AI models

The AI models behind Reigniter

No single model is best at everything. Auto select picks the best model for each shot: a people model for people, a reference model for your product and logo, a lip-sync model for a presenter who speaks. You can also choose any model yourself, scene by scene.

Auto select

How Reigniter picks a model for each shot.

These are the rules Auto select really applies. Every scene also has a model picker, and whatever you pick there is what renders and what you are billed for.

  1. 01

    People go to a people model

    A person, mascot or animal in shot goes to Wan: Wan 2.6 Flash on Draft and Standard, Wan 2.6 on Premium. In a product-led advert on Standard or Premium, MiniMax H3 takes the shot so your product photo rides along.

  2. 02

    Products go to a product model

    A product or object shot goes to a product model built for close-ups and reveals.

  3. 03

    Presenters get lip-sync

    A presenter who speaks goes to Wan, which moves the mouth to your voiceover. Walk-and-talk adds Kling and OmniHuman.

  4. 04

    Every shot starts as a frame

    Nano Banana 2 draws the first frame you approve, and the video model animates exactly that frame.

  5. 05

    Words go to the Motion Editor

    Lettering with nothing to copy from, and diagrams, are drawn by the Motion Editor instead of a model that would invent them.

Questions

About the models.

It reads each shot and sends it to the model that fits: people to a people model, a speaking presenter to a lip-sync model, your product and logo to a reference model, and lettering or diagrams to the Motion Editor. The model used is shown on the node once it renders.

The right model for every shot, on your brand.

Start with a brief. Reigniter plans the shots, routes each one, and applies your brand kit at render time.

Make an advert See plans