
Open an episode in a project that has the world model, select a segment that has not been generated, and look at the preview on the right. Or drag the demo in Try it here below.
While the segment generates, the progress shrinks to a small pill at the bottom, just above the timeline, so the set has the room. If the segment failed, its card sits at the bottom with the retry button and the error, as before.
The figures are deliberately simple. They show where people are and which way they face; how each person looks comes from their character image, not from the figure.
| To do this | On a computer | On a phone or tablet |
|---|---|---|
| Look around | Drag with the mouse; let go and it glides to a stop | Drag with one finger |
| Look by moving the device | – | Tap Tilt to look, then turn the phone. iPhone asks for motion access first |
| Take a small step (sets that allow it) | W A S D or the arrow keys, or Shift + drag | Drag with two fingers |
| Pause on one moment | Hold and drag the timeline cursor | Hold and drag the timeline cursor |
Leave it alone and the view turns slowly by itself. It stops turning while the shots are playing through or while tilt is on. Tilt to look only appears on touch devices with motion sensors, and it works from the way you were holding the phone when you tapped it. Steps are limited to about half a metre from where the set was captured, so the picture stays clean.
This is the demo from the SceneMixer homepage: one set, four characters, four cameras. Your own segments show your own set this way while they render; in the editor the people are drawn as simple coloured figures and the cameras as numbered wedges.
The preview is read-only: you cannot drag a figure to move it. Positions come from the storyboard text, so that is where you fix them. Say who sits where, who faces whom, and what they stand next to, then generate again once the current run has finished. Each generation is charged as usual; the earlier take stays in Version History.
How to write positions so they come out right: Write positions the world can read. To change just one shot: Edit one shot and regenerate.
| What you see | What it means |
|---|---|
| The room, no people | The set exists; the characters have not been placed yet |
| Figures, no cameras | People are placed; the shots are not planned yet |
| Figures, numbered wedges and a moving timeline | Everything is planned; the video is being generated from these cameras |
| The view turns and fades into a video | The clip just finished |
| The usual card, no set | Finished, a sample, a project without the world model, or the set did not load |
No. The preview is free and read-only. You pay for generating the segment, as always.
No. Positions come from the storyboard text. Change the text, then generate the segment again.
The characters have not been placed yet. People only appear once their positions are worked out; they are never guessed in the meantime.
A finished segment shows its video. The set is shown only while a segment is not generated yet, generating, or failed.
It only appears on phones and tablets with motion sensors. On an iPhone, allow motion access when asked; if you declined, reload the page and tap it again.
It shows the plan the video is generated from: the room, where everyone stands and where each camera is. The video adds the real faces, light and movement, so small differences are normal.
The set is captured from one point. Steps are kept to about half a metre from it so the picture does not stretch.
Projects created before the world model launched keep the earlier staging, and there is no switch. Start a new project on a paid plan to get it.
Start a project, generate a segment, and drag around the room while the shot renders.
Updated October 15, 2026
Pricing·Novel to Video·Guides·Samples·Alternatives·Answers·Support·Privacy·Terms·Legal·Contact·Affiliate program·© 2026 SceneMixer