Worlds

Estimated reading: 3 minutes

Worlds are the environments where your scene takes place.
You can generate a new world or choose one from your assets.

World Generation Models

You can create a new world in two ways:

Open the Worlds panel, select My Assets, and select Create World.
elect the circular Create button at the bottom center of the Viewport.

Both options open the same world-generation window.

To generate a world:

Select a generation model from the menu at the top of the window.
Enter a description of the world in the prompt field.
Optionally upload one or more images, a panorama, or a video as a visual reference.
Select the arrow button to start generation.

The generation task appears in the Queue. A world typically takes approximately two to three minutes to generate, although processing time may vary.

When generation is complete, the world appears under My Assets in the Worlds panel, ready to be added to the scene.

Best Practices

World generation works best when the prompt or reference media clearly describes an environment with recognizable depth, structure, and lighting.

Writing a Prompt

Describe a specific location and include the details that matter most, such as:

Environment type
Architecture and layout
Lighting and time of day
Materials and colors
Mood and visual style
Important foreground and background elements

When creating a world for animated characters, include open areas for movement and clearly defined surfaces where characters can stand.

Using Images

Use sharp, well-lit images with clear depth and perspective. Images that include floors, walls, ceilings, or ground planes help define the structure of the environment.

For best results, choose images with visible foreground, middle-ground, and background elements.

Avoid:

Blurry or heavily compressed images
Extreme close-ups with little environmental context
Very dark or overexposed images
Flat graphics, logos, or text
Images with visible borders
Images focused mainly on a person, animal, or individual object

The generated world may extend beyond the visible area of the reference image to create a complete, explorable environment.

Using Multiple Images

When uploading multiple images of the same location:

Capture the environment from different viewpoints.
Include some overlap between images.
Keep the resolution and aspect ratio consistent.
Use images with similar lighting and color.
Make sure all images show the same space.

Overlapping viewpoints help the model understand how the images connect within the environment.

Using a Panorama

Use a complete 360-degree equirectangular panorama with a 2:1 aspect ratio.

For best results:

Make sure the left and right edges connect cleanly.
Include enough sky and ground.
Avoid visible stitching errors.
Use a sharp, evenly exposed image.

Panoramas provide greater control over the layout of the generated world.

Using a Video

Record a static environment in one continuous take. Move the camera slowly and steadily while capturing as much of the space as possible.

For best results:

Cover approximately 180 to 360 degrees of the environment.
Avoid sudden or shaky camera movements.
Avoid motion blur.
Do not zoom while recording.
Keep the exposure consistent.
Avoid moving people or objects.

Camera shake, moving subjects, zooms, and exposure changes may create artifacts or unclear geometry in the generated world.