> -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Harman Bajwa > > Well, if you look here, someone in the linux community > has already (and succesfully) attempted to peek into > the RAW format. Take a look here: > > http://www.cybercom.net/~dcoffin/dcraw/ > http://www.aim-dtp.net/aim/digicam/dcraw/ > http://www.insflug.org/raw/software/tools/dcraw.php3 > > The source code file (dcraw.c), very clearly uses a > Bayer mask for getting at the info and if you look at > the history section, the author clearly mentions doing > interpolation which means the body is *not* applying > interpolation before creating raw file. IMO a RAW file > is really a representation of the raw sensor data with > the parameter settings in the header. This reinforces > what I was saying earlier on in that various camera > body RAW formats are highly dependent on the sensor > (RAW Luminance only for each channel in whatever > format the sensor is arranged) and that is why a DNG > is really a higher level of abstraction than a RAW > (for archival and perhaps greater compatibility in > future). Sort of like having different Machine level > instructions for different processors (RAW), while > having a high level language across formats C/C++ etc. > (DNG) as long as there are compilers available for > each hardware platform (RAW to DNG conversion > utilities). IMO They really complement each other. > > Another link here describes the raw data format in an > excel spreadsheet format: > http://www.wonderland.org/crw/ > > And and if you look here: > http://www.cybercom.net/~dcoffin/dcraw/secrets.html > > The "CRW Files" section mentions the fact that a RAW > is *not* a TIFF type file at all, although its > heraders are TIFF style (perhaps that is why others > refer to it as a TIFF type file, mistakenly). Although > it talks about the Canon Powershot 600 only, I am sure > the concept is same for other bodies, with compression > applied (as is evident in the source code for dcraw.c > accessible from the first link). The following steps > are elucidated here as neccessary tp process the CRW > file: > <quote..> > # Parse the CRW file and extract the pixel values. > # Interpolate them to produce green, magenta, cyan, > and yellow values for every non-edge pixel (thus the > output will be 852x611). > # Interpolate again to smooth out the color ratios. > # Compute RGB for each pixel to collect statistics. > # Compute RGB again, applying gamma correction and > other adjustments. > etc.... > </quote ....> > > Finally, if the RAW file really was a TIFF type file > (ie interpolation is already applied minus the camera > settings), Canon would not have to go thru the trouble > of supplying an FVU utility - they would just output > either a JPEG (lossy) or TIFF (lossless) - no RAW > neccessary. The DNG initiative would really be > unneccessary - the TIFF has been around for a while > and enjoys tremendous support with professionals and > is a very well known and flexible format to work with. > > > Think about it. At least, this is better than just a > guess :)
Thanks a lot for the links. I did not have time to look at it yet but most definitely will do so later. What I have been writing before pretty much matches with what you are saying here. The only thing we disagree is that I believe it is possible that the DNG can be made flexible enough to accommodate almost any sensor including exotics like Foveon. If that really is Adobe's goal and if they do try to do this I don't know. I would have to look at the specs to find that out. By the way, what I said before was a guess about CANONS's raw format. I have decoded Nikon's raw format when I owned a Nikon D1. Also I work with and process digital images (gray scale) on a daily basis... Robert * **** ******* *********************************************************** * For list instructions, including unsubscribe, see: * http://www.a1.nl/phomepag/markerink/eos_list.htm ***********************************************************
