Features
Every feature, organized by what it does for the work.
No marketing fluff. Each line below maps to something you can use today (or, where noted, on a near-term roadmap).
Generation
Reference-driven workflows on top of best-in-class models.
Reference images, first-class
Drop one or many. Per-image labels propagate to every generated output, so subjects stay searchable across thousands of frames.
Multiple models, one workflow
Switch Seedream variants per prompt without losing your reference set. Per-generation model id is recorded in metadata.
Image-to-video via Seedance
Send any still through to Seedance for HD clips. Resume-safe queues never re-bill in-flight tasks after a service restart.
Live SSE streaming
Watch results land in real time. Server-Sent Events fan out to every open tab and device, filtered to your account only.
Curation at scale
Built for the moment your gallery has tens of thousands of images.
Two-tier labels
People and things. Color-coded, propagated from inputs, sortable, filterable, and revocable per image.
Favorites with rich rating
Score 1–N, curation score for editorial picks, instant keyboard navigation. Auto-unfavorite at zero so noise self-cleans.
Prompt grouping & duplicate detection
Group every image by prompt to compare runs side-by-side. Hash-based duplicate detection catches near-identical outputs.
Rankings view
Sort the entire gallery by rating, curation score, or duration. Find the best of every batch in seconds.
Workspaces & ownership
One account, many isolated projects.
Multi-workspace projects
Spin up a workspace per client, pitch, or experiment. Each has its own gallery, prompts, labels, and S3 prefix.
Workspace switcher, one click
Switch via the user menu. Every grid, prompt list, and queue stream re-binds to the active workspace instantly.
Hard data isolation
Per-workspace database rows, per-workspace S3 prefix, HMAC-signed identity on every internal request. Zero cross-tenant leak.
Bring-your-own object storage
Self-host with MinIO or point at AWS S3. Object keys are stable; you own the data on disk.
Account control
Knobs you actually want to turn.
Hard spend caps
Per-account daily and monthly caps in integer cents. Enforced before every Submit; never overshoots, ever.
Persistent job queue
Backed by Postgres FOR UPDATE SKIP LOCKED. Restart-safe, multi-replica safe, never silently drops a job.
API keys (coming soon)
Bearer-token API keys with scoped permissions for headless workflows and CI pipelines.
GDPR-friendly account self-service
Profile edit, full data export as a zip, account deletion with a 72-hour grace period.