Re: Digital Camera Profiling
Re: Digital Camera Profiling
- Subject: Re: Digital Camera Profiling
- From: Igor <email@hidden>
- Date: Wed, 26 Sep 2001 23:36:31 +0200
John Gnaegy wrote:
Profiling is aimed at describing the behavior of a device. With a
camera, profiling describes the relationship between a color of light
entering the lens and its corresponding value written to the image
file, that's the basic idea.
Yes that'll tell us what the camera sees. But will it tell us what our
eyes see in the same circumstances? nope. Our eyes simply don't work
that way. Talking about our eyes in fact inevitably involves talking
about our vision system wich includes the brain. And the brain plays
many dirty tricks on colours.
Meaning given some bits in an image file, how can I recreate on
another device (a display) what that apple would have looked like if I
had been looking over the shoulder of the photographer.
There you have it: No way you can ever recreate on a device what that
apple would have looked like if you had been looking over the shoulder
of the photographer.
One of my teachers used to say: every photo is a lie. How true that
is....
Notwithstanding an often made comparison: our vision system does not
work quite the same way a camera does. And an eye is far from being a
colorimeter.
That seems to me an exaggeration given that profiling simply describes
the device. If instead you're trying to figure out what objective
color each object in a photo is, what Pantone chip represents that
leaf, etc, then I'd agree yes, that would be a hugely complex task
since reflected light from each object affects its surrounding
objects. But that's not "profiling" in the color management sense of
the word.
If we're not trying to figure out what the 'objective' colour of an
object is. Then what are we trying to figure out?, the 'subjective'
colour?
Let's take a coca-cola can. You 'see' coca-cola red, right? You would
like to reproduce this coca-cola red on a device. But the colour is
subjective: it only exists in your brain. (This is not some wild
hypothesis, I'm talking science here. I'm talking about the essence of
colour constancy) If it only exists in you brain, how can you reproduce
it on a device? You simply can't.
But yes, you can produce an image that exactly resembles a coca-cola
can. In fact you can make many different images that all exactly
resemble a coca-cola can. But if measured, you'll find that all cans are
of different colour. In fact: you're not likely to find the exact
measured value of coca cola red In any of these images. So wich image is
true? Wich image represents the right colour(s)? It's all strictly in
the eye of the beholder.
You can make a camera profile, and it might be usefull. But you can't
use it to determine wich colours are 'right'. And isn't that what
colour-management is all about? To reproduce the 'right' colours? If a
camera is supposed to replace our eyes, we should be aware that colour
management as we know it does not apply.
Igor Asselbergs