Re: colorsync-users digest, Vol 2 #553 - 15 msgs
Re: colorsync-users digest, Vol 2 #553 - 15 msgs
- Subject: Re: colorsync-users digest, Vol 2 #553 - 15 msgs
- From: Jack Bingham <email@hidden>
- Date: Mon, 01 Oct 2001 18:32:44 -0400
- Organization: Jack Bingham Studio
Just another thought on the subject of camera profiling:
Many of the posts indicate that a camera profile is useful as long as the
captured scene stays more or less the same. The more deviations in the scene
(e.g. lighting conditions etc.) the less useful the profile. Which follows
common sense.
So I wonder? What is profiled exactly when we shoot a color target and run
it through profiling software? I guess it's not the camera being profiled,
but the scene. Change the scene and the profile is useless.
Any comments? ARigorsselbergs
Once again, camera profiling has absolutely nothing to do with lighting
in any specific scene, and for that matter any specific color of light.
A camera profile maps the camera's ability to see the colors presented
to it on a target. It builds a set of corrections for those colors on
that target and thereby a set of corrections to use under all conditions
presented to the camera. Is it perfect, of course not. Does it correct
for a great deal of individual characteristics of a given camera thereby
making the final image more accurate. absolutely. Now lets take another
tack at this profiling each scene. An accurate profile is built by
comparing a data set for a specific target, with the data captured by a
camera from that target. If the target has matching white patches in all
four corners and those patches are not evenly lit, then the machinery
breaks down. If you have multiple lights on a set and they vary in color
temperature so one side of the target is warmer than the other, again
the machinery breaks down. The only way to build a valid camera profile
is with a perfectly evenly lit , or as close as possible, target with
one light or 2 color matched lights. Those white corner patches started
out matching each other on the target, and they need to match on the
final capture, or the data set will be invalid, and therefore will build
a bad profile. Given that it is impossible to just throw a target into a
scene and expect to build a profile. Now on the subject of film and
scanning I am stunned to hear anyone suggest that film sees things the
way our eyes do. My eyes must be much more tuned to digital cameras
because the colors "appear" to match the products I shoot much more with
digital captures, and my many clients seem to concur. As for profiling
cameras verses scanners it is exactly the same process. The light source
is irrelevant. We are mapping what the scanner can see from a given set
of colors. That's all.