- This topic has 116 replies, 12 voices, and was last updated 1 year, 3 months ago by
.
-
Posts
-
I am running into a problem that I am beginning to believe is an issue with the i1 (Calibrite ColorChecker) Display Pro when used with the LG 32EP950, which is a JOLED (RGB OLED, not WRGB) panel.
It seems that LAB readings for patches with low R values and zero for G/B are coming back as L=0,A=0,B=0 or other bad values.
This problem does not occur with my old SpyderX (which I replaced with the Display Pro until now!). And this problem does not occur with various LED panels I have, leading me to believe it isn’t the particular Display Pro unit that is defective, but more an incompatibility with JOLED panels.
I am attaching two back-to-back Measurement Reports comparing the identical LG 32EP950 measured first with DisplayPro then with SpyderX.
Look at patch 29 (R=34, G=0, B=0) reading LAB=0/0/0. Also patch 41 (R=68,G=0,B=0) and patch 75 (R=85,G=0,B=0).
In a test wtih LG’s own tool, there was a patch with R=65,G=0,B=0 reading LAB=0/0/0.
These patches all read OK with SpyderX.
Wondering if anyone else has seen this or has any insights.
-Ray
SpyderX Pro on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.The ony issue is in patch #29. #41 and #75 are measured properly.
Example (i1d3):
041 (68,0,0)
Predicted xyY 0.6818, 0.3165 1.3938
Measured xyY 0.7273, 0.2727, 1.1896
That means that display behavior is not linear in that zone and that a simple 1TRC matrix profile cannot predict its behavior.Regarding #29 check that computed calibration curves are not clipping black. Unload/clean VCGT, load a big patch with RGB (34,0,0) in MS paint, read it with ArgyllCMS spotread. Same with VCGT loaded.
You can use HCFR in free measurement mode with same CCSS , uses Argyll under the hood.OK…. the spotread with R=34 G=0 B=0 patch is very informative, but I think it just confirms an issue with the DisplayPro.
First using SpyderX:
With full calibration/profile: Result is XYZ: 0.952106 0.458261 0.010491, D50 Lab: 4.139452 20.461377 6.938925
With AppleRGB: Result is XYZ: 0.498402 0.228788 0.000000, D50 Lab: 2.066634 11.217904 3.563163
Then using DisplayPro — here it gets interesting.
The spotread hangs for several seconds and then returns: Result is XYZ: 0.000000 0.000000 0.000000, D50 Lab: 0.000000 0.000000 0.000000
This is with my profile or with AppleRGB!!!
So I moved it to a random place on the screen and that reads quickly just fine: Result is XYZ: 31.666800 26.687107 47.191896, D50 Lab: 58.683585 23.058471 -37.264380
If I switch over to an LED monitor (an LG 32UL950), though it isn’t warmed up, the same RGB patch reads using DisplayPro:
Result is XYZ: 0.080457 0.084680 0.149097, D50 Lab: 0.764912 -0.048130 -1.496103
Curious: How does one display the Predicted values in the measurement file?
I should add, the patch is visibly not black on either monitor. If anything, it appears (to me) closer to black on the LCD than the JOLED.
Unfortunately, I don’t have a Windows box, but I do not blame Argyll unless LG Calibration Studio licenses Argyll under the hood.
Using LG Calibration Studio’s validation with its test patch set, I see the same issue with R=63,G=0,B=0 reading Lab=0,0,0 (and working fine with Spyder on the JOLED…. both work fine on the LED…. they’re both LG monitors so the same tool runs on both).
Curious: How does one display the Predicted values in the measurement file?
https://argyllcms.com/doc/xicclu.html
or
https://argyllcms.com/doc/icclu.htmlforward or reverse (or the opposite, I have a batch so I do not have to remember). I do not remember exact output but RGB values may be 0-100 range.
Unfortunately, I don’t have a Windows box, but I do not blame Argyll unless LG Calibration Studio licenses Argyll under the hood.
Using LG Calibration Studio’s validation with its test patch set, I see the same issue with R=63,G=0,B=0 reading Lab=0,0,0 (and working fine with Spyder on the JOLED…. both work fine on the LED…. they’re both LG monitors so the same tool runs on both).
Strange behavior. Do you have a raspberry pi or can use a USB drive to boot a Linux with some ArgyllCMS installed? or request some windows laptop from sombody you know.
i1d3 is a HID device and does not need drivers. Just plug it to that windows laptop, download Argyll, unzip and run spotread on MS paint patch.
I will work on borrowing a Windows laptop and will report back. Meanwhile I wish Calibrite would make progress on my ticket. I miss Xrite.
Borrowed Windows laptop. Same results. 10 secs to spotread R=34,G=0,B=0 and gets 0’s.
spotread’ing R=0,G=0,B=34 for comparison takes 5 seconds and gets non-zero values.
Given that the same DisplayPro works fine with dark red-only patches on LED backlit panel, I am strongly suspecting it’s a tech incompatibility with JOLED panels rather than a bad colorimeter (still have heard nothing back from Calibrite support ticket). Alas, I can’t find anyone I know with their Studio photospectrometer unit to give that a try. SpyderX, though I had relegated it to storage in favor of the DisplayPro, works – and the dE2000s are repeatable/respectable enough that I may switch back to the Datacolor camp. Sad bec X-rite had previously served me well.
If there is a tech incompatibility as you claim, try to figure the boundaries at R=X,G=0, B=0 and R=X,G=Y,B=Y Y<X
Or run argyll in verbose/debug, or report these findings in Argyll maillist.I1d3 have been used for other RGB OLED and AMOLED and there was no such incompatibility
That is an excellent suggestion especially as DisplayCal seems to let one create a test chart with channel sweeps, though the sort order is slightly off. Attached are 3 Measurement reports.
JOLED with SpyderX – looks good
JOLED with DisplayPro – Red is dead <= 59,G=0,B=0 and shows high errors above that. (In an earlier LG Calibration Studio run, I get dead at R=63 as well but that was a slightly different profile).
LED with DisplayPro – looks good
A brief update:
To Calibrite’s credit, we have been engaged in trying to understand what is going on, so while it took a while, all good on the support front.
There are some interesting observations which I will take to the argyllcms mailing list shortly. The tl;dr is that -Y A (don’t use adaptive integration time) returns non-zero values, albeit with expected low accuracy even with the edr (converted to ccss) as recommended by xrite. This leads me to suspect that argyll is timing out or some such.
What is mysterious is why LG Calibration Studio also returns 0 values — perhaps their coding was inspired by Argyll’s….
Practically speaking, I’ve purchased an i1 Studio and it works great, even if slower.
