Display Calibration

[Asmodian]

That looks good, without finer steps it is basically a perfect BT.1886 gamma, but <10% is where gamma issues usually happen. ;)

Do not worry about it if it looks good.


That's for sure?


Yes, BT.1886 is pure power 2.4 gamma but for displays with black > 0 nits (it adapts to your display). When having madVR change your gamma mathematically to a different pure power gamma it should think your display is currently pure power 2.4. Do not use "color & gamma" to change your display to a BT.709 gamma (do not use that gamma period). If you are not changing your gamma this way it is not important.

[hajosattila]

That looks good, without finer steps it is basically a perfect BT.1886 gamma, but <10% is where gamma issues usually happen. ;)

Do not worry about it if it looks good.



Yes, BT.1886 is pure power 2.4 gamma but for displays with black > 0 nits (it adapts to your display). When having madVR change your gamma mathematically to a different pure power gamma it should think your display is currently pure power 2.4. Do not use "color & gamma" to change your display to a BT.709 gamma (do not use that gamma period). If you are not changing your gamma this way it is not important.

finally this is the correct setting: https://i.imgur.com/jOZ7EUZ.jpg | https://i.imgur.com/Hbl0xmi.jpg (inactive)

Would you confirm to me? :D:thanks:

[Asmodian]

Yes, both those settings pages are correct. :)

[hajosattila]

Yes, both those settings pages are correct. :)

Thanks for you're help!

[hajosattila]

Other (CALMAN) question: Which pattern size do you recommend for a 55' white LED display? (55H6500 TV/CALMAN Home Advanced)

Thanks!

https://i.imgur.com/IfCXPIE.jpg

[chros]

Other (CALMAN) question: Which pattern size do you recommend for a 55' white LED display? (55H6500 TV/CALMAN Home Advanced)
Hi Attila!

A very few people use Calman here, as I mentioned we usually use 3dlut created by DisplayCal. Even with that you cannot achieve perfect result (but good one), so you don't have to worry about the "old" way at all :)
I think it's good to go with half (or more) of the size of your screen.

[Asmodian]

Other (CALMAN) question: Which pattern size do you recommend for a 55' white LED display? (55H6500 TV/CALMAN Home Advanced)

Thanks!

I use 25% with Calman, Constant APL 25 is good unless your TV goes into a low power mode if the APL doesn’t change for too long.

Constant APL is not important for your TV, in fact pattern choice is not important for your TV, any of them will give you the same results.

LCDs without a dynamic backlight or local dimming are easy to calibrate. :)

Oh, you mean the slider? I believe I used 100? Does that setting exist with madVR’s pattern generator? :o

[hajosattila]

@Asmodian | @chros Thank you!!

Calibration is complete! but unfortunately there are a lot of blue and red colors.
I tried it several times, but I could not solve the problem.

https://i.imgur.com/NXSForX.jpg

ps.: I'll make a 3D LUT file later, but first I want to calibrate the TV. (learning)

[Asmodian]

A lot of blue and red colors?

Do you mean your red and blue primaries are off? Primary colors are often off, they generally are not changed by in device calibrations because they are dependent on the backlight. What color is the backlight if the LCD filters out all the blue and green light? The only way to control that is to filter out less blue and green, which is usually not the correction needed.

P.s. For some reason the image quality of your report link is so bad I cannot make out the gamut chart or deltaE values. :(

[chros]

I'll make a 3D LUT file later, but first I want to calibrate the TV. (learning)
Well, you can learn if you want, but this knowlegde/pain won't be needed there (unless you want to be a "professional" calibrator :) ):
- you only have to set the white point on the TV (which is pretty easy compared to the tedious task of manual calibration of all the colors)
- the rest is controlled via software completely

P.s. For some reason the image quality of your report link is so bad I cannot make out the gamut chart or deltaE values. :(
And I can't even open the link at my workplace :D I'll take a look at it from home.

[chros]

Calibration is complete! but unfortunately there are a lot of blue and red colors.
I can see that, but the CIE chart looks fine. Are they also visible?
Take a note of your settings of the TV then try out DisplayCal with 3dlut:
- reset all your config in the TV
- and only set the white balance nothing else when DisplayCal asks for it

[Asmodian]

[B][U]https://i.imgur.com/NXSForX.jpg

ps.: I'll make a 3D LUT file later, but first I want to calibrate the TV. (learning)

I finally looked at your chart on a desktop. That red and blue error is exactly what I was talking about.

I can see that, but the CIE chart looks fine. Are they also visible?
Take a note of your settings of the TV then try out DisplayCal with 3dlut:
- reset all your config in the TV
- and only set the white balance nothing else when DisplayCal asks for it

The new version of Calman (2017) has a very simple almost automatic workflow for creating very good 3DLUTs for madVR (Color Cube). To my surprise they do seem a little better than those created by Argyllcms. It supports madTPG too, of course.

The level of integration is impressive, it even names the 3DLUTs according to the color space and configures madVR to use them. It supports creating a 3DLUT for every color space madVR does (and more).

It you have a license for Calman you should use it but DisplayCal is a very good and much cheaper option.

[Oguignant]

I finally looked at your chart on a desktop. That red and blue error is exactly what I was talking about.



The new version of Calman (2017) has a very simple almost automatic workflow for creating very good 3DLUTs for madVR (Color Cube). To my surprise they do seem a little better than those created by Argyllcms. It supports madTPG too, of course.

The level of integration is impressive, it even names the 3DLUTs according to the color space and configures madVR to use them. It supports creating a 3DLUT for every color space madVR does (and more).

It you have a license for Calman you should use it but DisplayCal is a very good and much cheaper option.

True, it's much better Calman to create the 3Dlut. (the 2016 version of Calman is exactly the same as the 2017 version for creating 3Dlut).

I found a way to create a detailed 3Dlut of 17,576 points (Madvr limits it to 10,000 max points). If anyone cares, I can explain how.

https://i.imgur.com/ur5CjXOh.png?1

[Asmodian]

Did you notice a difference between 17 thousand and 10 thousand?! 3000 is usually plenty, 6000 is more than enough. :p

True, it's much better Calman to create the 3Dlut. (the 2016 version of Calman is exactly the same as the 2017 version for creating 3Dlut).

The quality is the same but the integration is much better now. :)

[Oguignant]

Did you notice a difference between 17 thousand and 10 thousand?! 3000 is usually plenty, 6000 is more than enough. :p

it is not only the quantity, it is different the method that uses to create the 3dlut. Looks better with these options. Try it and you'll see the difference, especially in the blackcrush

Did you notice a difference between 17 thousand and 10 thousand?! 3000 is usually plenty, 6000 is more than enough. :p



The quality is the same but the integration is much better now. :)

I do not want to be annoying, but I use Calman from the 2016 version and I do not change anything, the integration is the same. Note that the latest version 2016 is from August 2017

CalMAN 2016 Rev 2
Version: 5.7.3 Build: 2359
Official Release: August 7th, 2017

[hajosattila]

That red and blue error is exactly what I was talking about.

What do you think exactly?

[Oguignant]

@Asmodian | @chros Thank you!!

Calibration is complete! but unfortunately there are a lot of blue and red colors.
I tried it several times, but I could not solve the problem.

https://i.imgur.com/NXSForX.jpg

ps.: I'll make a 3D LUT file later, but first I want to calibrate the TV. (learning)

Did you create a meter profile for your Tv?

[hajosattila]

Did you create a meter profile for your Tv?

No, I did not make it.

[Oguignant]

No, I did not make it.

in the "Correction" option that you use?

[Asmodian]

What do you think exactly?

Your red and blue primaries are off, your green primary is very accurate.

You cannot change these with a calibration, they are not bad and I would not worry about them.

[hajosattila]

Your red and blue primaries are off, your green primary is very accurate.

Asmodian, what do you mean "OFF" because minimally adjustable... Can you explain more? Thank you

You cannot change these with a calibration, they are not bad and I would not worry about them.

But I've tried everything!
Luminance/stimulus: 75%/75% - 75%/100% - 100%/75% - 100%/100% but only minimally improved the colors, and I'm unable to set the red and blue colors to a lower value...

At weekends I look at the HCFR program.

One more thing: Does it matter, from image qualtiy point of view, which setting I use out of the two option during the calibration process?

CALMAN is 16-235 at every case!
Setup 1 (limited range)
TV: 16-235 | NVIDIA CP: 16-235

Setup 2 (full range)
TV: 0-255 | NVIDIA CP: 0-255

I would like to apologize for the wrong wording if there is any, but unfortunately there is 0 information about the subject in my native language, especially on this level, so I have to ask my questions on international portals.

[Asmodian]

By off I simply mean they are not accurate. It is tricky to develop white LEDs with exactly the correct pure red, green, and/or blue for BT.709. Most displays' primary colors are not perfect and your's are actually pretty good.

Trying to fix it will probably make a lot of other aspects of your calibration, that you are not currently measuring, worse.

But I've tried everything!

What about being unable to correct them with on-device calibration do you not believe. ;)

Your red and blue primaries are slightly under-saturated, that means a calibration would need to remove more than 100% of the green and blue from the red and more than 100% of the green and red from the blue. This is not possible, you cannot increase the saturation of a primary color with a calibration. Setting your display to a wider gamut, if available, would increase saturation but it would be MUCH higher, giving much higher dE values.

In fact it usually messes up a lot of other aspects of a calibration to even attempt to correct the primary colors. Even correcting the secondary colors usually goes bad, you can get 100% magenta looking correct but 50% magenta doesn't change so you end up with non-linear saturation ramps that are hard to deal with during a software calibration. Make sure to look at the saturation ramps before and after, don't make them look worse after calibration. Also Calman's "color checker" is very helpful when evaluating manual calibrations.

Before creating a 3DLUT I would only use the on-device calibration to correct the white point, reset color management to the defaults and then only use it to correct the white point.

One more thing: Does it matter, from image qualtiy point of view, which setting I use out of the two option during the calibration process?

CALMAN is 16-235 at every case!
Setup 1 (limited range)
TV: 16-235 | NVIDIA CP: 16-235

Setup 2 (full range)
TV: 0-255 | NVIDIA CP: 0-255

Are you using madTPG or something else? I have never tested using another pattern generator but this is dependent on that generator and your display.

madTPG will expand 16-235 to full range RGB so Setup 1 and Setup 2 give the same results (2 preferred, all else being equal). This assumes your display is set to full range in madVR.

If the pattern source does not expand the dynamic range (madVR set to limited range or another generator) it is important to use Setup 3: TV: 16-235 | NVIDIA CP: 0-255. Or switch Calman to 0-255 and use Setup 2 (Setup 1 works too), Windows is always 0-255 so when not using madTPG in Windows Calman should probably be set to 0-255.

You can use madTPG even when not creating a 3DLUT so it is strongly recommended.

Edit: I know it is hard to grasp but your Setup 1 and 2 look basically the same to calibration software. Setup 1; with the NVIDIA CP set to 16-235 the GPU compresses 0-255 into 16-235 so any test patterns are compressed, the TV then displays the software's pattern value of 0 as 0% black and what was a pattern value of 255 as 100% white. The data format is always 0-255, the GPU option 16-235 is an active range compression step while the option 0-255 does not do anything to the data. The GPU always gets full range RGB even with limited range set in Calman, all that changes is what Calman thinks is 0% IRE; binary (0000000, 0000000, 0000000) or (00010000, 00010000, 00010000).

[chros]

Edit: I know it is hard to grasp but your Setup 1 and 2 look basically the same to calibration software. Setup 1; with the NVIDIA CP set to 16-235 the GPU compresses 0-255 into 16-235 so any test patterns are compressed, the TV then displays the software's pattern value of 0 as 0% black and what was a pattern value of 255 as 100% white.
As Asmodian suggested: use setup 2. if it's possible.

[hajosattila]

Are you using madTPG or something else? I have never tested using another pattern generator but this is dependent on that generator and your display.

madTPG will expand 16-235 to full range RGB so Setup 1 and Setup 2 give the same results (2 preferred, all else being equal). This assumes your display is set to full range in madVR.

If the pattern source does not expand the dynamic range (madVR set to limited range or another generator) it is important to use Setup 3: TV: 16-235 | NVIDIA CP: 0-255. Or switch Calman to 0-255 and use Setup 2 (Setup 1 works too), Windows is always 0-255 so when not using madTPG in Windows Calman should probably be set to 0-255.

Thank you very much for the detailed answer!! :goodpost:

First step: Correct dynamic range TV: 0-255 | NVIDIA CP: 0-255 | CALMAN 16-235(?)

next step: I changed the program's internal generator. I use madtgp :)

If this is correct, then I'll start calibration.

sorry for my bad English

[Asmodian]

Yes, that is correct.

[hajosattila]

Finally, I decided to perform the measurement under HCFR, which I am completely satisfied with if the gamma did not slide in the 50-60-80 and 90 range.

How can I correct this?

https://imgur.com/a/iU3qV

[Asmodian]

You probably cannot, those errors are very minor and your display looks good overall.

Your TV would need a multi-point luminescence control to adjust that, some TVs do have luminescence controls as part of a 10 or 20 point white balance IRE adjustment but I get better results leaving that alone and letting the gamma curve be slightly imperfect. I get much better results using a software calibration to make those fine adjustments (a 3DLUT for madVR or an ICC profile for Windows).

[chros]

I think also that those results are good :) You'll never get perfect result only if you're lucky.

[chros]

How does "convert or process HDR content by using an external 3DLUT" work using an SD display? The description of those options aren't straightforward to me.

So, what should I do / which option to use if I wan to try it out using an UHD BD encode (x265 10bit HDR10) and a 3dlut calibrated FullHD display (Bt.709)?

Thanks

[huhn]

if you select on option in HDr there is an text that should make it clear what which option does.

process HDR content by using an external 3D LUt means the ct by using an 3D lut so it is still HDR.

on a not HDR display you should either select convert HDR to SDR by using pixel shader math or convert HDR content to SDR by using an external 3D LUT (if you can create of have access to a HDR -> SDR 3D LUT).

[chros]

process HDR content by using an external 3D LUt means the ct by using an 3D lut so it is still HDR.
Interesting. Is this setting for "profiled" display in HDR mode?

on a not HDR display you should either select convert HDR to SDR by using pixel shader math
That's what I use now but using GPU for this puts more stress onto my underclocked :) GPU, that's why i thought that maybe using a 3dlut for this would be better for me.
But on second thought: since decoding is done by CPU (that's not powerful at all either) then meybe it would result more stress onto the CPU if I use 3dlut for this.
Is the HDR-SDR 3dlut conversion done by CPU, isn't it?

or convert HDR content to SDR by using an external 3D LUT (if you can create of have access to a HDR -> SDR 3D LUT).
And which 3dlut should be created? BT.2020 or DCI-P3? Currently all the HDR contents are DCPI-3P.
Do you have a link how to create those 3dluts with DiplayCal?

Thanks for your help!

[huhn]

Interesting. Is this setting for "profiled" display in HDR mode?
it's the same as a normal 3D LUt just for the HDR mode.

That's what I use now but using GPU for this puts more stress onto my underclocked :) GPU, that's why i thought that maybe using a 3dlut for this would be better for me.
But on second thought: since decoding is done by CPU (that's not powerful at all either) then meybe it would result more stress onto the CPU if I use 3dlut for this.
Is the HDR-SDR 3dlut conversion done by CPU, isn't it?

HDR -> SDR conversation is done by the GPU.
And which 3dlut should be created? BT.2020 or DCI-P3? Currently all the HDR contents are DCPI-3P.
Do you have a link how to create those 3dluts with DiplayCal?

Thanks for your help!

http://www.avsforum.com/forum/139-display-calibration/1471169-madvr-argyllcms.html

just use the d65 Rec. 2020 / SMPTE 2084 madVR profile

[chros]

it's the same as a normal 3D LUt just for the HDR mode.
Yep, thanks, that's what I meant.

HDR -> SDR conversation is done by the GPU.
http://www.avsforum.com/forum/139-display-calibration/1471169-madvr-argyllcms.html
just use the d65 Rec. 2020 / SMPTE 2084 madVR profile
It's good news, looks promising, I'll give it a try then when I'll have time.

[Arm3nian]

Two questions:
Does lowering RGB levels in the monitor menu reduce gamut?

Which would give more accurate color:
1) no correction
2) spectral correction LCD White LED IPS (generic one in dispcal)
3) matrix correction for the same model colorimeter and display that I have but created and uploaded by someone else, so "different" hardware.

All three give different results, so how do I know which is the most accurate without a reference or more tools like a spectrometer of my own.

[huhn]

Two questions:
Does lowering RGB levels in the monitor menu reduce gamut?
there is no general answer to this. maybe.
Which would give more accurate color:
1) no correction
2) spectral correction LCD White LED IPS (generic one in dispcal)
3) matrix correction for the same model colorimeter and display that I have but created and uploaded by someone else, so "different" hardware.
3.
All three give different results, so how do I know which is the most accurate without a reference or more tools like a spectrometer of my own.
you don't.

[chros]

Which would give more accurate color:
1) no correction
2) spectral correction LCD White LED IPS (generic one in dispcal)
3) matrix correction for the same model colorimeter and display that I have but created and uploaded by someone else, so "different" hardware.

3.
I've been also thinking about the diff between 2) and 3): e.g. if you have a old spoctrophotometer that age (like my i1Pro) then who knows: maybe 2) would give better results :)

[e-t172]

Does lowering RGB levels in the monitor menu reduce gamut?

From the rest of your post it looks like you have measurement hardware on hand, so why don't you try it and see for yourself? That's the best way to be sure. (Using ColorHCFR for example, which will show you the gamut graphically.)

Every monitor is different in terms of settings, but in general, if the RGB settings don't mention anything about hue or saturation, it usually means it's an individual channel luminance setting. That affects the white point, but not the position of the primaries (which is what most people mean by "gamut", i.e. in 2D).

[Asmodian]

I have had R, G, and B move the primaries before (on a plasma TV) but they usually do not on anything with a backlight. They do not on my OLED TV either.

[hajosattila]

I get much better results using a software calibration to make those fine adjustments (a 3DLUT for madVR).

I would like to fine tune this setting: https://imgur.com/a/iU3qV (link) Can you help me with DisplayCal's settings? https://imgur.com/a/PRxXz

In my case, is it necessary to use the Interactive Display Adjustment option? I just want to refine my CMS because I've calibrated everything with BT1886 CALMAN software

Thanks for your help and I apologize for my weak English!

[Asmodian]

What do you want to fine tune about them? Why does your CMS need refining? Your first link shows everything pretty much perfect. You are not going to get noticeably better than that.

Your red and blue primaries at 1.2 dE is not fixable (see above) and you have 0.2 dE or lower everywhere else, 0.2 dE is very very good. Even 1.2 dE is good, below 2 dE is considered a good calibration in most professional color work, but that is for all colors.

You should measure your saturation ramps and/or a color checker to better judge your current calibration.

Edit: Hmm.. I think I misunderstood you. You can create the 3DLUT for madVR with Calman so I got confused. :o

If you simply want to create a 3DLUT for madVR you definitely do not need the Interactive Display Adjustment option. :)

One change I would make is to set the Whitepoint to "as measured" or "native" too. Your white point is already very good so tweaking in software lowers contrast for no gain. Otherwise your options look good for your display.

[hajosattila]

So you suggest I should leave it as it is and there is no use for fine tuning? Actually I am quite satisfied with the results as every REMUX is awesome on it.

[Asmodian]

A 3DLUT for madVR could be beneficial but you don't want to adjust your white point. I haven't actually seen your gamma results but a 3DLUT is a great way to get a perfect gamma. Only looking at primary and secondary colors and the gray scale is not enough information.

The settings you showed for DisplayCal were good, except I would change the gamma to BT.1886 and the white point to as measured (or native if that is what DisplayCal calls it). Unless your gamma is already good.

Even with all those options set to "as measured" a 3DLUT can be helpful, it won't make a big difference in all scenes but can help a lot in some.

However, if you do a "Color Checker" scan in HFCR and everything is below 2 dE, maybe don't bother with a 3DLUT. :)

[chros]

Can you help me with DisplayCal's settings? https://imgur.com/a/PRxXz
In my case, is it necessary to use the Interactive Display Adjustment option? I just want to refine my CMS because I've calibrated everything with BT1886 CALMAN software!

So you suggest I should leave it as it is and there is no use for fine tuning? Actually I am quite satisfied with the results as every REMUX is awesome on it.

Can you also remind us which TV do you have, Attila?
Well, here's what you can do: select a different user preset on your TV and reset everything back to default under it, than you can create a proper 3dlut with DisplayCal and you can easily switch between presets on the TV to see the diff between CMS and 3dlut calibration.

... except I would change the gamma to BT.1886 ...
Well, I'd rather go for 2.4 gamma if you can select it on your TV as well :) You can take a look here (http://www.avsforum.com/forum/139-display-calibration/1471169-madvr-argyllcms-139.html#post40108362), why, Attila.

As far as modding the default MadVR preset in DisplayCal, you only have to change:
- the amount of patches on Profiling tab (I used to create around 4000)
- and the above mentioned gamma settings on 3dlut tab

And yes, you have to adjust the white point of your TV!

When it's finished, you have to make sure that the generated 3dlut is properly set in MadVR, then you can go to the verification tab, and set the chosen gamma there as well, and select XXXL video profile.
And you can also make a profile with HCFR as well, as Asmodian said.

... and I apologize for my weak English!
Your English is fine there's no need for apologizing for it :)

[Asmodian]

And yes, you have to adjust the white point of your TV!

Their white point is already within 0.2 dE! Why would they want to use software to mess with the white point at all? Any change done in software would reduce contrast for nothing. :mad:

Well, I'd rather go for 2.4 gamma if you can select it on your TV as well :) You can take a look here (http://www.avsforum.com/forum/139-display-calibration/1471169-madvr-argyllcms-139.html#post40108362), why, Attila.

for evening/night usage (with just a dim environment light on): 2.4 like gamma with slightly elevated shadows at the lower end of greyscale

Guess what BT.1886 is. :p

[hajosattila]

However, if you do a "Color Checker" scan in HFCR and everything is below 2 dE, maybe don't bother with a 3DLUT. :)

As far as I remember the avg. in Color Checker was around 0,9.Including the skin tone, blue sky etc.

Thanks guys!!

[chros]

Their white point is already within 0.2 dE! Why would they want to use software to mess with the white point at all? Any change done in software would reduce contrast for nothing. :mad:
:D Well, then he doesn't have to change it then :)

[Arm3nian]

I notice the closer I get RGB to each other the less difference I see at the end of the calibration when I click preview. Essentially just adjusting those values is what you do on a display not connected to a computer, and can use HCFR. So is there an actual reason to "calibrate" with the software and have it adjust levels if not doing pro color work? I leave the settings on "as measured" anyway.

https://i.imgur.com/gHxN8kg.png

[chros]

So is there an actual reason to "calibrate" with the software and have it adjust levels if not doing pro color work?
Do you mean 3dlut?

[Asmodian]

If your white point is already as good as your example I strongly recommend you not calibrate it in software. I also recommend you NEVER calibrate the brightness with software, for any reason. :p

A 3DLUT can still be quite helpful, and even an ICC profile if your gamma or gray scale need work, but simply set the aspects that are already good to "as measured". The less you adjust in software the better. :)

[markanini]

Is it worth creating a 3DLUT for every color space ie Rec.709, SMPTE-C, EBU/PAL?