Guide

How MatchCut works

A match cut works because the eye locks onto something that does not move. In film, that is a shape or a face held across two shots. In MatchCut, it is your highlighted word — pinned to the center of the frame while everything else cuts around it.

The pipeline

1. You type a phrase. Any line up to 80 characters — a headline, lyric, joke setup, or caption.
2. You tap the focal word. Word chips appear under the input; the selected word renders with a marker highlight and stays in the same screen position across every cut.
3. MatchCut lays out variants. For each cut (4–24 per second) the engine rotates, scales, repositions and recolors the clipping card, cycles newspaper furniture, and fires a 1-frame white flash at the boundary.
4. It records what you see. The same canvas that drives the preview is captured by MediaRecorder, so preview and export are pixel-identical. MP4 where supported, WebM otherwise.

What is actually rendered

Every frame is drawn with the Canvas 2D API: a paper-grain background, an aged-newsprint clipping card, masthead rules, and heavy condensed type. The highlighted word gets a hand-drawn marker block that changes color per cut. A slow zoom push and optional blur sell the motion.

Where the file lives

Nothing is uploaded. Rendering and recording happen on your device; the finished blob is handed to your browser as a download. Close the tab mid-render and nothing was ever sent anywhere.

Sounds

Five optional sound effects — paper shuffle, camera shutter, film advance, Polaroid, flash pop — are synthesized with WebAudio oscillators and filtered noise at each cut boundary, then muxed into the exported file.