Calibration and Lut settings

[Vincent]

Does that tablet have some kind of RGB gains?
Maybe you are pushing RGB gains (or equivalent control) too much from Native to get D65. But this may drop contrast to 200:1 for example from a severe green-cyan tint ~7000K to D65.

• This reply was modified 1 week, 4 days ago by Vincent.

[Vincent]

I’d try to do a factory reset and measure tablet “as is”, USB-C and check contrast.

[Ben]

The last report is probably black level matched.  Measured black luminance:
0.1357 cd/m²     .      I aim for 0.1 or less and 100 lumananc or more for 1000 contrast.     Did you calibrate white to be 80 lumance on interactive step in displaycal?

It should fully cover the blue.  https://petapixel.com/2022/01/11/huion-kamvas-pro-24-review-a-flagship-4k-pen-display-for-a-lot-less/   .

Try movie mode shown in the video I posted.   It should have d65.  It liked not default nividia.  Not RGB.   It gave a higher gamma which makes things more contrasty.   I think your screen type is qled and needs a qled profile.   If I am mixed up sorry.

[Ben]

Kubi has the profile and noted it.   A big choice though with wide gamut screens.    displayspecifications.com  is a big help.     In displaycal the display and insturment tab has a link to give info on the display types at the bottom.

[G G]

Thanks Vincent, some small progress I think. the controls wheren’t that far from the middle but were indeed having an effect
I tried a few different settings, the best seems to be everything at midrange: contrast/RGB all at 50. max ratio I had was 851:1
then I lowered the backlight to a measured 110 cd/m^2. I figured this would help the range but we’re about there, at least is not blinding. it looks very cold/bluish

everything to 50 and 110 brightness
20:38:12,505 Black level = 0.1321 cd/m^2
20:38:12,505 50% level = 24.08 cd/m^2
20:38:12,505 White level = 110.74 cd/m^2
20:38:12,505 Aprox. gamma = 2.20
20:38:12,505 Contrast ratio = 838:1

there is an HDR setting on the OSD, I was wondering if their claimed 1200:1 is using hdr but turning that on doesnt seem to have an effect on the contrast.

then I recalibrated but running the verification shows it back to 664:1

• This reply was modified 1 week, 3 days ago by G G.

Attachments:

You must be logged in to view attached files.

[G G]

interesting @Ben thanks for the link!

I don’t think I set the luminance to 80, something around 100 is okish for this but its hard to get the black level low without making the screen look really dim.
for example the movie and game settings you mentioned boost up the backlight to 100, I get a black of 0.36. in those modes contrast ratio never goes above 850:1

you are correct about the Qled, I’ve been using correction available in display cal since it was mentioned yesterday.
not sure what you mean with Kubi has the profile and noted it? I did check displayspecifications.com but unfortunately there’s nothing for huions

[Ben]

I was rushing answering before work.    Kubi noted the display type first.   Not likely to have good HDR results.   My low nits HDR tv has black at .5 almost and its 220 lumance max also.   Guesing contrast is the key to get  contrast ratio up.  Do not crank up contrast till color clips or whites.    If color clips and have to lower contrast I actualy got to raise green to get its nits up and still had to lower blue.  Red is dies on my screen at high level and peaks out at 95% of the color white.   Green up boosted my contrast from 850 to 1030 .

Start with your RGB at 50,50,50 .   Backlight at 50 to.   Pushing backlight makes odd things pop up.   And under backlight makes white not white.   You can fix backlight later.   RGB gains might work better at backlight 50.       Wish there was a temperatur setting and was hopeing movie mode would do it.  I will try searching for more help.

I was wrong.  Petapixel used color temp 6500 and adjusted rgb gains.  There contrast was at 50.    backlight was 50 and there rgb gains 50.

• This reply was modified 1 week, 3 days ago by Ben.

[Vincent]

If native WP is too far from your D65 target as you measured, contrast drop is unavoidable. You’ll have to use that way.

[G G]

Thanks @Vincent I guess this is as good as it gets. I’ll revert to correcting white via hardware first as i seem to get about the same contrast.

regarding the blue error in the 3dlut, could I just double check my last settings are correct or if there’s anything else that can be done as I’m still getting it?
in the creation dialog I choose source profile sRGB with unmodified tone curve, apply calibration and absolute colorimetric with white point scaling.

[Vincent]

From your previous image, sRGB blue is out of gamut, non correctable.

It will be nice if somebody measured that model with an spectrophotometer to find what is the reason.

[Ben]

Whats the grey sticking up in the right bottom picture with the color bar on the bottom?   Mine has that too.

[Vincent]

sRGB and other ideal/content colorspaces have infinite contrast. RGB=0 —> L*=0

Display profiles can store actual black (RGB 0 => L* X value) or fake infinite contrast (RGB0 -> L*0), what DisplayCAL “black point compensation” option (BPC).

“Smart” color engines like photoshop can “bend” in a perceptual way L*0 from image colorspace to non zero L* for RGB0, that’s called too “black point compensation”

Non smart color engines just clip those blacks because they are OoG. (that includes Apple CCM used by macOS apps)

So on a well behaved display with high enough contrast (maybe 700+ but you can force lower) we use as display profile an idealized, matrix type, perfect grey (1 TRC curve) with BPC. That will be our main ICC for color managed apps.

When creating LUT3D usually you want to capture full irregular behavior so we use XYZLUT profile type without BPC.

On a preset/memory/calibration for Phophotop (or equivalent) meant for softproof/print to screen match on a well behaved display but with limited contrast (maybe lowered to 250:1 by your own choice) we usualy use as display ICC profile  a matrix (ideal additive behavior), 1 curve (ideal grey, to avoid some unwanted rounding errors by photoshop) and NO BPC (so actual black is stored in profile’s TRC).
Photoshop knows how to deal with that non infinite contrast using BPC on its engine. Also since actual black is stored in display profile TRC when you simulate “black ink” while softproofing images black won’t be innacurately lifted up (it will be lifted up to whatever value says printer profile, but no higher)

• This reply was modified 1 week, 2 days ago by Vincent. Reason: typo

[G G]

interesting about how photoshop deals with this.

I’ve been on a bit of a rabbit hole checking a smooth black to white gradient in photoshop with various settings. my XYZlut+matrix profile in windows gives banding, particularly near the blacks and with reddish tones. after reading around the forum I tried the single curve + matrix expecting it to fix the problem but doesn’t quite, it removes the color in the bands but they’re still visible, any calibration also alters the rolloff of the values in a way that feels off, it feels like the range gets compressed, lifted and somehow not as smoothly transitioning (when compared to the standard windows sRGB profile applied which has this beautiful roll up of values in the terminator from pure black to the dark grays)

the dithering in dwmlut is great on this to the point where it almost gets rid of the problem altogether and I actually prefer the look of the XYZlut+matrix to the single curve. the rolloff of the values is still different but after playing with the 3dlut maker it seems I can get a result that is pretty close to what you’d get in the default sRGB of windows by applying a custom 2.2 relative gamma with black output offset to 0%. it looks different from “unmodified” and “apply black output offset 100%”  options. I don’t quite understand what I’m doing but it looks right/better on my monitor. 🙂

• This reply was modified 1 week, 2 days ago by G G.

[Author]

Posts