Documentation
Nodes
Node categories, connector types, and where the detailed reference will live.
Nodes are the unit of work in VFX Loom. Each node takes inputs, does one thing, and exposes outputs that other nodes can consume.
Overview
Nodes are presented by role in the node menu. The categories below are the structure this reference will grow into:
| Category | Purpose |
|---|---|
| Input / Sources | Coordinate streams and deterministic value sources. |
| Generators | Nodes that create an image from nothing — noise, shapes, gradients, waves. |
| Transform | Nodes that change where a source is sampled from. |
| Color | Nodes that change colour, channels, and ramps. |
| Filters | Image processing — levels, blur, masking, resampling and similar. |
| Math / Vector | Scalar and vector utilities used to drive parameters. |
| Composite | Blending and layering of multiple images. |
| Animation | Time and motion sources that make a graph move. |
| Output | The node whose rendered image is used for preview and export. |
Category colours in the editor follow the same vocabulary, and are presentation only — they are not saved in projects and do not affect evaluation.
How it works
Connector types
Connections are typed. At a high level there are two families:
Image connections carry a rendered image. Images flow from generators, through transforms, filters and composites, to the output node.
Value connections carry numbers and vectors — scalars, Vector2 values,
colours and curves. Value connections drive node parameters: anywhere a node
exposes a parameter that accepts an input, a value connection can replace the
fixed value.
A socket only accepts a compatible source. When two types do not connect directly, a conversion node from the Math or Vector group bridges them — for example, splitting a vector into its components or remapping a scalar into a different range.
Output
Exactly one node is the render target at a time. Image nodes expose a Set Output action, and the current target is marked in the menu so the active output is always visible.
Notes
- Node evaluation is deterministic: the same graph and the same inputs produce the same image.
- Randomness is explicit. Nodes that produce variation take seed and sample inputs rather than using mutable random state or real-world time.