ReSpace: Text-Driven Autoregressive 3D Indoor Scene Synthesis and Editing

ArXi:2506.02459v5 Announce Type: replace Scene synthesis and editing has emerged as a promising direction in computer graphics. Current trained approaches for 3D indoor scene generation either oversimplify object semantics through one-hot class encodings (e.g., 'chair' or 'table'), require masked diffusion for editing, ignore room boundaries, or rely on floor plan renderings that fail to capture complex layouts.