Am 11.09.07, 17:20 +1000 schrieb Graeme Gill: > Kai-Uwe Behrmann wrote: > > Indeed, I looked at some available samples from other profiling packages, > > like PM and L-Prof, and found, the RGB 0...255 range where used for obvious > > reasons. > > Sorry, I don't find the reasons obvious. Many CMYK devices > also use 8 bits, but you won't find a CGATS file having CMYK > values of 0-255. There is also the possibility of driving devices > with 12 or 16 bits/component, so for consistency these should > be 0-4095, and 0-65535 ? I don't think so. > The CGATS standard has always allowed for real values, so > this can't be a factor either. My reasoning is just to understand what others do. The special 0...255 range is used for RGB as an exception, based on the experience that long long times mainly 8-bit systems where used for creation, storage and displaying of RGB data. I accept this as a obvious reasoning for the creators and users of RGB measurements files. I did not want to exclude any other approach. > > As well the #17 draft speaks explicitely about that range > > for RGB. > > This may be so, but it was added after I created the .ti3 format, > and ANSI standards are not easily come by as an individual (ie. > they are not generally available for download, only for a fee). To not to obtain such important information is a pain. > It seems that they have merely adopted the de facto usage of > other packages, without regard to the logic of it. > > I think it's a bad recommendation for the reasons mentioned above. Personally I'd prefere consistency. Just, I want my software to work with most files. > > Did you run a different profiling package with argyll's > > measurements, for instance L-Prof? > > Argyll's measurements are labelled as .ti3 for the specific reason > that they are setup for Argyll. Importing other measurement > sets into Argyll using (say) logo2cgats takes care of these scaling > issues. Argyll has from it's inception kept the device values in > the same file as the measurement values, unlike many other profiling > packages in which require tracking several files, so I've never > regarded such things as freely interchangeable. To keep things in one file seems very handy. To embedd these measurements on request as targ tag into the final profile, would even enhance argyll behaviour. > Graeme Gill. > > kind regards Kai-Uwe Behrmann -- developing for colour management www.behrmann.name + www.oyranos.org + www.cinepaint.org