The profile checker produces three curves:
LUT prelinearization (AToB)
LUT postlinearization (AToB)
The first two are, in my case, accompanied by text that says "Red Green Blue
and the last one by text that says L* a* b*
1. Do these curves represent the derived profile, or some kind of error analysis?
2. Every profile I have generated, regardless of dE, peak difference value, and std. deviation, looks smooth.
The first always looks to be a portion of a parabolic curve, concave up.
The second is always what looks to be a portion of a parabolic curve, concave down, but with less overall curvature.
The third is always a straight line from the origin (lower left) to the upper right corner.
What can one tell by looking at the curves?
3. Every profile I have generated has the different color curves on each plot essentially superimposed on top of each other;
regardless of how "good" the curve seems to be as described by the error statistics.
Anything to be learned here?
4. Do the curves represent the results at different stages of the profiling process, and if so, should looking at,
for example, the shaper TRC give one some insight in how to better set the profiling parameters?
5. Does the postlinearization LUT curve represent the final profile?
In all cases, the profiles show straight postlinearization LUT lines, totally superimposed on each other.
Does a profile ever generate anything different, and if so, what does that mean?
6. If a profile is "bad" in the sense of being an over-specified fit and not smooth, will that show up in any of the curves?
Finally, when I run a profile using an image of a W.Faust Chart shot at high noon,
bright sunshine, the profile reports a white point. I don't know what the
lighting conditions are for the "standard" target description file,
but I have been changing to use a reference file closer to the indicated white
point on the assumption this would be more accurate.
Doing so always produces a more accurate profile as reported by the checker.
For example, if I start using the standard reference file,
the profile reports a white point near 5033K or near D50.
I then switch to using the D5010 reference file.
Is this the right thing to do?
BTW, what determines whether the white point is reported as "near 5033K" or "near D50"?
How close does the white point need to be to 5000K to be "near D50"?
However, 5000K is well below what I think the conditions of the chart shot was;
6500K seems more reasonable.
Any helpful hints as to what is going on?
Is it normal for the computed white point to be so far off?
Log in to post a comment.