D50 vs D65

[Michael_Eddington]

I was comparing Lab measurements exported out of Profilemaker MeasureTool ast both D50 and D65, then attempted to do the same within a ColorThink worksheet from a single spectral data set, but got vastly different results. I would expect to get the same results.

In the first instance I open a spectral measurement file in MeasureTool and exported both D50 and D65 Lab values, then dropped in a Colorthink worksheet and compared them

Next, I dropped the spectral file in a worksheet, then dropped the same measurement file again, and changed the Color setup to CIEstdD65 (6430K) and compared. I was expecting near the same delta values as in the first method.

There appears to be a major difference between D65 out of MeasureTool and CIED65 (6430) in Colorthink…anyone care to comment?

Results below:

Exported from MeasureTool and compared in ColorThink:

dE Report

Number of Samples: 1617

Delta-E Formula dE76

Overall - (1617 colors)

Average dE: 1.49
Max dE: 3.70
Min dE: 0.00
StdDev dE: 0.82

Best 90% - (1454 colors)

Average dE: 1.32
Max dE: 2.61
Min dE: 0.00
StdDev dE: 0.69

Worst 10% - (163 colors)

Average dE: 2.95
Max dE: 3.70
Min dE: 2.61
StdDev dE: 0.27

---

Calculated D50 and CIEstdD65 within Colorthink and compared

---

dE Report

Number of Samples: 1617

Delta-E Formula dE76

Overall - (1617 colors)

Average dE: 10.07
Max dE: 16.69
Min dE: 4.20
StdDev dE: 2.84

Best 90% - (1454 colors)

Average dE: 9.50
Max dE: 13.92
Min dE: 4.20
StdDev dE: 2.39

Worst 10% - (163 colors)

Average dE: 15.10
Max dE: 16.69
Min dE: 13.96
StdDev dE: 0.81

---

---

[Patrick_Herold]

Steve responded to this question in the ColorSync mailing list. I will post his response here:

Hi Mike,

I think I can speak to some of the differences you are seeing.

When exporting Lab values from MeasureTool you are getting “raw” D50 and D65 flavors of Lab, as expected.

When they are imported into ColorThink (or any other compare tool, probably) they are treated as if they are the same flavor of Lab and delta-E is calculated. As far as I know, this is not correct.

ColorThink calculates D50 and D65 as requested and then the D65 values are adapted back to D50 for the dE differencing calculations. As a result you will also see the “D65” values in ColorThink are different than what you might see in another tool. They are post-adapted values. In this manner the dE calcs reflect the correct comparison (D50 → D50) and the numbers “make sense” on screen.

To be honest, this is off the top of my head. I believe that this is the way it was designed and implemented. I haven’t tested it recently to confirm that it’s working as expected.

I hope this sheds some light on the behaviors

regards,

Steve

At 11:33 AM -0400 6/2/09, Michael Eddington wrote:
Thanks for all the information.

Let me ask this question, what differences would one expect between D65 lab values derived from spectral data versus D65 values chromatically adapted from XYZ?

Good question. The answer lies in metamerism. (illuminant metamerism specifically).

Basically adapting from one XYZ flavor to another ignores any spectral interaction between the illuminant(s) and the sample.

Calculating D65 with the D65 illuminant captures the effect of the illuminant on the sample. The difference is obviously important.

The term Metamerism in this case is used a little loosely but I’m OK with it (others may not be). To be correct, this kind of calculation is part of determining metamerism with another illuminant measuring the same sample.

For example, say someone is measuring sample with a handheld with D65 as the user defined illuminant, versus, spectral data exported out of MeasureTool , PatchTool or ColorThink. If someone were to set up a workflow for proofing and process control utilizing D65, what kind of issues could they potentially run into?

So MeasureTool is deriving D65 from XYZ data rather than from spectral? Why then is this only an option when spectral data is present?

It uses the spectral data and gets D65-XYZ which is then converted into D65-Lab.

regards,

Steve