Cherry/red tint after calibration (understanding how calibration works)

[caiokao]

Hi everyone.
I’m new to this calibration world and I’m trying to understand how things work.

Here’s my equipment:

– Ozone DSP27 IPS (1440p, brand says 100% sRGB and 95% DCI-P3)
– LG 29UM69G-B (1080p Ultrawide, 98% sRGB)
– Windows 10, HDR Off

I work with video productions and editing.

What I did and my problem:

I’ve been using the LG for a long time. Got the Ozone as an ‘upgrade’.
Bought the SpyderX Pro and used it’s software to calibrate the Ozone.
I tried setting the white point as measured, 5000k, 5800k, 6500k, custom, but every single calibration ended up with this pink/redish white and the reds looking like cherry (don’t know if I’m using the best word to describe it. Sorry).
At the end of the calibration there’s an image to see the before and after, and the peppers looked soooooo cherry/pink.

Tried DisplayCal with the same results.

I’ve returned the SpyderX and now I’m using a i1 Display Plus (a lot of people in the foruns saying that was a better device)

Using the ccProfiler software, I’ve set the white point to D65, adjusted the RGB channels throught the OSD of the Ozone, but in the end of the complete calibration the image looked red/cherry/pink again.
When adjusting the RGB I’ve set the levels correctly following the instructions on the screen, and when the calibration ended I ran the quality check and the white point was really off.

With DisplayCal and i1 Plus, same result.

I noticed that when the calibration starts (white/black/grey squares, then colored ones) everything seems fine.
Near the end of the process (80~90%, right before the creation of the profile), I can see all the image on the monitor shifting to this red tint.

I attached a test with the white point set to 7000k, and got really close just using the OSD of the monitor. After the calibration I’ve ran a verification and the white point was 0,3046 x 0,3118
Unfortunately I made the mistake of deleting this profile, so I don’t have the report.

Sorry for the long post, but I’ve banging my head the past 5 days trying to understand all of this.

Has anyone know what could be wrong?

Attachments:

You must be logged in to view attached files.

SpyderX Pro on Amazon  
Disclosure: As an Amazon Associate I earn from qualifying purchases.

[caiokao]

Used DisplayCal again!

White point set to 6500k.

Attached the reports. I know it shows good results, but this time I got the most redish whites ever. The skin tones are just terrible.

Attachments:

You must be logged in to view attached files.

[Vincent]

You are not using a colorimeter correction.
For xrite’s i1d3 (without buying additional gear) it must be a CCSS file with the spectral power distribution of the LED backlight technology of each display, hence each display may need a different one.

Try:
Ozone: Panasonic VVX[…] WLED PFS 94%
LG: White LED IPS

• This reply was modified 2 years, 2 months ago by Vincent.

[caiokao]

Thanks Vincent. I’ll try that.

I don’t know how this correction works, so I’ve leave it ‘None’

I’m wondering if something’s wrong with Windows. I made a wrong calibration on purpose (white point at 10k), and when I switch the profile through Windows, nothing happens.

[caiokao]

OK, I figured out the problem with windows not switching the profiles (actualy I Googled)

That correction really made a difference. The tint is almost gone, the skin tones are better but not good yet

I ran two calibrations. 6500k and 7000k. The reports are attatched.

Forgot to mention that on the LG this tint and red problem doesn’t occour. DisplayCal measure it at ~98,5% sRGB. The skin tones are WAY more natural. Just to clarify, I’m not judging only by eye. Premiere Pro and DaVinci Resolve have a tool to measure the correct hue for the skin tones. I have an exported video using this “cientific skin tones tool”, and I’m using it to compare the image on the monitors. The LG seems natural, the Ozone seems red/cherry.

DisplayCal ‘instrument mode’ shows two options for i1d3: LCD (generic) and Refresh (generic). I’m using the LCD. Is that correct?

Any more ideas?

Attachments:

You must be logged in to view attached files.

[Vincent]

(GPU) Calibration corrects white & grey, then makes a profile which stores display behavior.

100% apps by default do not ask OS for display profile and most of them do not know or care about color management, hence if content is supposed to be sRGB/Rec709 (and that is a big IF) and you sed those RGB numbers to display, all color will look oversaturaed on a widegamut screen.

Use a LUT3D for resolve on Ozone (on LG too) to map Rec709 content to display Colorspace. Apply VCGT on LUT3D if you don’t want global grey calibration at oS level or the opposite if you want to keep global grey calibration.

You should not buy a widegamut without knowing how to use it (if you are a Windows user).

[caiokao]

Thanks for your help Vincent.
I saw many threads in this foruns that has your advices.

I undestood most of the things you said. I’ll do some research for the things I’m missing.

I didn’t know that in a widegamut screen, the colors of Rec709 shows up oversaturated. I edit and export in Rec709.
You’re saying that I need to use LUT3D to convert this widegamut to Rec709 so I can see the correct colors, the way they were created. Is that correct?

The OS uses a profile, apps just ignore it. Ok, got it. But you have any other thoughts on why the whites aren’t correct?
I know the reports shows they are correct, but if I use my wallpaper as a solid white, it’s definitely tinted in red.

[Vincent]

I didn’t know that in a widegamut screen, the colors of Rec709 shows up oversaturated.

Non color manages, yes. Color managed all shows as expected.

You’re saying that I need to use LUT3D to convert this widegamut to Rec709 so I can see the correct colors, the way they were created. Is that correct?

The oposite, LUT3D reencodes Rec709 content in your display colorspace. But, “yes”.

The OS uses a profile, apps just ignore it.

Some apps.

Ok, got it. But you have any other thoughts on why the whites aren’t correct?
I know the reports shows they are correct, but if I use my wallpaper as a solid white, it’s definitely tinted in red.

-Wrong spectral correction, “it does not mach your display SPD”
-metameric observer failure
-colorimeter’s firmware sensivity curves do not match actual filter sensivity curves.

From most likely to less likely.

[caiokao]

Non color manages, yes. Color managed all shows as expected.

But when I calibrate and use a software like DisplayCal, I should see everything correct, right?

Just for information, I didn’t install DisplayCal. I’m using the portable version just to create the profile.

The oposite, LUT3D reencodes Rec709 content in your display colorspace. But, “yes”.

Got it. Can you sugest a software that I can try?
I’ve created a LUT using the 3D LUT tab in DisplayCal, and now I’m using this file in an adjustment layer in Premiere. Now I’m able to, at least, edit the videos.

-Wrong spectral correction, “it does not mach your display SPD”
-metameric observer failure
-colorimeter’s firmware sensivity curves do not match actual filter sensivity curves.

-should I try another?
-This observer is me, right? hahahahaha

I’m 100% sure that the white is tinted.

Thank you @Vincent for all your help!

[Vincent]

Non color manages, yes. Color managed all shows as expected.

But when I calibrate and use a software like DisplayCal, I should see everything correct, right?

No. Non color managed = oversaturated, color manages = ok.

If you want that non color managed content shows liek in an sRGB display you’ll need to use system wide simulation, like AMD’s , novideoSRGB or DWMLUT

The oposite, LUT3D reencodes Rec709 content in your display colorspace. But, “yes”.

Got it. Can you sugest a software that I can try?

DisplayCAL for LUT3D creation.

I’ve created a LUT using the 3D LUT tab in DisplayCal, and now I’m using this file in an adjustment layer in Premiere. Now I’m able to, at least, edit the videos.

Adobe suite should be color managed by ICC profiles unless you are suing a ver very very old version

-Wrong spectral correction, “it does not mach your display SPD”
-metameric observer failure
-colorimeter’s firmware sensivity curves do not match actual filter sensivity curves.

-should I try another?

You try the proper one for your display, or create one with an spectrophotometer if you do not find any better CCSS. Do not use CCMX unless yu made them.

-This observer is me, right? hahahahaha

and I. Observer metameric failure = you and Std. observer do not match. Since std. observer is a “mean”… it’s somehow expected when display emitted SPD has very narrow peaks, like in WLED PFS or laser projectors. WLED PFS should not be a huge problem but may happen.

As I said on 1st message I assumed that Ozone “MAY BE” an WLED PFS. It may be something different like those YAG phosphor seen on other threads. You’ll have to find Sectral Power Distribution for that Ozone display, ot buy/rent spectrophotometer and measure it (make CCSS correction, then share with community).
Or use some “close” to what it should be and match white visually. You can’t use abs. colorimetric luts3d if you do that, only intents with relative whietpoint, otherwise you’ll loose the visually matched white.

• This reply was modified 2 years, 2 months ago by Vincent.
• This reply was modified 2 years, 2 months ago by Vincent.

[caiokao]

At this point, my limited knowledge is causing me confusion.

I thought that when I did this calibration with DisplayCal or ccProfiler, the created profile was acting like a color manager.

I’ll do some more research before I continue. Can you sugest a link to start, please?

[Author]

Posts