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.
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?
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.
displays for sure!
You can have my SIG when you pry it from my cold, dead hands.
Is space more valuable than time? CPUs are becoming faster, but storage is becoming cheaper too...
Don't forget that it's no longer just space/time tradeoffs. There's also the network bandwidth tradeoff. And network bandwidth is not on the same kind of curve as CPU's or storage at least for WANs.
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'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."