toreel · AI avatars

The presenter

Three ways to put a presenter in front of the camera

Somebody has to say the lines. The question is only how much of yourself you want in it — and all three answers set up once and then film every reel without you.

An AI avatar is not a cartoon of you. It is a likeness that can be handed a script it has never seen and deliver it to camera — the same face, the same voice, the same room, across fifteen reels filmed while you were asleep.

There are three routes to one, and they differ in exactly one thing: how much of you goes in at the start.

01 · From your own footage

A few minutes to camera is enough. The likeness is trained from that take, and after it every reel is filmed without you — your face, your voice, a script you have not read aloud. This is the route most people come for, and the only one where the person on screen is you.

It costs you one recording session, once. Everything after that is typing.

02 · From a single photograph

One still of a face is enough. It is animated and lip-synced to the script — no camera, no session, nothing to set up. The result moves less than a trained likeness does, which matters more on a long piece to camera than on a thirty-second reel.

Worth it when you want a presenter tonight and cannot film today.

03 · From the catalogue

Pick a ready-made presenter and start immediately. Nothing to upload, nothing to train, nobody's consent to collect. The face is not yours, which for some subjects is the point — a channel about a topic rather than about a person does not need a person's identity attached to it.

Consent is not a formality

Cloning a real person requires that person's recorded consent — theirs, on camera, given knowingly. That is not toreel being cautious; it is the rule the avatar engines enforce, and the reason a photograph of someone else is not a shortcut to a presenter.

A catalogue avatar needs none of that, which is what it is for.

Eye contact is the part nobody mentions

A generated presenter reading a script looks like a person reading a script: the eyes track the words rather than the lens. Every take in toreel goes through a second pass that corrects the gaze to camera, and it is the difference between a clip that feels addressed to you and one that feels recorded near you.