How to make a 3-hour ambience video without 3 hours of rendering
Consumer video editors re-encode an entire timeline on export, so a 3-4 hour video can take hours to render. FlipLetters instead renders one short loopable segment, then stitches copies of it together with a lossless stream-copy — no re-encoding — so a multi-hour ambience video assembles in under a minute.
Split-flap and fireplace-style ambience videos are popular precisely because they're long — people put them on a TV for background motion, not a five-minute clip. But most video editors were built around short-form editing, and their export pipeline reflects that: everything on the timeline gets re-encoded frame by frame when you hit export, whether you changed anything or not. Stretch that timeline to three or four hours and export time scales right along with it.
FlipLetters' rendering engine is deterministic — the same project and the same random seed always produce the exact same video, frame for frame. That makes it possible to render a short segment once (say, five or ten minutes) and treat the rest of the video as literal copies of that segment, rather than re-simulating and re-encoding the whole thing.
The assembly step uses ffmpeg's stream-copy concatenation, which stitches video files together without touching the compressed video data at all — no re-encode, no quality loss, and it runs in roughly the time it takes to copy the bytes. A 3-hour video built from an 8-minute segment repeated ~22 times assembles in well under a minute.
For a genuinely seamless loop, the trick is authoring the project so its last slide flips back into the first slide's exact content — the board settles into the same state it started in, so the loop point is invisible. On the Studio plan, you can also mux in your own audio track (rain, a lo-fi playlist, station ambience) across the full assembled video, which is a straightforward audio-only encode pass on top of the already-assembled video.