Scene representation
Keep spatial appearance and scene structure closer together instead of splitting them into unrelated layers.
Product
The world model direction extends scene representation toward simulation and runtime behavior, aiming to connect appearance, structure, and physical context inside one spatial stack.
Keep spatial appearance and scene structure closer together instead of splitting them into unrelated layers.
Support research paths where scene assets need to reflect more than geometry and visual output alone.
Frame spatial assets for workflows that may later require interaction, training, or simulation use.
This is a research-oriented product direction and is best understood through workflow discussion rather than a fixed feature checklist.