API Reference / RenderPlan · interface

Interface: RenderPlan

A machine-readable render contract for a hybrid pipeline: ask an image model for a body-free background plate, then composite the computed layers (bodies, stars, the Milky Way, overlays) locally at their exact pixels. The image model supplies atmosphere; Caelus supplies the physically correct objects. This is the structured complement to prompt, which is for humans and image models.

Properties

PropertyTypeDescription
<a id="animation"></a> animation{ notes: string; pole: CelestialPole; rotationDegPerHour: number; strategy: "static" | "sequence-composite"; }How to drive an animation. static for a single frame.
animation.notesstring-
animation.poleCelestialPole-
animation.rotationDegPerHournumberSidereal rotation of the sky about the pole, degrees/hour.
animation.strategy"static" | "sequence-composite"-
<a id="background"></a> background{ constraints: string[]; height: number; prompt: string; width: number; }The body-free sky/atmosphere plate to generate with an image model.
background.constraintsstring[]Hard constraints on the plate (no bodies, horizon row, even sky).
background.heightnumber-
background.promptstring-
background.widthnumber-
<a id="layers"></a> layersRenderLayer[]Layers to composite locally; do not ask the model to draw these.
<a id="postprocess"></a> postprocessstring[]Post-processing and grading notes for the composite.

Start building

Quickstart →