Slashdot Mirror
Microsoft Move to be the End of JPEG?
jcatcw writes "Microsoft Corp. will submit a new photo format to an international standards organization. The format, HD Photo (formerly known as Windows Media Photo), can accommodate lossless and lossy compression. Microsoft claims that adjustments can be made to color balance and exposure settings that won't discard or truncate data that occurs with other bit-map formats."
Not going to end jpg - everyone dissatisfied with JPG is already using RAW. Everyone satisified with jpg will stick with jpg.
This is going to enjoy the same sort of limited uptake as jpeg2000 vs jpg, mp4/wma/ogg vs mp3, png vs gif, etc.
Few other things to note:
1) The 'HD' doesn't stand for High Definition, it's just there to get the association with HD TV in consumers minds. *rolls eyes*
2) This technology is patented to the hilt & the licensing terms for the HD Photo Device Porting Kit 1.0 licensing terms specifically exclude copyleft (GPL style) licenses.
There are shills on slashdot. Apparently, I'm one of them.
I predict it will succeed in displacing jpg just like png displaced both gif and jpg.
If you're not discarding data when you're adjusting color-balance and other settings, you're by definition not compressing as much as you possibly can.
For example, if I desaturate a photo I'm throwing away tons of color information. If that color information is still being written to the file, the file isn't as small as it could be.
Aside from that, PNG should have dethroned JPG long ago for the very simple reason that it contains an alpha channel -- but I still see plenty of JPG's.
------ The best brain training is now totally free : )
JPEG and PNG are fine, if we want a HDR capable lossless image format we'll use OpenEXR (No George, we still don't forgive you for Jar Jar). Why do Microsoft have to keep re-inventing the wheel? OpenEXR has mad force powers, Microsoft image formats smell like Ballmers toe nail clippings. What have they patented or what DRM switch and bait are Microsoft trying to pull with this move?
The PNG format uses the DEFLATE compression algorithm to minimize its data size. DEFLATE is the same compression method used by gzip. We all know that for larger files, the bzip2 compression utility tends to obtain better compression ratios than gzip. So would it not be possible to use the bzip2 algorithm instead of DEFLATE when compressing the image data, to obtain a smaller image file size at the cost of greater compression and decompression times?
PNG is a replacement for GIF, if anything. JPG files are much smaller than PNG files for typical photographs (though can be smaller for line art and the like), which will always leave JPG as the favorite much like FLAC isn't replacing MP3 anytime soon. The alpha channel in PNG is absolutely a nice perk, but thanks to the dim people at Microsoft never supporting it right until IE7, there wasn't much benefit over using GIF files. (Even though PNG did bi-level transparency just as fine as GIF files - even better, you didn't lose 1 palette entry - but that as an aside.)
If you want a JPG replacement - a la OGG Vorbis over MP3 - try JPEG2000 or the lurawave stuff based on wavelets.
Doesn't do anything tiff can't
If this is the same as the last time around, they've just taken tiff, duplicated a bunch of the baseline tags for no good reason (other than to make it incompatible), added their own codec (which they could have done to tiff very easily), removed a bunch of useful stuff from tiff, and called it their own image format. It's a real hack job.
It's just MS being the MS we've come to know and love so well -- making their own binary formats in the hopes of extending their monopoly.
Ian Ameline
"Everyone dissatisfied with JPG is already using RAW"?
I don't know where you're getting that statement from. Everybody dissatisfied with JPG - which I can only imagine stems from the fact that it is lossy compression - is either using:
PNG - because it's common, free to use, etc. etc.
EXR - because it'll allow you to store whatever the hell you want
GIF - because it's ubiquitious and is free to use nowadays (not that too many people cared a few years ago)
'RAW' isn't used by anybody. 'RAW' does not exist. 'RAW' is a collective name for a shitload of formats by a smaller shitload of digital camera companies. And it is never "RAW".. it is never raw data.. it's compressed, stored integratedly or separately, encrypted or not (SONY, among other) and contains a bunch of camera data. The closest thing to a "RAW" format is, say, PFM (portable/pixel float map) or any other format that just stores every color(group) as a bunch of bytes in a long chunk with minimal to zero header/footer information whatsoever that you can only open if you know things like bitdepth and dimension. The closest thing to a unified 'RAW' format for cameras is Adobe's DNG (Digital Negative) - and that's finding slow (no?) adoption as it is. And the closest thing to a unified non-'RAW' format for cameras that isn't lossy compressed is TIFF. None of which you can toss on a website and make viewable in any of the major browsers without plugin installation (if even available!)
That said, I agree with all your other points, especially point 1. Microsoft should be kicked even when down for jumping on the HD bandwagon with a product (or format) that has nothing whatsoever to do with HD.
displays for sure!
You can have my SIG when you pry it from my cold, dead hands.
In reality, this doesn't mean anything, because there's insufficient information in the linked article.
Microsoft, just like any other vendor on the planet, is free to submit anything they like into standards bodies, and ask that they be accepted or considered for use in the world. If Microsoft's new format is useful, fantastic, we all should start using it.
But if, and only if, that format comes free from the burden of licensing or copyright. We've seen how damaging these restrictions can be to simple file format (remember ARC? And all the fun that went on with GIF?) - If Microsoft is releasing an idea for folks to use and adopt? Excellent. If they're pushing an internal format that they hold a patent on, and are requesting other vendors to adopt it? Then it's simply Microsoft once again trying to dick over the industry. And I can't see how it can possibly work under those circumstances.
They don't have the big stick they used to. This is no longer 2000, where the corporate juggernaut simply needed to wave it's financial might and the net doth tremble before it. Microsoft has to tread carefully on an increasingly powerful free software world.
We'll see how this goes. Me, I'm waiting to hear more information.
Event Management Solutions : http://www.stonekeep.com/
No. Documents you create with OO.org aren't all GPLed either. The GPL specifically applies to code. You own the copyright to anything you create, even if you use someone else's program to do it. Now, if you use someone else's work and modify it, such as modifying a GPLed program, THEN you have to abide by their copyright wishes.
I think, in part, you're confusing patents and copyright (for example, your discussion of MP3s), and I think, in part, you're trying to extrapolate the GPL as if it were copyright law.
So, let's step back a bit and try to untangle exactly what's going on. When you use a codec, you're using a piece of software. The codec itself is protected under copyright and possibly under patents. In any event, the actions the codec carries out are not in themselves creative. By this, I mean, the transformations are deterministic with an intended output; creativity could be said to be non-deterministic (ie, originality) with an intended output. Copyright only extends to works that are the result of a creative process. To that end, nothing a codec does could itself be copyrighted; if it could be, the codec itself would be the copyright owner, not the writer of the codec; of course, such a codec would seemingly fill the requirements of a partial AI, so I think the concerns of copyright would not exactly be high on the list of discussion.
Having said all that, we get into the issue of something like the MP3 codec. The concern with it, as related to the GPL, has more to do with the GPL having provisions about patents. Patents, as you likely know, apply to a process, not a specific implementation. This, of course, can be a huge issue with something like the GPL because a large point of the GPL is to allow for the redistribution of GPLed code. If only some people were allowed to legally redistribute the code, by paying patent royalties, then the "network" of involvement to improve GPLed code would be a lot less webbed and a lot more hierarchical (or, it'd be a lot more illegal). Because of this, the GPL requires that all distributed code that implements a patent include royalty-free redistribution covering that patent. Because the MP3 code is patented and there is no royalty-free redistribution allowed (no matter what is said about trying to include an exception for open source), gpled mp3 codecs are illegal, if for no other reason than the distributor of the gpled code is granting others a privilege he doesn't have.
Having said all that, there's nothing illegal about the mp3 format or inherently legal about mp3s themselves. But given the fact that you can't include an mp3 encoder or decoder with a totally GPL software distribution, MP3s have been frowned upon in the free/open software world. On top of that, of course, is the excessive piracy of music (and note, this is further proof that codecs don't change copyright; if they did, the codec maker would be the one suing over all the CD->MP3ed music, not the RIAA and its members) in MP3 format, as it was the first to make it readily possible to share music (commercial and otherwise) over the internet that has basically made MP3 synonymous with pirated music. The last thing many in the free software world want is to have the appearance that the GPL is all about "getting free (as in beer) stuff", even if it's through illegal means.
PS - Things like the gcc include an exception about the GPL not applying as a result of using gcc to compile a program precisely to avoid confusion over the issue; this is somewhat humorous as there are many places were a transformation application of one sort or another will copy small fragments of itself into the destination application, which you seem to recognize. The one overriding principle to always remember is that copyright applies first. The GPL is subordinate to the rules of copyright. So is every other, proprietary license. Now, if you wanted something with more fuzzy lines, one could discuss the linking of libraries. But, that's a whole other discussion.
Eurohacker European paranoia, gun rights, and h
I don't know where you're getting that statement from. Everybody dissatisfied with JPG - which I can only imagine stems from the fact that it is lossy compression - is either using: (PNG, GIF, EXR.)
You don't understand what "raw" images are used for. They're used PURELY in the acquisition phase. There isn't a (non-webcam/hideously-dumbed down) camera in the world that records to GIF, I don't know of a single camera on the market that records to PNG, and EXR is a very specialized format used mostly in "film" (ie movie production.) No still digital cameras on the market record to it.
'RAW' isn't used by anybody. 'RAW' does not exist. 'RAW' is a collective name for a shitload of formats by a smaller shitload of digital camera companies.
No, it's not. RAW = Canon's "raw" image format. "Raw" image formats are produced by many higher-end digital cameras. I'm sorry you don't understand the distinction between RAW and raw, but it does make it painfully obvious this isn't your area of expertise. It is mine: I've shot RAW images on my Canon dSLR for fun and profit for several years now. I shoot exclusively in RAW format because of the extra bit depth which makes adjustments much more 'transparent' (a level adjustment won't cause as much problems wit 10-12 bit data as it will with 8 bit, and you also have no compression artifacts.) I archive everything in the original Canon RAW format.
Your characterization that "raw" formats are used by a "shitload of smaller digital camera companies" is also completely wrong. Canon's RAW and Nikons's NEF are by far the largest, most commonly used "raw" formats. Phase1 is probably up there with their digital camera backs. I'm now guessing, but Fuji is probably next (Fuji dSLRs were very popular a few years back, in part because the Fuji SuperCCD was superior to almost everything else on the market at the time), followed by Panasonic/Leica, followed by Pentax.
Many point-and-shoot consumer cameras these days are incapable of shooting in a RAW mode; it's left to the "prosumer" models by most manufacturers.
And it is never "RAW".. it is never raw data.. it's compressed, stored integratedly or separately, encrypted or not (SONY, among other) and contains a bunch of camera data.
It most certainly is raw image sensor data; that's the whole point. "Raw" camera formats all use LOSSLESS compression. Yes, all of them contain incredibly useful EXIF-like data in them. This is not, despite your rant, a negative to anyone I know. Few manufacturers encrypt the data; Nikon encrypts the white balance info on one or two models (which happen to be the several-thousand-dollar professional digital SLR bodies.)
In most cameras (certainly the Canons and Nikons), it is, in fact, "raw"; it represents the closest you can get to the original sensor data, with little or no processing (on Canon cameras, I believe they don't even do thermal noise subtraction prior to writing the RAW file; the file even contains the "dead" area of the sensor used for such compensation), and anywhere from 10 to 12 bits per channel precision. No white balance, brightness/contrast, gamma, or sharpening adjustments are applied before the data is recorded.
Please help metamoderate.
Before slamming the format, please read more about it. Regardless of what you think about Microsoft, I think it has great potential. Some highlights:
Also, take a look at http://labs.live.com/photosynth and http://blogs.msdn.com/billcrow. To quote one thing from his blog:
IMHO this seems like a well-balanced format that has most of the advantages of a cornucopia of different formats (JPEG, JPEG 2000, RAW, TIFF) without the corresponding disadvantages. If it's not successful, I at least hope something equivalent is!
I am talking about the communication from the OS to the videocard to the display.
Before Windows Vista, the OS was limited to 8 bits per channel (RGB) OUTPUT for the video card. The video card will only get 8 bit of data per channel from the OS, so even if you have a nice ATI card that can do 10 bit per channel (RGB) output from the port, it's still being fed 8bpc data.
Cards from Matrox that can output 10 bit grayscale for 10 bit monochrome displays use DirectX and special drivers to overcome this limitation. Matrox video cards also support 10bpc in Photoshop using a special plugin/driver. However, you have to run the plugin and switch away from the Photoshop interface to see the extra bit of colors.
I know that OpenGL can do high bit rendering, like in the case of the nVidia Quadro cards, or just using floating point representation. The Quardo uses 128 bit precision for all the fancy 3d effects. However what you're seeing on screen is limited by 8bpc output of your video card (though a quadro supports 12 bit output)
Windows Vista supports 128 bit at the OS level. That means you can have a video card that can output 10bpc (for 30 bits total) and it will contain real information that let's say a nice HDTV can read (using HDMI). Or you can just open a regular RGBA image (32 bits) and using a some sort of 3d program to do fancy compositing using different textures and store the information in 128 bit (or the lesser formats; look at MSDN for the various encoding schemes) for speed.
The point is, Vista has the headroom to really display images that contain more than 8bpc (RGB). I'm hoping that Linux would follow suit (it will once HDR displays become commercialy viable) and I believe Mac OS X Leopard will also have this high bit output support (though I have not found any evidence of that yet.)
I'll preface this by saying that on the topic of file and data formats in general, I am intensely conservative. I think it's ridiculous to switch to a new format or compression scheme, unless the benefits are massive -- in particular I've never understood people who seem to gleefully parade from one file compression system to the next every few years, abandoning perfectly good and well-understood formats for ones that don't have decent, widely-available reference implementations; but I digress -- but I'm rather bullish on DNG.
I don't know whether Adobe will pull it off, but I hope that it succeeds, or at least survives.
TIFF is a huge mess. Let's face it; it's a gigantic cockup. Anyone can write TIFF files, but they're nearly impossible to "read" in the sense that a user is going to expect: if I say that my application will "read TIFFs," they're going to expect that anything with a TIF extension is going to get read. And that's almost never the case; you can pack just too much stuff into the container.
(Although container formats have a certain elegance to them from a geek perspective, I'm not sure they're all they're cracked up to be. The number of times I've gotten a video file that I don't have a codec for, but have no way of knowing about until I try to open it, because the codec is concealed inside the MOV or AVI container, or similar problems with TIFs, is beyond number. There's some good sense in eliminating container formats, or at least tying the file extension and other metadata, not to the container, but to the codec inside.)
What I hope that Adobe can do, is give us some neutral ground that the various camera manufacturers can agree to use, so we can break away from the per-manufacturer RAW file formats, and the TIFF morass for interchange.
DNG already has support in probably the biggest single application of consequence, and that's Photoshop, and now they've got quite a few camera manufacturers on board, and the specification is open so there are FOSS implementations. Ed Hamrick's excellent VueScan scanning software produces them, too, and perhaps SilverFast will join the party sometime soon. If they can get the middle-market of consumer and prosumer cameras on board, then I think it will have a chance at achieving dominance from the imaging sensors on down the chain.
There's a lot to be said for it; anyone can implement it, but at the same time, there's some centralized control over the format, so that every Tom, Dick, and Harry can't build on their own crappy extension to the format and create the sort of Balkanization that's plagued TIFF. Hopefully, this will mean that people can implement it, and be confident that if they say that their app will 'read DNG,' that it will actually read all the various types of DNG files that users will throw at it.
If that's the only thing that DNG did, it would be a huge step forward.
"Ladies and gentlemen, my killbot features Lotus Notes and a machine gun. It is the finest available."
"PNG restarts the compression on each row"
That is absolutely not true, and would be madness if it were. From the specification, section 4.5.5:
The rest of your post is suspect now, of course.itsatrap.