Pages: [1]   Go Down

Author Topic: Scanner profiling with DCSG - homebrewed individual reference data  (Read 1607 times)

Stephen G

  • Full Member
  • ***
  • Offline Offline
  • Posts: 173
Scanner profiling with DCSG - homebrewed individual reference data
« on: September 08, 2011, 07:27:50 am »

I've borrowed a Gretag Digital Colorchecker SG in order to profile my scanner (epson V700, Vuescan, 'raw' tiff output). Seems to work quite nicely, building profiles with PM5 and the generic reference file that it uses.

I want to build a reference file for this specific Colorchecker so using my Spyder3Print Spectrowhatsit I've taken Lab measurements from each patch on the target. 5 measurements per patch, so I can average them.

I'm stumped with the next step: how do I turn my little .txt file into a reference file that PM5 will recognize? (maybe something Argyll can deal with?)

And: have I captured enough data to do it? is my data good enough?

Have to give the target back later tomorrow morning, so the data I've got is it, for now. this is a clip of what it looks like:

1   96.76   -0.27   3.60
2   96.62   -0.45   3.59
3   96.80   -0.37   3.51
4   96.78   -0.47   3.54
5   96.64   -0.60   3.55
6   8.72   -0.45   -0.60
7   8.60   -1.12   -0.81
8   8.60   -1.12   -0.81
9   8.62   -1.18   -0.78
10   8.60   -1.12   -0.79

1-5 being patch A1
6-10 being patch A2, etc.
« Last Edit: September 11, 2011, 07:16:04 am by ElSteev »
Logged

Stephen G

  • Full Member
  • ***
  • Offline Offline
  • Posts: 173
Re: Scanner profiling with DCSG - homebrewed individual reference data
« Reply #1 on: September 15, 2011, 07:07:42 am »

Got it working in Argyll, but the profile generated was rather disappointing. Perhaps the SpyderSpectro is just not accurate enough. The PM5 profile is so accurate that I hardly have to touch my scans and they print accurately.
Logged

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
Re: Scanner profiling with DCSG - homebrewed individual reference data
« Reply #2 on: September 16, 2011, 04:55:00 am »

Patches build with more colorants could help but I have my doubts on too many colorants, all with different "metamerism" issues + the scanner an observer with specific RGB sensitivity and not a full spectrum lighting must lead to strange profiles and scans. Patches made with colorants that come closer to the colorants used in the originals to scan would be better. Multi spectral scans are a better answer. The Epson profiling is not that bad. I made a scanner profile some time ago that did it quite good with some odd colored wax sketches. But putting a Z3100 printed calibration target on the scanner, 6 pure hues + black and three hues were way off.

Good read:
http://www.image-engineering.de/index.php?option=com_content&view=article&id=57&Itemid=91
http://lists.apple.com/archives/colorsync-users/2011/Jun/msg00013.html
http://lists.apple.com/archives/colorsync-users/2011/Jul/msg00040.html


met vriendelijke groeten, Ernst

Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Scanner profiling with DCSG - homebrewed individual reference data
« Reply #3 on: September 16, 2011, 09:29:34 am »

Its either the Spyder data or some error in the way the reference was built (something you can build in Excel too). You say the original ref is better, use that. Better still would have been using an IT8 target which is better designed for a scanner profile. The target you specify is for creating camera profiles. The it8 is printed on some kind of photo paper material and if that’s what you are actually scanning, more better <g>.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Stephen G

  • Full Member
  • ***
  • Offline Offline
  • Posts: 173
Re: Scanner profiling with DCSG - homebrewed individual reference data
« Reply #4 on: September 16, 2011, 12:32:14 pm »

Its either the Spyder data or some error in the way the reference was built

Pretty much what I figured, and I don't have the skills/time/equipment to figured out what exactly and improve the process. Why bother when the PM5 process works so well?

If I was scanning printed photographs then an IT8 would probably work very well. I'm scanning botanical watercolour paintings, however, and the paper white on the paintings is much brighter than the white of a photographic target. I've GOT a Hutchcolour reflective 5x7 target and it does not cut it. Using it creates a profile that describes a gamut much smaller than the scanner's actual gamut. As soon as I apply it to a scan the paper whites and anything close to them get clipped off. Deep shadow colour is not so accurate either.

The DCSG describes a much larger gamut and has saved my butt. I'm still getting clipping in the highlights, but it's borderline stuff so I can cheat a little and pull the curves down a teensy bit before applying the scanner profile. probably loose a little accuracy here, but the reprints look OK.

I have this sinking feeling that I'm heading in the direction of custom targets.... asking watercolour artists to copy the DCSG in their pigments etc etc  gonna need a decent spectro for that
Logged
Pages: [1]   Go Up