Which color profiles for game compatibility

Whoisthisreally said:
High CCT is normal for LCD blacks. There is nothing to be done.
Click to expand...

CCT?

I don't think the amount of blue on this is normal... I had about 10 other LCD's in the past and this is the worst for "blue blacks" ... So I think it should be fixable somehow.
 
Here is an image of the TV without any adjustments on "warm2".... All I did is change to warm2 and leave RGB gain and bias at default.

This amount of blue is ridiculous for warm2 profile?!

bluerueueue.jpg
 
And this is the best I can get it, when the blacks are still blue tinted, but the whole screen is not green and the blacks are not "sick" brown color. This is warm2 with blue gain -8 and RGB bias -2 -2 -2, still very blue... but actually looks better than the more accurate looking greyscale because of the problems with brown colored blacks and the whole screen looking green.

sdfdsfsdfsdfsdfsdfsdfsdfsdff565656565656.jpg


It should be possible to get something better with a profile?

I would be happy with slightly blue tinted blacks (currently VERY blue tinted) as the TV has a blue tint due to the LED backlights anyway, 1886 gamma and normal colors above 30%.

using the TV controls my options are : blue tint OR green tint with brown blacks.
 
It would be much better to show this sort of image:

2e68020.png


As for what a profile is vs calibration, this post may help.

Also, need to see your primaries (Graphs/CIE diagram).
 
This is the best results WITHOUT getting either sick / green colored blacks OR a full screen green tint, I have a quite accurate IPS next to it so I have an idea what should look like.

454545.jpg


I can get the greyscale a lot more accurate than that, but the actual screen looks very green and the blacks looks green/brown so it is actually worse than this result even thought this measures a lot worse. I want to minimise the blue tint, but I would still prefer blue tinted blacks to green or red tint.

here is the CIE diagram of the blue tinted one I posted last. As you can see it is not accurate... I can get it more accurate.... But because of the lack of controls on this TV it ends up looking worse... because its very green tinted and the blacks are brown / green. The color controls on this TV are terrible for the price.

cie.jpg
 
Can you please just show me the results (grayscale tracking, primaries) of the combination of a hardware and dispcalgui adjustment. I don't care about blue tint, I just want to see the results for now.
 
This is the most accurate I can get it to measure although in reality as I said the whites have an obvious green tint and the blacks look brown / green color. The gamma is not set up I am just giving you an idea of what I can get with the TV controls only. I will install the profile I made the other day and post that as well.

image.jpg


image.jpg


image.jpg
 
To my eyes most of the image looks slightly green, main problem is that to fix the blacks it has almost halved the contrast. The IPS I have here does not look green compared to the TV. For example the light blue sky on the wallpaper I have looks more green than it should and greys look slightly green. On the CIE diagram the greens are wonky and the triangle does not line up with the reference.
 
Last edited:
I am still confused about what parts actually work in games and if they will fix the blue tint, I assume the gamma settings that correct the gamma to BT 1886 will work in games but will the part of the profile that fixes the blue tint work in games? Is the part that would fix the blue tint not the profile part? eg. "xxx color should be xxx" or is that loaded into VCGT with the 1DLUT?... Is the 1D LUT the calibration which will fix colors and gamma etc. as much as possible... then the profiling part is a profile of the calibration which corrects it further (but the 2nd part will not work in games). Why wouldn't using more patches help with the accuracy of the 1DLUT? Is there any way of improving the 1DLUT after it has been created? I will post some screenshots of HCFR without a profile later on, maybe you can help, but I have literally tried every setting and combination of settings possible, except the service menu and getting a profile that fixes the problems / works in games.
Click to expand...
I'm very busy at the moment. Will answer your questions as soon as possible but it may take one day - but already one short hint: If you carefully look at the calibration process you will see that it consists of two parts: The calibration itself (will only display neutral tones) which linearizes the display accordant to the chosen parameters (whitepoint, tonal curve) and the characterization afterwards (this data will be represented by a matrix-shaper or CLUT configuration). The vcgt consists of the calibration data and will lead to a neutral reproduction with a defined tonal curve. But it can't realize color space transformations.
 
Last edited:
I've just created a 5% gray pattern for you to test.

According to your measurements, this should not be visually different with and without the dispcalgui generated LUT.

Please confirm.
 
It looks similar with and without, except darker without the profile and slightly more green possibly. about the same not a massive difference.
 
Sailor_Moon said:
I'm very busy at the moment. Will answer your questions as soon as possible but it may take one day - but already one short hint: If you carefully look at the calibration process you will see that it consists of two parts: The calibration itself (will only display neutral tones) which linearizes the display accordant to the chosen parameters (whitepoint, tonal curve) and the characterization afterwards (this data will be represented by a matrix-shaper or CLUT configuration). The vcgt consists of the calibration data and will lead to a neutral reproduction with a defined tonal curve. But it can't realize color space transformations.
Click to expand...

Ok thanks, I don't mind if the colors are not 100% accurate if that is not possible, but if I could fix the greyscale and have blacks that are close to black (slightly blue because of the LED backlight, not VERY blue, red or green) while still keeping 2500:1+ contrast with BT1886 gamma, that would still be a huge improvement over what it is like now.
 
spacediver said:
Ok, repeat the same thing, but with a pure black screen.
Click to expand...

I deleted the profile so I cannot test now, from what i remember one is green tinted (no profile) and the profile looks better (slightly blue which is not a problem) its easier to see all squares on the black test but obviously the contrast is down to 1700:1. The profile looked ok except, slightly green tinted light colors and obviously the contrast. I don't know what I should be setting in dispcalgui to fix the blacks without ruining the contrast ratio... losing 10% is ok but losing 40% is not. using "auto" on black correction ends up with 1700:1 contrast.
 
Nicholars said:
. I don't know what I should be setting in dispcalgui to fix the blacks without ruining the contrast ratio... losing 10% is ok but losing 40% is not. using "auto" on black correction ends up with 1700:1 contrast.
Click to expand...

I explained to you that this is often a natural tradeoff, and it is quite well explained in the documentation (which you claim to have read).
 
spacediver said:
I explained to you that this is often a natural tradeoff, and it is quite well explained in the documentation (which you claim to have read).
Click to expand...

I have yes but 1700:1 is a massive difference... if it went down to 2300:1 or something that would be ok but 1700:1 is a massive difference from 2700:1 without a profile. It should be possible to get better than 1700:1?

I don't know :

a) what TV settings would be the best to calibrate from (dark colors above 100% cannot be corrected as far as I know, dark colors below 100% can be corrected by adding other colors, but reduces contrast ratio)

b) what the best settings to use in dispcalgui to get it as accurate as possible without destroying the contrast ratio.
 
It would have helped if you were able to compare the pure black with and without the applied LUT.
 
spacediver said:
It would have helped if you were able to compare the pure black with and without the applied LUT.
Click to expand...

Yes stupidly I deleted it, from memory with the profile applied the blacks were much closer to the correct color, slightly blue tinted (normal) but the contrast was much lower, the blacks without the profile are green / brown colored because of the big dip in red at the bottom, but that is the closest it will go without going over the green 100% line. I will do another profile and see if it makes any difference, do you know what I should set the black point correction to? try 10% instead of auto? Thanks for the help BTW... Any idea what I should set for the next profile that might give a better result?
 
Last edited:
you can experiment yourself with different settings to adjust the amount of tradeoff between contrast and black accuracy. A software adjustment can only do so much to compensate for inaccurate black point. I recommend getting your blacks as good as possible with hardware adjustment alone, and then using dispcalgui.

beyond that, there's nothing you can do.
 
Nicholars said:
I will take some screenshots, I am still confused about what parts actually work in games and if they will fix the blue tint, I assume the gamma settings that correct the gamma to BT 1886 will work in games but will the part of the profile that fixes the blue tint work in games?
Click to expand...

The only thing you can do is a 1D LUT, and that will have an effect in windows (desktop, browsing, etc.). Some games will respect this LUT, others won't. There are various workarounds for some games.

And as for profile, it's utterly and completely irrelevant to you. Read my icc post that I linked earlier. Your profile will not have an effect on anything you do. The ONLY thing that has an effect is the calibration you do (calibration is DIFFERENT from profiling, as you should know by now from the dispcalgui documentation).

Clearly, any calibration you do through the TV OSD will have a global effect, but the LUT calibration may not be respected in certain games.
 
spacediver said:
The only thing you can do is a 1D LUT, and that will have an effect in windows (desktop, browsing, etc.). Some games will respect this LUT, others won't. There are various workarounds for some games.

And as for profile, it's utterly and completely irrelevant to you. Read my icc post that I linked earlier. Your profile will not have an effect on anything you do. The ONLY thing that has an effect is the calibration you do (calibration is DIFFERENT from profiling, as you should know by now from the dispcalgui documentation).

Clearly, any calibration you do through the TV OSD will have a global effect, but the LUT calibration may not be respected in certain games.
Click to expand...

So to calibrate my TV for gaming and desktop etc. I need to set up everything in the "calibration" tab of dispcalgui as well as possible to get the best possible result there, with gamma and removing tints etc.... eg. calibration speed low, highest settings etc.

Then make the "profiling" tab as basic (calibration speed fast, single curve + matrix lowest amount of patches) as possible (or turn it off if possible, I have not used the new version of dispcalgui much, it used to look different and had a "calibrate" or "calibrate + profile" button) basically ignore the profiling tab and concentrate on the calibration tab for what I am trying to do? Adding loads of patches and having profile quality high etc. will make zero difference to what I am trying to do which is 1DLUT for windows and gaming?

Another question, why does my triangle on the CIE diagram not line up with reference? Is it because my screen has a slightly different color gamut to the reference and cannot be fixed to line up with reference? Also how can I get the greens and purples straight (closer to reference)? hue control on the TV? or that a problem with the TV that cannot be fixed (I have made a lot of profiles in the past and the greens never line up with the boxes, the triangle also never lines up with the reference)

So this is what I need to do:

1) Set up TV controls best possible
2) concentrate on the calibration tab, get the 1DLUT which is loaded into VCGT as good as possible, make the profile tab as basic/fast as possible as it will not work anyway for games and will be wasting time for this part.
3) create a 3DLUT for MADVR with the existing (best I can get) 1DLUT / calibration loaded in VCGT (do I create a 3DLUT with the existing calibration loaded? update calibration?) for the 3DLUT I want to use lots of patches etc. as the profile will be used for MADVR 3DLUT. But for the first part (1DLUT) that will work in games, the amount of patches, profile type etc. is irrelevant.

I understand most things about calibrating, except was confused about the 1DLUT vs profile vs 3DLUT, profile types and what works where etc... But I think I have got that now?

Also why is "do not use video card gamma table to apply calibration" checked by default? Do I need to uncheck that?

As profiling is irrelevant for the 1DLUT, why has dispcalgui lost the option to just calibrate without making a profile?

Also for the first part what is the best way of loading the calibration in windows? The windows color management seems quite useless and does not work properly, dispcalgui gives you the option to load via dispcalgui, should I use CPkeeper and capture calibration then load that on startup? To force the calibration?

Final question is anyone know best values for : black output offset, black point correction and correction rate? to fix blue tint without lowering contrast more than about 10%?
 
Last edited:
Are you sure your meter is reading correctly? Do you have another one to borrow, preferably a different model?
 
meter is fine. NIcholars, you're overthinking this. Your results are fine. Nothing you do on the software level is going to fix the contrast/accuracy tradefoff you're forced into. Your primaries are good enough. I'm not familiar with dispcalgui - I use ArgyllCMS command line. I believe # of patches will have an impact, because the more patches the software can use to test your display, the more accurate a model it builds of it, and can generate a more appropriate LUT.

But, as I said, your results are fine. Your delta E's are all very low, and your primaries are almost perfect.

As for madvr and 3d LUT stuff, this is the thread you want.

The rest of your questions can probably be figured out through some diligence, reading, and experimentation. The author of dispcalgui frequents AVS forums.
 
spacediver said:
meter is fine. NIcholars, you're overthinking this. Your results are fine. Nothing you do on the software level is going to fix the contrast/accuracy tradefoff you're forced into. Your primaries are good enough. I'm not familiar with dispcalgui - I use ArgyllCMS command line. I believe # of patches will have an impact, because the more patches the software can use to test your display, the more accurate a model it builds of it, and can generate a more appropriate LUT.

But, as I said, your results are fine. Your delta E's are all very low, and your primaries are almost perfect.

As for madvr and 3d LUT stuff, this is the thread you want.

The rest of your questions can probably be figured out through some diligence, reading, and experimentation. The author of dispcalgui frequents AVS forums.
Click to expand...

Right so the amount of patches used in the profiling section (which on dispcalgui is a separate tab to the calibrating section) does actually effect the accuracy of the 1DLUT? What other people said made me think the patches in the profiling section are just for the profile and will not affect the 1DLUT made in the calibrating section (which has options for calibation already)
 
I'm honestly not sure.

All I'm familiar with is dispcal. It may be the case that # of patches in dispcalgui has zero effect. In dispcal, you can adjust quailty, but this changes the number of "test readings and refinement passes", which I'm not sure is the same as # of patches in the context of dispcalgui.

Anyway, for you, it's completely moot, unless you are going to be using madvr and a 3DLUT.
 
spacediver said:
I'm honestly not sure.

All I'm familiar with is dispcal. It may be the case that # of patches in dispcalgui has zero effect. In dispcal, you can adjust quailty, but this changes the number of "test readings and refinement passes", which I'm not sure is the same as # of patches in the context of dispcalgui.

Anyway, for you, it's completely moot, unless you are going to be using madvr and a 3DLUT.
Click to expand...

I will do one for MADVR after I have got the best possible for windows / games. I am not happy with the one I posted because of the contrast ratio so I will try and get something better eg. 10% contrast lost instead of 40%
 
if you want to minimize contrast lost, tune the color temperature closer to your target(6500k) first with the tv internal control.

dispcal correlate to the calibration tab on dispcalgui, the speed slider is the number of patches(sort of, actually is the quality option ) for dispcal.

For the profile tab, just leave it on default. Profiling is quick anyways (with default 200-300 patches).

And to you main problem: comparing to a IPS screen and finding it green. It just doesn't' really work that way~ .
 
Last edited:
color temperature (actually, correlated color temperature) is not really what one should be concerned with. Chromaticity is the key here, in particular D65.
 
Nicholars, if you're having to struggle this much with calibration in spite of having a colorimeter, I'd suspect your meter has some egregious inaccuracy right out of the box. I tried two seperate i1 Display Pros (cousin to the Colormunki display) - the first gave a notably greenish cast and the other noticeably reddish. Gave up on X-rite and tried a Calman RGB package with a C3 meter. It accurately saw and corrected the imbalances the X-rite thought were fine and I couldn't be happier. Surprisingly accurate results for the money out of the box.
 
I'm pretty sure the colorimeter is fine. The issue was the inaccurate blacks when trying to preserve the full contrast of the display. Even a lab grade spectoradiometer can't help there.
 
You must log in or register to reply here.
Back
Top