Library & stock
A library that finds things for you.
The library is shared across your whole team and organized by you — folders, saved searches, tags, and bulk operations. Stock from Pixabay, Pexels, and Wikimedia is built in, and everything that lands in Aperture runs the real ingest pipeline: storage, proxies, catalog, transcript.
Collections
Folders for work. Smart collections for logic.
Collections are folders — drag assets in, share tenant-wide. Smart collections are saved searches that update themselves: "needs review", "interviews tagged product", "duplicates". Build once, they stay current forever.
Save any search
Filter by kind, tag, status, or transcript hit — then save it as a live collection.
Kind-aware views
Video, audio, image, and document tabs keep the grid honest.
Shared, not siloed
The library is tenant-wide — one truth, per-collection permissions on top (Team plan).
Tags, bulk operations, and a library that cleans itself.
Metadata is the difference between a library and a dumping ground.
Shift-select anything
Multi-select across the grid, then act on all of it at once.
Bulk tag, collect, delete
Tag, favorite, add to collections, or remove — one action, hundred assets.
Auto-tagging on ingest
The AI writes tags and descriptions for new uploads. Your library organizes itself while you sleep.
Duplicate surfacing
Every file is fingerprinted at ingest (sha256). The duplicates view groups them; merge in a click.
Stock, built in
Search three providers without a tab switch.
Pixabay, Pexels, and Wikimedia Commons — searchable from the library panel, filterable by provider and kind. Import is not a download: it's a real ingest into your library, with proxies, catalog, and transcript if there's audio.
Provider filter chips
Compare results across providers in one grid. Pick the shot, not the site.
Admin-managed keys
API keys live server-side in settings — never shipped to the browser.
Stock audio too
Wikimedia audio search feeds the music lane; auto-duck keeps it under the voice.
Ingest that happens while you're not looking.
Point a watch folder at an S3 prefix. Files that appear there flow through the full ingest pipeline — storage, proxies, transcripts, auto-tags — and show up in the library as notifications. Uploads from the browser are resumable and chunked, so a dropped connection never costs you a take.
- Watch folders: drop files into an S3 prefix, they appear processed every 60 seconds
- Proxy ladder on every ingest — 720p + 540p editorial proxies and a scrub rail; playback never touches the master
- Resumable growing uploads — chunked sessions via API survive flaky connections; finalize when ready
- sha256 duplicate detection with merge — the same footage never piles up twice
- Version stacks with a side-by-side compare player
- Everything searchable the moment it lands — transcripts included, semantic search across the whole library
- Windowed list rendering: 10 000 assets scroll as fast as 10 — the library never turns into a loading screen
- Per-user storage quotas enforced at upload (admins set the gigabytes, ingest does the arithmetic) with real usage per account
- Recorder takes arrive as layers — screen lane + camera lane, re-layoutable, never a flat file
- Templates + brand kit: new projects seed lanes and canvas in your colors (Blank / Podcast 9:16 / Social ad 1:1 / Explainer 16:9)
- ⌘K palette jumps to any sequence or asset from anywhere
A library that works
while you create.
Collections, smart views, and stock on every plan. No credit card.