Confused about which delta-e formula to use when calibrating and profiling?

[Ellis Vener]

∆E (delta-E) 1976 or ∆E 2000 (1:1:1)?
According to http://www.colorwiki.com/wiki/Delta_E:_The_Color_Difference

"One problem with dE76 is that Lab itself is not "perceptually uniform" as its creators had intended. So different amounts of visual color shift in different color areas of Lab might have the same dE76 number. Conversely, the same amount of color shift might result in different dE76 values. Another issue is that the eye is most sensitive to hue differences, then chroma and finally lightness and dE76 does not take this into account (since Lab does not take this into account)."

While
"Delta-E 2000 is the first major revision of the dE94 equation. Unlike dE94, which assumes that L* correctly reflects the perceived differences in lightness, dE2000 varies the weighting of L* depending on where in the lightness range the color falls. dE2000 is still under consideration and does not seem to be widely supported in graphics arts applications."

but further down it adds:
Finally, which equation should be chosen and how should it be used?

for basic / fast calculations, you can use dE76 but beware of its problems
for graphics arts use we recommend dE94 and perhaps dE-CMC 2:1
for textiles use dE-CMC

Editor's note:
Since this article was written, dE2000 has become the industry standard formula to use. dE2000 is recommended for all calculations except textiles which still use dE-CMC."

[Mark D Segal]

Hi Ellis,

It depends on the purpose being served. If all you want to see is "raw" differences between file reference values and the rendered, correctly measured values on prints of target patches with known reference values from whatever paper/profile/printer combination you are using, and printed with Absolute R.I., dE(76) is the most direct measurement of those differences available. If, however, you want to see these differences corrected for human visual perception, dE(2000) is now the accepted standard.

[Doug Gray]

Hi Ellis,

It depends on the purpose being served. If all you want to see is "raw" differences between file reference values and the rendered, correctly measured values on prints of target patches with known reference values from whatever paper/profile/printer combination you are using, and printed with Absolute R.I., dE(76) is the most direct measurement of those differences available. If, however, you want to see these differences corrected for human visual perception, dE(2000) is now the accepted standard.

dE2000 and dE1976 are both functions of two L*a*b* triplets. dE1976 has the advantage that you can get close by inspection of the two L*a*b* triplets as it is just the sqrt of the sum of the squares of the differences of each component. dE2000, OTOH, is great at providing a perceptual difference but is complex enough that it's really hard to just look at the L*a*b* values and get anywhere close.

One of the more interesting things to do is look at the dE2000 for a change of 1 dE1976 on a circle of constant L* as you rotate around the a*b* plane. Quite a change depending on hue.

[Jim Kasson]

Ellis, it's not clear to me what you want to use a color-difference metric to accomplish here/ If you could explain in detail exately what you're trying to do, it would help.

I don't think either measure is intended to be used for large differences, although we do that all the time in photography.

Jim

[Ellis Vener]

Ellis, it's not clear to me what you want to use a color-difference metric to accomplish here/ If you could explain in detail exately what you're trying to do, it would help.

I don't think either measure is intended to be used for large differences, although we do that all the time in photography.

Jim

I was in the middle of calibrating and profiling my display when I decided to dig in a bit deeper into an explanation of the two ∆E  options in i1 Profiler. Mark Segal (Hi Mark!) confirms that using the ∆E 2000 is the proper way to proceed.

[GWGill]

As a general observation, DE76 over estimates delta's of more saturated colors. DE94 and DE2000 make adjustments that leave delta's of neutral colors largely unchanged compared to DE76, while reducing delta's of saturated colors.

[digitaldog]

I'm almost always looking for smaller dE differences between device values in ColorThink Pro or BabeColor (CT&A) so I stick with dE 2000.

[Mark D Segal]

I'm almost always looking for smaller dE differences between device values in ColorThink Pro or BabeColor (CT&A) so I stick with dE 2000.

Again, it gets back to the basic purpose of the exercise. If it's to better replicate differences correlating as well as possible with human visual perception of those differences yes, dE2000. If it is to note simple differences of values between one measurement and another, dE(76) is straightforward and does that.