Reference

Node graphs and node reference

Learn Oraphim node-graph fundamentals, typed connections, debugging, persistence, and how to use the exhaustive 462-node runtime reference.

Updated August 28, 2026

Node graphs and node reference

Oraphim uses node graphs when explicit data flow is more useful than a flat stack. The operating model is simple: choose a node, connect compatible sockets, edit its controls, preview the result, validate the graph path, then save and render.

For individual nodes, use the complete Oraphim node reference. It contains a dedicated guide for every node currently registered by the Oraphim 1.0.0 Beta runtime.

Typed connections

Each socket carries a registered data type. Connect an output only to an input that accepts compatible data. Image, color, geometry, vector, scalar, string, object, volume/grid, and other graph values are not interchangeable just because two sockets happen to be visually close.

When a connection is rejected, open the individual node pages for both ends and compare their Inputs and Outputs tables. Those tables are generated from the runtime registry rather than a manually maintained list.

Finding the exact node

Search the manual using the visible node name, such as Accumulate Field, or the exact registry ID, such as geometry.accumulate_field. Exact IDs are especially useful when two registered nodes share a similar display name.

Select a node before editing its properties. Drag nodes to organize the graph, pan or zoom the canvas when working on larger systems, and keep the path to the viewer/output visually understandable. If a drag behaves like a canvas pan, establish focus on a clear node or canvas target and retry.

Complex graphs are easier to maintain when branches have one purpose, important nodes have meaningful names, and duplicated branches are checked for independent node identities and intended connections.

Debug a graph from the output backward

  1. Confirm the correct graph/resource is active.
  2. Start at the viewer/output and follow connections upstream.
  3. Stop at the first node whose output differs from what you expect.
  4. Check that node's required input types and mode-dependent controls.
  5. Test the node in a smaller graph if its behavior is unclear.
  6. Compare preview with a short export when the operation affects final rendered pixels.

Persistence check

For a graph you intend to reuse, save the project, reopen it, and verify node IDs/names, positions, connections, edited control values, and the resulting output. The individual node pages give you the exact registry schema to compare against when something appears different after reopening.

The 462-node catalog is also the fastest starting point when you know a node exists but do not yet know what it connects to or which controls it exposes.