- This topic has 83 replies, 5 voices, and was last updated 9 months, 2 weeks ago by
.
-
Posts
-
Vincent wrote:
-That CR200 is not useful anymore, at least for dark readings.
It is a CS-200 and at $14,000 US a pop, I tend to trust it more than any of my other instrument.
Could be I have a defective monitor.
If two or more of these othe “lesser devices” measure <<<expected>>> contrast ratio at factory settings… the chances are not good for that device WHEN MEASURING BLACK
i1Pro2/3 are not so great for dark readings, but they will certainly able to measure 270nit /1000:1 CR.Even if it is actually not able to measure dark patches accurately it can be used as reference device for colorimeter corrections as long as it is able to measure properly on white.
So an accurate solution for your displays is feasible in $200-300 range, much more if you want a Klein.Could be my GeForce GTX3080 Ti is fucked.
“Misconfigured”.
Google about video levels output instead of full 0-255 range when your display expects 0-255 (typically, some displays offer video level input switch on OSD, but usually in different market segment).
There are visual test images that you can open on paint to check is something is not OK.Here is some measurements to chew on…
Using SPOTREAD -e
on a Black patch ————————————
i1pro 2002 : 0.001611 0.038651 0.374939
i1pro : 0.25 0.23 0.53
i1pro UVcut: 0.24 0.22 0.52
i1pro2 : 0.25 0.23 0.54
CS200: 0.62 0.59 0.85
i1d3: 0.28 0.26 0.58
Monaco Optix: 0.30 0.28 0.60on a White patch ——————————
i1pro2002 : 76.70 79.79 66.50
i1pro : 83.02 85.56 73.40
i1pro UVcut: 77.45 79.41 69.60
i1pro2 : XYZ: 79.42 81.71 72.50
CS200: 87.76 89.64 76.72
i1d3: 83.92 87.32 74.04
Monaco Optix: 83.96 85.43 73.93I should probably send in my CS-200 for recertification but at $1,000 USD a pop, considering I’m not doing consultant work anymore and my name is not Elon Musk… I’m hesitant., Vincent.
Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Here is some measurements to chew on…
Using SPOTREAD -e
on a Black patch ————————————
i1pro 2002 : 0.001611 0.038651 0.374939
i1pro : 0.25 0.23 0.53
i1pro UVcut: 0.24 0.22 0.52
i1pro2 : 0.25 0.23 0.54
CS200: 0.62 ****0.59**** 0.85
i1d3: 0.28 0.26 0.58
Monaco Optix: 0.30 0.28 0.60on a White patch ——————————
i1pro2002 : 76.70 79.79 66.50
i1pro : 83.02 85.56 73.40
i1pro UVcut: 77.45 79.41 69.60
i1pro2 : XYZ: 79.42 81.71 72.50
CS200: 87.76 89.64 76.72
i1d3: 83.92 87.32 74.04
Monaco Optix: 83.96 85.43 73.93XYZ? that would reduce to half the error we see from manufacturer spec (~1000:1). But going from ~350:1 to native white static contrast ratio there are other additional sources of error (the non exclusive error list shown in previous posts).
As said many times in this thread GPU levels misconfiguration on your OS may account for a big bite of the missing contrast. Google about it. It’s up to your GPU config and may vary between versions.Once fixed, if it is a GPU levels misconfiguration, redo the readings with the other devices and see how blind is that CS200 in the dark when the others read closer to 1000:1.
If that CS200 are still bad in comparison but you wish to use it for the full calibration process (unwise if you had other devices at your disposal), and calibration (VCGT applies to GPU on that monitor output) leaves no visible color tint (so most of the error is in “Y” of xyY, no visible color tint on dark greys with calibration applied, non color managed like in MS paint with lagom lcd gradient), ****since Y readings would be fake near black****, then use BPC compensation on profile. Let color managed apps believe display black is an ideal black like in any other >600:1 display with default profile configuration in most calibration apps (display RGB 0 => L* 0)
Actual black level stored in display profile is useful if it is high and it is to be trusted, if it is an accurate reading. If it s not, it’s a problem that you can easily get rid of.This will save you from recertification fee, sort of.
I forgot it...are you measuring your current setting ~90nit with your RGB gains applied? the static contrast ratio would be lower.
Better do these readings at factory settings to avoid taking account of display (mis)configuration. Write down your current setting and do a factory reset. Otherwise you may not be able to appraise if there are more errors in the chain, like the GPU levels issue.Vincent wrote:
Leaving profile type (not calibration) with BPC will store RGB 0 = 0nit so when you softproof
Wire wrote:
So where do you get 0nit from?
Vincent replied:
All these idealized colorspaces have RGB 0 -> L* 0. A linear function Y= A*X gives you 0 at 0.
The reason I asked is that I have been carefully looking at the definitions for these transfer functions and I notice that sRGB, Rec.709 (1886) and L* all include an assumption of a contrast window, which results from a power curve being spliced to a linear slope near “black” where black is not zero nit, but subjective to a reference white luminance.
This implies these three transfers express a specific dynamic range in each case:
For sRGB roughly 500:1 against a standard display white.
For 709 (1886 Reference EOTF) very roughly 5000:1 against a standard display white (it’s hard to be sure because the 1886 reference EOTF is subject to a user-supplied black which will always be non-zero)
L* (Y) roughly 5000:1 against a user-supplied reference white.
In each case , zero input to TF never corresponds to zero nit.
So maybe you meant “idealized spaces such as sRGB” tend to 0 L* due to BPC in CMM, or is that not what you are saying?
In each case , zero input to TF never corresponds to zero nit.
So maybe you meant “idealized spaces such as sRGB” tend to 0 L* due to BPC in CMM, or is that not what you are saying?
No, by colorspace definition. I showed the source before, there is not need to show it again. Go to that folder, open file, dump TRC.
BPC in CMM deals with out of gamut dark colors in device colorspace. If my display profile “lies” about “infinite contrast” (BPC in displaycal, not in CMM), there is no need of BPC in CMM (and some CMM do not even have this feature, as explained several times before)… that’s the reason it’s the default configuration in lots of cal+profile apps, and its not a real issue with typical CR. But if your display ***truly*** has rised blacks you may need to store actual black in profile. That’s the difference between recommending BPC ON for displaycal if your CR200 has fake Y readings in blacks versus BPC OFF if your CR200 was accurate in blacks and display was actually 150:1
Sorry about the delay. Had to solve access to my MYIRO-1 back from VisualStudio.
So, please find attached Black and White measurements.Perhaps you are right.
I did take a reading with my PR-705 yesterday but did not bother to keep the spectral data…if you were sending RGB0 and RGB 255 to display calibrated @D50, as said previously, you are adding additional “noise” to the problem we were talking about, if display or computer is minsconfigured somehow.
You know that CS200 was adding noise in black but it does not account for all missing range.
Is that ~350:1 caused by RGB gains alone or do you have also video/full level issue in GPU configuration? That’s why it is important to take those measurements at native white & factory settings. If it is still~400-500:1 too, that is far from device’s spec =>most likely 16-235 on GPU athough there may be other additioal issues.How can you say “You know that CS200 was adding noise in black but it does not account for all missing range.”?
No “16-235” on GPU. I’m not into video.
How can you say “You know that CS200 was adding noise in black but it does not account for all missing range.”?
It’s written all over this page…
No “16-235” on GPU. I’m not into video.
Did you check on GPU confuguration? or open an test image on MS paint?
Which page are you alluding to?
I enclose my “GPU configuration” for your perusal.
Which “test image” should I open in MS paint?
Which page are you alluding to?
This one, 5 and maybe 4th too.
Again, as said before it at native white + factory settings on a reliable device like most of the ones you used for measurement you get ~350:1 on a display that its spec are ~1000:1 statci contrast ratio… there must be something messing there. It could be “display quality” yes, but it is not imposible to have this kind of hdmi range issue
This does not mean that you have to work at 1000:1 or 600:1, it is legit to seek 250:1 or something like that for the reasons explained in this thread (and better BPC off on display profile for those task).
I’m just saying that if at native white, factory settings you cannot get even close to 700-1000:1 (display spec + a generous margin) with a reliable measurement device, there is something wrong, be it display itseft or 16-235 range in GPU (even control panel is saying full range), or other unknown issue.You got about 350:1 with the Myro-1 at your warm white calibration target, RGB gains may drop it a CR little but it should not drop that much.
Re try readings at native white , factory setting with the other spectros or colorimeters, no VCGT data. If you get 700+ or maybe even 600:1, there seems to be no issue in your computer, no miscomfiguration, display behaves that way with RGB Gains and nothing can be done.
But if you still get 350:1 – 450:1 at those native white factory serttings … it should be clear for every body that something is off.I enclose my “GPU configuration” for your perusal.
Which “test image” should I open in MS paint?
You can find them in AVSForum although I do not remember thread name, it’s a set of patches from 0 to 32 and from 200 to 255 in a scale. If you see them cliped from black to RGB 16 and from 235 to 255… there is an issue. I think that HCFR or Pi Generator have this test images built in, but I’m not sure about that cb I’ve not use them recently.
Image is simple so maybe anybody could make its own in Paint. Paint is not color managed so it should render on device colorspace (so you can test this in Paint) but maybe I’m wrong and you need a direct display rendering like HCFR.
For better support about detecting video/full level issues you can try AvsForum, but add to your question that you want to test it with free software only or you may get redirected to propieraty paid software.
