[argyllcms] Re: colormouse vs. digital camera
- From: "Alastair M. Robinson" <blackfive@xxxxxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Sat, 22 Oct 2005 00:13:45 +0100
Hi,
Stephan Bourgeois wrote:
On the other hand I did a profile for a D88 using pigment based inks
(Epson DuraBrite) with 350 patches. In that case, the colormouse got it
right, and the digital camera method was off, especially in the shadows.
Yeah - inks do vary a lot. Ironically the pigment-based inks are said
to suffer far more from metamerism when using a scanner as a colorimeter
- looks like that applies to the digital camera method too - yet the
colourmouse has no problem...
I've generated profiles for Tesco "General Use Picture Paper" 120gsm
matte, and 7DayShop's own 270gsm Glossy - two very different papers -
and I'm now getting a very good match when I print the same image on
each type...
How do you do that? Do you manually add data into the .ti1 and .ti2
files, and obviously more readings into the .ti3?
Yes, I add them to the .ti1 file (the XYZ values can just be filled in
as zero - they're not needed.) then use targen and printtarg as though
preparing for a scanner.
I've used a sort of "helix" of near-grey patches around the RGB grey
axis. The set I currently use is as follows;
100 4.8 6.3 5.8 0 0 0
101 9.6 9.6 9.6 0 0 0
102 14.4 16.4 14.4 0 0 0
103 21.7 19.2 22.2 0 0 0
104 27 26.5 24 0 0 0
105 28.8 28.8 30.8 0 0 0
106 37.1 33.6 33.6 0 0 0
107 38.4 40.9 40.4 0 0 0
108 43.2 43.2 43.2 0 0 0
109 50 48 50.5 0 0 0
110 54.8 53.8 52.8 0 0 0
111 57.6 57.6 60.6 0 0 0
112 64.4 62.4 62.4 0 0 0
113 67.2 69.7 70.2 0 0 0
114 72 72 72 0 0 0
115 78.8 76.8 78.3 0 0 0
116 83.6 84.1 81.6 0 0 0
117 86.4 86.4 89.9 0 0 0
118 94.2 91.2 91.2 0 0 0
119 96 96.7 96.7 0 0 0
The idea being that this will "enclose" the true grey axis. I haven't
done any formal testing to see whether this performs better than just
using R=G=B patches...
I usually go through the motions of the scanner method to create a .ti3
file and then replace the .ti3 file's XYZ values with the measured
values using oocalc.
Can I also ask if you are using the C84 or C86 with Linux?
The C86 was with Linux. Both printers are at work, but I took in my
laptop for that test.
All the best,
--
Alastair M. Robinson
Other related posts:
