Home Products Download Order Contacts


Subject: Re: OpenRAW article: "DNG is not the answer"!

There are 2 separate issues here: does DNG recognise the special nature of the DMR's AA, and what should raw converters do about it?

First, yes DNG does recognise it, and either the DMR back itself or the DNG Converter (for other cameras) does something about it. Here are values for DNG's "AntiAliasStrength" parameters for some cameras:
Leica DMR back: 0.00
Nikon D70, D70s, D50: 0.25
Virtually all other cameras: 1.00

The problem with raw converters in this case isn't DNG. Given that the Leica DMR back has a weak or non-existent AA filter, it would still pose precisely the same problem if it used a proprietary raw format. Similarly, those Nikons present the same problems whether a raw converter handles them via NEFs or DNGs. (Raw Magick use NEFs from a D70 to illustrate the superiority of their anti-Moire anti-colour-distortion demosaicing algorithm, and Raw Magick doesn't support DNG).

You make the point "... support for the Modul-R was not deemed cost effective". And that is precisely the issue here, which is why OpenRAW's assertion that open documentation is sufficient to meet their objective is wrong - it isn't sufficient.

That doesn't mean that DNG by itself solves the problem. It removes several of the obstacles faced by raw converters, but leaves one fundamental problem - you still have to write special code if a camera+sensor inherently needs special code! If the sensor has offset pixels, (Some Fuji cameras), DNG can tell you it has, but you still have to write the code. Ditto non-square pixels, (D1X). Etc.

There appears to be a partial "get-out", quite deliberatly placed there by Adobe - "Linearised" DNG. Converting a Fuji S2 raw to "Linear DNG" removes the "CFALayout" parameter, presumably because the result is now rectilinear. It doesn't remove the "AntiAliasStrength: 0.00" from the Leica DMR back, perhaps because linearising DNG doesn't get as far as dealing with the AA issues. But - I know of people who would prefer to deal with those in Photoshop when they arise, rather than always have AA. (Leica users tend to be a bit fanatical about sharpness!)

As I said - this isn't about DNG. It is about a product supplier's cost-benefit analysis for the code needed to handle special features of a camera. DNG helps - it eliminates unnecessary differences, and avoids the need for reverse-engineering. And some problems can be side-stepped with linearised DNG. But some innovations still need someone to write some code!

Frankly, (and I am not the first to say this), Bibble appear to spend more time dreaming up reasons not to support DNG than it would take to support it!


View All Messages in adobe.digital.negative

OpenRAW article: "DNG is not the answer"! =>


Copyright 2006 WatermarkFactory.com. All Rights Reserved.