- This topic has 11 replies, 3 voices, and was last updated 9 months, 3 weeks ago by
.
-
Posts
-
Hello,
I am trying to perform hardware calibration of my ViewSonic VP2785-2K. The latest version of ViewSonic’s Colorbration+ software finally doesn’t crash with my Calibrite ColorChecker Display Plus colorimeter, and gives relatively normal results (previous versions used to crash most of the time, or produce horrible results, when worked). Depending on target luminance, the Delta E of my calibration efforts as reported by Colorbration+ after the calibration completion varies between 0.84 and 1.2, which theoretically should be a pretty acceptable result. The thing is that when I compare my calibration results with the factory Adobe RGB setting from monitor’s display, it becomes evident that the factory calibration is better – a black to white gradient with factory calibration doesn’t have any color casts and the gradient is smooth, while my calibration has a slight overall reddish cast and some stripes in the gradient are even more red or greenish, also has a slight banding in the shadows. The report from factory calibration I got, when bought the monitor, says that the average Delta E of factory calibration was 0.64, so even the numbers support my visual observations. The report from my calibration shows that the gray samples produce highest Delta E – up to 4, while in the factory report all grey samples were below 2, so it is understandable that there are issues with achieving neutral greys. Factory Adobe RGB setting though is a little darker for my taste, that’s why I would like to make a custom calibration with a brighter target luminance. Also I would like to justify the purchase of the pretty expensive colorimeter.
Tried also a software calibration with Calibrite Profiler, but it produced even worse results.
Is it normal that I cannot achieve at least the same results as factory calibration, or it is ViewSonic’s sh***y calibration software?
OS is Windows 10, latest update, latest monitor firmware, latest software versions, also I make sure ViewSonic’s vDisplay Manager is not running, when calibration is running. Tried different numbers of test patches, different target luminance, but no combination went below Delta E of 0.84.
Calibrite Display Plus HL on Amazon Calibrite Display SL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Hello,
I am trying to perform hardware calibration of my ViewSonic VP2785-2K. The latest version of ViewSonic’s Colorbration+ software finally doesn’t crash with my Calibrite ColorChecker Display Plus colorimeter, and gives relatively normal results (previous versions used to crash most of the time, or produce horrible results, when worked). Depending on target luminance, the Delta E of my calibration efforts as reported by Colorbration+ after the calibration completion varies between 0.84 and 1.2, which theoretically should be a pretty acceptable result. The thing is that when I compare my calibration results with the factory Adobe RGB setting from monitor’s display, it becomes evident that the factory calibration is better – a black to white gradient with factory calibration doesn’t have any color casts and the gradient is smooth, while my calibration has a slight overall reddish cast and some stripes in the gradient are even more red or greenish, also has a slight banding in the shadows. The report from factory calibration I got, when bought the monitor, says that the average Delta E of factory calibration was 0.64, so even the numbers support my visual observations. The report from my calibration shows that the gray samples produce highest Delta E – up to 4, while in the factory report all grey samples were below 2, so it is understandable that there are issues with achieving neutral greys. Factory Adobe RGB setting though is a little darker for my taste, that’s why I would like to make a custom calibration with a brighter target luminance. Also I would like to justify the purchase of the pretty expensive colorimeter.
Low quaility panels and low quiality QC on backligfht assembly like those from Dell, Benq, LG, Asus, Viewsonic.. on their WIDEGAMUT displays need more measurement steps on uncalibrated panel behavior. Also WP drifts as you move brightness setting.
Most vendor HW calibration solutions fail to recognize the low quaility materials they are working with, so they usually assume a near ideal behavior of uncalibrated display that doe snot match reality.
Tried also a software calibration with Calibrite Profiler, but it produced even worse results.
Unlikely (22 steps in grey ramp), but possible. The culprit is usally your graphics card, buy an AMD and use displayCAL calibration loader.
Xrite GPU calibration loader (windows) truncates calibration to 8bit causing banding, those color strips. AFAIK newer M GPUs from apple do not cause such errors as AMD ones.An universal alternative in Windows is to use DWMLUT, which as AMD cards do dither, but DWMLUT does dither on GPU shaders and does not use GPU LUTs, hence can avoid HW limitations in some GPUs like intel iGPU.
You need to create a LUT3D for DWMLUT, destnation colordspace is custom profile (for Xrite profiles do not use abs colorimetric intent), source colorspace is target calibration. If you need to simulate native gamut HW cal, use as source profile an idealization of custom profile (syth profile, matrix, single curve/gamma TRC).Is it normal that I cannot achieve at least the same results as factory calibration, or it is ViewSonic’s sh***y calibration software?
It’s a mix as explained above, 45% bad monitor, 45% bad HW calibration software from vendro, 10% unsuitable GPU for calibration
OS is Windows 10, latest update, latest monitor firmware, latest software versions, also I make sure ViewSonic’s vDisplay Manager is not running, when calibration is running.
try DWMLUT as explained. Also disable xrite gamma loader form startup.
Tried different numbers of test patches, different target luminance, but no combination went below Delta E of 0.84.
Your issues are not related at all with a mean error of 0.84. They are related to bad “grey range” and teh cause is explained above. DIsplayCAL HTML report (grey balance+ RGB) or prad.de review show it.
Note: when using DWMLUT you must set as default display profile the ICC profile you are trying to simulate (source colorspace).
Tried also a software calibration with Calibrite Profiler, but it produced even worse results.
Unlikely (22 steps in grey ramp), but possible. The culprit is usally your graphics card, buy an AMD and use displayCAL calibration loader.
Xrite GPU calibration loader (windows) truncates calibration to 8bit causing banding, those color strips. AFAIK newer M GPUs from apple do not cause such errors as AMD ones.My graphics card is Radeon RX 6400, so may I try first this solution, as it seems far easier than the DWMLUT route, which brings a lot of unknowns for me, reading what you have written further down? I guess I need third-party loader only when doing software calibration either with Calibrite Profiler or Cargyll/DysplayCal? I was not aware to this moment that this loader is needed and running- I was thinking that the software calibration process only creates a profile, which contains both monitor profiling data and LUT, instead of a profile for hardware calibration, which contains only the monitor profiling data, and I was thinking that Windows is responsible to load that data from the profile. How using a third-party loader differs from what I described in my previous sentence?
OS is Windows 10, latest update, latest monitor firmware, latest software versions, also I make sure ViewSonic’s vDisplay Manager is not running, when calibration is running.
try DWMLUT as explained. Also disable xrite gamma loader form startup.
I guess it is XRGamma.exe, located in ViewSonic/Colorbration folder – this is the only loader I see in startup, and as I mentioned, I was even not aware it is there. Strangely, it is under the ViewSonic folder, instead of the X-Rite folder.
Tried different numbers of test patches, different target luminance, but no combination went below Delta E of 0.84.
Your issues are not related at all with a mean error of 0.84. They are related to bad “grey range” and teh cause is explained above. DIsplayCAL HTML report (grey balance+ RGB) or prad.de review show it.
Which prad.de review do yo mean? “Bad grey range” – this is related to monitor’s limitations, monitor manufacturer calibration software limitations, graphic card limitations or calibration device limitations?
Thank you for taking the time to reply extensively!
Tried also a software calibration with Calibrite Profiler, but it produced even worse results.
Unlikely (22 steps in grey ramp), but possible. The culprit is usally your graphics card, buy an AMD and use displayCAL calibration loader.
Xrite GPU calibration loader (windows) truncates calibration to 8bit causing banding, those color strips. AFAIK newer M GPUs from apple do not cause such errors as AMD ones.My graphics card is Radeon RX 6400, so may I try first this solution, as it seems far easier than the DWMLUT route, which brings a lot of unknowns for me, reading what you have written further down? I guess I need third-party loader only when doing software calibration either with Calibrite Profiler or Cargyll/DysplayCal? I was not aware to this moment that this loader is needed and running- I was thinking that the software calibration process only creates a profile, which contains both monitor profiling data and LUT, instead of a profile for hardware calibration, which contains only the monitor profiling data, and I was thinking that Windows is responsible to load that data from the profile. How using a third-party loader differs from what I described in my previous sentence?
Disable all Xrite/caibrite “gamma loaders” (Task manager/Startup). Calibrate with DisplayCAL, install profile and use recommended settings when asking for loader at the end of process so DIsplayCAL install its loader.
Also, due to Windows OS limitation if display goes to sleep/standby, it is truncated to 8bit, no solution unless restart, although DIsplayCAL loader should be less banded even in that situation.
Displaycal LUT loader will load 16bit VCGT grey calibration into GPU LUT. Your AMD card will load that tables to its >8bit LUTs and then dither, then to output. It shloudl be smooth (non color managed gradient) unless there are somo other issue in your system.
VP2785-2k looks like a WLED PFS, adobeRGB flavor (missing in i1Profiler, Calibrite and Viewsonic) so use bundled correction HP Z24x G2 for xrite i1d3 colorimeter in DisplayCAL. Import them (auto) if you do not have them, then select ot manually. Otherwise measurements will be wrong and whitepoint will be different taht what you asked. (note: Colorimeter correction is not causing issues in your grey ramp)
OS is Windows 10, latest update, latest monitor firmware, latest software versions, also I make sure ViewSonic’s vDisplay Manager is not running, when calibration is running.
try DWMLUT as explained. Also disable xrite gamma loader form startup.
I guess it is XRGamma.exe, located in ViewSonic/Colorbration folder – this is the only loader I see in startup, and as I mentioned, I was even not aware it is there. Strangely, it is under the ViewSonic folder, instead of the X-Rite folder.
Old Dell HW cal solution and Viewsonic are based on i1Profiler, so it is using its loader even if it is using HW cal (so VCGT data to LUT is linear)
Tried different numbers of test patches, different target luminance, but no combination went below Delta E of 0.84.
Your issues are not related at all with a mean error of 0.84. They are related to bad “grey range” and teh cause is explained above. DIsplayCAL HTML report (grey balance+ RGB) or prad.de review show it.
Which prad.de review do yo mean? “Bad grey range” – this is related to monitor’s limitations, monitor manufacturer calibration software limitations, graphic card limitations or calibration device limitations?
All of them combined. Your grey A could be less than 1dE to ref and your grey B could be less tahn 1dE to reference, but color (exclusing brightness) difference between A and B is high, hence the rainbow in uncalibrated or poorly HW calibrated.
Color banding after GPU calibration may be not noticeable as grey range measurement because it may not measure those greys. It is spotted in a “no color managed gradient” (like MS paint opening a bandless gradient).
Color bands in color managed software like PS could bve caused by rounding errors, hence calibration in i1Profiler to “matrix profile” or “matrix + single curve /gamma” in DIsplayCAL should minimize color bands (ypu keep the color bands caused by GPU calibration or poor HW cal). Non color bands in gradients in color managed software is due to rounding errors and cannot be avoided unless that app offers dithering (ACR, LR, C1) or driver assited dither (like 10bit OpenGL output in PS, dithered in driver to an 8/10bit connection)
Thank you for taking the time to reply extensively!
VP2785-2k looks like a WLED PFS, adobeRGB flavor (missing in i1Profiler, Calibrite and Viewsonic) so use bundled correction HP Z24x G2 for xrite i1d3 colorimeter in DisplayCAL. Import them (auto) if you do not have them, then select ot manually. Otherwise measurements will be wrong and whitepoint will be different taht what you asked. (note: Colorimeter correction is not causing issues in your grey ramp)
Does it mean, that when I use both ViewSonic software or Calibrite software, they are missing the correction information for my monitor, which could be one of the reasons for poor calibration results? If that is the case, it is outrageous! How could I prove that? Look at the .EDR files or something else? Could I actually try to fix that somehow and make their software work with the applicable corrections, so I still stay on the hardware calibration route?
“HP Z24x G2 correction” is bundled where, and how do I import it in DisplayCAL?
Color bands in color managed software like PS could bve caused by rounding errors, hence calibration in i1Profiler to “matrix profile” or “matrix + single curve /gamma” in DIsplayCAL should minimize color bands (ypu keep the color bands caused by GPU calibration or poor HW cal). Non color bands in gradients in color managed software is due to rounding errors and cannot be avoided unless that app offers dithering (ACR, LR, C1) or driver assited dither (like 10bit OpenGL output in PS, dithered in driver to an 8/10bit connection)
I am looking at the B/W gradient only in PS, and there were both color and non-color bands with my software calibration. With factory Adobe RGB calibration there are no color bands and almost no non-color bands. My graphic card output is set to 10 bit.
What mode should I select in the monitor, if using only software calibration? I guess the best thing would be to do a full reset on the monitor?
VP2785-2k looks like a WLED PFS, adobeRGB flavor (missing in i1Profiler, Calibrite and Viewsonic) so use bundled correction HP Z24x G2 for xrite i1d3 colorimeter in DisplayCAL. Import them (auto) if you do not have them, then select ot manually. Otherwise measurements will be wrong and whitepoint will be different taht what you asked. (note: Colorimeter correction is not causing issues in your grey ramp)
Does it mean, that when I use both ViewSonic software or Calibrite software, they are missing the correction information for my monitor, which could be one of the reasons for poor calibration results?
Read what I wrote, wrong EDR => wrong whitepoint, this is the expected biggest error. This is a measurement error
Grayscale issues, color tint in grey different than white, is not caused by EDR. It is not a measurement error. It is a calibration software error (usually too few measurements on bad quality displays)
If that is the case, it is outrageous!
Yes, it’s common
How could I prove that? Look at the .EDR files or something else?
Yes, but i have not checked latest “Colorbration” solution. Latest time i checked it used GB-LED EDR (RG_phpshor)
To formally prove that you’ll nee an spectrophotometer and measure white, or trust community and take a look on displaycal colorimeter correction databaseCould I actually try to fix that somehow and make their software work with the applicable corrections, so I still stay on the hardware calibration route?
You’ll need full spectral data replacement (Eizo APAC way) or try a replacement with same file name but different spectral data size.
“HP Z24x G2 correction” is bundled where, and how do I import it in DisplayCAL?
Tool / correction / import, auto for i1d3. Then choose that HP.
Color bands in color managed software like PS could bve caused by rounding errors, hence calibration in i1Profiler to “matrix profile” or “matrix + single curve /gamma” in DIsplayCAL should minimize color bands (ypu keep the color bands caused by GPU calibration or poor HW cal). Non color bands in gradients in color managed software is due to rounding errors and cannot be avoided unless that app offers dithering (ACR, LR, C1) or driver assited dither (like 10bit OpenGL output in PS, dithered in driver to an 8/10bit connection)
I am looking at the B/W gradient only in PS, and there were both color and non-color bands with my software calibration. With factory Adobe RGB calibration there are no color bands and almost no non-color bands. My graphic card output is set to 10 bit.
As explained before, calibration bands needs to be insected non color managed.
“non color bands” asuming perfect factory AdobeRGB mode + AdoberGB as display profile + PS, it is caused by rounding errors on undithered output and it’s unavoidable as long as display TRC and image profile TRC do not match with 8bit image.
What mode should I select in the monitor, if using only software calibration? I guess the best thing would be to do a full reset on the monitor?
I’d try DisplayCAL on user/custom color mode, do full DisplayCAL calibration and choose “matrix + single curve” as display type.
Also I’ll put here a report, maybe your overreacting regarding grayscale (but whitepoint error may be real)
Installed DispCAL, imported the HPZ24x G2 correction (along all other corrections, available in Colorbration+ and Calibrite folders), and executed calibration with default settings and the HP correction. Didn’t find how to choose “matrix + single curve” as display type so it was with the default option there. Left the option to automatically load calibration at start-up, although I don’t see any new item in Windows startup, which would load it, unless it is implemented as a Windows service, instead of an application. Calibration results look better than software calibration with Calibrite Profiler, and I would say, maybe even than the hardware calibration with Colorbration+. Highlights and midtones show only slight color banding, but the transition between upper shadows and lower midtones is quite abrupt and has a reddish tint + banding. It is maybe closest to the factory Adobe RGB setting from all my efforts so far, but still a tad worse. Attached here are initial settings with which I start the calibration, available profile corrections, what should be the report (profile info), the end screen after the report and the profile gamut.
OK, so I found the “matrix + single curve” setting and used that one in a new calibration attempt. Also set the speed to “slow” and used also black and white level drift compensation with HPZ24x G2 correction (see the attached screenshots). Now the gradient is almost as smooth as the factory Adobe RGB setting, but still there is an overall reddish cast, which is noticeable in both B/W gradient, as well as on a regular color photo (confirmed by my wife, so not only my visual system issue here). I was able to get rid of the reddish cast by manually reducing red gain from 100 to 94 – now my calibration result and the factory Adobe RGB look almost identical. Unfortunately, although last nigh I saved the report after the calibration, now Windows finds it, but is not able to open it, as if it is missing. If needed I can do the calibration again and upload it. Wondering also to try some other correction (see the attachments in my previous comment).
Then use community CCSS for ViewSonic VP2785-2K: https://colorimetercorrections.displaycal.net/ instead of generic one. if whitepoint does not match use visual apporach as you did.
Also all you have wrote does not mean that AdobeRGB preset’s whitepoint is better. it could be actually greener than D65.
On a well behaved i1d3 applying custom CCSS 3nm and measurong will solve the issue. For better assesment you’ll need a high end spectro and then it looks redder but reads D65… it’s you and your wife visual system “fault”. On a typical normal distribution (for each cone) there will be a significative amount of people further than 1 std dev.Anyway if it looks redder/greener to you then use a visual approach like you did, but do not use abs colorimetric LUT3d with that profile.
WHen using visual approach for WP you must ignore “assumed vs measured dE” in reportsdue to Windows OS limitation if display goes to sleep/standby, it is truncated to 8bit
I was wondering why the hell Displaycal reports 8-bit seemingly randomly and why I need to reboot every time I calibrate. Useful piece of info here Vincent, thanks. That’s pretty broken from Microsoft, although not entirely surprising
