FS7: Lut first or adjust clip first? Slog2 custom match Lumix in Davinci

Publimix

Well-known member
This is my workflow after testing a lot of others:

camera 1. Fs7 recording in 4K slog2 custom
camera 2. Lumix GH7 4K

Whitebalance both cameras
In post use Lut sonyslog2rec.709
Both camera's matched. No problem smooth and easy.


Earlier I had a post about coloring with Davinci. One of the comments was : first adjust the clip and then add the lut.
But then the clip is too bright. My workflow: apply Lut and then change brightness.

Any thoughts?

Lut first or adjust clip first?
 
As far as exposure, always adjust it first and then apply the LUT.

If the clip is too bright then lower the brightness after the exposure is controlled.

This might be enough when dialing down exposure before the LUT, but if not then it could also look like this in the pipeline:

Main Exposure
LUT
Brightness

[Usually WB is done first with exposure as well, but it's not as imperative as exposure if the footage is of a higher-quality because then it just become color grading if it's done after the LUT.]
 
Thank you NoBro

It is a bit counter intuitive
The clips are well exposed. No adjusting needed. When I apply the LUT the clip is overexposed .

I gave it a try: turned down gain then applied LUT.

The result is the same. Both pictures are exactly identical (as far as I can see).

But when you start adjusting the Gain, how do you know it is the right amount of gain?
With the 'LUT first' method you see what you get.

Is there any info on the Web?
 
Last edited:
Conversion LUT as final node in Resolve for me. All colour/exposure adjustments beforehand. This means you can swap out the final lut for rec2020 or similar and maintain your grade (within reason) if you go back to it and update.
 
Thank you NoBro

It is a bit counter intuitive
The clips are well exposed. No adjusting needed. When I apply the LUT the clip is overexposed .

I gave it a try: turned down gain then applied LUT.

The result is the same. Both pictures are exactly identical (as far as I can see).

But when you start adjusting the Gain, how do you know it is the right amount of gain?
With the 'LUT first' method you see what you get.

Is there any info on the Web?

There's a ton of info on YouTube and dedicated [paid] services.

You can learn to become one of the best at almost any computer-related tasks these days if you dedicated your life to it.

As far as adjusting gain, the results shouldn't be the same. How much depends on your taste and direction, and the tools you use.

There are color engineers and specialists who are much more qualified to explain a lot of this in depth (not just people reading a script), but my rudimentary understanding is the chain of command matters because it affects the computation of the mathematical values, the input and output.

___

Here is a slightly overexposed clip in order to amplify the results. It's in FCP X but all NLEs do the same thing (just some have more tedious control for certain tasks than others, like Resolve with pinpoint color manipulation).

I used ALEXA's famous LogC LUT which is known to really blow out every camera but its own.

In order of appearance:

(1) Original.
(2) Only the LUT has been applied (which naturally overexposed the clip even more).
(3) The exposure has been adjusted AFTER the LUT (exposure setting, -40).
(4) The exposure has been adjusted BEFORE the LUT (exposure setting, -40).

Visually, you can see the whites have turned into paste in (3) and all information has vanished.

(5) The waveform is immediately clipped (with exposure adjustments after the LUT).
(6) The waveform is naturally producing a sine wave (with exposure adjustments before the LUT).

___

The final image is just for fun, and has nothing do with any of the workflow above.
___

1_Original.jpg

2_LUT ONLY.jpg

3_Exposure AFTER LUT.jpg

4_Exposure BEFORE LUT.jpg

5_Waveform AFTER.jpg

6_Waveform BEFORE.jpg

x7HvYeA.jpg
 
Last edited:
The idea is your first Node is for the exposure , your 2nd node is for LUT . Then you go back to the first node and adjust exposure to make the LUT look right. sounds as if you had the LUT and gain on the same node, then your right it wouldn't make any difference. LUT is your last node. keep it seperate.
 
If you have lut first and it pushes the shot to clip you can’t recover that later in the node tree - hence lut is used typically near the end of the tree
 
Thanks to all,

NoBro the waveform tells the story, I see, thanks.

Had this misconception in my head for years. I now realize how and why I did this for the last few years. Until now I didn't use nodes. Put all the corrections in the basic node. That is, gain and LUT, and sharpness no more, no less. It works but it is not ideal.

But I do realize now that adding a node is better. And that changing the gain in the first node will be vissible after node 2.

So my workflow now will be: add last node for LUT. Adjust gain in first node and see the effect too.
 
Just to add on, yes order of operations is really important. If you are a bit OCD like I am, I tend break everything up into separate nodes. However there is an order of operations WITHIN nodes, so knowing where an operation happens within a node can guide you on which things you would prefer to do within a single node. Gain occurs before a 3D LUT in the order of operations so you could perform both operations within a single node and get the same result. But I still tend to separate everything even when I know what comes before what because it gives me the flexibility in turning things off individually either to see the effect or go in a different direction.

I believe there has been an updated diagram for the order of operations in Davinci Resolve ever since they added new controls but I can't seem to find it online, might have to take a look in the manual. See below for a version from about 2015:

http://vanhurkman.com/wordpress/?attachment_id=3331
Davinci Resolve Order of Operations 2015.jpg

Juan Melara had a good post about this a few years back:
https://juanmelara.com.au/blog/basic-resolve-node-structure-and-order-of-operations
 
But I still tend to separate everything even when I know what comes before what because it gives me the flexibility in turning things off individually either to see the effect or go in a different direction.

That's how I like to work as well, and it's one of my favorite features in FCP X...the ability to see a simple vertical checklist of "effects".

The disabling, moving around, swapping; all so quick.

FCP X_Effects.jpg
 
That's how I like to work as well, and it's one of my favorite features in FCP X...the ability to see a simple vertical checklist of "effects".

The disabling, moving around, swapping; all so quick.

View attachment 142612

I've never used FCP X, but that vertical checklist that you posted looks like a really great thing! Maybe one day Resolve will "borrow" that idea, lol. Although I'm not sure how well it could be implemented seeing that Resolve is node based with layer, parallel and even RGB split nodes being different methods of combining and routing nodes. Or is that a checklist of effects within a single node or layer? Is FCP X node or layer based?
 
Honestly, I can't pretend that I know what exactly a node is, lol.

But based on the vertical checklist in FCP X, any "effects" there appear to be layers all working together. You can't isolate anything without having the other effects not be impacted or have an impact.

The first one at the top is your starting point and everything you "layer" below moving forward affects the image. Anything you move up or down changes the results immediately (of course some changes are more aggressive than others like flipping two very diverse LUTs).

People who appreciate a lot of meticulous control will not be impressed, but for copying-and-pasting and trial-and-error-like workflows for blazing your way through projects, it's ideal for me.
 
It’s is important to understand which actions are destructive.

For example
Node1 sat =100
Node 2 sat =-100

Does this have any affect on the image? Or do they cancel out? What about the other way round?

Once one has a handle on understanding these transforms the order becomes obvious.
 
Last edited:
Honestly, I can't pretend that I know what exactly a node is, lol.

But based on the vertical checklist in FCP X, any "effects" there appear to be layers all working together. You can't isolate anything without having the other effects not be impacted or have an impact.

The first one at the top is your starting point and everything you "layer" below moving forward affects the image. Anything you move up or down changes the results immediately (of course some changes are more aggressive than others like flipping two very diverse LUTs).

People who appreciate a lot of meticulous control will not be impressed, but for copying-and-pasting and trial-and-error-like workflows for blazing your way through projects, it's ideal for me.

Ah ok, yeah sounds layer based. I started in Adobe Premiere and actually used to do my color grading layer based in After Effects using the Color GHear (not a typo) presets and LUTs for anyone who remembers the Color GHear product. I moved onto Resolve when BMD released the free version and then based everything that I was doing on Juan Melara's tutorials which I always reiterate is one of the best resources when learning to color grade. To be honest I had only used serial nodes in Resolve for a long time and I didn't really start using the layers, parallel or RGB splitter nodes until about a year and a half ago and it opened up a much bigger world of color grading for me.

Anyways to get back on topic, yes as Morgan mentioned above me, knowing which actions are destructive is key as well. I believe as it is operating at 32bit floating (someone more technical can please correct me if I am wrong) just about every color effect in Resolve is non-destructive EXECPT for LUTs that are on nodes. If you clip the LUT you cannot pull it back in following serial node, which is why it is typical a good idea to use it last.
 
Back
Top