- This topic has 25 replies, 3 voices, and was last updated 3 months, 1 week ago by
.
-
Posts
-
Yes, profile was active, but I don’t know what exactly happens with linear / missing LUT.
The laptop is a M1 Macbook Pro, so Apple Silicon mac.I told him to calibrate grey in hardware.
Does PA279CV support HW calibration? Asus web says no.
And since mac has system-wide color management, a profile is enough to correct grey system-wide.
I does not, that’s the key = 1 TRC. Profile assumes perfect grey color, but maybe not the expected gamma value.
If you stored 3TRC from actual measurements even on a well behaved display like that Apple CMS may go crazy as PS does.It’s not chance. Did you look at the report? Grey is perfect.
“I doubt the monitor is naturally that good.”
It is. That’s the point. It is written on the report. That’s why it works.
He said he had banding with the VCGT and none without. I think that speaks for itself, don’t you?
That says nothing about monitor calibration of of the box, it says something about GPU capabilities and LUT loader.
That graph looks excellent. You can’t really correct levels that low. The backlight is probably blue-ish (normal for LED). I would expect near-black to look blue-ish though, not brownish. Maybe it’s slightly correcting blue to brown. It’s probably no fault of the meter. Just a natural side effect of the backlight. I’d be pretty happy with those results. You could try profiling with more dark emphasis if you want
Yes, profile was active, but I don’t know what exactly happens with linear / missing LUT.
Input = output, do nothing
The laptop is a M1 Macbook Pro, so Apple Silicon mac.
Strange, maybe they simplified all this in new models or macOS versions.
Anyway, just for comparison, an AMD GPU using a properly behaved LUt loader should not show banding induced by VCGT calibration. >8bit luts and temproal ditheirng… it’s like display has HW calibration for grey.
Also LeDoge’s DWMLUT for WIndows does dithering on GPU shader so in troublesome GPUs, like some laptops with iGPU ( 8bit undithered), using a LUT3D to simulate an ideal version of display itself (a 1TRC matrix profile) that removes banding too.Dither is the key for this stuff to work like if was HW calibration.
I thought profiles captured grey behavior and CMS can use that for correction, no? It’s what I always do and it works. (Similar to DWMLUT). I’m thinking of 3D LUT’s, btw, not curves. Not sure about curves, but I thought they’d work the same.
And yes, I was referring to VCGT handling on that last point, not monitor calibration.
The other parts I agree with you on. I assume you know better than me, but the language barrier is making things hard.
Basically, I thought VCGT was unnecessary in color-managed environments. Am I wrong? Why would you need a 1D LUT when you already have a 3D LUT that captures the same information?
https://www.portrait.com/dtune/asc/enu/index.html
This looks like Dell Display Manager or some similar software in Benq or MSI. It’s a fast OSD switch without touching OSD controls + ICC profile association for each preset.
It does not look like HW cal mini apps provided by Portrait DIsplays for MSI or Dell monitors with HW cal…Yeah, maybe he can’t further calibrate 1D in hardware, but it shouldn’t matter with a 3D LUT, right?
I thought profiles captured grey behavior and CMS can use that for correction, no?
Yes, but you’ll suffer rounding errors from app color management engine. That’s why if display is well behaved in grey, out of teh box or after GPU calibration, better use 1 TRC “single curve + matrix”. Let color managed apps work like if was an idealized display.
***IF*** color managed app & engine uses dithering like C1/LR/ACR in windows using 3 different TRC for each channel should not cause banding. But all the other apps.
It’s what I always do and it works. (Similar to DWMLUT). I’m thinking of 3D LUT’s, btw, not curves. Not sure about curves, but I thought they’d work the same.
DWMLUT uses dithering hence no banding. That’s whay it resembles HW calibration. It’s super cool. As I said before it’s like having HW cal on any monitor.
Dither+high bitdepth is the key, no matter if it is inside monitor, on GPU 1D LUTs or running on your GPU shader/compute units like madVR or DWMLUT or inside an app color management.
Unfortunatelly some apps and engines in that last part oversimplify things and truncate to 8 bit without temporal dithering and that is the source of banding. Same on intel iGPUs or Windows OS LUT loader or i1Profiler “XRgamma.exe”: better to use DisplayCAL loader and disable the others.Semi offtopic: if you install calibrite profiler it lacks of lut loader of it’s own so it enables Windows OS GPU LUT loader which disables DisplayCAL loader. Beware & disable it on Classic Control Panel/ color managemt / advanced
Yeah, maybe he can’t further calibrate 1D in hardware, but it shouldn’t matter with a 3D LUT, right?
AFAIK there is no system wide LUT3D running on shaders for macOS liek DWMLUT. It there is one I would like to test it.
Whole Apple CMM runing on desktop is based on simple (I pressume GPU assisted) transformations between matrix profiles with 1TRC. This oversimplifies everything and can run very fast…. but only holds if display can be modeled accurately by “single cuve + matrix, black point compensated” ICC profile. That’s why DisplayCAL warns you if you change those settings.
It works out of the box for apple displays or monitors with reiable HW calibration because after calibration behavios is described that way: single curve + matrix + BPC. But if you create a profile with actual L* brightness on RGB 0, for example an ICC porfile for a HW cal on an Eizo 200:1 contrast for softproof…. then Apple CMM does not knwo how to handle out of gamut blacks and clips instead of doing the potentially more complex “black point compensation” like Photoshop does.
I think that Firefox suffers the same on Windows if you create a profile with no BPC. They are meant for speed after all.It’s a cheap display from Asus, it doesn’t support HW calibration. I mean it came from factory with a HW calibration report, so maybe it has some locked-down LUT inside which cannot be written by the user. That app is Windows only and as I see, it doesn’t really do anything, just switches profiles.
In the OSD I can access:
– Six-axis Hue
– Six-axis Saturation
– Gain
– Offset
– White point (already set at 6500K)All I did was set the RGB Gain, nothing else.
So as I understand now, basically I’m using this display uncalibrated and I’m just lucky that it’s mostly linear. But for a different display, I’d probably need to use calibration, even if it introduces banding, is that right?
Attached is a gray report.
It’s a proart! Not what I’d call cheap, but that’s relative. By, “calibrate grey in HW,” I meant as much as you can, given whatever controls you have access to. White point, gain/offset, and EOTF (either preset or 10/20-point). I wasn’t thinking of uploading a LUT, but of course if you can, even better. Luckily, your monitor’s grey is already well-calibrated in HW, so the lack of controls isn’t a problem. I was assuming the profile would take care of it, anyway, which it does (?), albeit roughly (if I’m understanding Vincent). For me, rough is good enough for system-wide and the apps that need accurate color (mainly mpv in my case), I assume use competent CMS’s.
I wouldn’t call it “uncalibrated.” It’s factory calibrated pretty well and you’ve already lightly calibrated by setting the color temp/RGB gain (maybe try doing offset too). If macos is using the profile, it’s also calibrating, just maybe not grey and not very well (did I get that right, Vincent?)
For a poorly-behaved display (like mine), I do exactly like I’ve recommended here. I do the best I can for grey in hardware and just profile (XYZ LUT). I live with the inaccuracies on the desktop because it usually doesn’t matter much. Banding (IMO) is more inaccurate than small grey errors. I use nvidia, so VCGT causes banding. You can hack dithering, but it’s flakey. DWMLUT (AFAIK) is just one hardcoded transform for the whole desktop regardless of source color space, incurs a significant performance penalty, and can trigger anti-cheat. So that doesn’t work for me either. AMD cards are fine, to my knowledge, but i use nvidia.
Now I just have to figure out if mpv’s CMS uses the XYZ LUT information or if I need to make my own 3D LUT’s. This stuff is all waaay too complicated
