Can someone explain to me what the profile checker figures actually mean?
The checker gives results for
dE, Peak Value, Min Value, and standard deviation
I can come up with two possible interpretations:
1. The statistics reflect how far the *actual* patches on the test shot are from the *generated curve*;
in which case they are a measure of how well the test curve fits the actual test shot.
2. The statistics reflect how far the *corrected* patches on the test shot are from a curve for a theoretically "perfect" target,
in which case they are a measure of how well the test curve actually corrects.
There may be other interpretations.
Can someone tell me what the statistics really are about?
Thanks,
Gary
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Input to the profile checker is an ICC profile, a set of device values (e.g. RGB numbers) and a corresponding set of "true" XZY or CIELAB values. Basically the checker compares the given "true" CIE values to the predicted CIE values (predicted by applying the profile to the given device values) and reports the color difference for each sample, and also accumulated statistics of the color differences like average, min, peak, standard deviation, etc. over all samples.
Additionally there is an option to compute a simple correction model (e.g. luminance scaling, 3x3 matrix in XYZ space,…) and to apply this correction model to the predicted CIE numbers, before comparing them to the "true" CIE numbers. The idea behind this functionality is to find out whether the profile has possibly a systematic error which can be described by one of these simple models.
Regards,
Gerhard
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Can someone explain to me what the profile checker figures actually mean?
The checker gives results for
dE, Peak Value, Min Value, and standard deviation
I can come up with two possible interpretations:
1. The statistics reflect how far the *actual* patches on the test shot are from the *generated curve*;
in which case they are a measure of how well the test curve fits the actual test shot.
2. The statistics reflect how far the *corrected* patches on the test shot are from a curve for a theoretically "perfect" target,
in which case they are a measure of how well the test curve actually corrects.
There may be other interpretations.
Can someone tell me what the statistics really are about?
Thanks,
Gary
Input to the profile checker is an ICC profile, a set of device values (e.g. RGB numbers) and a corresponding set of "true" XZY or CIELAB values. Basically the checker compares the given "true" CIE values to the predicted CIE values (predicted by applying the profile to the given device values) and reports the color difference for each sample, and also accumulated statistics of the color differences like average, min, peak, standard deviation, etc. over all samples.
Additionally there is an option to compute a simple correction model (e.g. luminance scaling, 3x3 matrix in XYZ space,…) and to apply this correction model to the predicted CIE numbers, before comparing them to the "true" CIE numbers. The idea behind this functionality is to find out whether the profile has possibly a systematic error which can be described by one of these simple models.
Regards,
Gerhard
Thanks, that clears it up.