[argyllcms] RAW camera matrix profile from emissive display captures

  • From: János, Tóth F. <janos666@xxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Thu, 26 Jan 2012 12:10:41 +0100

First of all: This is is only a cheap camera (Olympus E-PL1) for
absolutely casual purposes, not even enthusiast hobby photographing
But you know... if you have some knowledge about color management and
you have a spectro at home you start thinking whenever you need to
work with colors. :)
So, when I happened to pick up relatively decent camera tomorrow, I
started to think about the possibilities of some kind of calibration.

Again, I am not an enthusiast hobby photographer, so hard-printed 300$
calibration charts are out of question...
But what if I use an emissive display device to take some shoots and
also use a spectrophotometer to get the spectral distribution of the
captured colors...?

I searched a bit with google and found this:
They say they realized some kind of improvement in almost every
scenarios (even with uncalibrated displays).

They used an LCD, but I happened to have a PDP display which is more
evenly saturated than most of the LCDs.
But I also have a LED LCD with narrow gamut, so it would also be
possible to capture colors from displays with very different spectral
characteristics (in case that could help...)

I think I managed to fully understand the characterization process of
output deices by now, but I can't tell the same about the input
devices (I have some idea about it, but there is a little hole there).

If I think about my RGB camera as a tri-stimulus colorimeter then I
realize that one matrix profile if valid for one given spectral

But what about these new X-Rite i1d3 colorimeters with spectral
sensitivity curves and on-the-fly created corrections? Can we do the
same with a camera?
Let's say I use a PDP display to capture and measure some patches and
then I use my spectro with the ambient light diffuser to measure the
lighting conditions and then create a matrix profile for those light

But I am not sure if some measurement from a display is enough to
spectrally characterize an input device, even if I can use different
type of displays with various spectra... And what about the tools...?
I guess it would require a new software (like a new ArgyllCMS tool

After all, I would ask these questions:

- Did any of you tried this?

- Does anybody have a test chart for this?

- If I create a test chart (for a simple matrix profile with assumed
linear response in RAW mode) then should I use RGBW only or should I
add CMY as well?
Does colprof really benefit from CMY measurements when I use an RGB
device as a light source and I create an RGB matrix profile?

- Should I blur the camera shot with wrong lens focus to even out the
RGB sub-pixels and the dithering patter of the display device as more
homogeneous patches?

- Do you thin it would worth a try? (If I need to learn how to create
ArgyllCMS compatible test charts...)

Other related posts: