Exporting LAB LUTs from Photoshop: Does it work?

Do LAB LUTs exported from Photoshop work properly (in DaVinci Resolve)?

In an earlier Insight of this LAB color space series, a Member commented asking if we could export a LUT from Photoshop and use it in DaVinci Resolve (or any other app that accepts Look Up Tables)?

It’s an interesting question because LAB lets us make some pretty extreme corrections, some of which look like pure ‘secondary’ corrections—corrections where we isolate a single color and change it.

Typically, we want to avoid secondary corrections when using LUTs to share corrections because they tend to break (resulting in terrible looking images). But some types of isolations can be carried across by a LUT and so it’s a question worth exploring:

Do LAB LUTs from Photoshop work properly, even the extreme corrections?

Note: LAB LUTs don’t actually exist

We first need to define what I mean by ‘LAB LUT’.

The thing to remember, the LAB color space is a working color space – not a viewing color space. When we work in LAB our images eventually need to be converted out of LAB and back into RGB for our monitors to properly display the image. When we’re working in LAB, we’re always still viewing the image in RGB!

The exact same theory goes for the LUTs we create.

The LUTs you and I share with each other are always assuming RGB color channels. Our LAB manipulations are first converted back to RGB before the LUT gets exported. LUTs do not manipulate anything other than RGB values.

When I say I’m exporting a LAB LUT, what I’m saying is: I’ve made color manipulations in the LAB color space and exported a LUT… but the LUT itself is exported only AFTER the image has been converted back into RGB. It’s a LUT just like any other LUT.

But because LAB can manipulate images in ways that are impossible in RGB, the question becomes:

Can LAB corrections manipulate images in such a way that RGB LUTs can’t properly express those manipulations?

4 LAB Corrections, 4 LUTs and DaVinci Resolve

In this Insight I’m going to create four different corrections in Photoshop CC 2014—in the LAB color space—and I’ll export them as .cube files to use in DaVinci Resolve. I’ll also export out JPEGs of the images to import into the Gallery so we can compare the images created by Photoshop and look for any variances.

Each of the four LUTs will get more and more extreme as I try to ‘break’ the LUT and find the limits of the technology.

The Limits of this Insight

If you’ve been following this series, you know that I’m sharing my Insights on this topic as I learn about them. This means I’m very likely to change my mind, workflow or recommendations as I gain more experience working in this color space.

After watching this Insight, keep in mind – you may find a very particular situation in which my recommendation doesn’t hold true. That’s okay! Really. It’d be awesome if you shared this situation with the rest of us here in the comments.

And you can be sure, if I run across such a situation – I’ll be creating an Insight for all of us to learn.

Enjoy the video below, as we dig into the LAB / LUT workflow because remember:

This LUT workflow could open up the LAB color space manipulations for apps that don’t support LAB… watch to find out if that’s the case.

– pat

Comments

Homepage › Forums › Exporting LAB LUTs from Photoshop: Does it work?

• Creator
  Discussion
• January 30, 2015 at 10:58 am #33658
  
  Patrick Inhofer

  Keymaster

  This is the discussion thread for the Insight: Exporting LAB LUTs from Photoshop: Does it work?

• Creator
  Discussion

• 
  Greg Greenhaw

  Member

  January 30, 2015 at 3:56 pm

  Hey Patrick thanks for the insight.

  Do you feel the photoshop lut is more color accurate? Does resolve’s curves force a slight color shift in lab?

  If you use a lut you move from a 32bit to a 10bit colorspace right? So you would have to be more strategic
  on where you place the node.

  I have you compared applying your AB3 powergrade with the intensity sliders dropped down compared to say your AB2.2 powergrade.
• 
  Patrick Inhofer

  Guest

  January 30, 2015 at 8:59 pm

  Greg – I can’t say the Photoshop LUT is more accurate… but I think it’s a ‘purer’ contrast expansion. Whereas in Resolve we have to add points to replicate Photoshop’s linear contrast expansion – which doesn’t make me happy when all I want is a pure contrast expansion at the endpoints (it’s simple request, IMO).

  RE: 32bit to 10bit colorspace – ‘Color space’ is the wrong term to use with bit depth. The two are independent of each other. LUTs – by definition – are much less granular representations than the corrections that generate them… so yes, in the points plotted out in a 32 point cube any variations between two points from the original correction will be lost. But it’s not a function of bit depth so much as trying to represent a very finely granular ‘transform’ with a much less precise representation of that transform (a LUT). Make sense?

  RE: The Photoshop AB3 LUT matches pretty close to my original Resolve AB2.4 PowerGrade. As mentioned (but not explained as well as I would have liked) there’s a slight variation between the two – but that fine, IMO. The fact they were so close just proves that an RGB LUT can represent a LAB color contrast expansion with precision. To me, that’s the big revelation of this Insight and if you have Photoshop CC 2014 – then you can bring LAB corrections to other apps besides Resolve.
• Author
  Replies

Log in to reply.

• Public
• All Members
• Only Me

• Public
• All Members
• Only Me

• Public
• All Members
• Only Me

• Public
• All Members
• Only Me

• Public
• All Members
• Only Me

• Public
• All Members
• Only Me