Nodes

This page explains node types you will use on the canvas.

1. Prompt node (core)

Prompt nodes call models and produce outputs.

What you can control

• output modality: text / image / video / audio
• capability and tier (when route groups are enabled)
• parameters like aspect ratio, resolution, and duration
• batch size: x1 / x2 / x4

Inputs and outputs

• Inputs: text, reference image, reference video (depends on selected model)
• Outputs: text/image/video/audio (based on generation result)

Usage tips

• Run x1 first to validate direction, then scale to x2/x4.
• Change only a few parameters per iteration.

2. Input nodes

Input Text

Store prompts, scripts, or context. Reusable across multiple Prompt nodes.

Input Image

Upload or drag/drop images. Common for image-to-image and image-guided video.

Input Video

Upload or drag/drop videos. Common for remix, extend, and enhancement flows.

Input Audio

Upload or drag/drop audio. Common for music continuation and speech workflows.

3. Result node

Displays generation output from Prompt nodes.

• auto-adapts viewer by content type
• supports download and chaining to next steps
• can be reused as upstream input

4. Split Output node

Splits long upstream text into multiple segments.

Typical use case:

• split a storyboard script into scenes, then route each scene into downstream generation.

Split rules:

• if numbered (1. / 1)), split by numbering
• otherwise split by blank lines

You can also add, remove, and edit segments manually.

5. Global Context node

Stores project-level defaults and shared constraints.

Common use cases:

• style constraints and naming conventions
• combined with workflow settings so new Prompt nodes get consistent defaults

6. Group node

Organizes complex graphs.

• move related nodes together
• drag a child outside the group to detach
• batch generation can create groups automatically