Re: [Lcms-user] Camera Color Calibration

[<jc...@gm...>]

On 10/20/07, Mark <de...@ci...> wrote:
> But then, I have created profiles (LUT and matrix) for some
> industrial cameras - RAW image directly from sensor, debayered,
> linear = gamma 1.0 - and I can only say that I see a very huge
> difference in image quality between LUT and matrix profiles. Almost
> all packages for profiling digital cameras create LUT profiles
> nowadays. Why?

I guess because a LUT profile will always work, no matter how strange
the camera (fuji's two sensors per site, nikon's CMYG filters, sony's
RGBC), whereas a matrix profile requires camera filters which are
close to a linear recombination of XYZ and not too much unknown
processing before the profile is applied.

I don't know why you saw a quality improvement with a LUT profile.
I've always had better results with a matrix on the cameras I've
worked on. Image quality is hard to quantify.

> As for scanners - which really are digital cameras with one fixed
> illuminant - almost all profiling packages also create LUT based
> profiles. Why?

Most scanners are designed to scan photographic or print material
where your profile has to fix up those (very non-linear) processes as
well as the CCD + filter. Most flatbeds have narrowband filters,
rather than the broad filters in a camera, because they are designed
to measure dye density rather than colour.

John