- This topic has 8 replies, 3 voices, and was last updated 1 month, 2 weeks ago by
.
-
Posts
-
on the measurement report there is a check box saying “absolute values” when I check it the reported data seems to be worse than if it was kept unchecked as the values change drastically.
what is meant exactly by that? and which stats should I consider when trying to evaluate a calibration checked or unchecked absolute values?
By default comparisons are made in PCS, all color cocordinates are transformed to PCS, “lingua franca” for profiles.
Absolute does not do that so White 255 @ D65 will report as L100 b* -20 a* ~0. It’s easuer to make comparisons with coords tranlated to PCS with D50 where Lab is 100 0 0IDNK what you meant for “worse” if you meant decimals nobody cares, if you meant order of magnitude seems unlikely.
I mean that when the box of “absolute values” is kept unchecked all the color measurements seem to have just 0.0 DE00 color difference. However, once it is checked the color difference value would start to jump up and be definitely above 0.0 up to 10.0 in some cases!
and hence I am not sure how to make my judgement
(I do not have a screenshot right now to show the results, maybe I can do that at later time)
I mean that when the box of “absolute values” is kept unchecked all the color measurements seem to have just 0.0 DE00 color difference. However, once it is checked the color difference value would start to jump up and be definitely above 0.0 up to 10.0 in some cases!
and hence I am not sure how to make my judgement
(I do not have a screenshot right now to show the results, maybe I can do that at later time)
It seems that you may be using Erkan’s port to python 3 instead of official python2 build. That seems a bug.
I mean that when the box of “absolute values” is kept unchecked all the color measurements seem to have just 0.0 DE00 color difference. However, once it is checked the color difference value would start to jump up and be definitely above 0.0 up to 10.0 in some cases!
and hence I am not sure how to make my judgement
(I do not have a screenshot right now to show the results, maybe I can do that at later time)
It seems that you may be using Erkan’s port to python 3 instead of official python2 build. That seems a bug.
yeah that is true, I am using the adapted version by Erkan as the older version did not seem to run on my Linux machine
Any tips?
I mean that when the box of “absolute values” is kept unchecked all the color measurements seem to have just 0.0 DE00 color difference. However, once it is checked the color difference value would start to jump up and be definitely above 0.0 up to 10.0 in some cases!
and hence I am not sure how to make my judgement
(I do not have a screenshot right now to show the results, maybe I can do that at later time)
It seems that you may be using Erkan’s port to python 3 instead of official python2 build. That seems a bug.
yeah that is true, I am using the adapted version by Erkan as the older version did not seem to run on my Linux machine
Any tips?
It’s a know bug, hence not related to actual relative (PCS) vs absolute measured values behavior explained above.
For verification, since grey calibration is loaded by OS, consider running vanilla DIsplayCAL on a Windows virtual machine (guest) on your linux host. Virtua machines (VMware/virtualbox) can access USB devices.
Remember to install your host ICC display porfile in the guest through DisplayCAL interface, even copying full displayCAL storage folder fot that specific ICC to WIndows guest, so DIsplaYCAL will belive that it’s one of its profiles.if I understood your reply correctly…. you mean to say that it is rather a bug of representation and not of actual measurements?
and a way to verify that is to take the same measurements under different OS so to check the behavior of the grayscale in the report, if it is the same or not?
I think that’s what he means, yes. I think the measurement *report* is bugged in the py3 version
and a way to verify that is to take the same measurements under different OS so to check the behavior of the grayscale in the report, if it is the same or not?
Run verification on a Windows virtual machine (argyll 32 bit + python2 DisplayCAL), installing the display ICC profile from host OS into guest OS (the virtualized one) through DisplayCAL.
Virtualized OSes can use USB ports as of they were its USB
