Semantic video search
When a video library grows, tags like b-roll, interview, product, or reference become too broad. ShotMind adds a local semantic search layer so analyzed shots can be found by what is visible, how the camera moves, and what the shot feels like.
01
Manual tags are useful, but they often stop at broad categories. A tag can say product or city, while the actual need is a close-up hand picking up a product, or a rainy neon street with a slow push-in.
02
Semantic search looks for meaning in the analyzed shot description. In ShotMind, selected shot clips are analyzed, then descriptions, thumbnails, and a local searchable index are saved back to the local library on your Mac.
03
The searchable layer is meant to sit beside your existing folders. After analysis, you can browse and search already-analyzed shots locally without turning your private video library into a cloud media repository.
04
Semantic search is a retrieval layer. It helps you find likely shots faster, but important matches still deserve human review before client work, licensing decisions, or final delivery.
"rainy neon street, lone person walking, slow push-in, wet reflections"
Find a mood and camera move even if no one tagged it that way.
"quiet office detail shot, warm window light, no people, slow lateral move"
Recover a reusable cutaway from old project footage.
"close-up hand placing a device on desk, clean highlight, shallow depth of field"
Search by action and framing rather than by shoot folder.
It means searching analyzed video shots by visual meaning, such as subject, action, scene, light, mood, framing, or movement, while the searchable index lives in your local ShotMind library.
They work together. AI descriptions help create the searchable layer; semantic search is how you later find matching shots with natural-language queries.
No. Full source videos stay on your computer. Only selected shot clips are temporarily sent to cloud AI when you choose to analyze them, and the results return to your local library.