[argyllcms] Re: Trying to generate icc profiles for camera and printer

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Tue, 05 May 2015 10:33:39 +1000

Bernard wrote:

Hi,

2. to attempt generating a color profile for my printer (an old HP Deskjet
1120C)

I think I got fair results in the first point, but I doubt if I have in the
second one.

Generating an icc profile for my camera, showed a 'peak error' of 7.42 on a
new
photography or the target, was 6.45 on using last year's image. At first
sight, I don't
suppose that this somewhat higher peak error reveals a result of a quality
significantly
lower. The average errors were 1.81 and 1.70 respectively.

Sounds about the same then.

I then printed the image that generated the icc profile (on matte photography
paper), and
shot the print with said camera, then I calculated a new icc profile from the
chosen
image. The peak error was 67.53 and the avg error 15.32.

Hmm. Are you attempting to use the camera to measure the printed output,
rather than using a color measurement instrument ?

The results from this approach have always been rather limited in
my experience, probably due to camera spectral sensitivities being
too different to the human eye, combined with pigment differences
between the photo test chart and the printed output.

Do you think such results had to be expected on a printed target ? Could
this be due to a
defect pertaining to the printer, i.e. if the print heads spray irregularly
from left to
right of the page ? Might the results have been better with a laser printer ?

But your self fit errors are extreme, which may indicate some other issue,
such as being out of exposure range, some automatic adjustment
being done by the camera, mis-registration with scanin, or some sort
of procedural error.

Do you think that the printer icc profile that I have obtained can still be
of some use in
printing my photographs ?

No, I wouldn't trust it.

Or should I rather try anything else ?

It's an investment, but something like a ColorMunki spectrometer
will give a very much better result.

Graeme Gill.


Other related posts:

  1. Home
  2. argyllcms
  3. Archive
  4. 05-2015