How to Separate Scenes from a VHS Recording

When you digitize a VHS tape, you get one long file — but that file almost certainly contains dozens of separate recordings. Every time someone pressed stop and then record again, a new scene began. A two-hour tape might have thirty distinct events on it: birthdays, holidays, weekend trips, random moments in between. They all run together in a single unbroken file.

Separating those scenes into individual clips is the step that makes your digitized footage actually usable. Here's how it works and how to do it without watching every second of footage manually.

Why VHS recordings merge separate scenes into one file

VHS tapes don't store chapters or metadata — they're a continuous strip of magnetic signal. When someone stopped and restarted recording, the camera simply picked up where the tape left off. There's no marker in the recording to say "a new scene starts here." From the tape's perspective, it's all one recording.

When you digitize the tape, the capture software faithfully reproduces that continuous signal as a single video file. So a tape that was recorded across thirty separate occasions becomes a single MP4 with no internal structure — just raw footage from start to finish.

What a "scene boundary" looks like

When a camcorder stopped and restarted, it left a characteristic signature in the recording: a brief flash of black or static, a sudden change in scene content, and often a small glitch in the video signal. These are the scene boundaries — the moments that separate one recording from the next.

Some tapes also have fade-outs, title cards, or deliberate transitions between scenes. Others cut abruptly with nothing but a frame of noise. The pattern depends on the camera and how the person using it operated it.

The manual approach — and why it takes so long

You can separate VHS scenes manually using any video editor. The process is: watch the footage, find each transition point, mark an in and out point, and export the segment as its own file. Repeat for every scene on the tape.

For a tape with thirty scenes, that means watching at least some of each one, making sixty precise edit points, and exporting thirty files. On a two-hour tape, the viewing time alone approaches the length of the tape — you can't scrub too fast or you miss transitions. For a box of ten tapes, the manual approach is measured in days.

How automatic scene detection works

Scene detection software analyzes the video frame by frame, looking for the visual signatures of a stop-start boundary: a sudden drop to black, a large change in pixel content between consecutive frames, or a shift in the overall color and brightness profile of the image. When it finds a candidate boundary, it marks it as a potential scene cut.

VHS tape introduces extra complexity because the signal degrades over time — noise, static, and tracking errors can look like scene boundaries to a naive detector. VHS Scene Detector handles this by running analysis entirely in your browser using the WebCodecs API, then using Gemini to verify each candidate cut before it's confirmed. The result is a list of real scene boundaries with very few false positives.

How to separate scenes using VHS Scene Detector

1. Get your digitized file. You need the tape as an MP4 or MOV file first. This comes from a capture card setup or a digitization service like Legacybox.
2. Drop it into VHS Scene Detector. Open the app in Chrome or Edge and drag your file onto the upload area. No account required to start, and nothing is uploaded to a server — analysis runs locally in your browser.
3. Let it detect the boundaries. The detector analyses every frame and identifies where each scene begins and ends. On a two-hour tape this typically takes a few minutes.
4. Review and adjust the cuts. You'll see a list of detected scenes with thumbnails and timestamps. Drag boundaries to fine-tune any that landed slightly off, and delete any false positives.
5. Label each scene. Name each clip while you're looking at it — "Christmas 1988", "Dad's birthday", "Lake trip". This is the most valuable step and takes a few seconds per clip.
6. Export. Download each scene as a separate MP4, or send them directly to Google Photos with the correct original recording date embedded in the metadata.

What you get at the end

Instead of one two-hour file you can't navigate, you end up with a folder of short, labeled clips — one per event. Each clip has a meaningful name, a thumbnail that represents its content, and (if exported to Google Photos) a date that places it correctly in your family timeline. The footage is now browsable, shareable, and findable.

This is the step that turns a digitization project from "the footage exists somewhere" into "I can find the Christmas 1991 clip in thirty seconds and share it with my sister." The rest of the post-digitization workflow — culling, archiving, backing up — builds on top of this step.