Camera Profiling: What's wrong with this picture?
Camera Profiling: What's wrong with this picture?
- Subject: Camera Profiling: What's wrong with this picture?
- From: email@hidden
- Date: Mon, 21 Apr 2003 10:55:52 EDT
A request (plea?) to the camera manufacturers and camera profile package
vendors that frequent this list to improve reproduction and help workflows.
And to help us mere mortal photographic practitioners understand current
limitations. The recent discussion seems to be largely described near two
extremes:
1) One size fits all, that is one well-created camera profile will work
optimally for all conditions.
2) Camera profiling doesn't work well at all.
One size fits all: nonsense. Most of the camera profiling packages I've
tested tend towards relative colorimetric reproduction. Even if this were the
goal of digital photography (which it usually isn't), cameras are not
colorimeters. A camera and a person have different color responses so when
they "look" at a scene; they "see" it differently. Generally speaking, a
profile can at best provide a metameric match for one scene condition, and in
this sense the profile is optimal for that condition. The same profile
applied for images captured under different conditions will not be optimal.
That does not mean the profile won't work well, it just won't be optimal. How
well the profile works depends on how "colorimetric" the camera is, how
optimal the profile is, and how different the scene conditions are from those
used to create the profile. It is fundamental colorimetry.
Camera profiling doesn't work well at all: nonsense. Even in its present
state of development, camera profiling can indeed be useful. Making different
cameras perform similarly, improving egregious color errors, getting camera
reproduction to a quasi-stationary starting point, and using this starting
point for editing images to get the desired look (rendering), to name just a
few.
In my opinion, the real issue is the rendering. Understanding camera
colorimetry is critical to achieving accurate color reproduction. But
accuracy is not enough; rather it is just the input to the rendering process.
ICC profiles support multiple rendering intents, but camera profiles don't
seem to take this far enough, at least not yet. As a band-aid, perhaps camera
profiles should support various styles of perceptual rendering designed for
us old-fashioned photographers who are used to film and consider that kind of
reproduction satisfactory. Different renderings and/or profiles to emulate
different film-based workflows might be a reasonable transition.
But this is way to limiting and we need to think outside the film box since
digital photography offers so much more than film-based color rendering.
There is no way camera profilers can know a priori all the possible different
renderings users require. But they can make a very good start. Just as some
cameras offer scene modes, camera profilers could offer different renderings
for a number of easily identifiable classifications of imagery or workflow.
Taking it one step further, rendering tools could be built into camera
profiling packages to customize the rendering for a particular
classification, workflow, or client preference. This is more than just
profile editing.
Not only would this improve user satisfaction with digital photography but it
also would make for a huge workflow improvement. Rather than manually render
images in Photoshop to get the desired look, rendering preferences could be
built into the profiles. Of course it won't solve the problem entirely, but
it could improve productivity drastically. Some might argue that this kind of
rendering is an editorial function and not a profiling one. But then, so is
the choice of film type b&
Eric Walowit
Tahoe
_______________________________________________
colorsync-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/colorsync-users
Do not post admin requests to the list. They will be ignored.