Collections

Collection settings

Open a collection's settings to rename it, control who can see it, tune the processing pipeline, manage its images, and roll back to an earlier version. Each area lives on its own tab.

Last updated 2026-05-29

Reach collection settings from the collection card’s menu, or from the gear/settings control in the viewer. Settings are organized into tabs.

The collection settings tab bar: General, Tags, Sharing, Hero, Data, Processing, Actions, Versions
The eight settings tabs.

General

Name and describe the collection, and set its attribution — the source label, URL, logo, licence, and any required statement shown to viewers. This tab also shows the created / updated timestamps and the collection ID.

Tags

Collection-level tags used for discovery and organization (distinct from per-image tags, which you add in the Inspector). Tags help collections surface in browse and search.

Sharing

Control visibility — private, unlisted, or public — set a custom slug for a friendly URL, and (for curators) feature the collection on the public home page. Public collections show a page-view count here. Full detail in Sharing & publishing.

Hero

Pick the hero image shown on cards and the public detail page. Choose a frame from within the collection or upload a custom image.

Data

Add images, replace the metadata sheet, and run per-image actions (replace, rotate, flip, remove) on upload-based collections — or re-import a catalogue-sourced collection. This is the home for any change to the collection’s content; it has its own full guide: Data tab.

Processing

Toggle the optional pipeline features (Parquet, KTX2, CLIP, UMAP, TDA, visual features, GEMMA) and re-run them. Turning a feature on stamps the collection for reprocessing; the new columns/artifacts appear when the run finishes. Each step — what it produces and when to enable it — is covered in Processing & analysis.

Actions

Operational controls: cancel a running pipeline, move the collection to another workspace, re-process from scratch, download the raw data, and delete the collection. Deletion is recoverable from the workspace Trash for a window — see Workspaces.

Versions

Each processing run is a version. This tab lists the history and lets you roll back to a previous version if a re-process or data change went wrong.

See also