How To Make Resolve’s ‘Color Match’ Easier with the ‘Brightness First’ Rule

December 25, 2014

DaVinci Resolve's 'Color Match' feature can be tricky. But if you follow the 'Brightness First' Rule, you'll probably get better results.


Series
Day 25: 25 Insights in 25 Days Holiday Marathon

Why the ‘Brightness First’ Rule Makes DaVinci Resolve’s Color Match Easier

In Part 1 we examined what DaVinci Resolve’s ‘Color Match’ feature is trying to do with test charts.

We also took a guess at what seems to be the problem many users are experiencing when it fails them.

I made the observation that we’re asking ‘Color Match’ to also match exposure… and that’s likely where the workflow is tripping up.

Un-tripping ourselves with the ‘Brightness First’ Rule

In this Insight, I’m going to put my theory to the test.

We are going to do an overall luminance adjustment to get our two test shots to line up on the Waveform Monitor. We’ll also fix overall color balance problems as revealed in the OneShot’s black/gray/white chips.

THEN we’ll see if we get a closer match. But remember…

The goal is a Color Match, not a perfect match

The key here is: We’re looking for a workflow that allows two different cameras to match each other… not get a perfect match from the same exact camera with two different recording profiles.

At the end of this Insight you’ll notice there’s slight differences in saturation. But if these were two cameras from two different angles? A slight saturation adjustment in the shadows on one camera and we’d be good to go!

Try this workflow and let me know how it goes?

As I say in this Insight… color charts are not something I get very often. But if you’re a DP or do a lot of dailies work, you may find this workflow takes a ton of frustration out of matching cameras… especially if you shot test charts for the specific purpose of getting Resolve’s ‘Color Match’ feature to do the heavy lifting.

Please use the comments to let me know if this workflow is working for you! Or if you have any refinements you find to increase the reliability of this workflow.

I’m calling this Insight Series a work-in-progress... for now, at least.

– pat

Member Content

Sorry... the rest of this content is for members only. You'll need to login or sign up to continue (we hope you do!).

Membership options
Member Login

Comments

17 thoughts on “How To Make Resolve’s ‘Color Match’ Easier with the ‘Brightness First’ Rule”

  1. Well THIS is brilliant. Solves a headache I’ve been having of late. Not sure why I didn’t think about focusing on the gamma problem first THEN the color issue. Duh, smart Patrick. You guys just keep on bringing more and more goodness. Thank you.

      1. Hi Patrick. Great insight.

        By the time Art Adams published the DSC Labs OneShot, he specifically said that the chart was made for Rec.709. However, I see that through out this series you are using BMD Log footage to work with the chart. What is the reasoning in your approach using the chart in a non Rec.709 gamut? Also, I would like to know if the DSC Labs OneShot cinema version has been released? I haven’t been able to find it.

        Thanks in advance.

        1. All Art is saying is that his targets don’t go outside the Rec709 gamut. He’s not saying you need to record to Rec709 for his targets to work.

          I personally get nearly identical results using Color Match before and after the contrast expansion into Rec709… except that I find I get more visually pleasing results with Color Match as the first node in the node tree.

          Luckily enough, where to put the Color Match node is super-easy for you to test. And you *should* test it. What I offer is a starting point. Feel free to start where I suggest and then completely redesign your workflow should you decide you like the results better (in fact, nothing would make me happier).

          RE: BMD Log footage – I just happen to own a Pocket Camera, so that’s the easiest for me to test these workflows with. Also, much of the footage we have access to use on our Insights is either BMD Log or Arri Log. But the concepts will translate to other cameras.

  2. The latest version of resolve is significantly better at color matching.

    Why do you think saturation is not as important? In your finial grade the reds were way of.
    Seems like the goal is to remove any camera bias? Since the rgbcym swatches were at the same
    saturation would it not make sense to match their saturation? However I wonder if removing the cameras

    look/biases in the end might be a bad thing. Seems like we end up pushing and pulling the image around

    so much that skin tones can end up looking less flattering.

    In this case the only difference between the shots was the gamma, shot on their cameras. Seems weird they ask for the gamma but then can’t match their own cameras under ideal circumstances. I guess you could apply their film to rec709 lut before hand?

    Wonder how the light color temp affect this as well. If you have a weak channel in one shot could you ever match it?
    http://www.dvinfo.net/article/ultra-hd/dragon-color-does-tungsten-or-daylight-make-a-difference.html

    Can you do a chart match by hand? Wonder if you think this match feature saves any time?

    1. You make a bunch of points here and I’ll see if I can hit them all:

      • It’s Color Match, not Camera Match or Exposure Match. As you saw, it does a great job lining up the color test patches on the required vectors and even tries to match saturation… but the price is overall color balance.

      • I wasn’t worried about the saturation in the lows because I can fix that. Again, the goal is a perceptual match between cameras… typically cameras are shooting different angles and different framings, a saturation mis-match in one part of the tonal range may not be a problem at all. And if it is, I’d just fix that in a separate node.

      • If you’re using a test chart to match different types of cameras, then one thing that helps to get them to match together is to remove their biases. In this case, Color Match does a nice job of it.

      • Think of Color Match as your Primary Correction. It’s job isn’t to get flattering skin tones. You will probably need to do that in a Secondary. Even when matching a one camera shoot between the wide and medium or close-up, pulling skin tones to make them more flattering is Standard Operating Procedure. Much of this depends on the DP and how the lighting works between the angles.

      Make sense?

  3. I think this is smart enough that BMD should probably add this to their instructions for Color Match. I’d like to see a Part 3 where Patrick does the exact same adjustment manually — which I still say can be done every bit as accurately as Color Match. It does help to have the charts. And there will always be the issue of charts that are shot badly.

    1. I have manually replicated exactly what Color Match is doing. Kind’a fun and interesting to do so. I’ll do a Part 3 on this.

      Of course, in the normal course of a job, I don’t have Color Match (since I don’t have charts) and do this by eye… which is what I think you’d like to see me do Marc, yes?

    1. Nick – Generally speaking, if the brightest elements in the waveform should be balanced to white and the darkest elements in the waveform should be balanced to black AND there is no clipping at either end—auto balance does a decent job. I’d do the auto balance in one node, then follow up with the Color Match in another (keeping Color Match’s machinations to just working on color and not contrast).

      1. Hey guys – sorry about that! This must have fallen through the cracks in our switch over to the new site and or patrick didn’t see Anton’s comment a couple weeks ago. Regardless, our apologies. I’ve fixed the streaming video (download for premium members was ok)

Leave a Reply

Hundreds of Free Tutorials

Get full access to our entire library of 900+ color tutorials for an entire week!


Start Your Free Trial
Loading...