Hello,

Trying to follow the ICC display calibration workflow (copy from the guide below), ***
I'm having trouble removing the 1D components of my 3D LUT.

- I managed to neutralize it using the Filters --> Axis blend tool - but that keeps the general luminance correction of the 1D LUT.
- I managed to subtract from it a 1D LUT created from the same original 3D LUT using Maths --> subtract: But the result is not perfect and there is still R G B variation on the 1D axis.

But I think there must be an easier way that I didn't think of.

*** Here is the part in the ICC guide that I can't manage follow.
" If you want to use the VCGT for OS wide gamma/grey scale/white point calibration, the doubling-up can easily be avoided by generating a second 3D LUT within ColourSpace that has no 1D component. The original 3D LUT with the 1D component can then be imported into the ICC profile, with the Build VCGT Tag option enabled to correctly build the VCGT. The second 3D LUT exported from ColourSpace without the 1D component can then be imported into the same ICC with Build VCGT Tag disabled, replacing the 3D LUT component of the ICC for accurate gamma correction, removing the doubling-up of the 1D LUT data. "

You would have to provide the actual original profile, and the values used to generate the LUT, including the LUT Generation Size setting, etc., for assessment.

Steve

Of course,
Please find in the attachment:

1. The original profile
2. The calibration LUT: LUT is a BMD 33^3
3. Values used to generate the calibration LUT: Source: sRGB - Target: original profile (in attachment) - No luminance limit - No gamut mapping - Peak Chroma - normal range.

Attached file: Orifinal Profile

Attached file: Calibrtion LUT 33^3

Ignoring your issue on the 1D stuff, why is the profile so crap?
The Gamut is TINY...
And what patch set have you used?
There are strange points that are very close together, and the places with points lacking...

Steve

It is because the display I use to test this workflow is an especially bad one.
It is the integrated LCD display from an Asus ExpertBook laptop with a very tiny space (as you can see).

But my goal here is more about learning the process than achieving a good calibration.
I wanted to understand how to remove the 1D component of a 3D LUT in the hypothetical scenario of having to calibrate a monitor using only ICC + VCGT.
It can be the case in some editing facilities where I would find an editing station that has several UI monitors (for user interface) and a reference monitor for only video.

Eventually, my approach in that case would first be to manually correct the monitor using its own integrated settings (manual calibration guide) before using the ICC + VCGT (which have the limitations we know).
But in the specific case of the Asus laptop I use, there isn't any other option than ICC + VCGT.

I would totally understand if you don't have time for that, of course.

Start with a better patch set.
Then work from there.

Steve

Yes, that's exactly what I'm looking to understand as well.
In Calman, we can enable 'Set Grayscale to Unity' to prevent the 3DLUT from affecting the grayscale axis.
How does ColourSpace achieve this? It is critical for workflows where both a 1DLUT and a 3DLUT coexist in the pipeline

Unfortunately, what Calman does is wrong/fake.
It just deletes the grey scale points, and doesn't actually remove the 1D component from the full 3D LUT volume.
That means the 3D LUT volume still has the effect of the 1D component throughout the gamut.

You cannot therefore then apply a separate 1D LUT correctly, as the colour volume will be distorted incorrectly.

The process we define for ColourSpace to remove the 1D LUT from the 3D LUT does exactly what should be done mathematically correctly.
But, if the 3D LUT has issues, such as with the example here, the maths to remove the 1D component is virtually impossible to apply cleanly.

The example from raphdub used a poor patch set, as well as being a particularly bad display.
But the result of removing the 1D LUT component is not that bad for such a bad LUT, as can be seen below.
You can see the 1D component has been removed from the full 3D LUT volume, not just the grey scale points deleted/nulled.

But, if you really want to ONLY remove the grey scale points from the LUT you can use Point Adjust.
Just be aware that what you are doing is not removing the 1D component correctly.

Steve

1D LUT Component Removal

Hello,
Thank you for the further question/explanation.
And sorry for not replying earlier.
(I'm currently working on some grading projects and had no time to produce a cleaner profile with a better patch set.
I'll be pleased to do so as soon as I'm done with my current projects.)

In the meantime, wouldn't it be possible to theoretically describe the process of correctly removing the 1D component from the 3D one?

I totally get the volume distortions that can be remnants if you only take out the 1D grayscale corrections.
I guess removing them involves some sort of volume adjustment around the grey axis, correlated to the inverse of the 1D corrections.

https://lightillusion.com/advanced_lut_swap.html

Steve

This is excellent,
Thank you!

Always good to RTFM

Steve