PCG graphs

A PCG graph is a binary procedural generation network: sampler and filter nodes with their settings, connected pin to pin, plus input and output terminals. You normally read it by tracing edges in the PCG editor. PCGIR writes the same graph out as line-based text: every node with its settings and edited properties, every connection, and the terminals. You can compare it, review it, and give it to your assistant as context. PCGIR is read-only. Writing text back is planned.

What it gives you

Diffable in version control

The same reader writes a pcgir.txt asset-dump file, so a tuned scatter density or a rewired filter shows up as a few changed lines that git log -p and grep can read.

Reviewable like code

You read which node uses which settings, what each edited property is set to, and how pins connect from input to output. No clicking from node to node to work out what the graph builds.

Readable for your assistant

One consistent format for the whole graph. Each node is named by its settings class, and subgraphs show inline as a subgraph = "…" property. Your assistant reads the graph and reasons about it.

Faithful and consistent

The live read and the pcgir.txt file share one reader, so they always match. The input and output terminals are included, and each connection appears exactly once.

What the text looks like

The first line names the graph. Each node gets a short name and a settings class, an optional @(x, y) sets its position, edited properties sit in braces, and connect lines join output pins to input pins, including the N_Input and N_Output terminals:

entry pcg `/Game/PCG/G_FoliageScatter` {
    node N_Input = `/Script/PCG.PCGGraphInputOutputSettings` @(-400, 0)
    node N_Output = `/Script/PCG.PCGGraphInputOutputSettings` @(800, 0)
    node N1 = `/Script/PCG.PCGSurfaceSamplerSettings` @(0, 0) {
        PointsPerSquaredMeter = 2.5
    }
    node N2 = `/Script/PCG.PCGSubgraphSettings` @(400, 0) {
        subgraph = "/Game/PCG/SG_BoulderScatter"
    }
    connect N_Input.In -> N1.In
    connect N1.Out -> N2.In
    connect N2.Out -> N_Output.Out
}

A subgraph appears as a subgraph = "…" property on its parent node. Ask for the subgraph separately to read it. Node names (N1, N2, …) are stable within one read, but they can be renumbered after an edit.

How to ask for it

Ask about the graph, or ask for a change. Your assistant reads it as text to answer or to plan, then makes edits in the editor. PCGIR itself doesn't write the asset. More examples are on the Procedural generation (PCG) page.

You: What's scattering points in G_FoliageScatter, and does it call a subgraph?

G_FoliageScatter scatters with a Surface Sampler at 2.5 points/m², then feeds a
Subgraph node that calls /Game/PCG/SG_BoulderScatter.