Data source: none external. Everything on screen is computed live, in your browser, from the program you enter — space names, areas, floors, departments and the adjacencies you draw. Nothing is uploaded; there is no database, API, or outside dataset. This is a design-thinking instrument, not a measured or code-checked result.
Forces (v2) — a damped force-directed layout. Related shapes attract, every pair repels within a spacing margin so they keep apart, Avoid pairs get an extra push, and a boundary force plus a hard per-tick clamp keep every shape inside the site. Positions integrate with velocity damping until motion falls below a threshold, then it rests. Shapes are rigid and a shape's drawn size is its true area.
Soft-pack (v1) — a Kangaroo-style overdamped soft-body relaxation (position-based dynamics). Each space is an 18-point ring pulled by three self-consistent goals — edge length, true area and shape — plus hard non-overlap (vertex-in-polygon push-out) and the site boundary, which never fade, so blobs pack against each other with clean edges. The area goal is (near-)conserving: where a space is squeezed by its neighbours it bulges into free space to keep its true ft² rather than shrink — so a crowded plan reads like packed soap-bubble cells, each still sized to its area. While arranging, related blobs pull toward a shared edge and Avoid pushes apart; those arranging goals fade once the layout is placed, so sliders and dragging re-settle it in place instead of reshuffling. The result is an area-proportional packed "bubble" diagram.
0.82). FAR = gross ÷ site area. Built volume = Σ (area × height).The physics is a stylized packing model for exploring program & adjacency — not an architectural layout, and not a structural, daylight, energy, or code analysis. Bubbles show relative size and wanted relationships; they don't resolve circulation, structure, or egress. Use it to think, then draw for real.
Müller, Heidelberger, Teschner & Gross, Meshless Deformations Based on Shape Matching (SIGGRAPH 2005) — the shape-matching goal. Bender, Müller & Macklin, Position-Based Simulation Methods (2015) — the PBD constraint solver. D. Piker, Kangaroo (2013) — the projective particle-and-goals model this engine echoes. Fruchterman & Reingold, Graph Drawing by Force-Directed Placement (1991) — the Forces engine's lineage.