What is Digg?56 Comments
- Lagged2Death, on 10/12/2007, -0/+7I'm skeptical. There's something about data compression (like perpetual motion, or water-into-fuel schemes, etc) that attracts cranks, charlatans, scam artists, and miracle-seekers.
There have been quite a few announced "breakthroughs" in data compression that never went anywhere. Remember the Windows RAM compression utilities that were supposed to be almost as good as doubling your PC's RAM, but which turned out to do nothing? Or the announcement of Stacker-like HD "expanders" that (supposedly) worked recursively - cutting the size of your data by half, then running again to cut it to a quarter, and so on for as long as you wish?
The association with the medical profession is suspicious, too. Medical systems aren't built with super-duper latest-greatest tech, but with proven tech, for obvious reasons. The most obvious early-adopter of an algorithm like this would be the Web, with next-generation video (no need for HD-DVD or Blu-Ray!) a close second. If none of that's happening, then associating with doctors (a notoriously gullible bunch, considering their education level) looks a lot like a stab at scoring some much-needed credibility. - chmilar, on 10/12/2007, -0/+7Looking at the white paper, the only part of their algorithm that is explained is the so-called BitMatrix. This is merely variation on "run-length encoding" (RLE). For RLE, you save a pixel value and a count of how many pixels the value continues for. When you get a new pixel value, repeat the process.
The BitMatrix does the same thing, except that a 1 is stored for each repetition, and a 0 to indicate a new value. This is more efficient than RLE for runs shorter than 8 pixels, and less efficient for longer runs.
For runs of one or two pixels, BitMatrix is less efficient than no encoding at all.
For the type of monochrome x-ray image they show, a run-length type of encoding will be very efficient. The image has long runs of black. In fact, for the x-ray image, a standard RLE will be _better_ than BitMatrix.
However, in a more "typical" photograph, run-length encoding is not very good. The pixel runs are usually too short for RLE to give much compression.
Then, the white paper becomes very unclear. They seem to imply that the resulting BitMatrix and PixelValues are compressed using some other, unspecified algorithm. There is also some blubbering about quantization of floating point, which is off-topic since they are dealing solely with integral pixel data to begin with.
Since the BitMatrix is _not_ a high compression algorithm, the Unspecified Algorithm must be very magical!
It really looks like they have merely developed a variation on run-length encoding which works well for a specific type of image (x-ray). It would not be very efficient for more general images.
The rest of the paper is just marketing-speak, trying to make a simple algorithm sound revolutionary. - oepapel, on 10/12/2007, -0/+4"I predict this gets bought up and buried by some corporation. Why? Well if that little extremely overpriced memory stick that comes with your digital camera suddenly is capable of holding 15 times as many photos then you wouldn't have to spend an ungodly amount of money on a larger, actually useful memory stick."
Are you high? It's a hoax. The claims are unable to hold up to scrutiny. They aren't the first people to claim magical compression. Everything they have shown is useless and certainly not as good as JPEG or even standard lossless compression (pkzip, and others).
Nobody will buy a company without due diligence. That includes evaluating their assets of which in this case is intellectual property. No sane person with even a passing acquaintance with compression will EVER say that this stuff has any merit.
Go home. There's nothing to see here. Move along people. - inactive, on 10/12/2007, -0/+4------------------------------------------------------
Welcome to the 2nd edition of In-Digg-Nation.
------------------------------------------------------
The in depth editorial analysis and commentary
of Digg stories IN Digg Stories.
------------------------------------------------------
If you knew anything whatsoever about compression you would have realized that this story is fake. You cannot compress a file, with lossless quality, smaller than a compressed lossy file. File size reduction is the result of compromises made with the level of detail you want to preserve. To suggest that this new compression algorithm not only compresses images 15% smaller than the JPEG algorithm and preserves the original image is impossible. - oepapel, on 10/12/2007, -0/+3The description in the PDF file on the website is pure nonsense. Their claims of discovery in the "technical" explanation are a joke! They are trying to pass off run length encoding plus a little networking mumbo jumbo as a breakthrough. The numbers that they quote are meaningless. JPEG and lossless JPEG are orders of magnitude more complex and effective.
This is pure snake oil.
This guy is just trying for his 15 minutes of infamy. - chmilar, on 10/12/2007, -0/+3A quick google reveals this choice quote:
'ABO, Mr Thiagarajan explained, does not eliminate data. "On the contrary, it achieves significantly higher compression ratios by value-adding to data in such a way as to permit superior speed and security without data degradation."'
The phrase "value-adding to data" should make you want to run away. Quickly!
Another quote: "ABO is an optimisation algorithm with very high compression ratios at mathematical lossless quality."
Note the phrase "mathematical lossless quality". It sounds like the algorithm is not actually lossless, but rather "appears to have the same level of quality as lossless", ie. it changes the data, but the changed data is statistically similar to the original. - oepapel, on 10/12/2007, -0/+2"REPETITION CODED Compression invented by MatrixView assigns a value of `1′ for successively similar values (pixels) in an image."
Also known by the much more common name Run Length Encoded or RLE!
No invention here, your Jedi Mind tricks won't work with me! - sleepless, on 10/12/2007, -0/+2demo page to upload your 1bit/8bit tiff file to test...
http://dv1.matrixview.com/Compress.asp - Lagged2Death, on 10/12/2007, -0/+2Wow that white-paper is like a nonsensical entry to a computer-science name-dropping competition. They start out by talking about the OSI 7-layer networking model. WTF?
Some of us have joked about a porn angle - well, they managed to get Lenna (Miss November 1972 - http://www.cs.cmu.edu/~chuck/lennapg/lenna.shtml) into their white paper. So they're ahead of you on that one. - oepapel, on 10/12/2007, -0/+2"Heck, RLE would work incredibly well on X-Ray images."
Actually, it doesn't work at all since the RLE file size is usually the same size or larger than the original uncompressed file size. X-Ray images are high bit depth grayscale images (10-12 bits) and have a relatively high poisson-type noise component. That means that it is very unlikely for 2 adjacent pixels to have the exact same value. - pfister_, on 10/12/2007, -0/+1Sounds like a hoax to me. I see nothing on the MatrixView website to contraindicate. No details on the algorithm anywhere.
- VickieDavis, on 10/12/2007, -0/+1It seems it only works with Please note that only 1-bit B/W or 8-bit grayscale uncompressed TIF images can be used." So the web site said when I tried to compress a jpg picture.
- owlet, on 10/12/2007, -0/+1I can't believe they found funding. The limits of image compression have been proven ages ago.
The only really usable development is to optimize for specific sets of images, like x-ray images or finger prints.
Generic image compression has hit its peak with the solutions used in PNG, JPEG and JPEG2000. (I really like JPEG2000... to bad it didn't become popular.) - theantidote, on 10/12/2007, -0/+1@dude3609: Read the article. It would revolutionize the healthcare industry who need to send millions of high quality MRIs and such to doctors and other hospitals.
- hax0r, on 10/12/2007, -0/+1Lovely... seems like a bunch of crap to me!
HTTP 500.100 - Internal Server Error - ASP error
Internet Information Services
Technical Information (for support personnel)
* Error Type:
(0x800706BE)
/Compress.asp, line 226
* Browser Type:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.10) Gecko/20050716 Firefox/1.0.6
* Page:
POST 750915 bytes to /Compress.asp
* POST Data:
error '80020009'
Exception occurred.
/iisHelp/common/500-100.asp, line 223 - lazerfred, on 10/12/2007, -0/+1Note the source: Business 2.0, which is the lamest source of McBusiness News. Pretty much everything they publish is guilty until proven innocent in my eyes. So I too call BS.
- zone, on 10/12/2007, -0/+1I'd like to test it myself, a demo or something, but the PDF on their website is the only interesting thing I saw so far..
http://www.matrixview.com/
http://matrixview.com/files/ABO%20white%20paper.pdf - digital.dreamer, on 10/12/2007, -0/+1That would be nice to see this technology work with digital cameras and other devices. Good technology for camera phones. Hope this goes somewhere.
- WillyWonka, on 10/12/2007, -0/+0It's better then JPEG but is it better then JPEG 2000? Also, can you decompress it? :) I can get 100:1 compression with my algorithm. I just haven't figured out how to decompress it yet.
- stanleyfresh, on 10/12/2007, -0/+0Hmmm.... that much smaller than a JPEG but lossless? Pure awesome. This revolutionizes not only the medical industry but the pr0n industry. Now I can store 15 times the T&A, lol!
- procras, on 10/12/2007, -0/+0business 2.0 is a reliable publication, i would trust that this is not a hoax, even if we don't have great details about it
- pondster, on 10/12/2007, -0/+0"uhmm.. what industry what it revolutionize? :P"
Uhmm... if you read the article you would see the medical industry for starters lol
:-) - spyres, on 10/12/2007, -0/+0I seem to recall this as a hoax from a while back.
- antiTRACE, on 10/12/2007, -0/+0"uhmm.. what industry what it revolutionize? :P"
Prob. banking too, taking pics of old fashined checks take a lot of resources at a large level - Lagged2Death, on 10/12/2007, -0/+0Whoops, sorry, here's a working Lenna link:
http://www.cs.cmu.edu/~chuck/lennapg/lenna.shtml - rhyno2000, on 10/12/2007, -0/+0"Right now it's negotiating with chipmakers to embed the technology in cameras and fit more files on storage cards. MRIs today, vacation snaps tomorrow"
Can't arrive soon enough! Because I can only fit 650 5MP highest-quality .JPGs on my S2 IS...
Seriously, it's cool stuff, especially the fact that it's lossless. But it will take years to arrive, and that's only in niche industries. Forget about seeing this in any consumer product in the next five years.
We were all supposed to be using JPEG2000 & MP3Pro by now as well... - davidg, on 10/12/2007, -0/+0Hmmmm.... in the link to the LinuxWorld Australia article the chief scientist says that it works by eliminating identical pixels and looking them up from a common place. Isn't that just a palette? And isn't a palette mostly useless for the kind of photographic images that JPEG is used for?
He also says that they can compress beyond the theoretical limits. Presumably the ones from information theory?
Sounds very iffy to me. - SkeletaLlama, on 10/12/2007, -0/+0Why does everything have to revolutionize an industry? Can an industry honestly be revolutionized twice a year or so? Wouldn't that just be progress and not a revolution? This isn't electricity or e=mc2, it isn't going to revolutionize anything.
- chmilar, on 10/12/2007, -0/+0If the algorithm is truly "universal", it could be coded into library, just like libjpeg or libtiff, which could easily be easily distributed and added to any application (Photoshop, image viewers, etc.).
The talk about "targetting" healthcare companies sounds very suspicious. Either the inventor is hiding something, or he has a very bad, greedy business plan, which is likely to fail.
If the algorithm is as good as he says, he could be licensing a library to Apple, Microsoft, Adobe, etc. - LeegleechN, on 10/12/2007, -0/+0Sounds like a hoax to me, especially with the way they describe the algorithm. Remember folks, yoyu can't get something for nothing! All photographic images have some amount of uncompressable entropy.
- TKDWILSON, on 10/12/2007, -0/+0""""""@dude3609: Read the article. It would revolutionize the healthcare industry who need to send millions of high quality MRIs and such to doctors and other hospitals."""""""
Wow. I don't think I want a doctor looking at my MRI under compression no matter how good it is supposed to be.
Eric Wilson - Otto, on 10/12/2007, -0/+0It's quite possible that they have developed a compression method that is better than JPEG and lossless, when you use it on specific types of images. That's not revolutionary. Heck, RLE would work incredibly well on X-Ray images.
If they're claiming a general purpose algorithim for images though, then they're definitely full of *****. - digitalhome, on 10/12/2007, -0/+0MatrixView is listed on the Australian Stock Exchange.
http://www.asx.com.au/asx/research/CompanyInfoSearchResults.jsp?searchBy=asxCode&allinfo=on&asxCode=MVU&companyName=&principalActivity=&industryGroup=NO
Certainly made some good gains from the IPO price a year ago. - Trepan, on 10/12/2007, -1/+1Sounds like a hoax to me. I see nothing on the MatrixView website to contraindicate. No details on the algorithm anywhere.
pfister_ posted by pfister_ (0)
contraindicate
v : make a treatment inadvisable [ant: indicate]
nice usage of the english language - noof, on 10/12/2007, -0/+0No, you can't make an image 15 times smaller (sizeof(jpeg) < sizeof(raw), 15*sizeof(abo)=sizeof(jpeg) => sizeof(raw) > 15*sizeof(abo)) without loosing information. For example, how would you possible compress an 15x1 image with totally random colors into 1 color-triple? You can't!
I would say that it's pure BS:
".. ABO restructures these algorithmic data structures in such a way that they are greatly simplified, resulting in hitherto unachievable levels of data compression."
And btw, don't forget the "adaptively optimized binary data transmission". I would like to see a large network running adaptively data transmission in real time at 100mbits :) - wilsonics, on 10/12/2007, -0/+0I guess they worked so hard on the compression, they forgot to watch for aspect ratio of the images on their site:
http://www.matrixview.com/en/solutions/bg_menu/solutionsmenu_r1_c1.jpg
dude's head looks a little stretched
;) - TimmyK., on 10/12/2007, -0/+0I predict this gets bought up and buried by some corporation. Why? Well if that little extremely overpriced memory stick that comes with your digital camera suddenly is capable of holding 15 times as many photos then you wouldn't have to spend an ungodly amount of money on a larger, actually useful memory stick. They would lose lots of money on their insanely priced memory cards and sticks.
- oepapel, on 10/12/2007, -0/+0Which JPEG quality are they comparing to? You canchange the "Quality" parameter when saving a JPEG and get a bigger or smaller file. Heck, I can compress it down to 1 byte if you are happy with a Quality of 0.
The only thing these guys invented is their brochure. - binarypower, on 10/12/2007, -0/+0Vaporware at it's finest
- geminitojanus, on 10/12/2007, -0/+0I call *****. They give no demo, they give no explaination of how the technology works, and they're not shipping yet. Biggest vaporware ever.
- oepapel, on 10/12/2007, -0/+0"Wow. I don't think I want a doctor looking at my MRI under compression no matter how good it is supposed to be. "
Too late! Compression is ALREADY used on these images. But not all compression is bad. Lossless compression is perfectly fine when it is transparent to the process. No data is changed with lossless compression. Even Lossy compression CAN be fine as long as it is applied correctly and it's effects minimized. A single pixel is rarely useful. In fact, most systems compensate for single pixel anomalies. Fine structure is what is usually considered important. As long as that is not compromised, compression can help remote and/or slow links in a system.
You just need to know what you are doing. - mopenstein, on 10/12/2007, -1/+0shannanigans!
- spyhunter, on 10/12/2007, -2/+1FAKE
The only news here is that Business 2.0 was suckered in. - surfing, on 10/12/2007, -1/+0thank goodness, faster porn!
- bacon_skoda, on 10/12/2007, -1/+0cellphones
- dude3609, on 10/12/2007, -1/+0uhmm.. what industry what it revolutionize? :P
But i know what ya mean. heh - naing_oo_1, on 10/12/2007, -1/+0that is great!
good times - aggrazel, on 10/12/2007, -1/+0I call BS
- TheRepublic, on 10/12/2007, -1/+0interesting enough. If it's true its a new img revolution. :D
- eigh, on 10/12/2007, -2/+11. hate to say it, but ive read this here before
2. singapore company? im pretty sure the last time i read this it was proven that these guy were full of it.
3. they are full of it -
Show 51 - 55 of 55 discussions
Check out the new & improved 
© Digg Inc. 2009
— Content created and posted by Digg users is 