Pages: [1]   Go Down

Author Topic: LUT color calibration for DaVinci Resolve 14.2  (Read 805 times)

The View

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1250
LUT color calibration for DaVinci Resolve 14.2
« on: June 16, 2018, 05:12:03 PM »

DaVinci Resolve needs a LUT calibration as it doesn't accept .icc profiles.

 https://www.youtube.com/watch?v=ugfcvfXNsqc


What worries me: this LUT calibration starts with manually balancing red, blue, and green channel on the monitor via buttons on the monitor, only then the automatic process with hundreds of patches (we know from .icc calibration) would start.

So this would likely mess up the .icc profile  (as it seems to adjust the monitor's basic settings). So am I alright to create a new .icc profile after that LUT calibration and can these two calibrations co-exist?

Has anyone done a LUT profiling of his monitor? Does the manual change of the rgb colors give a problem for an .icc calibration?


I'm finishing up two projects and losing my color management for Photoshop would be a nightmare.


« Last Edit: June 16, 2018, 07:08:07 PM by The View »
Logged
The View of deserts, forests, mountains. Not the TV show that I have never watched.

The View

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1250
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #1 on: June 16, 2018, 07:24:06 PM »

can you convert a .icc into a 3DLUT?
Logged
The View of deserts, forests, mountains. Not the TV show that I have never watched.

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 14214
    • http://www.digitaldog.net/
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #2 on: June 16, 2018, 11:21:20 PM »

An ICC profile can be a LUT based or Matrix based (simple) profile. I can't fathom how one could convert a Matrix profile into a LUT.
Logged
Andrew Rodney
Author “Color Management for Photographers"

Czornyj

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1808
    • zarzadzaniebarwa.pl
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #3 on: June 17, 2018, 05:57:37 AM »

Having PA271W you don't have to make 3DLUT calibration. Just calibrate the color space of the display to the color space of video content, and that's it.

smthopr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 457
    • Bruce Alan Greene Cinematography
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #4 on: June 18, 2018, 03:51:23 AM »

DaVinci Resolve needs a LUT calibration as it doesn't accept .icc profiles.

 https://www.youtube.com/watch?v=ugfcvfXNsqc


What worries me: this LUT calibration starts with manually balancing red, blue, and green channel on the monitor via buttons on the monitor, only then the automatic process with hundreds of patches (we know from .icc calibration) would start.

So this would likely mess up the .icc profile  (as it seems to adjust the monitor's basic settings). So am I alright to create a new .icc profile after that LUT calibration and can these two calibrations co-exist?

Has anyone done a LUT profiling of his monitor? Does the manual change of the rgb colors give a problem for an .icc calibration?


I'm finishing up two projects and losing my color management for Photoshop would be a nightmare.

I have an Eizo that does not have 3d LUT storage built in.  I saves a 1D LUT + matrix.  An associated .icc profile is also created with the profile for this set-up which will tell Photoshop what the display color space is for proper photoshop use.

So, to create the 3D LUT for Resolve, and the .icc profile for photoshop, I run Colornavigator software to set up my display to D65 white point, Native color space. (full gamut that the display is capable of)  This creates a .icc profile for the operating system (this profile doesn't effect the state of the display, it's a null correction with the display data).

Next, I create a 3D LUT using Resolve and Displaycal.  (you can use Calman or Lightspace also)  This LUT limits my display to REC709 colorspace and sets the correct gamma for color grading. (either 2.4 or 2.2 gamma.  It's your choice)

When running Resolve, the 3d LUT is applied to the video output of my decklink device to the display and I'm ready for color grading video.

When using photoshop through the displayport, my display is in full color gamut and all files are displayed correctly.

It's all pretty easy, except for making the 3d LUT for Resolve.  That has a bit bigger learning curve than the typical i1, or spectraview display calibration software.
Logged
Bruce Alan Greene
www.brucealangreene.com

smthopr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 457
    • Bruce Alan Greene Cinematography
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #5 on: June 18, 2018, 04:01:51 AM »

I have an Eizo that does not have 3d LUT storage built in.  I saves a 1D LUT + matrix.  An associated .icc profile is also created with the profile for this set-up which will tell Photoshop what the display color space is for proper photoshop use.

So, to create the 3D LUT for Resolve, and the .icc profile for photoshop, I run Colornavigator software to set up my display to D65 white point, Native color space. (full gamut that the display is capable of)  This creates a .icc profile for the operating system (this profile doesn't effect the state of the display, it's a null correction with the display data).

Next, I create a 3D LUT using Resolve and Displaycal.  (you can use Calman or Lightspace also)  This LUT limits my display to REC709 colorspace and sets the correct gamma for color grading. (either 2.4 or 2.2 gamma.  It's your choice)

When running Resolve, the 3d LUT is applied to the video output of my decklink device to the display and I'm ready for color grading video.

When using photoshop through the displayport, my display is in full color gamut and all files are displayed correctly.

It's all pretty easy, except for making the 3d LUT for Resolve.  That has a bit bigger learning curve than the typical i1, or spectraview display calibration software.

Please be warned that the youtube video that you linked to is not very thorough and The Displaycal settings are quite important.  The default settings create an sRGB gamma curve that I do not recommend for color grading movies.  Unfortunately, the Displaycal manual is quite difficult to figure out. And the terminology for the settings is a bit strange.

Lastly, when using Displaycal, loading the correct offset for your probe/display technology is critical and not always clear.  You NEC display calibration software does this automatically, but it's a manual setting in Displaycal.  I bought an i1 spectro to create my own offset and this is really the best way if you are going to create 3d LUTs for Resolve. (And the spectro makes great printer profiles too!!!!)
Logged
Bruce Alan Greene
www.brucealangreene.com

Czornyj

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1808
    • zarzadzaniebarwa.pl
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #6 on: June 18, 2018, 04:31:23 AM »

Why you’re not setting the color space and TRC using internal, high bit LUT instead of creating 3DLUT for Resolve, which gives worse quality?

smthopr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 457
    • Bruce Alan Greene Cinematography
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #7 on: June 18, 2018, 06:08:40 AM »

Why you’re not setting the color space and TRC using internal, high bit LUT instead of creating 3DLUT for Resolve, which gives worse quality?
When creating the 3d LUT, it is best to start with a color space that more than covers REC709.  I do set the gamma to something slightly brighter than I will use to make sure nothing is crushed at the low end before creating the LUT.  If you think about it, if I use color navigator to make the REC709 calibration it will start from the native state of the display, and then try to limit it.  If it limits too much any color, than it can not be addressed properly in the 3d LUT.  I also set the peak white a little higher than I will be using so that any correction of the white point does not dim the display further than the target peak white level.

Also, using the native color space of the display as the base from which to build the LUT, allows one to use the same base and .icc profile when using photoshop, which is what the original poster is trying to do.

Hope this make some sense to you.
Logged
Bruce Alan Greene
www.brucealangreene.com

Czornyj

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1808
    • zarzadzaniebarwa.pl
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #8 on: June 18, 2018, 06:37:51 AM »

ColorNavigator uses internal 16bit 3x3Matrix LUT to simulate the editing color space, and it also allows you to choose any TRC and wtpt as calibration target.

The 3DLUT applied on native color space of the display will introduce more rounding errors and will work on 8bit content, so all you acheive is deterioration of quality rather than any improvement.

The switch of calibration and profile is just two clicks on ColorNavigator icon, or a press of a button on PA271W housing, so it's not a big problem.
« Last Edit: June 18, 2018, 06:45:18 AM by Czornyj »
Logged

smthopr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 457
    • Bruce Alan Greene Cinematography
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #9 on: June 18, 2018, 07:06:00 AM »

ColorNavigator uses internal 16bit 3x3Matrix LUT to simulate the editing color space, and it also allows you to choose any TRC and wtpt as calibration target.

The 3DLUT applied on native color space of the display will introduce more rounding errors and will work on 8bit content, so all you acheive is deterioration of quality rather than any improvement.

The switch of calibration and profile is just two clicks on ColorNavigator icon, or a press of a button on PA271W housing, so it's not a big problem.
My Eizo display does not load or save 3d LUTs like some of the more expensive CG models.  So, I'm going this route.  And I'm also pretty sure that the CN calibration on my display is not as accurate as the 3D LUT created with a proper white point offset from the spectro.  For whatever reason, the CN/REC709 calibration shows a slight red saturation push leading to desaturated images in a cinema.   Almost all colorists grading on a display that is not a DCI compliant cinema projector are working with a 3d LUT calibration, usually using a LUT box in the display output to correct the display.  If you are an expert at display and print profiling and calibration for still photographers, you may not have come across the issues involved with color correcting movies.

I am not familiar with the NEC displays and calibration options, and maybe they behave better than their Eizo counterparts.  And the 3d LUTs don't seem to degrade the image on the display in any way that I can see with my eyes.  All the math, before the display output, is done in 32bit floating point, and that's what's really important here.  So, even an 8 bit display output is quite acceptable here.

Grading for HDR deliverables... Now that's a whole more complicated set of issues where the display depth is really going to be visible.  But one is getting into displays here that presently cost $30,000 or more...
Logged
Bruce Alan Greene
www.brucealangreene.com

Czornyj

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1808
    • zarzadzaniebarwa.pl
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #10 on: June 18, 2018, 12:31:43 PM »

I'm pretty sure you're wrong ;)

You can also use wtpt offset created with a spectro in ColorNavigator, but popular spectrophotometer is less accurate than i1Display Pro, see evaluation that I prepared with my friend Maciej Koper:
https://translate.google.pl/translate?sl=pl&tl=en&js=y&prev=_t&hl=pl&ie=UTF-8&u=https%3A%2F%2Fhdtvpolska.com%2Fforum%2Ftopic%2F35765-test-kolorymetrów-spyder-5-vs-i1-display-pro-vs-nec-pa242w%2F&edit-text=&act=url

FWIW - Maciej is THX/ISF certified color expert that works with film studios:
https://skalibrujtv.com/what-makes-us-better/#our-equipment

DaVinci can't create "3D LUT box" out of nothing, so it only can apply correction 3D LUT to 8bit content, so such calibration must be less accurate than properly calibrated high bit LUT display with 3x3 Matrix LUT. And even external high bit 3D LUT doesn't provide more precision than high bit 3x3 Matrix LUT in simulating synthetic color spaces. And even if it would, NEC PA271W features internal 17x17x17 3D LUT (like all NEC PA series displays)

smthopr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 457
    • Bruce Alan Greene Cinematography
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #11 on: June 18, 2018, 02:39:21 PM »

I'm pretty sure you're wrong ;)

You can also use wtpt offset created with a spectro in ColorNavigator, but popular spectrophotometer is less accurate than i1Display Pro, see evaluation that I prepared with my friend Maciej Koper:
https://translate.google.pl/translate?sl=pl&tl=en&js=y&prev=_t&hl=pl&ie=UTF-8&u=https%3A%2F%2Fhdtvpolska.com%2Fforum%2Ftopic%2F35765-test-kolorymetrów-spyder-5-vs-i1-display-pro-vs-nec-pa242w%2F&edit-text=&act=url

FWIW - Maciej is THX/ISF certified color expert that works with film studios:
https://skalibrujtv.com/what-makes-us-better/#our-equipment

DaVinci can't create "3D LUT box" out of nothing, so it only can apply correction 3D LUT to 8bit content, so such calibration must be less accurate than properly calibrated high bit LUT display with 3x3 Matrix LUT. And even external high bit 3D LUT doesn't provide more precision than high bit 3x3 Matrix LUT in simulating synthetic color spaces. And even if it would, NEC PA271W features internal 17x17x17 3D LUT (like all NEC PA series displays)

Color navigator only will use spectro offsets with the CG models that upload 3d LUTs...  For my model, spectro offsets don't appear as an option in the CN software.  I'm using an iOne 2 spectro.  It's really not so much about the absolute accuracy of the offset, it's about not choosing the wrong offset when not using CN software :)

Davinci will apply the 3d LUT in software, before it gets to the decklink device so I'm guessing that it's not 8 bit at that point.  But that's not really a big issue for me anyway. I'm not seeing any banding in my output now, and if I did, I can always turn off the LUT, switch to 10 bit output and use the REC709 CN calibration that I've also got saved in my display and check for banding in more detail.  8 bit vs 10 bit display accuracy is not going to change the grade that I will apply and does not effect the rendering of the movie.  The very small, but  just noticeable difference between the 3d LUT calibration and the CN calibration does effect my color decision, so that is more of an issue.
Logged
Bruce Alan Greene
www.brucealangreene.com

Czornyj

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1808
    • zarzadzaniebarwa.pl
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #12 on: June 18, 2018, 04:22:10 PM »

Actually you can calibrate the display using i1Pro2, and then read x,y coordinates with i1Display Pro sensor to create CN offset target. The resulting calibration should give same effect as 3D LUT, but should introduce less banding.

It's still less accurate method than calibrating with i1Display Pro, that has spectral calibrations done with higher resolution spectroradiometer (Konica-Minolta CS-1000), but if you prefer the result, you can achieve it that way.

The View

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1250
Re: LUT color calibration for DaVinci Resolve 14.2
« Reply #13 on: June 18, 2018, 06:16:19 PM »

I have an Eizo that does not have 3d LUT storage built in.  I saves a 1D LUT + matrix.  An associated .icc profile is also created with the profile for this set-up which will tell Photoshop what the display color space is for proper photoshop use.

So, to create the 3D LUT for Resolve, and the .icc profile for photoshop, I run Colornavigator software to set up my display to D65 white point, Native color space. (full gamut that the display is capable of)  This creates a .icc profile for the operating system (this profile doesn't effect the state of the display, it's a null correction with the display data).

Next, I create a 3D LUT using Resolve and Displaycal.  (you can use Calman or Lightspace also)  This LUT limits my display to REC709 colorspace and sets the correct gamma for color grading. (either 2.4 or 2.2 gamma.  It's your choice)

When running Resolve, the 3d LUT is applied to the video output of my decklink device to the display and I'm ready for color grading video.

When using photoshop through the displayport, my display is in full color gamut and all files are displayed correctly.

It's all pretty easy, except for making the 3d LUT for Resolve.  That has a bit bigger learning curve than the typical i1, or spectraview display calibration software.

I just heard back from a post I did on the DisplayCal forum.

Because I'm not using a decklink and just one monitor I was advised to convert the .icc profile into a 3d LUT - which DisplayCal can do.

I think this way the wide gamut color space of the display will not - as you mentioned - be crushed and my .icc profiles will stay untouched.  (The fiddling with the RGB channels via hard buttons on the display is also being avoided this way)

To be safe, I will first complete a client project and only then calibrate the GUI viewer on DaVinci. (I changed the color space from 2.4 gamma to 2.2 gamma and now it is quite close to what FCP X shows when addressing .icc profiles).
Logged
The View of deserts, forests, mountains. Not the TV show that I have never watched.
Pages: [1]   Go Up