I'll often end up with a LUT that looks great except for a bad patch or two. I'm usually doing at least an 18pt calibration, and the other patches all seem right. And repeating the calibration means the error doesn't usually repeat in the same spot.

In cube view, it will look like a random dot thats just way out of line.
Putting up a multi-color gradient test image, the gradation will look clean for the whole image, and then have a blip of weirdness around that color value.

In the past, I've used the LUT filters to smooth it out. Which is doable, but not ideal. Is there a way to delete these bad points? (i.e. Sorting by Delta-E, exporting it, deleting, re-importing?)

Thanks for reading.

You can open the profile in any text editor and edit the patch data.
Just make the values -0.000 (negative zero).
(You can find the patch data by searching for the RGB Triplet values, or the Yxy values, as seen in LightSpace when you double click the point in the CIE diagrams, when opened via the 'Manage Colour Spaces' menu.)

That way LightSpace will 'ignore' the reading.

Hope that helps!

But, I would look into why you have bad readings.
That suggests an issue with your workflow (or the probe), especially if they only happen 'sometimes'...

Steve

Yes but I don't know which reading is bad in the CIE diagram. When mapping a wide gamut display to a narrowed gamut target, almost every reading will have a delta greater than 5. So there's no way to weed them out.

What I want to do is to remove a bad point on a LUT. I'm less inclined to believe these occasional bad patches are a hint of a bigger problem. Because the problems don't repeat in the same places. They have occurred occasionally across 4 different probes I've owned (including my K10a) and 3 different display technologies (Plasma, OLED, proj.) The only common link is that resolve feeds them. So I'm more likely to write the errors off to latency issues.

If you have 'bad points' in the LUT, then you have bad points in the profile, and they will be obvious to see as they will be well outside the normal readings as seen when you use the 'filters' to select different patches to view on the CIE diagrams.

If they are not 'obvious' they are not bad readings.

It actually sounds like the readings are good, but the 'patch' measured is incorrect.
That can happen if Resolve 'freezes' and the patch change is no longer in-sync with LightSpace.
(Latency issues as you say)

In a situation like that you have to fix the real issue, not the end result!

To test, use a different patch generator.

Steve