Analyse once,
search forever

Indexing

Four steps from raw footage to searchable video library

Select files or folders, and FrameQuery handles the rest. Frames and audio are extracted on your device and sent for analysis, nothing else. Your originals never leave your machine, and the extracted data is discarded the moment analysis completes.

On-Device Recognition

Find anyone by face or voice

Tag a person once and FrameQuery finds every appearance across your entire library, by what they look like and what they sound like. Both models run 100% on your machine. Your biometric data never leaves your device, and embeddings and labels are never included when you share an index.

Search

Instant, offline, free forever

Once indexed, your data is stored locally using Tantivy (a Rust port of Lucene). Search runs entirely on your machine: no internet required, no API calls, no per-query cost. Results are ranked with BM25 relevance scoring and field-based boosting across transcripts, scene descriptions, objects, and filenames.

Keyword search is only half of it. Semantic search lets you describe what you're looking for in plain language. Try “sunset over a mountain,” “person celebrating after a goal,” or “empty boardroom at night,” and FrameQuery surfaces clips that match the meaning, not just the words. It works across the visual content itself via CLIP embeddings, and across the text of your transcripts, scene descriptions, and detected objects. So “telling a story about a road trip” finds dialogue about travel, “camping gear” surfaces clips tagged with tents and backpacks, and “quiet morning light” picks up scenes described as calm and golden hour. All of it runs locally.