Where the build stands

A note from inside the private beta. The match engine runs end to end on a Mac; the one-click timeline apply is the piece still in active development. Here is the honest state of it.

MOVON Match is in private beta and pre-release. No public build has shipped yet, so instead of a version number, here is the honest state of the work.

What runs today

The match engine works end to end on a Mac. Point it at a multicam timeline and it does the full chain: it samples frames from every clip, runs them through the two AI models, groups the clips by camera on its own, picks the hero, and generates a standard 33³ .cube LUT for each camera group. All of it local. Nothing uploads, no account, no connection.

That chain (analyze, cluster, match) is the hard part, and it is solid.

What is still being built

The last step is the one being wired up now: applying every generated LUT across the whole timeline in a single click, as one undo step. The match output is ready; the part that writes it back onto each clip through MOVON's own effect is in active development.

So if you are in the beta today, the engine matches your cameras. The one-click apply is the build we are finishing.

On speed

Honest answer: the inference path is not fast yet. The models run, the output is correct, but the timing is not where it needs to be. The target is a full multicam timeline matched in about a minute on an M-series Mac, and getting there is a dedicated optimization pass, not a number we can claim today.

When the numbers are real and measured, they go here. Not before.

What's next

Auto / Manual mode, per-clip match confidence, then the one-click apply landing for real. The roadmap tracks the order. No dates promised. Just the order things land in.

— The MOVON Labs team