The node tree I use on every job, in the same order every time

Open ten projects from ten different people and you will find ten different node graphs. Open ten clips inside one of those projects and you will often find three. That is the real cost: not the grade itself, but the fact that every clip starts with an architectural decision.

A fixed tree removes the decision. Every clip gets the same nodes in the same order, whether it needs them or not. Unused nodes sit there doing nothing, and that is fine — an empty node costs you no render time and saves you the question.

The tree below is four clip nodes and two timeline nodes. It has survived commercials, showreels and long-form work without changing, which is the only argument for it that matters.

Resolve gives you a timeline-level grade that sits above every clip. Put your colour space transforms there and nowhere else.

• CST_IN — source colour space and gamma into your working space.
• CST_OUT — working space to your delivery, usually Rec.709 Gamma 2.4.

The clip grade now happens between two fixed walls. You are always working in a wide, predictable space, and the output transform is applied once for the whole timeline rather than eighty-six times with eighty-six chances to be inconsistent.

On the clip, four serial nodes, always the same four:

1. EXPO — exposure only. Get the image sitting at the right brightness before you judge anything else. Nothing here touches colour.
2. WB — white balance and overall cast. Now that the level is right, neutralise.
3. FACE — skin. A qualifier or a tracked window, and small moves only.
4. LOOK — the creative grade. Contrast, curves, whatever the project's identity is.

The order is not decoration. Exposure changes how you perceive white balance, white balance changes how skin reads, and skin decides how far you can push the look. Working out of order means revisiting earlier nodes, which is exactly the loop the fixed tree exists to break.

The tree is only useful if you can look at a node name and know what is inside it. That requires discipline that feels pedantic on clip one and pays off on clip forty.

If a clip needs a second window, it becomes a layer or a parallel branch attached to the node it belongs to — never a new serial node appended to the end. The chain stays four long forever.

When you inherit a project three months later and something looks wrong, you can go straight to the node responsible. That is the entire return on the discipline.

Build the tree on one clip with every node present and empty. Grab it as a still, or save it as a preset, and apply it to the whole timeline before you grade anything.

Now every clip in the project has the same four nodes waiting. Grading becomes filling in a form rather than drawing a diagram.

Right-click the clip → Grab Still
Gallery → right-click the still → Apply Grade to All Clips

Keep the preset outside the project too, in a folder you back up. A node tree you have refined over years is a real asset and it lives in exactly one place by default.

A fixed tree has a second benefit that only shows up later: it is machine-readable. When every clip has the same four nodes with the same names, a script can find the exposure node on every clip in a timeline and nudge it. A variable tree cannot be automated at all.

That is the bridge to scripted grading — batch exposure matching, automatic face-node placement, look propagation across a set. None of it is possible until the structure is predictable.

Start with the discipline. The automation is a consequence of it, not a replacement for it.