Gaussian splat tours + AI image generation create spatial reimagining pipeline

This experimental pipeline merges 3D Gaussian splatting with generative AI to reimagine real spaces. The process starts by capturing a physical environment (e.g., a room) and reconstructing it into a Gaussian splat, a lightweight 3D representation that preserves high visual fidelity. That splat is loaded into Spatial Studio, a tool for building interactive walkthrough tours. A user navigates the scene, selects a camera pose, and sends that exact 2D frame to an image generation model (likely Stable Diffusion or similar). The AI produces an edited version of the view—changing furniture, colors, or decor—while the original 3D splat remains untouched.

The next experimental step, called 'Spatial Props,' takes the AI-generated elements (e.g., a new sofa or lighting fixture) and layers them back into the 3D tour as interactive objects, rather than keeping them as flat 2D edits. This allows potential buyers or designers to walk around a reimagined space in real time. Use cases include property staging, event planning, renovation previews, and marketing content creation. The approach avoids directly modifying the original splat, preserving the real capture while exploring what-if scenarios.