Adobe Releases DNG SDK

Question

Adobe just released an SDK for the DNG format. What does this mean fro better DNG support in Aperture if anything? Are we to assume that Apple hasn't had access to this before now? I'm hoping what this means is that in the future we'll have better support for DNG which means it will support virtually any RAW format. Does anyone know, if Apple didn't have access to this before how long this might take to get working? Is it a matter of a couple months, or would we have to wait until 2.0?
http://www.macworld.com/news/2006/04/18/sdk/index.php?lsrc=mwrss

Answer 1

Well, first off, dcraw produces pretty remarkable
conversions. Perhaps that's why so many products out
there (including ACR) use the code as part of their
converter.
No argument that the output is decent, but I tend to think the adoption of DCRaw is because it is (a) Free (as in beer or speech), and (b) way easier than doing it yourself from scratch. Did I mention it was free?
I mean, lets say you are writing "Bob's RAW Readin' App". What else are you really going to use? If you're running on OSX you could rely on the OS conversion but many of these apps want to support windows as well. DCRaw is just sitting there, quite free for the taking, and the "Readin'" part of that app is done. Now that Apple's conversions are better that idea may take root a little more than it has in the past.
I honestly don't know that there is even another library around, commercial or not, that does what DCRaw can do.
Secondly, it's rather spurious reasoning to assume
that Apple is only capable of producing broken
renderers (though, I'll admit that between
image-quality problems in 1.0 and performance
problems in 1.1 I don't have a lot of confidence) -
and it's absurd to keep adding support for
oddball/exotic cameras while ignoring DNG.
Here's the problem though. In order to support something like the Fuji, they are going to have to be able to parse that offset image sensor data (rotated 45 degrees) wether or not it comes from a Fuji RAW or a DNG converted from a Fuji. Thanks to DCRaw we all know right where that RAW data is held, so it's not a problem to reach it even without assitance from Fuji. The trick is producing a decent conversion. Even now there's a thread in the Lightroom forums complaining about the Fuji S2 conversion being not that great:
http://www.macromedia.com/cfusion/webforums/forum/messageview.cfm?catid=589&thre adid=1125459&enterthread=y
and I'll bet ACR has put more thought into Fuji RAW than even DCRaw has so far! Is it not fair to say that if Aperture had the same problems with conversions we'd have a hue and cry across the land? Far better to say you can't do something and have people know not to buy the product, than to lead people into thinking you can do something when some people are going to be very dissapointed in your "support". Even now I feel Aperture should simly remove "DNG" from the list of supported formats, because it does not meet what I consider to be the absolute minimum for that format - DNG Linear.
I actually partly agree with you in that I think they should add DNG Linear support first, which would placate a lot of people and be super easy to add since there is only one form of DNG Linear. Then they can finish support for the more offbeat formats.
As for why to use DNG - it's simple - there are a LOT
of cameras out there that are not supported
otherwise, and there always will be. It's wonderful
to have a simple, compatible format.
Except that every true variant of a Bayer array (Canon and Nikon are not variants that differ significantly in this regard) requires different processing techniques. The Fuji 45 degree rotation is replicated exactly (well, as exactly as it can be) in that DNG file and requires different techiniques to work with and get a good conversion. I shudder to think how the SuperCCD SR data from the S3 is represented in a DNG file since it has two sensors at each pixel - I suspect it is probably stored as a linear DNG file (in short it demosiacs the RAW data before storage and does not really hold the raw data [unprooved speculation]).
It's bizarre that you'd advocate for not supporting
this, it could only be a good thing.
It would be an awesome thing if it really worked. When I first read about DNG I was very, very excited! It was only after I read the specs and realized you still would need a different conversions algorithm for any of the numberless variants of RAW data that DNG is capible of holding that my joy turned to anguish.
The reality on the ground is that pretty much all cameras that do not have bog-standard bayer CFA grids will be shunt off to the Island of Misfit Cameras if DNG were really to take hold, because most programs would really only bother to include conversion algorithms for the most popular kinds of formats held in DNG. I dearly love the Fujis and Sigmas and Sonys (cyan color in grid) of the world and want to see alternate choices flourish.
The problem is not exactly with DNG so much as that what DNG is trying to do (pretend to normalize all sensor data as if it were the same) is simply not possible, and in my mind not even desireable because, as I said, I want to keep sensor diversity alive and alternate sensor designs coming down the pipe.