[argyllcms] Re: Camera matrix profile, adding ti3 perfect white data set

  • From: "Iliah Borg" <ib@xxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Fri, 22 Jan 2010 11:39:44 -0500

Dear Elle,

> profile check complete, peak err = 33.155823, avg err =
> 7.494565

The peak error value you are reporting is untypically high.
I never got more than 20, average 12, and max of 7 with
modern high-end cameras.

There might be one or several problems with the workflow:

- target is unsuitable
- target reference file is not accurate
- flare

To get to black point, the best method I know is to add a
black trap to the target.

To provide white point adding simulated white reading is a
common method, however I prefer to add a white matt backlit
patch wich reads 1/3 stop higher than the white patch of
the target.

Scaling of data for camera profiling is not a good idea at
all.

The quality of the shot is more important than the number
of patches, a well-shot CC24 target (especially the new
Passport or the first releasde of original ColorChecker)
results in better matrix profiles than many other complex
targets. If you mount the target on a grey card (that is
sticking out of the target by 2-3") and ensure the
brightness and white balance near to the four corners are
close within +/- 2 RGB you will get very good results.
White target patch must have 95% value, also +/- 2 RGB.

For more complex profiles the only method that I know
working is projecting the rainbow from a diffraction
grating onto the sensor.

--
Best regards,
Iliah Borg

Other related posts:

  1. Home
  2. argyllcms
  3. Archive
  4. 01-2010