RE: Ooops - D65 != D65 ???
RE: Ooops - D65 != D65 ???
- Subject: RE: Ooops - D65 != D65 ???
- From: tom lianza <email@hidden>
- Date: Mon, 18 Jun 2007 21:00:32 -0400
Hi Uli,
You wrote:
"Soon after starting to create test profiles, I realized that even if
I set the 3 different X-Rite products all to exactly the same
whitepoint/gamma settings, e.g. D65 G1.8, they produced profiles with
*very* different whitepoints."
How did you determine the white point in the profile?
Here are some points to ponder
First, no matter how you set your display, the only valid white point in a display profile is D50. By definition, a profile transforms from the display space to the PCS (profile connection space) which has a defined white point at D50. This process requires a chromatic adaptation step. In a version 2 profile, the adaptation is performed using something called the "Wrong Von Kries" transform. In a version 4 profile there are a number of different adaptation methods. In order to get to the original white point data, you need to know how the the RGB 2 XYZ scaling was performed. This is described in section 6 of the ICC specification http://www.color.org/ICC1V42.pdf Appendix D of the ICC spec is also informative, but make sure you read it shortly after waking up because it certainly represents one of the most boring and arcane documents ever written. Take a moment to read appendix F, if you are still awake after Appendix D
Monitor profiles can have two basic forms, matrix-shaper and 3-d lut. All monitor profiles must have a matrix shaper section, but some manufacturers prefer to generate a 3-d lut form profile. All ICC compliant applications must be capable of using the matrix-shaper construct of the profile. Some applications will use a 3-d form if present.
I haven't looked at the profiles generated by all our products, but I know that Huey produces a profile using the Bradford Transformation Adaptation in a matrix-shaper form. I believe, but am not certain, that Profile Maker generates a matrix shaper section using Wrong VonKreis, but also generates a 3-d representation using a more optimized process. When you go back through the proper scaling you should be very close to the exact white point in all our products. This would be true even if the hardware products were vastly out of calibration. In this case, the displays would look different, but the profile conversions should be correct.
You also wrote the following:
"One additional remark: I was amazed that even in the expensive Eye-
One Pro package, the software is nothing more than a black box that
somehow produces a profile. There's no way to access the and work
with the physical measurement data at all. "
Actually, we make a free product http://www.xrite.com/product_overview.aspx?Industry=1&Segment=11&ID=719 which allows you to play with the data to your hearts content. Steve Upton also makes a nice product called ColorThink ( I think) that takes a rather honest look at profiles.
One final point: Xrite products currently have DNA from a number of different companies which have recently merged, hence there are certain differences found between applications. I am the wrong guy to be talking about product specifics, but if you have some information that you would like to share you can contact me at tlianza AT Xrite DOT com.
Regards,
Tom Lianza
Director R&D
Video and Motion Picture Technology.
X-rite
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden