[argyllcms] Re: Question regarding White Point and spotread
- From: Richard Kirk <richard@xxxxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Sun, 9 Nov 2014 09:56:28 +0000
Hi.
I've been seeing these white point posts, and I feel I ought to speak up for
Graeme, here. I have been doing commercial colour calibration software since
about 1985. If predicting a visual match with mixed media was
'press-the-green-button-to-calibrate' easy, that's what we our software would
do.
Supposing I calibrated a printer (that's where I started) and a user complains
that they have used their software, and their print does not match what they
are seeing on the monitor. I reply 'What monitor is this? Have you calibrated
it? Are you sure no-one has fiddled with the contrast or gain settings?'.
In general, the reply fell into one of three types...
(A) Gosh, I never thought of that.
This is rare, but very gratifying when it happens. I have seen otherwise very
intelligent scientific people being astonished byt the obviousness that the
printer calibration software in fact has no idea of where the contrast and
brightness settings are on their monitor. Or, at a less trivial level, that our
eye and brain happily and automatically correct for the illuminant white with
reflection objects, but make an exception for self-luminous objects such as
displays; which is why you can happily compare images under D32
incandescent-matched lighting with a D65 monitor, and hardly notice the
difference. Or that the paper image has a white surround, and the monitor
surround is black, so you see more shadow detail on the monitor. Or a CRT
monitor, a black is flat and the surface is shiny, where projected film has
film grain roughness; so what is measurably the same black has a very different
sensation.
(B) Yes, but it looked wrong so I tweaked it.
This is surprisingly common. It looked sort-of right but the reds were not
vibrant enough, so I used the D55 ICC when displaying my images. The neutrals
were not looking quite right, but if you changed the gamma from 2.4 to about
2.1 it looks a lot better. But I find the blue-greens and pinks and now looking
off. Can you give me something to correct for that? The thing I often gave them
back then was a big label over their monitor controls, sighed and dated when
the calibration was done. Everyone who is in the colour business probably has
two colour measuring instrument built into their face, and have been comparing
colours all their lives. We can measure XYZ, but they still know it is not the
'real colour' they have in their head. It is not bad, but they remember that
the colours were different in the original. That way lies madness. Yet on
several major film projects, I have had to clear out layer upon layer of tweaks
and edits, and once you have done that, the point where the errors are coming
in (if you have not already fixed the problem by clearing out the tweaks)
becomes something that shows up in the figures.
(C) You had one job...
The software is supposed to make the images look right. I used it. The images
look wrong. Look at my monitor! Do you call that a match? I don't. I am
disappointed in you. Not only have you failed to make my monitor match, you
have also wasted the expensive time of a professional person. I don't care
about your excuses about the monitor settings. I am not even going to listen to
that. The software should have done the job. If it is not up to the job, then
don't waste our time.
Yes, I have seen quite a few of type C. You get used to it. You go limp and let
all the bile just wash over you. Eventually, when they cannot fix it
themselves, they often send some minion to ask the same questions in a more
polite way without them having to lose face.
A real example:
About five years ago, the commonest complaint I get is that someone set up
(say) a Sony BVM LCD to match their old Sony CRT. They set both monitor to D65
with their Philips monitor, or Truelight probe, or whatever, and the LCS looks
orange. We have had good instruments sent back for recalibration. The fact is,
if you are looking at a whole screen of white rather than the original 2-degree
target the CIE spaces were based upon, then you are using more of your
peripheral vision, which has more blue cones, and less macular dye. We clearly
see big patches of colour differently to small patches. However, a big patch
and a small patch on the same monitor will look the same colour. Your eyes are
not uniform, and your brain is doing a flying repair on the data so the
contrast is usually not seen.
In this case...
(A) will understand, and live with the difference
(B) will tweak the LCD to match the CRT (probably harmless)
(C) will send their monitor probe back, get another, send that back, then send
the monitor back, and all the while wonder why all colour scientists are fools,
and no-one can make an instrument that is as good as their eyes.
Cheers.
Richard Kirk
---
FilmLight Ltd, Artists House, 14-15 Manette Street, London W1D 4AP
Tel: +44 (0)20 7292 0400 Fax: +44 (0)20 7292 0401
Other related posts:
