[argyllcms] Re: watercolor profile
- From: "Richard Kirk" <dmarc-noreply@xxxxxxxxxxxxx> (Redacted sender "richard" for DMARC)
- To: argyllcms@xxxxxxxxxxxxx
- Date: Sun, 19 Aug 2018 13:06:41 +0100
On 18 Aug 2018, at 23:35, Lars-Daniel Weber <Lars-Daniel.Weber@xxxxxx> wrote:
I’ve got a i1Pro (without UV-cut), calibrated four months ago and I want to
try something with my CCD scanner (no professional use): profiling using 36
watercolor-pens (+ 4 others).
I’ve drawn 40 patches with watercolor- and some other pens. Sure, the
pressure isn’t perfect and I’ve used dry watercolor, but I just want to give
it a try (I might try wet paint later on a better paper).
I have actually done this myself, ages ago, when I worked for Canon. I begged
one of the painted test charts off Windsor & Newton, and from Rowney, and a
measured them in a photocopier scanner and using a PR-650. I was then able to
scan in watercolours and produce realistic scans. I still have a copy of one of
my grandfather’s watercolours, and it was hard to tell from the original even
when it was next to it (there was a bit f green vegetation, and a bit of red in
a door that was out of gamut, but everything else worked. I was using my own
software because Argyll didn’t exist then but the principle is the same.
The Windsor and Newton chart had a graded set of tints so I could do several
densities. I think I had to make up a set of neutral tints but blacks are
pretty black over all the visible wavelengths.
Why does this work? You are going from a 40-dimensional pigment space to a 3
dimensional scanner space. However, most colours have only one or two sensible
ways of mixing them using the nearest saturated colours, and black if you want
a darker tone. You could mix grey from yellow and blue, for instance, but
no-one does. There are one or two odd colours - Vandyke Brown, for example, is
a very saturated yellow but very dark too, so you can’t use it in most
mixtures. Mars red is more saturated then I expected. So, if you measure a set
of patches and interpolate varying additions of black, the interpolation copes
with the mixture of one or two colours.
Oddly enough, some artists who put their paintings through my process actually
preferred the original photocopy output. The print tended to look cleaner and
brighter, and this was enhanced by the print which has a white paper surround
rather than a yellowed piece of watercolour paper. There were some metameric
shifts, but nothing too far out. I think the artists just liked seeing their
work afresh with new colours, and preferred it to the accurate copy.
I had the list of colours and their L*a*b* values for ages. I think I finally
lost it when my machines at work and at home simultaneously had disc failures.
I doubt if it would have helped you anyway, as you would have to measure the
same chart in your scanner and your spectrometer.
It also works with acrylics. You have much the same set of pigments but also
mixtures such as flesh tones. It even works with the fluorescent acrylics. This
sounds quite mad as the scanner cannot produce the fluorescence, but if
something is the right pink or yellow for a fluorescent colour, then we still
see it as fluorescent - a high-vis jacket still looks high-vis in a photo,
after all.
NB: If you are printing, you may want to turn off under colour removal, so you
can get saturated deep colours.
Good luck!
Richard Kirk
Other related posts:
