Film · The Signal
Simulon lets filmmakers act out a scene before its characters exist
Divesh Naidoo's new Reality-Aware Rendering demonstration uses a phone performance to guide AI-generated characters inside a real environment.
Reporting updated Nov 1, 2026
From the creator
Simulon lets filmmakers act out a scene before its characters existView original post on X ↗
Simulon is exploring a new way to direct generated characters. Creators place a digital subject in their surroundings, perform the scene with an iPhone and let the system generate a finished take that responds to their movement, dialogue and environment.
WHAT HAPPENED
Filmmakers normally need a character in the scene before they can shoot around it.
Simulon's Reality-Aware Rendering reverses that sequence.
In an October 28 demonstration, Divesh Naidoo shows how a creator can position a neural asset with a phone, move through the location and perform the scene.
After recording, Simulon processes what happened and generates a new video in which the digital subject responds to the captured action.
This is not live rendering. Naidoo says processing can take several minutes, and the system relies on cloud computation.
Simulon 2.0 is available on iPhone, with Android support planned.
WHY IT MATTERS
The performer and camera operator can make physical decisions before committing to the generated result.
Instead of trying to describe every movement, a creator demonstrates the timing and blocking.
That could make AI-assisted previs and short-form production feel more like directing.
VEYR ANGLE
The intriguing idea is that a person can perform the instructions.
The prompt becomes a scene you act out.
Focused takeaway
Create
Make or select a neural asset.
Place
Position it in the real environment.
Perform
Record actions, movement and spoken direction.
Render
Let Simulon generate a scene informed by that recording.
Naidoo confirms that the process is not real-time and currently requires cloud rendering.
Source excerpt