Re: [colorvision_group] Meaning of delta-e
2006-03-17 by CDTobie@aol.com
I'm reposting this with an appropriate subject line; if we don't start doing that, no one is ever going to be able to use the archives to find things!
In a message dated 3/17/06 11:38:06 AM, jvlist@... writes:
My question is with delta values: Is there a minimum delta value that is visible to us humans?
Thats how delta-e values are defined: with the smallest change perceivable to the (typical, healthy) human eye under a specific set of conditions to be defined as one delta-e. These conditions are for colors in the 2 degree area of our most accute vision, patches seperated from each other, not touching. Seamlessly touching colors could be distinguished to less than a delta-e (say half a delta-e), moving a square of one color around on screen, on top of a patch of another color, adds motion to the mix, so perhaps a quarter of a delta-e could be distinguished. But earlier delta-e forumli are not perceptually uniform, so newer attempts to get the level of difference to reflect what the human eye can distinguish more closely throughout the colorspace have been created. At the moment the most reasonable compromise is Delta-e94. Anything thats fractional in Delta-e94 can pretty much be ingnored.
So, for instance, reading all the patches on the same Color Checker with a new EyeOne spectro, and a DC 1005, then calculating delta-e 94 values for all of them shows an average difference between the EyeOne and the PFP spectro of less than a delta-e94... so I'd say pretty much ignorable.
C. David Tobie
Product Technology Manager
ColorVision Business Unit
Datacolor Inc.
CDTobie@...
www.colorvision.com
In a message dated 3/17/06 11:38:06 AM, jvlist@... writes:
My question is with delta values: Is there a minimum delta value that is visible to us humans?
Thats how delta-e values are defined: with the smallest change perceivable to the (typical, healthy) human eye under a specific set of conditions to be defined as one delta-e. These conditions are for colors in the 2 degree area of our most accute vision, patches seperated from each other, not touching. Seamlessly touching colors could be distinguished to less than a delta-e (say half a delta-e), moving a square of one color around on screen, on top of a patch of another color, adds motion to the mix, so perhaps a quarter of a delta-e could be distinguished. But earlier delta-e forumli are not perceptually uniform, so newer attempts to get the level of difference to reflect what the human eye can distinguish more closely throughout the colorspace have been created. At the moment the most reasonable compromise is Delta-e94. Anything thats fractional in Delta-e94 can pretty much be ingnored.
So, for instance, reading all the patches on the same Color Checker with a new EyeOne spectro, and a DC 1005, then calculating delta-e 94 values for all of them shows an average difference between the EyeOne and the PFP spectro of less than a delta-e94... so I'd say pretty much ignorable.
C. David Tobie
Product Technology Manager
ColorVision Business Unit
Datacolor Inc.
CDTobie@...
www.colorvision.com