Display Calibration

[chros]

I don't think so. I have 2 for rec.709 (night, day), and similarly 2 for hdr->sdr conversion.

[Arm3nian]

If your white point is already as good as your example I strongly recommend you not calibrate it in software.
I use the powers of :devil: to resurrect this quote!...

What if the white point is not exact? :)

On my new laptop, you can't adjust RGB levels. The display by default is really good, 3840x2160 wide gamut IPS. The white point was off quite a bit tho, delta E of ~7. Do I leave it as measured or target 6500k? Targeting a value actually influences decisions made during calibration right?

I tried the Dell Premier Color app. It has a bunch of profiles:
https://i.imgur.com/UYVQaxY.png
https://i.imgur.com/8rA5g0S.png

It seemed to brighten things up a bit, but the actual white level didn't change much. The white point changed very little also. The app has a mini calibration tool that uses the i1 display pro to adjust the profiles based on the specific display. After doing that the delta E was ~0.6, but I'm guessing it's because it actually calibrated something. Dispcal and madTPG methods of disabling existing calibrations don't seem to have any effect on the Dell app. You can have both the Dell app and a separate calibration running at the same time. I calibrated, profiled, and then made a 3d-lut for madVR while the Dell app was on the DCI-P3 setting. The gamut coverages were really bad, and so was the overall picture. I'm not really sure how the app works but I'm probably going to uninstall it in favor of the more manual method.

For madVR calibration, dispcal targets P3 white point coordinates by default, so I'm guessing I should leave it alone. White level and gamut is set to as measured. I made the 3D-LUT using DCI-P3 as the source color space which targets a 2.6 gamma. The Dell app also targets a 2.6 gamma with that profile selected, both for obvious reasons. Maybe I should target 2.6 in the calibration when I remove the app, because the gamma will shift, probably to 2.2.

[Asmodian]

When calibrating a modern SDR computer display I would target the BT.709 gamut. It is very important to always set white level to as measured unless you are trying to perfectly match two different displays for some reason.

Calibrating the white point is more nuanced. I would target D65 (~6504K) to go with BT.709 but if your white looks white to you I suggest you not use the normal 3 hardware 1DLUTs of the video card or madVR's 3DLUT to adjust it. However, that dell app sounds like it could be a very good thing, it is likely setting up some hardware LUT in the panel itself; using this to correct the white point is similar to using RGB controls. I would put it on the BT.709 setting though, it sounds like your display has poor coverage of P3.

Then create a BT.709 3DLUT for madVR, targeting the native white point and level. Feel free to use the 3DLUT tool to also create LUTs for the rest of madVR's supported color spaces afterward but do the profiling with the understanding that your display is BT.709. Don't try to squish DCI-P3 into your display's gamut with a calibration.

[Arm3nian]

If anyone is interested, I calibrated a single RGB LED to a pretty good white Delta E :). R, G, and B separately look good, and so does white, but a combination of 2 of them looks pretty off. R and B don't look purple, etc.

Changing the voltage to one leg of the LED by 1 millivolt resulted in a ~1% change in a level in HCFR. It's impressive to me that the i1 can detect such a small change. The limiting factor, if anything, was the precision of my power supply.

https://i.imgur.com/8vSKbpxl.jpg

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

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

https://i.imgur.com/rbIabb5m.jpg

[Asmodian]

Haha, very nice. :)

You discovery that purple still doesn't look right demonstrates the major issue with RGB type fake color. Our hardware is now much more accurate than the differences between how we perceive mixtures of three frequencies at different intensities. Every individual will match a particular mix of a particular red, green, and/or blue primary color to a different place on the visible spectrum, and this is discounting things like color blindness.

We are currently using a "standard observer" from a 1931 study that is an average from many individuals. This study has been redone several times over the years and the results are similar, this is why you sometimes have options for other observers, but almost everything is still designed around the 1931 results. Ideally one would do this study on oneself, using the same primary colors as your display, and use the results as a "personal observer" instead of using a standard observer.

Edit:
Also, the camera generating the content has the same issue. It does not record the actual spectrum of light but only three intensities based on the response of the sensor. It is impossible to recover the real frequencies so it will never look the same as it would if you were in the room.

However, because we experiance color in such a relative and imprecise way all this is not in anyway obvious to us unless we start doing very careful comparisons.

[Arm3nian]

I had a chance to play with a new Alienware laptop, and noticed similar things on the LEDs you can control on it. All the colors look nice, but white doesn't really look white. I guess it really does depend on the hardware used. I wanted to try some higher quality LEDs to see if they perform better.

Anyway, for my XPS 9570 laptop, there is an option for "Color Accuracy" in the Intel graphics control panel that is set to on by default. I turned it off and the display's true potential was released :devil:

The colors, specifically red, which looks a bit orangish now, are really saturated, but personally I don't think they're "over-saturated", even though I guess technically they are. It looks much better than whatever gpu altering things "color accuracy" was doing.

You were correct about the Dell App potentially being a good thing. It essentially does modify hardware LUTs as no software affects it. There are many modes for it:
1) Clean install of windows (app not present anymore). Delta E of 3.2, white point around 6750-6800k. Massive color gamut.
2) App installed and reset to factory settings (gives greenish tint and delta E shoots up to ~8).
3) When preparing to calibrate with the i1 but not actually pressing the button. Can even exit the program at this point and it keeps the settings. Delta E now down to ~3.2. Now that I think of it, it's probably the same as just disabling the app all together.
4) Actually calibrating the built-in profiles with the i1. This is sometimes random I think, but generally gives excellent results. Sometimes Delta E of .8, other times .4, and if you get lucky 0.0 or 0.1. They really should increase the amount of patches it reads as what they've implemented has a lot of potential. It basically, as you mentioned, adjusts the RGB levels for you. Sometimes down to exactly 6501k and 2.20 gamma.

Right now, I am wondering which mode to use. Without the Dell app the gamut volume is crazy, and the coverage is almost 100% on both sRGB and Adobe RGB. DCI-P3 coverage stays around the same value in any of the cases. The volume changes by ~3%. The Dell App has profiles you can select but I'm fairly sure it only adjusts the gamma. It's kind of funny because the DCI-P3 profile actually shifts the picture of the CIE diagram from Adobe RGB primaries to DCI-P3, but the coverage measured with displaycal stays the same. I don't think you can change the actual primaries of your display without severely altering everything.

Most seem to recommend prioritizing white level and gamma over gamut coverage. So I guess just calibrating with the dell app and profiling with displaycal and creating 3D LUTs is the best choice. I don't think there is any need to calibrate, as everything is spot on after running the calibration in the app. I've tried it once and it made no difference to what I can see. The white level and gamma are a bit off without it, and the diagram has squiggly curves near the primaries. If using the app, it is sharp.

Is ~93% coverage enough to try and watch UHD blurays with? The display has more gamut volume than both Adobe RGB and DCI-P3 but the primaries are obviously Adobe RGB.

All the pictures are with no app installed:

Diagrams compared to display
sRGB: https://i.imgur.com/FRR63g6.png
Adobe RGB: https://i.imgur.com/D5Fse6q.png
DCI-P3: https://i.imgur.com/12oj040.png

Gamut Coverage and Volume (ran with 425 auto-optimized patches): https://i.imgur.com/QXpzFUT.jpg

White level and Gamma: https://i.imgur.com/m7vQFrM.jpg

[Asmodian]

Great to hear you got good results updating the hardware 3DLUT with the i1. :D

I don't think you can change the actual primaries of your display without severely altering everything.

It is entirely dependent on the backlight. You can only reduce saturation after the light has been created.

Most seem to recommend prioritizing white level and gamma over gamut coverage. So I guess just calibrating with the dell app and profiling with displaycal and creating 3D LUTs is the best choice.


I agree, good advice.

I don't think there is any need to calibrate, as everything is spot on after running the calibration in the app.

Yes, do not run a software calibration if you already have a hardware 3DLUT you can update. :)

Is ~93% coverage enough to try and watch UHD blurays with? The display has more gamut volume than both Adobe RGB and DCI-P3 but the primaries are obviously Adobe RGB.

Not without a 3DLUT for madVR, but it is fine if you do have a 3DLUT.

All the pictures are with no app installed:

Too bright! :p

But that is a great Adobe RGB gamut. It is too bad Adobe RGB is out of date now. It was designed for matching printers for professional publishing (CYMK). There is a use for it still but it is not a great "wide gamut" for general consumers.

[Arm3nian]

Great to hear you got good results updating the hardware 3DLUT with the i1. :D

Yes, do not run a software calibration if you already have a hardware 3DLUT you can update. :)

Well, it was working on my previous windows install, but I'm not getting good results this time. The calibration through the Dell App is essentially doing nothing now. Same measurements before and after.

The "default" state of the display isn't too off, about 2.2 dE. I'm guessing this is where a software calibration would actually help, due to the fact that it's not too off that it would destroy the color. I calibrated (high speed) to 6500k and 2.2 gamma, and the gamut coverages and volumes are the same as with the app, but the plot has the rounded edges near the primaries. Why is that?


Not without a 3DLUT for madVR, but it is fine if you do have a 3DLUT.

I created a 3D LUT from the calibration and profile with source space of DCI-P3 and the entire UHD bluray is purple :p. UHD bluray colorspace is actually BT.2020 correct? I noticed madVR converts this to DCI-P3 with the default settings as full coverage of BT.2020 is hard to come by on a consumer display. With BT.2020 as the source for the 3D LUT, the colors look good. What should the tone curve be? The calibration is done at 2.2.

With "disable calibration controls" in madVR the colors look over saturated. With "this display is calibrated to BT.2020" the colors look washed out. I am guessing this is because my display's volume is ~18% larger than DCI-P3 but smaller than BT.2020. If I set to calibrated to DCI-P3 then it looks decent. But my display coverage is 100% adobe RGB coverage not 100% DCI-P3, so it doesn't look optimal. The 3D LUT looks the best and the most accurate.


Too bright! :p

With a laptop moving from place to place, the ambient light brightness and temperature varies greatly. I guess 120 cd/m^2 would be good, but it's too dim during the day. Going for the average brightness I use can also work as the further you drift from the brightness it was calibrated at the worse it looks.


But that is a great Adobe RGB gamut. It is too bad Adobe RGB is out of date now. It was designed for matching printers for professional publishing (CYMK). There is a use for it still but it is not a great "wide gamut" for general consumers.
Adobe RGB is actually slightly bigger than DCI-P3 correct? So there's some benefit. DSLRs can output to the Adobe RGB color space as well.

[huhn]

adobe RGB is mostly using bigger green. well is was created by typo...

[Asmodian]

I calibrated (high speed) to 6500k and 2.2 gamma, and the gamut coverages and volumes are the same as with the app, but the plot has the rounded edges near the primaries. Why is that?

Probably because it is trying to adjust the color near the primaries. Gamut coverage and volume is controlled by your primaries, no GPU 1DLUT calibration is going to change it, that takes a 3DLUT.

I created a 3D LUT from the calibration and profile with source space of DCI-P3 and the entire UHD bluray is purple :p. UHD bluray colorspace is actually BT.2020 correct? I noticed madVR converts this to DCI-P3 with the default settings as full coverage of BT.2020 is hard to come by on a consumer display.

Yes, I use BT.2020 3DLUTs too, DCI-P3 LUTs don't seem to work correctly a lot of the time and UHD is stored on the disc as BT.2020.

A 3DLUT is the only option to get good color for video on your display since no consumer video is Adobe RGB and madVR doesn't have a "this display is already calibrated to Adobe RGB" option.

Adobe RGB is actually slightly bigger than DCI-P3 correct? So there's some benefit. DSLRs can output to the Adobe RGB color space as well.

No, DCI-P3 is bigger than Adobe RGB. Adobe RGB has the same blue and red as BT.709/sRGB (less saturated than DCI-P3) but a more saturated green than even DCI-P3.

http://i1222.photobucket.com/albums/dd496/asmodian3/Fig02-AdobeRGBDisplayP3sRGB-300x300_zpsdvbpsnim.jpg

[Nastenka]

Hello, I had a couple questions regarding the 3DLUTs created by DisplayCal.

- According to and following the various and official tutorials on how to create a madVR 3DLUT, during the workflow to create a madVR 3DLUT near the end madTPG should pop up and carry out further measurements. However, for me the 3DLUT is created after DisplayCal is done measuring, and while madTPG is launched, nothing is shown there or measured. Is this normal?

- I have profiled and created 3DLUTs for two monitors, however, when DisplayCal saves those 3DLUTs in its default location, it's called BT.709 for every monitor. This will lead to the files overwriting each other when profiling one after the other. I've furthermore noticed that the file sizes for different monitors are exactly the same, 96 MiB. I had the same question as above, have I set something wrong in the calibration options? I meticuously followed the official tutorial.

Thanks so much.

[huhn]

- According to and following the various and official tutorials on how to create a madVR 3DLUT, during the workflow to create a madVR 3DLUT near the end madTPG should pop up and carry out further measurements. However, for me the 3DLUT is created after DisplayCal is done measuring, and while madTPG is launched, nothing is shown there or measured. Is this normal?

this can be normal. you can verify it manually.

- I have profiled and created 3DLUTs for two monitors, however, when DisplayCal saves those 3DLUTs in its default location, it's called BT.709 for every monitor. This will lead to the files overwriting each other when profiling one after the other. I've furthermore noticed that the file sizes for different monitors are exactly the same, 96 MiB. I had the same question as above, have I set something wrong in the calibration options? I meticuously followed the official tutorial.

the same file size is normal and the overwritng too is a result from making displaycal easier to use. i totally gave up on the "easy" way and do it manual now.

you can create a new 3d with a custom name and your old readings using a tool called 3dLUT maker it's part of displaycal GUI.

[Nastenka]

Thank you for your response. It looks like everything might be alright after all. I tried the 3DLUT Maker, but wasn't sure if it would do the same as unchecking "Create 3DLUT after profiling".

[huhn]

it is the same program but it let you save the 3d LUt with it own name.

if you don't have 2 different 3dLUTs than you are not fine.

[Nastenka]

I've had a look at the 3DLut Maker, but it differs from the old method I used in that I can now choose gamma for Rec.1886, the default is 2.4 with my profile loaded.
During the regular calibration process however, I originally chose a Tone curve of 2.2.
Should I adjust the Rec.1886 gamma for the 3DLUT to 2.2 accordingly or should I go with 2.4 nonetheless?

[huhn]

it doesn't matter what you choice before you are just reading your screen the 3d LUT will use a source profile like bt 709 and map that to the readings of your screen.

or in short just select the gamma you want.

[Asmodian]

Simply agreeing with huhn.

If you can change options on your screen to more closely match your desired output, before profiling, you might get slightly better results but changing settings during profiling will not make a difference.

I don't think the color patches used for profiling change at all when changing options like gamut and gamma. At least I did some testing with my TV (LG OLED C7P) and I seem to get identical results (within my measurement variability) if I select either DCI-P3 / BT.709 or 2.20 / BT.1886 when profiling using Calman 2017 (the TV was always in the same mode, P3 BT.1886) and creating a BT.709 or BT.2020 3DLUT. Profiling is simply making a map of color value sent v.s. the color displayed so it makes sense that there aren't any options that affect the results. We are mapping the pure native RGB color space of the panel.

[Nastenka]

Thanks you two for your input. Left it on 2.4.

[Arm3nian]

No, DCI-P3 is bigger than Adobe

If so, why does Displaycal report the gamut volume of DCI-P3 as larger than Adobe RGB?

[huhn]

except for the bigger or larger word both you say the same not?

[Arm3nian]

I'm trying to measure HDR peak brightness to create a 3D LUT, but need some help to load a test pattern. I use HCFR and set the reference to UHDTV DCIP3 and the gamma to 2048. I set the generator to madTPG. The problem is the white it displays isn't white, it's a dark gray?

To get my TV to switch into HDR mode I start an UHD bluray and go fullscreen. It would be nice to skip this step entirely but I don't know if that's possible without madVR.

[Asmodian]

There is a button in madTPG to enable HDR so you shouldn't need to launch a UHD video to switch to HDR. Have you used it? This was all I needed to measure my displays peak white in HDR.

[Arm3nian]

The button does not do anything, with or without a test pattern. Also, when I switch by playing a movie, and measure a white, it's just the peak brightness of a non HDR white. I feel a test pattern with HDR metadata has to be sent for it to switch, and for it to actually display using HDR.

[Asmodian]

To get HDR to trigger with the newest Nvidia drivers (417.35) I need to set madVR to use FSE and output 10 bit as well as madTPG set to full screen. When I do this the 255,255,255 pattern from HCFR is at my HDR's peak white.

[Arm3nian]

I was on the previous drivers (416) when I received my TV. Updating to 417 changed nothing for me. Both windowed and FSE still trigger 12 bit HDR output in madVR. The button did nothing on 416 or 417 however. If I switch using madVR, and then output white in HCFR with madTPG fullscreen, it does give me the peak white.

I did this the other day though using some test patterns I found on avs. The TV sustains ~1930. After setting the optimal settings and adjusting the white point it drops to 1597 sustained. So I should use this in the 3DLUT menu as target peak luminance? The madVR HDR preset doesn't calibrate the gamma, only the whitepoint. So I was thinking to skip it all together, adjust it to 6500k manually using the RGB gain sliders, and just profile. The TV has gamma settings as well, if I really wanted to I could adjust that, but it's quite a bit of work.

[Damien147]

How do you guys adjust backlight on your TV?Is there a standard in cd/m^2 or it's personal preference?

[Asmodian]

It is personal preference, and the ideal level depends on the light level in the room. I use 100 cd/m² for a dark room.

The SDR calibration guidelines suggest between 80-160 cd/m².

[Damien147]

I was with 200 cd/m² these days for night and day.I am gonna try close to 160 then.Thank you.

[chros]

I was with 200 cd/m² these days for night and day.I am gonna try close to 160 then.Thank you.
There was a THX value back in the day (105 nits, in theatres), but due to lot of complaints (too dark) they raised it to 115-120 nits. That is for night usage.
I use:
- night: ~120 nits with 2.4 gamma
- day: ~140 nits with 2.2 gamma
Hence I had to create 2 3dluts and change them (with hardlinks in a script).

But you have to be aware that when you increase the blacklight of most of the displays (that doesn't have perfect black) the black level will also be raised (that's not a good thing).

[Damien147]

edit:I am a noob and for 2 years I got used to setting the backlight full.
I am gonna try your settings but from what I can understand it needs some time to get used to it.Thank you both.

[chros]

it needs some time to get used to it.
Yes, but it will be only days :) Especially if you created a 3dlut for it/them.

[Damien147]

I don't have a calibration device to do that.
I am using another guy's TV settings that calibrated in bt.1886 and know some values in cd/m^2 for backlight.I know that each calibration is unique but these settings look better than default.
I've got a cheap 6 series Samsung with good contrast,still don't know if it's worth it giving 1/3 of its price for a colormunki display.(or it pays you back?I watch 3-5 movies during the night at winter)
Back to our subject ~120 nits to match my monitor doesn't sound bad.I tried it yesterday and I think I am gonna settle like this. :)

[huhn]

how do you even set the brightness of a device without having a measuring device.

[Damien147]

Wasn't planning 100% accuracy,just copied measurements but long story short I found a colormunki display in a more logical price and ordered it to measure things myself.:p

[chros]

Wasn't planning 100% accuracy,just copied measurements but long story short I found a colormunki display in a more logical price and ordered it to measure things myself.:p
You can buy those colormunki/id3pro devices second hand as well, it would be cheaper.

[Damien147]

I think it's hard to find something like that second hand in Greece.
225 euro(256 usd) was the first price I've seen for a colomunki display and said no.After some googling yesterday I found it at 150 euro(171 usd) and said why not(?):)
Got some reading to do now,the only thing I know is that I have to use displaycal.

[huhn]

they nearly doubled in price in the last past years.

and i'm not a huge friend of used colorimeter they drift and the i1d3 is out for a long long time now.

[Damien147]

They translate 100 euro to 100 usd usually here(...and sometimes if you are lucky enough..).:confused:

[chros]

and i'm not a huge friend of used colorimeter they drift
I don't think that a colorimeter can drift in time, a spectrophotometer can (includes some gas in it), I think we talked about this about 3 years ago :)
E.g. on UK ebay (https://www.ebay.co.uk/sch/i.html?_osacat=0&_odkw=i1d3+pro+colorimeter&_from=R40&_trksid=p2334524.m570.l1313.TR1.TRC0.A0.H0.Xi1+display+pro.TRS0&_nkw=i1+display+pro&_sacat=0).

[huhn]

colorimeter are known to drift.
http://www.spectracal.com/downloads/files/Website/Website%20Articles/Maintaining%20Your%20Colorimeter.pdf

the i1d3 is supposed to be far more stable than older device but it will still drift.

[chros]

colorimeter are known to drift.
http://www.spectracal.com/downloads/files/Website/Website%20Articles/Maintaining%20Your%20Colorimeter.pdf

the i1d3 is supposed to be far more stable than older device but it will still drift.

Oh, ok, so it can. i1d3 has sealed filters so it should drift less than others that don't.

[Damien147]

~30cm next and ~30cm behind the TV I have my monitor facing the same direction.Is it ok to have my monitor on and use it while calibrating the TV?

[huhn]

if your meter is directly on the screen ambient light shouldn't matter much.

[chros]

~30cm next and ~30cm behind the TV I have my monitor facing the same direction.Is it ok to have my monitor on and use it while calibrating the TV?
You can turn its brightness down to minimum if you don't want to turn it off.

[Damien147]

That makes sense,thanks guys.
Meter is gonna take some time to come because there is no availability.Till then I am trying to learn as much as I can to start straight away.

[Damien147]

When creating an HDR to SDR lut for madvr you have to put tone curve in calibration tab to gamma 2.2?

[huhn]

you select "as measured" the option under this tab are not important for the 3D LUT.

[Damien147]

I've seen gamma 2.2 being mentioned for hdr randomly and probably misunderstood it.I did it right then,just changed target peak nits.Thank you.
Did my first calibration yesterday.TV's PC mode and monitor went fine but failed at regular TV calibration for channels.Specifically with HCFR and primaries and secondaries.
I opened a 75% test pattern and tried to set R G B in HCFR to 100% but it doesn't work.Can someone correct me what to do?

[huhn]

your SDR 3D LUT "needs" to be gamma 2.2 when madVR is doing the HDR->SDR conversation but that doesn't count for a HDR->SDR 3D LUT it will take care of the gamma problem on it's own.

[Damien147]

I see,all good then.Thanks again.