HCT User Guide
page 29
Copyright © 2001 HutchColor, LLC
User_Guide_43.docx 2/10/19
Profile testing
Quick visual test
Assign the profile to the target scan used to build the profile. If the image looks like the target on a
properly profiled monitor, the profile is probably good. Scan some live images and view them the
same way. You may need to set white and black in Levels for best results.
NOTE
: This test assumes you have a perfect monitor profile, which is seldom the case.
Printed test
To eliminate possible monitor errors, print the target scan on a well-profiled output device. Make two
conversions, one using Intent: Perceptual and one using Intent: Absolute Colorimetric, both with
Black Point Compensation OFF.
The perceptual conversion tests the profile’s general performance but may not reproduce saturated
colors or contrast accurately due to gamut compression in the output profile. The Absolute
conversion should reproduce in-gamut grays, in-gamut colors and mid-range contrast accurately
but may not look pleasing due to gamut limitations in the output device.
If at least one of the conversions looks like the target under proper viewing conditions (D-50
transmission and overhead), the scanner profile is probably good. Repeat with live images.
If neither conversion is acceptable (allowing for limitations in the output device) there may be an error
in EITHER the scanner OR the output profile. We need to resort to more scientific testing.
CIELab test
Open the target scan in Photoshop 6 and convert it to Lab Color, as follows. Record the Lab values
in several key patches including middle grays, pastels and saturated colors.
Converting the target scan to CIELab in Photoshop
Open the reference data file in Microsoft Excel, BBEdit or another text editor and compare the Lab
values for the same patches with the values recorded in Photoshop. Ideally all Photoshop samples
should have a* and b* values within about + or – 1.0 of the reference file. Allow more tolerance in L*
values, e.g. + or – 2.0, but all L* errors should be in roughly the same direction (e.g. all high or all
low.) Extremely dark patches may show more errors due to approximations in the profiling software
or limitations in the scanner itself.