About FrameDigest
What FrameDigest does — and what stays local
FrameDigest turns a short video or screen recording into timestamped visual evidence and compact AI-ready screenshot packs. The selected video is processed in your browser rather than uploaded to FrameDigest for extraction.
How it works
From local video to portable visual evidence
- 01Choose a video
Select an MP4, WebM, MOV, or M4V file from your device.
- 02Scan locally
Your browser decodes the video, captures timestamped frames, and measures useful visible changes.
- 03Review moments
Inspect representative visual states and add any exact moments you want to preserve manually.
- 04Export the right pack
Take a compact AI Handoff or the chronological Full Evidence archive, depending on the job.
Privacy boundary
The media work stays on your device
Your device
Media processing
- Selected video bytes and filename
- Browser-native video decoding
- Captured frames and visual-state scanning
- Manual selections
- ZIP and output generation
FrameDigest infrastructure
Only the application boundary
- Static application files needed to load FrameDigest
- Bodyless anonymous aggregate Sessions and Videos digested counter requests
Selected media, generated frames, filenames, and archives do not cross this boundary.
Scope
What FrameDigest does — and what it deliberately doesn't
FrameDigest does
- Extract timestamped visual frames.
- Preserve chronological 1 FPS Full Evidence.
- Find representative visible changes for a compact handoff.
- Preserve every moment you select manually.
- Generate visual outputs locally in the browser.
FrameDigest doesn't
- Upload your selected video for frame extraction.
- Send your frames to ChatGPT, Claude, or another AI API.
- Transcribe or invent spoken audio.
- Store projects or media in an account.
- Pretend still images reproduce every aspect of motion or sound.
Outputs
Two outputs for two different jobs
AI Handoff
AI Handoff
A manageable visual package you can take to ChatGPT, Claude, or another vision-capable tool yourself.
Compact timestamped visual sheets, every manually selected detail frame, context.md, and timeline.json inside the ZIP.
Full Evidence
Full Evidence
Predictable chronological visual coverage when you need the broader source trail.
1 FPS timestamped JPEGs, contact sheets, frames.csv, context.md, and timeline.json.
Current envelope
Clear limits are part of the product contract
- Current production limit: 180 seconds and 32 MiB.
- MP4, WebM, MOV, and M4V containers are accepted; actual codec decoding depends on the browser.
- Full Evidence is preserved automatically at 1 FPS, up to about 180 frames at the current duration limit.
- Production transcription is deferred, so FrameDigest does not create transcript files or invent speech.
FAQ
Questions people usually have before using FrameDigest
Does FrameDigest upload my video?
No. The selected media is decoded and processed locally in your browser for frame extraction, visual selection, and output generation.
Does FrameDigest send frames to ChatGPT or another AI service?
No. FrameDigest creates files you can take to a separate AI tool yourself; it does not call ChatGPT, Claude, or another AI API.
Does FrameDigest store projects or files after I leave?
No project account or hosted media store exists in the MVP. Generated outputs remain in your browser until you download or discard them.
Does FrameDigest transcribe the video's audio?
No. Production transcription is currently deferred, and FrameDigest does not fabricate transcript files when speech has not been transcribed.
Which video formats and browsers work?
FrameDigest accepts MP4, WebM, MOV, and M4V containers. Whether a particular codec decodes successfully depends on the browser and device.
What are the current video limits?
The current verified production envelope is up to 180 seconds and 32 MiB.
Ready when you are