The cut is the test

Nobody hires a colorist for one good-looking frame. They hire you because four hundred shots have to cut together without the audience ever feeling the camera change. The scene depends on those matches holding: A-cam to B-cam, morning pickup to afternoon master, wide to close.

In Resolve, start on a representative wide shot when it contains the scene's main faces, neutrals, and lighting relationships. Grade it, then use Grab Still in the Gallery to keep the approved result. A signed-off shot overrides that default. It is the master even when another frame would have been easier to grade.

Judge the cut, measure the delta

Your eye adapts within seconds, while scopes remain stable. Use the cut to judge continuity and the scopes to measure why it still calls attention to itself. Neither replaces the other.

Start with corresponding reference points. Compare the same face or neutral across the two shots. A practical light, wardrobe color or shadow region can work too. A wide and a close-up contain different proportions of the scene, so their whole-frame floors, ceilings, and trace shapes do not need to be identical.

{
  "kind": "trace",
  "shape": "healthy",
  "ghost": true,
  "regions": [
    {
      "band": "floor",
      "label": "floor, matched"
    },
    {
      "band": "ceiling",
      "label": "B-cam · delta",
      "accent": true
    }
  ],
  "caption": "Two traces, the master's and the B-cam's. Their floors already sit together, and the gap left at the top is the delta exposure has to close."
}

Then color. Sample or isolate corresponding subjects. The parade helps identify channel differences; the vectorscope helps compare hue direction and saturation reach. Take out the supported difference instead of forcing two globally different compositions into the same scope shape.

The master is not a clean plate

One thing changes for good in this room. Every reference you have matched to so far was the plate itself. The answer was always to undo whatever had been done to it. Here the reference is another shot from the same scene, already graded: the master the client signed off on this morning. Undo the B-cam's fault and you are still short, because the master sits somewhere the plate never did.

So you are measuring a distance rather than repairing a mistake, and you make the same read either way. Find where your trace sits against the master's and close that gap, in whatever direction it turns out to run.

Run that read on the next session. The B-cam came in brighter and flatter than the master. The same reference points sit higher and closer together, so exposure comes down and contrast restores their separation. Then cut between the shots. If the transition still calls attention to itself, inspect the remaining tonal, color, or local difference.

Which shot is the master

All of that assumes somebody already decided which frame the rest of the scene is measured against. On a real scene nobody has. That decision comes first and it is yours.

The shot you grade properly is the one the scene is judged on, and the others inherit from it. You copy the master's grade onto them and trim each one until the cut disappears. You do not build every shot from neutral and hope they end up in the same place.

Two things override the default. A shot the client has already signed off is the master whatever else you would have picked, because you cannot ask them to un-approve it. And a shot that was lit differently may not reach the master at all. Take that one as far as it will honestly go and say so, then let the edit and the client know before they find it themselves.