Reference

Settings Overview

All settings are in the settings view of the panel (gear button). This page lists them in short form.

Generation

Sample all layers (default on): The model sees everything visible in the target area. When off, it sees only the active layer.
Prompt enhance (default on): Shows the Enhance button.
Thinking level (default medium): Effort level of prompt enhancement, from none to high.
Library context (default enhance): How the text notes of a selected reference library are used. See the "Prompt & Reference Libraries" page.

Placement

Output mode (default Smart Object): Smart Object, New Layer, or New Document.
Placement (default Contain): Contain, Cover, or Fill. How results are fitted into the target area.
Mask to selection (default on): Mask new layers to the active selection.
Fit to canvas (default off): Scale results to the canvas when no target exists.
Auto Mask (default off): Object mask after generation, with separate switches for Create and Edit, a mask feather (default 2 px), and a smart edge feather. See the "Auto Mask (Beta)" page.

Color and data

Preserve color space (default on): Convert results correctly into the document color profile.
Layer provenance (default on): Store generation settings with each created layer.
History size (default 10): Length of the prompt history, from off to 100 entries.
Detailed logging (default on): Writes technical information to the console. Useful when you contact support.

Backup service

If a Google or OpenAI model does not respond, Seam can run the request through a backup service instead. By default Seam asks you first. In that dialog you can also choose to switch automatically in the future, or never.

Reset

"Reset Plugin" signs you out, restores all defaults, and clears the prompt history. Layer provenance records are kept. A confirmation dialog appears first.