[argyllcms] Re: Absolute colorimetric - dark saturated colours excessively light
- From: "Alastair M. Robinson" <blackfive@xxxxxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Sat, 04 Feb 2006 14:57:10 +0000
Hi,
Graeme Gill wrote:
Hmm. It's hard to say much about these images, without anything to
compare them to. The last one for instance, might be quite reasonable,
if the plain paper has a poor black point (was this running RGB to
the printer, or CMYK ? If the latter, it would be worth double checking
the ink limits used for the test chart, and during profile creation.)
The black on the plain paper is extremely poor (L* around 36) - I'm
including it mainly because it provides a particularly stark example of
something that shows up to a lesser extent on better papers.
I would need the .ti3 & compile command line for (say) the plain paper to
make any judgement as to whether things are working as expected here.
.ti3 file, profile and makefile here:
http://www.blackfiveservices.co.uk/R300_Plain_20060122.zip
The general behaviour of the grey wedge is consistent with an absolute
mapping,
in that it (plain paper) tracks the original down to wedge 12, and then
flattens out (clips).
Agreed.
The Proofing Prism result is completely inconsistent with
T8_Tesco_Proof.jpg.
By the latter, the black point on plain paper has an L of 22-23, but by the
former, it is 4! Which is correct - can you measure the black L value
independently to determine this ?
I should have been clearer here. Prism uses the scanner-as-colorimeter
trick, so this profile, unlike the others, was built using a scanner;
prism may well be over-optimistic about the strength of black (and a
print won't match this proof nearly as well as those created with the
Argyll-generated profiles)
The Prism profile is an old profile generated over a year ago. Since
then, I'm using different ink, the driver has changed and I can't even
swear which paper was used. I included it because I couldn't lay my
hands on a better "control" profile. At the top edge of columns 2 & 3,
though, I think it's doing marginally better than even the Glossy
profile created with Argyll.
Perhaps these two are better control subjects:
ISOuncoated:
http://i14.photobucket.com/albums/a346/robinsonb5/IT8_ISOuncoated.jpg
and
ICONSnewspaper26v5:
http://i14.photobucket.com/albums/a346/robinsonb5/IT8_Newspaper_Proof.jpg
The gamut-mapping strategy used by these two profiles appears to be to
preserve hue angle at all costs, and weight the saturation/luminance
balance in favour of preserving luminance.
Is there anything I could tweak to weight Argyll's strategy more towards
preserving luminance?
All the best,
--
Alastair M. Robinson
Other related posts: