What If You Could Film a Video Without Ever Hitting Record?

Video has long been the most resource-intensive format in content creation. From hiring talent and booking locations to post-production edits and licensing, every second of footage comes at a cost.

But Captions, a company known for automating editing workflows, is now asking a more radical question: What if the camera was optional? Their answer is Mirage—a generative video model that can create hyper-realistic videos from just an audio file or a written script. With it, creators can produce a full talking-head video—complete with expressive AI actors and dynamic backgrounds—without shooting a single frame.

How Mirage Works

The user experience is straightforward: upload an audio track or type out a script. Mirage then generates a video featuring an AI-generated person delivering that content, complete with synced lip movements, facial expressions, and body language. The actor appears against a stylized background, also generated by the model.

There’s no need for a green screen, camera, microphone, or editing suite. Everything is rendered and composited by AI.

Designed for Scale

Mirage is aimed at creators, marketers, educators, and anyone producing frequent video content on a tight budget or timeline. Since users retain full commercial rights to their videos, the tool fits seamlessly into professional workflows without the usual licensing hurdles.

It also supports over 29 languages, making it possible to generate multilingual versions of the same video with different voices or AI presenters.

What It Changes

Generative video tools have been steadily improving, but most still require pre-recorded video or a real actor as a starting point. Mirage skips that step entirely, allowing users to move from concept to full video in minutes. This removes a major barrier to entry for content creators and teams working across multiple markets.

The result isn’t just faster production—it’s a new kind of video creation, where the production process starts with the idea, not the camera.