ブログ

From a Single Room to an Entire Building: Exploring Spaces with Freakout AI

From a Single Room to an Entire Building: Exploring Spaces with Freakout AI

Freakout AI / Spatial stories

A small shop and an apartment building have something in common: people need to understand the space before they commit to it.

In a shop, that might mean seeing how furniture fits together. In property, it might mean understanding how an apartment sits within a building, where its balcony faces, or how the different floors connect.

Interactive 3D gives those questions a place to be explored. Freakout AI’s canvas brings that approach from individual rooms and retail spaces to larger architectural scenes—with AI assistants able to help build and edit the environment through MCP.

01 / The interior

Start with a room

Yoghurt settles into a golden armchair while the Freakout robot checks the room layout on a tablet.

A room is a useful starting point because its purpose is easy to understand.

Place the walls, arrange the furniture, adjust the lighting, and explore the result. A retailer can present products in context. An interior designer can communicate a layout. A customer can see the relationship between objects and the space around them.

Through Freakout’s Model Context Protocol (MCP) connection, an AI assistant can work with the canvas directly: inspecting the scene, placing objects, changing materials and lighting, and preparing edits for visual review.

The conversation becomes part of the design workflow. A request to rearrange a display or change the atmosphere can become a preview in the actual scene, ready to inspect and refine.

02 / The bigger picture

Scale up to the building

Yoghurt curls up inside a miniature cutaway apartment building while the Freakout robot studies a floor plan.

Block 1E at North London Business Park shows how that canvas can extend beyond a single interior.

Block 1E at North London Business Park, an unfurnished eight-storey apartment block built from the PLUS Architecture planning drawings.

The example represents an unfurnished eight-storey apartment block, comprising a ground floor and seven upper floors. Its scene description identifies the PLUS Architecture planning drawings as the source for the floors, stair and lift cores, doors, windows and balconies.

That changes the scope of the experience. The subject is a building with repeated architectural elements and relationships across multiple levels.

For property developers and real estate teams, this opens a useful way to present a project: give people a shared spatial reference alongside drawings, descriptions and still images.

03 / Built to repeat

Repetition makes scale practical

The Freakout robot stacks matching balcony modules while Yoghurt sits inside one.

Buildings contain a great deal of repetition: windows, doors, balconies, structural elements and floor details.

How a scene represents that repetition matters. The Block 1E scene contains more than 5,800 repeated geometry copies, organised through the canvas’s repeat structures.

Freakout’s canvas supports repeated geometry, shared materials, adaptive rendering and culling. Its MCP tools also let an assistant analyse scene complexity and identify potential optimisations.

These capabilities make performance part of the authoring process. As the project grows, the assistant can help examine how it is assembled and where rendering work can be reduced. Actual responsiveness still depends on the scene, viewing conditions and device.

04 / A shared perspective

A different way to communicate property

For a property team, the opportunity is to make a project easier to understand.

An interactive scene can provide context for conversations about layouts, circulation, interiors and the relationship between individual units and the wider building.

Freakout also supports named spaces and inspection tools, including cutaway, isolation and top views. These offer ways to focus attention within a larger model when those spaces have been authored into the scene.

The same canvas can therefore serve different scales of discussion: a furniture arrangement within a room, an apartment within a floor, or a building viewed as a whole.

05 / The creative loop

Build through conversation, review in the canvas

MCP connects the AI assistant to the editable scene.

Yoghurt chases the cursor while the Freakout robot previews an armchair move in a room scene.

The assistant can read the current environment, prepare changes, validate them and show a preview. People can then inspect the result and decide what to save.

For larger projects, that creates a practical collaboration loop: describe a change, see it in context, refine it, and keep building.

From a small retail space to a multi-storey property demonstration, the underlying idea remains consistent: make spaces easier to create, explore and explain.