[argyllcms] Re: Calculate RAW importer profiles ?
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Mon, 27 Jul 2015 18:18:53 +1000
Adriaan van Os wrote:
But many cameras do have an option to output both RAW and JPG. And the JPG
does reveal the
secret of how the imported RAW should look like (for a specific set of camera
So, why not take the JPG and the RAW imported TIFFs, compare them and
calculate a "raw
importer" profile from both. That profile, when assigned to the RAW imported
then represent the best possible color match with the JPG (for a specific set
Would this be possible with argyll ? If so, it would offer users a quick way
to create a
"raw importer" profile from a (statistically relevant) set of RAW and JPG
images. Or at
least it would be better than the current (rather dissatisfactory) situation.
I suspect that one of the key pieces to doing this with any convenience or
is probably missing :- a TIFF/JPEG to RAW converter. How else would you create
a RAW format test chart to feed into the two different RAW photo processors ?
You could try using a real photographic test chart like the ColorCheckerSG,
but this has it challenges, and would limit the gamut and detail possible
in characterizing the two RAW processors.
Putting aside that problem, the general workflow would be to create
an ICC profile for both profiles, interpreting the JPG/TIFF output
with the nominated output space profile (i.e. sRGB, AdobeRGB, ProPhoto),
so giving you two RAW->XYZ data sets to make the profiles from.
The final transform (device link) would be created by linking the
two profiles. A problem in all this is likely to be any clipping
introduced by the source profile. Once colors have been mapped
into the same output, you can't un-map them to something different.
Overall it seems like a task with a few "gotchas", and therefore
would be a project. So that raises the question of whether it is
really worthwhile. If you like what a particular RAW processor does,
why not just use it ?
Other related posts: