Light Illusion is renowned for the level of support it provides to all its customers, and strives to provide an unmatched level of assistance.

These forums are provided to enable open discussions on all aspects of the Film and TV industry.

 | Forums | Register | Reply | Search | Statistics |
Display Calibration Light Illusion Forums / Display Calibration /

alternative options to clean bad points on a LUT?

Author blasetheodore
#1 | Posted: 22 Sep 2016 00:33 
I've got a LUT that looks pretty solid except for one random bad spot. I've tried to use the smooth filter to notch it out, but couldn't hit it precisely enough.

Are there any other way to clean a bad spot on a LUT other than those filters?

Thanks for reading.

Author blasetheodore
#2 | Posted: 22 Sep 2016 00:34 
(screencap of bad spot on 3D cube in link)

Author Steve
#3 | Posted: 27 Sep 2016 11:06 
Looking at the image I suspect you have a display that is clipping.
You will need to address that, rather than attempt to 'fix' the LUT, as the LUT looks to be doing what it needs to to deal with the profile data it has.


Author blasetheodore
#4 | Posted: 29 Sep 2016 21:38 
I've gotten what looks like a very clean profile now. Used the display itself to hone in on 100cd, rather than using the LUT.

When I generate a rec709 LUT, the 3D cube looks pretty clean. And the picture looks close to reference grade. But the shadows are very red. And a 10bar greyscale has pink tones on bars at 3, 6,7.

The relax tool makes the cube look pretty, but makes the picture look very violet and wrong. Even light adjustments do harm, rather than good. Not sure what it should be used for.

The "smooth" tool seems broken. I can make the 3D cube look better. (specifically to smooth a red bulge in the shadows.) But when I apply the LUT, it looks like the image has been color inverted. Even on tiny tiny changes.

I'm considering either manually modifying the shadows of the LUT via TIFF, or maybe running a full profile of the display with the LUT, and combining the 2nd LUT with the first?

Anything obvious I'm missing?

- All non-essential functions disabled
- white point is correctly balanced via display controls
- luminance at 120cd via display controls
- fully light controlled room

- DaVinci resolve via network link to LS
- patch size determined by your gradient technique (though below 120cd, patch size irrelevant)
- 21pt profiles

- i1D3
- i1pro for offsets

Thanks for reading,

Author Steve
#5 | Posted: 29 Sep 2016 22:08 
You have to remember that any LUT LightSpace generates is a correction for the profile data.
So, if the LUT generated looks wrong, the probability is the profile data is wrong.
And the profile data is the raw readings from the probe.
So that suggests the probe is providing poor data.

The LUT Manipulation tools work as they should, but again if the profile data is poor there is little any manipulation tools can do to make what would be accurate improvements.
Having said that the Relax Tool cannot make the image colour change that is technically impossible.
It is equally impossible for the Smooth tool to invert the image - again technically impossible.
Such issues suggest a ore major problem with your profiling.

What you need to do is send us the raw profile data, and the LUT you have generated.
we can then assess what is likely the issue.
But, I suspect issues outside the control of LightSpace.


Author Steve
#6 | Posted: 29 Sep 2016 22:19 
Oh, and obviously if the display is not pre-set correctly, so the there is clipping in the blacks or the whites, no post calibration can ever work - as the various User Guides explain.


Author blasetheodore
#7 | Posted: 1 Oct 2016 20:22 
Hi Steve,

Here's all the profiles, matrices, LUTs, checkLUTs, and a step by step breakdown. Would really appreciate if you see something I'm missing.

Thanks for reading,

Author Steve
#8 | Posted: 1 Oct 2016 20:40 
Your integration time is way too low, at the minimum of 0.25.
You should be using a much higher value. That will be a key issue, as you have a number if invalid probe readings as the probe is struggling to get valid readings.

(0.75 is probably the minimum - 1.25, or higher, will be better)

Also, a drift value of every 10 patches is way too short - 100 is a more realistic value.

You should also be using AIO mode with the i1D3, as per the above link (assuming the i1D3 is a RevB).

And if the white point is spot-on, you should set the display to the desired peak luma value (100 nits) as there will be no reduction when the LUT is made. That will only happen if the white point is not accurate, as the user guides explain.


Author blasetheodore
#9 | Posted: 2 Oct 2016 05:08 
Thanks for the reply Steve.

Ok thanks. I thought the integration time was set by the probe automatically. The patch timings seem to be variable based on the patch luminance. (Dark patches take about 10 seconds, and bright ones take 2 seconds max.)
I'm running a profile now with 1.25, though the patch timings seem about the same as before.
I don't have the revB i1d3, so it doesn't support AIO.

I'll report back if the profile is any different with the new integration time setting.

Author Steve
#10 | Posted: 2 Oct 2016 09:42 
Further, using your profile (4K-w50c45o100-i1d3-21pt-02) as Destination, with a Rec709 2.2 Gamma Source, and Fit Chroma, I get this result.
And testing the LUT shows no artefacts at all.



Author Sanjin_Svajger
#11 | Posted: 25 Nov 2016 12:25 
I've got exactly the same problems. Maybe peak chroma is the culprit, I don't know. Will try fit chroma. Also regarding the probe integration time, I've always had the default, which is 0.25 and we used Closed loop, so there were no problems. I am trying a 1s integration time now if that changes anything. But I was under the impression that when in closed loop the probe says when it has read the patch to the CMS, so there is no need to manually set integration time?

Author Steve
#12 | Posted: 26 Nov 2016 09:46 | Edited by: Steve 
The Integration time setting sets the maximum time taken for the probe readings, so will always have an effect on the time taken and the number of individual readings that are 'integrated' for the final value read, so therefore will effect the quality of the final reading!

It has nothing to do with using DIP mode or not.

Also, I see you have found part of your problem here: t=2412&user=1


Author Sanjin Svajger
#13 | Posted: 27 Nov 2016 11:58 | Edited by: Sanjin Svajger 
So with a longer integration time, the probe will take multiple readings and the final value will be an average of the sum of all the readings?

Yeah, I fixed that. I was really doing the calibration as a side, third, thing and wasn't really giving it all the attention But the dE readings in the calibration interface are still way "off" compared to readings when doing a "characterisation"

Author Steve
#14 | Posted: 27 Nov 2016 12:06 | Edited by: Steve 
Yes, integration time acts exactly as described

And if you have differences with delta-E between different modes of profiling/patch reading you potentially still have an issue that is outside of LightSpace...

(Do remember that when performing a Quick Profile the data is always for the previous patch, obviously as the data can only be shown 'after' the patch has been read! And if taking 'Manual' readings you MUST set the Min/Max Luma in the Options page, as they are required (well, black is) to correctly calculate the delta-E and Y target values. All this info is in the various User Guides.)


Display Calibration Light Illusion Forums / Display Calibration / alternative options to clean bad points on a LUT? Top
Your Reply Click this icon to move up to the quoted message

Only registered users are allowed to post here. Please, enter your username/password details upon posting a message, or register first.
Online now: Guests - 1
Members - 0
Most users ever online: 411 [17 Mar 2015 10:47]
Guests - 411 / Members - 0
© Light Illusion - All rights reserved - Privacy Notice