Product visualizer
Shoppers see your products at true scale in a photo of their own room, with no app to download. A 3D product visualizer for Shopify. Book a demo.
World model
A world model is an AI system's internal model of how an environment works, which it uses to predict what happens next, often after an action, so it can plan before acting.
A world model is an AI system's internal model of how an environment works, used to predict what happens next so the system can plan. The term covers model-based reinforcement learning, video and world simulators, and spatial 3D world generators. Imersian is not a world model: it uses spatial AI to read a shopper's real room from one photo and place products at true scale.
This is the short version. For examples, sources and the full background, read What is a world model?
Three meanings of world model in AI today
1. Model-based reinforcement learning
The modern use comes from David Ha and Jürgen Schmidhuber's 2018 paper World Models, in which an agent learned a compressed model of a game and was trained inside that model (arXiv:1803.10122). DreamerV3 learns a model of its environment and improves its behaviour by imagining possible futures (Hafner et al., 2023). Meta's V-JEPA 2 applies the idea to robots, planning actions by predicting their result (Meta AI).
2. Video and world simulators
Large video and physics models that simulate the world, often to make training data for robots and self-driving systems. NVIDIA's Cosmos platform offers world foundation models for physical AI (arXiv:2501.03575). Google DeepMind's Genie 3 generates interactive environments in real time that you can move through (Google DeepMind).
3. Spatial and 3D world models
Models that create explorable 3D spaces. World Labs' Marble builds 3D worlds from text, images, video or a rough 3D layout (World Labs).
World model vs large language model
A large language model (LLM) predicts the next token of text. A world model predicts the next state of an environment. They overlap, since text describes the world, but they are trained on different data for different jobs.
Why it matters for home retail
A shopper buying a sofa or a rug needs to know whether it fits their real room. A generated world is designed to look plausible, not to measure anything, so it can't answer that. What retail needs today is perception: reading the real room accurately and placing the real product in it at true scale.
Is Imersian a world model?
No. Imersian uses spatial AI to understand a shopper's room from one photo, its floor, walls, depth and scale, explained under room reconstruction. It then places the real product at true scale. It doesn't predict how the room changes over time and doesn't invent a new room.
Frequently asked questions
Where does the term world model come from?
The idea of a mind carrying a small model of reality is older than AI, but the machine-learning use comes from Ha and Schmidhuber's 2018 paper World Models.
Is a world model the same as spatial AI?
No. Spatial AI perceives a space: where the surfaces are and how big things are. A world model also predicts how the space changes, often after an action.
Can a world model show furniture in my room?
Not reliably today. World generators create plausible scenes rather than measuring a real one. Room visualizers start from the shopper's own photo so size and placement are real.
Key Takeaways
- A world model predicts the next state of an environment; an LLM predicts the next token of text.
- The machine-learning use comes from Ha and Schmidhuber's 2018 paper World Models.
- Three meanings: model-based reinforcement learning, video and world simulators, and spatial 3D world models.
- Retail needs perception of the real room, not a generated one.
- Imersian is not a world model; it uses spatial AI to understand a shopper's room from one photo.
Related terms
Where this shows up in the product
What Is a World Model? A Clear Definition, With Examples
A world model is AI that learns how an environment works so it can predict what happens next. The three meanings of the term, named examples from the labs, and where spatial AI in retail fits.
3D model from photo
Make a 3D model from a product photo, with no scanning or 3D team. Rug models are free; new accounts get free credits to try furniture. Works with Shopify.