- This topic has 10 replies, 3 voices, and was last updated 1 year, 10 months ago by
.
-
Posts
-
Hello all,
I recently moved from a 27″ NEC display to a 43″ Dell U4320Q display. (I know, i need glasses…) I realize I went from a ‘good’ monitor to a ‘less good’ monitor (but I was not ready to go to a 42″ LG OLED — too affraid of burn in issues). Of course, I would love to be able to calibrate this “beast” but it is “weird”, to say it politely? The OSD only allows to select a White Point, which, in my case, is around 5000K / D50. Alas, the Black point of this monitor is “bluish” and remains independent of white point selection? Here are the best I was able to obtain using software brand “X” :
https://1drv.ms/u/s!AkD78CVR1NBqmOMv00phf6QJjugXsA?e=bKLvn5
My question : how far can DisplayCal go, in terms of calibration onthis Dell monitor? Can it iterate the R=G=B grays chromaticities all the way down from White point to Black point?
Please excuse my ignorance.
My question : how far can DisplayCal go, in terms of calibration onthis Dell monitor? Can it iterate the R=G=B grays chromaticities all the way down from White point to Black point?
Depends on calibration speed. From faster to slower it will go over an iterative process at 12, 24, 48 & 96 points in 256 greyscale.
On a 1000:1 display with the fast i1DisplayPro it may take up to 30min the slowest approach (4 iterations)
That grey calibration will be stored in VCGT of resulting ICC profile, extrapolating 96 nodes to 256.
When you make a 65point per cobe edge in a LUT3D that wikl be cipped to 65 points in diagonal. It you make a 17x17x17 LUT3D for an external lutbox greyscale correction will be clipped to 17 nodes. That is the reason why monitors with HW cal have a separate (somtimes from factory) 1DLUT to correct grey and an additional LUT3D on top of that.
Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Vincent,
17 nodes, out of an L=100 possibility means each node will represent and increase of 100/17 = 5.88, korrect?
No because monitor TRC would be closer to g2.2 or g2.4 on most scenarios, not L*. Hence deltaL increment won’t be constant in same delta RGB (/256) increment.
Anyway if you plan to run a 17x17x17 on resolve through a common GPU you can correct first grayscale (AMD gives you dithering) then apply LUT3D (but do not embed VCGT). That way you get grey correction from VCGT + GPU LUTs and full colorspace correction from software LUT3D.
If you run monitor through a decklink I’m afraid VCGT tracik won’t work but IDNK.Vincent,
Thanks for the help. My needs aren’t as sophisticated as “video”. I only live within the confines of Adobe Creative Cloud. I’m mostly concerned with viewing color “accurately” within Photoshop.
I have the possibility of calibrating to L*.
Care to explain how that will be different compared to calibrating to a 2.2 gamma encoding?
On color managed apps it does nt matter which TRC you choose… it’s color managed. “Usually” you choose the closest to most used/probable content colorspace to minimize rounding errors since rendering is done in limited precision in PS (not in LR or C1 or ACR) unless your GPU drivers can enable 10bit link between PS and OpenGL driver (just that, no 10bit end to end is needed, GPU driver can dither afterwards… the key is to avoid Photoshop low precision truncation).
Hence calibrating to L* is useful if your works are in eciRGBv2 or ProStarRGB or some colorspace close to that. Otherwise 2.2 gamma is usually closer to native TRC in monitor and closer to sRGB or AdobeRGB content.
Vincent,
I confess I am a little confused. I confess I have yet to turn on a full 10bit worflow on this Dell monitor. Right now, this is where I’m at :
https://1drv.ms/b/s!AkD78CVR1NBqmOM9tpX6TTuMiniMng?e=tnLvr0
You seem to imply that there would be no advantage of turning 10bit video on?
There is an advantage… on photoshop, because photoshop does not dither when rendering image content to color management engine to display.
But there are other ways that do not require 10bit open GL driver or 10bit end to end (for SDR images, like photos). For example if you have LR open “10bit test ramp” image (google) and go to develop module… it does need 10bit. Same for Capture one, or even Photoshop ACR if you turn off 10bit.
They do tempral dithering as an universal solution for all gpus and all monitors.Since PS does not dither is better to have that hook to OpenGL 10bit working properly (enabling it does not mean that works).
Anyway… for photography it usualyy does not matter, there are little to no smooth gradients to notince this, maybe sky.
Illustrators are more likely to need it to avoid banding and Illustrator doe snot have 10bit and does not have dithering… so bad luck. For those people better strick to 1 curve + matrix profile if their display is good enough to be described accurately that way.I don’t use Lightroom. Where would one know for sure “enabling 10bit workflow” actually works? Any ideas? On Windows 11? I have code that peeks into the gammaramp structure :
https://1drv.ms/u/s!AkD78CVR1NBqmONBvGZlOSeTO7A9Qg?e=vugDvS
As you can see in this screen capture, these 1D LUTs were created by “Profiler X” in an 8bit regime: how would you expect things to change in a 10bit regime?
Could the horizontal axis change to 1024, instead of 256, to reflect 10 bit? And how about the vertical axis : would it remain 16 bit (0-6535)?
I don’t use Lightroom. Where would one know for sure “enabling 10bit workflow” actually works?
Google “10 bit test ramp PSD TIF”, open on Photoshop or any of these editing apps.
Regarding VCGT IDNK the details, interpolation seems my best guess.
