Composer Block

What It Is

The Composer block (compositor) is a visual layout workspace inside the canvas.
You can stack layers, place 3D references, add text, and build a clean composition before turning it into a new rendered or extracted image block.

What You Can Do

Layer Image and 3D Sources

Connect Image and 3D blocks to Composer. Connected sources are ingested as layers you can arrange.

Add Text Layers

Create text overlays directly in Composer for labels, callouts, or layout notes.

Add Prompt Context For Render

Connect a Prompt block or Prompt Agent block into Composer's left-side Prompt input handle.
These prompt connections do not become layers. They are passed as additional context when you click Render.

Control Frame and Resolution

Set output aspect ratio (1:1, 4:5, 3:2, 16:9) and output scale (1K, 2K, 4K).

Create A New Image Block

Render or Extract creates an image block to the right of Composer.
Use that resulting block for downstream connections, further generation, or export.

Prompt Input Behavior

Use the Prompt input handle when you want Composer to keep the same visual layout, but give the AI extra instructions for the final render.

1. Connect a Prompt block for a reusable prompt.
2. Connect a Prompt Agent block if you want a richer text block to feed render instructions.
3. Click Render to merge the composition guide with that connected prompt context.
4. Click Extract if you want a flattened image only, with no AI reinterpretation.

Connected prompt text is appended on top of Composer's built-in render system prompt.
It affects Render only. It does not create a layer in the frame, and it does not affect Extract.

Render vs Extract

Both actions create a new image block to the right of Composer, but they do different jobs:

Render

Render sends the flattened composition guide through AI image generation.

1. Preserves the layout and framing from Composer.
2. Uses Composer's built-in render system prompt.
3. Also uses any connected Prompt or Prompt Agent context.
4. Produces a polished AI-finished image.

Extract

Extract does not use AI generation.

1. Flattens the current visible layers exactly as they are.
2. Ignores connected prompt context.
3. Keeps the current composition as a direct composite.
4. Produces a new image block you can reuse downstream or archive as-is.

Typical Workflow

1. Add a Composer block from the Utility section.
2. Connect Image and/or 3D blocks (or drag image blocks over Composer to ingest).
3. Optionally connect a Prompt or Prompt Agent block into the Prompt input handle for render instructions.
4. Arrange layers: move, resize, rotate, reorder, hide/show, lock.
5. Add text layers for notes or art direction.
6. Set aspect ratio and output scale.
7. Use Render for AI-finished output, or Extract for a clean flattened composite.
8. Use the resulting image block for downstream connections or further generation.

Use Cases

1. Build character + location composites before final rendering.
2. Create shot layouts that combine multiple references.
3. Prepare a clean visual base, then use the created render/extract block downstream.

Pro Tips

1. Keep Composer for layout decisions; do style/detail refinement in the image blocks created by Render or Extract.
2. In Focus Mode, generate fast variants first, then add the best assets to Canvas and compose.
3. Use text layers as temporary direction markers, then remove them before final output.
4. Use connected Prompt or Prompt Agent blocks when you want render-specific art direction without adding visible text into the composition.