Introduction ​

Caddis is a professional motion design tool that combines layer-based timeline editing with node-based compositing. The guiding idea is simple:

Caddis is a compositing engine with a timeline editor — not a timeline app with effects bolted on.

That means the full power of a node graph is available on every layer, and the timeline is how you animate it over time. You don't choose between the two paradigms; you use both on the same composition.

The hybrid model ​

A composition in Caddis has a timeline of layers and, behind each layer, a node graph that produces its image.

  • Timeline view — arrange layers in time, trim and slide clips, set keyframes, and scrub the playhead. This is where animation lives.
  • Node graph view — wire together sources, effects, generators, and compositing nodes to define what a layer actually renders. This is where looks live.

Both views describe the same underlying data, so a change in one is reflected in the other.

Core concepts ​

Compositions ​

A composition holds a stack of layers, a resolution, a frame rate, and a duration. Compositions can reference each other, so you can build complex scenes from reusable pieces.

Layers ​

A layer occupies a clip range on the timeline (start / end / offset) and points at a node graph. Its Output node is the compositing interface — opacity and blend mode — while spatial transforms happen inside the graph. Layers can also drive each other through track mattes and cross-layer references.

The node graph ​

Each layer's graph is a directed, acyclic network of nodes connected by edges. Evaluation is deterministic: the same inputs, time, and parameters always produce the same output, which keeps scrubbing, caching, and export predictable.

Ports & data types ​

Nodes pass typed data along their edges — not just images. Beyond the primary Image buffer there are value types (Scalar, Vec2, Color), geometry (Points, Shape), Text, and a family of Field types that effects and points can sample. See the Node Reference for the full type system.

Keyframes ​

Most parameters are animatable. Keyframes carry an interpolation mode — hold, linear, smooth, or bezier with editable tangents — and can be edited in the dope sheet or the graph editor.

What makes it different ​

  • Compositing-first — every layer is a node graph, so effects compose cleanly instead of stacking as opaque filters.
  • GPU-native — effects, blends, transforms, and fields run on the GPU via wgpu (Metal / Vulkan / D3D12).
  • Color-managed — buffers carry color-space metadata and conversions are explicit, with a linear-light working space.
  • Deterministic — (inputs, time, parameters) fully determine a frame.

Who it's for ​

Motion designers, compositors, and technical artists who want the immediacy of a timeline with the depth of a node compositor — for titles, motion graphics, generative and procedural work, screen replacements, and finishing.

Where to go next ​

© 2026 Caddis