is claiming infringement is not using the patent to generate revenue (excluding law suits). It seems like there should be a "minimum reasonable usage" clause in the patent law. By "minimum reasonable usage" means - you as a patent holder are using the patent in your livelihood or the corporations livelihood.
This is dumb, and some would argue anti-competetive monopolistic behaviour. You have a patent on something cool. You let people use it without any royalties; it becomes popular. Really popular. Then, all of a sudden, you start charging royalties, and everyone is trapped. It would not have become that popular if royalties had been there in the first place.
This is reminiscent of two things: Microsoft (slightly different modus operandi), and drug dealers (the first one's free kiddies).
Should be that if you don't enforce your patent within a reasonable time frame, you lose the right to do so. In a perfect world. Which we are far, far, away from.
"I think it would be a good idea" Gandhi, on Western Civilisation
I assume that they filed under the current system since, if the old system of 17 years after the granting of the patent were in place at the time of filing, the patent would have already expired over a year ago.
U.S. patents are issued with a term of grant + 3.5 years. Three renewals are available: grant + 7.5 years, grant + 11.5 years, and the full term. For this and other U.S. patents subsisting as of mid-1996, when the Uruguay Round Agreements Act came into effect in the United States, the full term is the longer of filing + 20 years or grant + 17 years.
Conversion from truecolor to indexed PNG is lossy.
And so is conversion from truecolor to 256-color BMP, and so, in fact, is cropping a BMP to include the middle 27.6% of the image. Is BMP lossy?
Of course not. You can do things to any file to lose information, and reducing the number of colors in an image is obviously one of them. You can't say text files are lossy just because you can convert a Unicode text file to ASCII and lose some characters in the process.
Then what Free format was designed to replace JPEG?
I don't think there are any formats comparable to JPEG currently, and I know that there certainly aren't any in widespread use. That's the whole point of this article. If JPEG users get screwed, then we're in serious trouble. Think about how many photos on the web are in JPEG format, and how many photos on people's computers are JPEG.
Not exactly. In fact JPEG's and GIF's are targetted at two different kinds of images.
Gifs (and later png's) work better for images with large areas of constant color (cartoon type images) where JPEG's are better for photgraphic like images where the the shifts in color are more gradual.
Also JPEG is usually a lossy format (there is a lossless mode, but it's essentially a totally different form of compression) where as GIF and PNG are lossless.
Using the wrong one can result in HUGE filesizes compared to using the right format for the job. Some apear to think a JPEG will always be a smaller final file because lossy should be smaller than lossless, but for drawings and cartoons this is often false. I've seen images (real images from real sources, not some 'ideal' image, or cherry picked image) that are much smaller in png than jpeg unless you turn the quality on the jpeg incoder so low that you can't tell horse from a house.
Mycroft
https://signup.leagueoflegends.com/?ref=4c3ed6600b6ea
The problem is most good image codecs are a patent mindfield [e.g. wavelets]. You can still do things like 5/9 [or whatever] and then typical entropy encoding. I think ... not 100% on top of the graphics scene.
Though a simple Haar wavelet can be effective [and with a tweak lossless].
Actually you can perform bincoding and/or lifting to most domain transforms [e.g. DCT] and wavelet based codecs to get a transform that works with integers only and can be lossless. The "bindct" papers of a few years ago are a good example. They showed how to do DCT type IV [i think, whatever JPEG uses] using only integer transforms [shifts,adds,subs] that got coding gains close to the traditional DCT.
For raster images PNG is as good as it gets at the moment. You could do a block sorting codec to get a slight better compression ratio but not by much [and it wouldn't be good for progressive images].
As for truecolour images there really isn't much unfortunately.
Tom
Someday, I'll have a real sig.
This type of stuff only reinforces the need to free software patents and helps the intiatives spearheaded at nosoftwarepatents.com
\
FWIW, Adobe's PDF includes embedded images using DCT or discrete cosine transform for compression. If you extract the image sections from the document, you end up with a JPEG file.
.... however it is copyrightable.
Copyrights last alot longer.... infinity according to the changing of laws to extend then before they expire..
Now the thing is, can you write an algorythim/program to do the same thing in such a manner that a picture compressed by it can be uncompressed by jpeg engines, without infringing upon the copyright?
if you can't then....still do the right thing and oppose the fraud of software patents.
Patent law needs to be made like trademark law: Use it or lose it. Basically if you are going to patent something, you need to either exert that patent, or the patent should lose standing. I would say something like if you fail to contact a company for a period of 6 months after a product using your patented technology is marketed on a level that you should be reasonably aware of it, you lose your patent.
So you are still protected, if someone is just developing something in secret, no problem, even if someone releases a product is a very small market that you wouldn't reasonably know about, no problem. However if your patent is in a product being sold on a wide scale, you have 6 months starting from when it is to contact the company. Failure to do so nullifies your patent.
I think that would keep it fair, such that people could patent and profit from ideas, but that you can't just patent something and wait for someone to be huge and then sandbag them.
Forgive me if I am wrong, but I went through the patent: Motion compensation, intraframes etc... isn't it the MPEG format we are talking about? JPEG is the static picture encoding format that is based on the cosin transform. MPEG also uses cosign transform but many other techniques among which motion compensation DPCM etc.
If you read the pubpatent filing, their main point is that an earlier patent, issued to the same company, is prior art for all the points in the '672' patent. The earlier patent was filed more than a year (plus one month) prior to the filing of the '672' patent, which makes it legally prior art.
Anyway, the sucker has less than a year to run, as it was filed in October, 1986. Probably why the lampreys at Forgent are pushing so aggressively. It'll only be a cash cow for another 11 months.
Interestingly, I could have been a target of Unisys, except they couldn't have gotten much blood from this stone. I was the original author of the "compress" program, which turned into an early "open source" effort (although the term hadn't been invented at the time). Compress was an implementation of LZW, based on Welch's 1984 paper in Computer. Only later was I informed that it was patented. After it had been incorporated into Berkeley Unix releases and into the GIF format. I was happy when that patent finally expired, but I had absolutely no doubt of its legitimacy.
As for the claimed superiority of PNG over JPEG, I'd say it depends on the application. JPEG was designed precisely and specifically for the purpose of compressing photographic images. Such images
- Do not compress well using techniques like LZW and Huffman coding
- Have intrinsic variation in pixel values due to noise in the recording process
- Don't have precisely straight and sharp edges
These characteristics make them poorly suited to lossless compression techniques, and also mean that a lossy technique will not degrade the image further than the original noisy recording method did. (Unless you turn up the loss level too high.)Because of the "if you have a hammer, every problem looks like a nail" principle, people have used JPEG in applications that it's not suited for -- applications where the lossy compression DOES degrade the image quality, and where a different method (LZW, for example) would in fact give a smaller file. Then other people point at these examples and say "PNG (or GIF) is better than JPEG!" My toolbox has hammers, screwdrivers, wrenches, etc. I try to pick the appropriate tool, and don't hammer with a wrench, for example. The same should be true of our computer tools.
Copyrights and patents aren't two names for the same thing. Inventions can be patented; the creative expression of an idea can be copyrighted. The idea itself cannot; see Feist v. Rural .
Laws do not persuade just because they threaten. --Seneca
We've seen this before.
When RSA got popular and people realized that it was patented, there was a large effort to switch to DSA. Right about the time that all the pieces of DSA support were in place, the RSA patent expired so people just kept using RSA.
When GIF got popular and people realized that LZW was patented, PNG was created. By the time PNG was actually supported more-or-less correctly in browsers, the LZW patent expired.
I suspect if this JPEG madness keeps up, people will try to switch to JPEG 2000 (which is still patented, but at least the patent holders appear friendly). But it looks like the JPEG patent expires around 2007, which does not leave enough time to switch to anything.
Think of all the Japanese digital camera manufacturers that sell cameras in the U.S.
Real products are a distraction for these people. Forgent got millions out of various companies without developing anything. Since the legal fees (costs) are much less than the licensing revenue, it's a self-perpetuating system. The RIAA settlements are the same way; each settlement pays for N new lawsuits to be filed and the profit rises exponentially.
PNG and JPEG fill different niches. They're different tools for different jobs. PNG is designed to be lossless and was meant as a successor to GIF rather than JPEG. JPEGs are a bit smaller because they throw out a lot of the information (hence are lossy) that isn't as high priority. It's a format for distribution rather than one that you'll want to work in while you manipulate images. For that you should probably be using TIFF or PSD. If you are constantly makinge changes to a set of images, the increase in storage space, while significant percentage-wise shouldn't make that much of a difference.
GIF, JPEG and PNG are meant for distribution of images over networks where banwidth usage matters. JPEG compresses photographs well. GIFs and PNGs are better for rasterizing text and vector graphics for diagrams and logos... usually things that aren't high in complexity but have a more extreme contrast than you'd find in a vacation photo. If you're seeing a big difference in quality, your JPEG compressor is probably set to being too lossy.
Technical issues aside it would definitely be nice to have an unencumbered lossy format to replace JPEGs but there are so many ridiculous patents nowadays that you'd probably invest most of your time figuring out what algorithms are safe to use.
CLI's aggressive assertion ...
I didn't know CLI has anything to do with JPEGs. I get pics in my CLI like this "cat ascii_art.txt".
w00t
They have already been challaged by many, but for the first time someone has a concrete case with 'prior art'. You can read more on this at The Data Compression News Blog
White Knight Charges Forgent http://www.c10n.info/archives/246
[Disclaimer: Shameless self promotion]
Software is indeed patentable. This was decided in the Diamond v. Diehr case over 20 years ago. Thousands of software patents have been granted since then, and the requirements have gotten weaker over that time. Whether they ought to be allowed is very debatable, but the fact of the law isn't.
"Diamon v. Diehr" is only law in the US and does not apply to rest of the world, which is what I suspect the other poster is trying to say. The long term effects that software patents will have on our country's economy is going to be devastating if our developers have to fear being sued for using ideas here that programmers elsewhere are free build upon.
Time is what keeps everything from happening all at once.
People at the USPTO need to go to jail for granting patents like these. They are crippling, willfully, the US economy. A bad patent has all the effect of sinking an oil tanker, and the effects are longers lasting. We still don't have encrypted email because of the willful sabotage committed at the USPTO by granting the RSA patent.
Throw a couple of examiners in jail. They'll read the next patent more thoroughly then.
The primary examiner on the patent was Howard W. Britton. He was either incompetant, or willfully neglegent when he granted this patent. If I make a mistake this serious at my job, I can be held seriously accountable for it. So should Mr. Britton.
I am in no way exaggerating or being sarcastic. In my honest opinion Britton deserves to go to jail for the damage he has done to the economy, and the disgrace he has brought on the entire patent system.
May the Maths Be with you!
Take a look at the real patent system,
When you take someone to court with a patent that will not hold up, not only is there patent tossed out, but there are penalties. The other party will also counter sue and make the holder of the bogus patent pay dearly.
I remember reading somewhere it's like 5x damages and expenses or something like that written into the law, just to make it risky to actually take people to court frivolously.
But what you miss is many lawyers are playing poker and not law.
By this I mean they will intimidate people into paying just so they can avoid a fight, but in practice they can't afford a fight any more then the other guy, maybe even more so. Just ignoring them, or counter bluffing probably wouldn't have any repercussions.
But you never know what cards they may be holding.
The problem is the Engineer's mentality. I know because I used to have it, and still do to a point. I have also had many partners with it.
They always play devils advocate, asking what's the worst cast scenario, how can I over engineer to make sure my bridge will not collapse, or my power supply will not burn down someone's house. Overly cautious.
But in law, and business and business law matters, a totally different approach is needed.
It's about Risk management, the risk reward tradeoffs, Poker, social interactions, and BALLS.
Most engineers panic with any legal things, always assume worst case, and constantly keep putting rules and boundaries where there really aren't any. (Thinking in the BOX) and making the box ever smaller.
"Oh is has to be done like this", or "you can't do that", "the law says this or that".
Well The Laws are only 1/3 of the legal story, Case precedence another large part. What did the courts rule in similar cases? How are these cases handled?
But most important are what are the practical reality of a given situations. Can these guys afford to sue, Can you stall the legal actions in the courts for 20 years till the guy keels over dead from old age. Can you even collect if you win?
I have come to find that many of these "business Guys" will blatantly cross a line. It's only the ones that do this as their full time MO for years before it will catch up with them.
In practice, it's like running a red light; it's only illegal if the cop see you. The differnece is police usualy have time, where District Attorneys are overwhelmed and can only go after the top 10% of what's out there, and only when everything is laid out simple even a five year old can what took place.
So if we had a race, the real business guys will cautiously run the lights, where the engineers wait till the light is completely green. The engineers will almost always loose the race.
I recently had a corporation in California. We had a business guy whom we gave a few shares to about two years into our startup, under 25%.
About a year later realizing that most of us in this venture after 3 years were low in cash, he just decided the company was his.
He closed down the office, took all the contracts, papers, and hardware home with him.
He had secretly filed papers with the state listing him as the sole owner and just deleted us like were never there.
GoDaddy even took the Domain and web site from me! (it was in my name)
And started to present my Inventions (patent pending) to Sun, HP and Intel as well as many investors without us!
We talked to the state, they just file corporate shareholder and offer filings, there is no verification or even protection from Fraud. This means in I could just file papers claiming I own Intel! The State would mindlessly file this, no questions asked. It's up to the courts to clean up any problems that my arise.
Of course someone like Intel would send hordes of Lawyers at
I am always doing that which I can not do, in order that I may learn how to do it. - Pablo Picasso
Huffmann coding was not considered patented at the time. The LZW algorithm had patent issues, but I reckoned at the time we could do LZW-compatible coding with a two-pass algorithm: the only unique feature of LZW was the reoptimization of the compression tables as the data was being processed.
There is a 1935 Italian patent on the compression of images, including colour images, for fax using variable length codes for different picture elements. The claims are sufficiently general the invalidate a lot of run-length compression schemes.
That's just what I can remember while writing this. I am not impressed by their patent. There are lots of dud patents in there. It is easy to search the patent database, and become convinced there is a claim. Look at prior art in products and in theliterature, and it is a different story.
PNG is lossless, so it looks good. JPEG is lossy, so it's small. You can get much better compression on photographic (lots of colors, few uniform color areas, few if any geometric features--this is only my crude layman's impression) images using lossy compression with JPEG, and furthermore, it's adjustable so if you need less quality, you can recover more bandwidth. PNG can only compress such images so far, and no farther.
By contrast, PNG can represent alpha transparency, so (if your browser supports it.. hope IE7 is out soon) you can get neat effects with PNG. PNG is also great at presenting images, created with vector graphics, to software that doesn't do vector rendering. That means logos on web pages, line drawings of all sorts such as scientific plots, etc. But the reason it's good for such things is that such images benefit far more from lossless rendering (than photographs or 2d art or the like).
PNG is great technology, but you can't simply put aside JPEG yet. Like PNG, JPEG is good at what it does, what it's intended for.
It's rare that you're presented with a knob whose only two positions are Make History and Flee Your Glorious Destiny.
Or better yet, make the patent examiner financially liable for wrongfully issued patents. If it's overturned, the restitution costs and penalties come out of his paycheck. That would likely be a stronger threat than jailtime, and would also help reimburse those injured by bad patents.
~REZ~ #43301. Who'd fake being me anyway?
Actually that's not exactly correct, GIF can support more than 256 colors, it's just that using the fomat to do so requires a bit of 'finess'. That and the fact that for the longest time most programs didn't support creating >256 color images (why bother when most computers couldn't display them).
If your currious about it, or want evidence the try these two links: http://phil.ipal.org/tc.html and http://en.wikipedia.org/wiki/GIF
If you want to be pedantic about it ANY digitized image is lossy in the manner you speak of (though at about 21bits or so most of us can't tell with just our eyes), nature's bpp AND resolution is much higher than scanners and thus even at the bit-depth of high end scanners and digital cameras some data is lost. At some point (several usually) between the original light bouncing around and the stored image file you WILL lose data. Lossless generally means the data put into the compressor can be exactly reproduced by the compressor.
Mycroft
https://signup.leagueoflegends.com/?ref=4c3ed6600b6ea