/ Video tools / Highlights Extractorđź”’ On-device editorTools
VIDEO CREATOR STUDIO
Video Highlights Extractor
Discover visually active moments, fine-tune every cut, and create a shareable compilation. No uploads, accounts or paid APIs.
Standard scanning uses scene and motion changes. Optional local vision AI can match described visual subjects, but neither method understands dialogue or subjective importance.
Source & preview
No video loaded
Ready
Import a video to start.
Highlight editor
Review each candidate, adjust in/out points, reorder, or add your own moments. The graph shows relative visual activity, not semantic importance.
Suggested highlights appear here after analysis. You can also add them manually.
How it works
01 · Import
Load your video locally. The file stays in your browser.
02 · Discover
Scan sampled frames for scene and motion variation. Preview and choose moments yourself.
03 · Refine
Adjust every clip’s start/end, ordering, and title, then export a WebM compilation or CSV list.
What is and isn’t supported
Great for action, sports, screen recordings, and visually changing footage. Dialogue significance, emotional moments, and audio peaks are not detected. The scan now explains scene-change and motion evidence, and you can keep your existing cuts when rescanning. WebM export can include the source soundtrack when audio capture is supported by the browser. For frame-accurate editing, soundtrack preservation, or MP4 output, use a dedicated video editor.
Exporting your compilation
Rendering video…
Video Highlights Extractor could not start
A required JavaScript file failed to initialize. The source file and your edits have not been cleared by this warning.
Using Video Highlights Extractor
Create an editable highlight reel from sampled visual activity or optional local image/text matching. Trim and reorder candidate clips, choose an aspect ratio and crop focus, then export a browser-recorded compilation or edit list.
Load a video and scan for visual changes, or explicitly set up the optional CLIP model and enter short descriptions of the moments you want to find.
Review each suggested range, set precise in/out times, reorder or add clips, and preview the final aspect ratio and static crop focus.
Export a WebM compilation, CSV edit list or JSON project. Include source audio only when the browser’s recording support makes it available.
Example: For a product demo, ask the optional model to match “hands assembling the device” and “finished product close-up,” review each proposed frame, trim the clips and build a 9:16 reel under 45 seconds.
Limits and privacy: Source video is capped at 1 GiB and two hours; the editor supports up to 40 clips and 180 seconds of compiled output. Visual scores are relative suggestions, not editorial judgments. CLIP compares sampled still frames with your text and does not analyze dialogue; its runtime and weights download from external hosts after setup. Crop focus is static, and WebM capture/audio support varies by browser.
Frequently asked questions
What does the optional AI use to find moments?
The local CLIP model matches sampled video frames against short descriptions that you provide. It does not listen to dialogue or decide what is important, so review and edit every suggestion.
Does the tool automatically make a finished highlight reel?
No. Scanning creates candidate ranges. You choose the in/out points, order and crop, then start the compilation export or save an edit list.
