+1
I hadn't read this post while writing my short one. Thanks for your time for writing something so informative.
I hadn't read this post while writing my short one. Thanks for your time for writing something so informative.
As I recall some of the older discussions when DNG was being rolled out, what Carsten is commenting on is correct. The DNG file format can hold whatever the camera maker wants it to hold, with respect to both the RAW file and any other notes and information, even proprietary stuff that the manufacturer wants to include. That is sort of where the sticking point comes in. In order to make use of any of that proprietary information contained within the file, there has to be something to decode, read and apply it. Essentially, that means the camera makers would need to provide how their recipes for processing are to be handled in the larger DNG world, otherwise, you just get the same RAW file as you do now with a more generic conversion. To me this has always been the problem of the design. As Marc points out, the only way to get to a more universal use is for the camera makers to have all of their secret sauce corrections done in the camera and written to the RAW file that is then just held in the DNG container in a format that is readable by a one size fits all DNG converter. That is about the only way it can work.
This is sort of the way folks are trying to use it now....do the processing with all the corrections and then embed that more final version into a DNG container for longer term storage and more "universal" access. Problem is that does not really work that well. Programs like Phocus from Hassy that have built their own corrections that may be lens and camera specific, can only apply those corrections after the RAW file is converted, but before it is output. If the corrections could be made before the RAW version is written, such as in the camera, a DNG file would be easy to produce, since it would just be a more universally readable RAW file.
As it now stands, there are so many different iterations of RAW files from so many manufacturers, that it is hard to keep up with. The basic algorithm for reading and converting the file is really not that complex.....IF the RAW file is written to DNG and contains all of the corrections and adjustments from the start. The easiest way to do this would be to have the camera write a RAW file into a DNG format that already makes all the corrections. The problem there becomes larger lookup tables and faster processor needs built into the camera to accommodate those corrections before the file is written.
In the case of the S2, as is mentioned, the size of the lookup tables needed to make the corrections is very small right now, since there are few lenses involved. In the case for others, where there are lots of different lenses needing different adjustments that are easy to do in software with a bigger computer, the requirements to stuff all of that into the camera are less attractive. The only other option would be for the camera makers to somehow supply their proprietary processing adjustment data to anybody wanting to put out a RAW development algorithm for more universal file conversion use. That would mean Hassy, Leica, Phase, Nikon, Canon and the rest supplying a continuously updated proprietary processing formula to anybody that asked for one in the DNG program group, so that their files could be read and converted properly. The more corrections the camera makers start to introduce, the more complex that becomes. It is far easier for them to build their own software and supply and update just that, even if it is not ideal for workflow. They then have the freedom to build in the more complicated changes and correction needs into their app, rather than having to supply that information, even if proprietary, for use in all DNG conversions by others.
This is how I have grown to understand the entire DNG problem. When DNG was first proposed by Adobe, there were almost no exotic tweaking of things as we have today. The general RAW conversion algorithm pretty much worked for all files, only needing camera specific profiles to make things correct with respect to colors and stuff. Things have gotten a whole lot more complicated and demanding now with the plethora of cameras, versions, lenses, and contant tweaking and updating by manufacturers as new tech and methods become available. The basic RAW conversion algorithms have not changed a whole lot from the start, but all the potential sauces that can be used have grown nearly exponentially. How and when those sauces get applied has now become the bigger issue. As mentioned above, the best place to do that for the most universal file usage is in the camera at the time the RAW file is being written, but that requires more horsepower in-camera.
We are actually probably closer to a point now where things like that are more possible, but almost all camera makers already have their own software divisions that are easier to control and make updates on their own on the fly, than to be constantly releasing stuff for all DNG program subscribers to incorporate into their stuff. I still think the best way is to write all the corrections to the RAW file at time of capture, but that can create problems beyond just in-camera processing needs. It would not permit any minor user preference adjustments as is now possible in many apps. It might also cause some file distortion that could not be undone if so desired. Things like that.
Bottom line from how I see things now....DNG is a very good concept, but things have advanced quite a bit since its inception. The information needed now to get the best results are not as easily incorporated and could compromise some of the "competitive advantage" that some camera makers now have over doing the best conversions for their files. Most of this could have been avoided had more manufacturers really signed on and helped push DNG from the start, but the security and uses at that time were very unknown and questionable. The advances in in-camera processing are making it now possible to do a lot more to files before they are written, but do not permit much user adjustment if needed yet. This is all evolving in ways that few could have really predicted when DNG was introduced. The entire transition to digital is not directly comparable to how film was shot and processed, where one had a few film types and their favorite processing labs that could soup things the way they preferred, yet that seems be the model that DNG is trying to follow....one universal development formula that could be tweaked a bit with some proprietary input. Things are much more complicated than that at this point.
Anyway, just a few thoughts on this before I have that next espresso ;-)
LJ
