Frame extraction
How to extract frames from a video in your browser
FrameDigest turns moments from a local video into timestamped JPEG evidence without uploading the selected clip for extraction. It preserves chronological Full Evidence separately from its compact adaptive handoff.
Workflow
Extract frames without sending the source away
- 01Choose the source
Select a supported local video up to the current 180-second / 32 MiB envelope.
- 02Process locally
The browser decodes the file, captures 1 FPS Full Evidence, and scans visible changes at higher frequency.
- 03Review and export
Keep representative states, add exact manual moments, then download the output that matches your task.
Two extraction views
Chronological coverage and adaptive visual states are separate
Full Evidence
Full Evidence
A predictable one-frame-per-second chronological archive. A 3-minute source produces about 180 timestamped Full Evidence frames.
Adaptive visual handoff
Adaptive visual handoff
A separate higher-frequency scan looks for visible UI, text, layout, scene, and other state changes so the compact handoff can reduce redundant automatic frames.
What to expect
The current browser-local extraction contract
- Neither output is a lossless reconstruction of every source frame.
- Timestamp-aware JPEG filenames and frames.csv preserve chronological order.
- Selected video bytes and generated media are not uploaded to FrameDigest application servers for extraction or ZIP generation.
- Transcription is deferred; visual extraction does not fabricate speech or transcript files.
FAQ
Frame-extraction questions
Does FrameDigest extract every frame from the source video?
No. Full Evidence preserves one frame per second, while the compact handoff separately scans for representative visible changes.
Why use 1 FPS Full Evidence?
The fixed rate creates a predictable chronological archive while keeping local resource use bounded at the current production envelope.
Can I preserve a specific moment manually?
Yes. Manually selected moments are retained in the AI Handoff rather than silently dropped by automatic compaction.
Is the video uploaded before frames are extracted?
No. Frame extraction and package generation run locally in the browser.
Ready when you are