If you build products, you already think in constraints: inputs, edges, and what breaks under load. A home layout is the same kind of system. The difference is that most people only evaluate it with taste—“does this look nice?”—instead of asking whether the geometry supports how humans actually move, rest, and cook.
This post is a practical checklist for turning photos and floor plans into spatial feedback. No classical school jargon required. Think of it as debugging a floor plan the way you would debug a UI flow.
Why photos beat memory
You live in a space, so your brain fills in gaps. A doorway that always feels “a bit tight” disappears from conscious notice. A photo from the entrance does the opposite: it freezes the first frame of the experience.
Capture three shots before you move anything:
- Entry frame — stand just inside the main door and shoot straight ahead.
- Primary rest zone — the sofa or bed from the opposite corner, so you see both the seat and the door.
- Work zone — kitchen prep line or desk, facing the direction you usually face while working.
Name the files by room and date. When you compare layouts later, you want timestamps, not vibes.
Floor plans as graphs, not drawings
Treat walls as edges and openings as nodes. Circulation is a path through that graph. The questions that matter:
- What is the shortest clear path from door to bedroom, bathroom, and kitchen?
- Does that path cross a rest zone (sofa, bed, dining table) or skim past it?
- Are there dead ends where junk accumulates because nothing “routes” through them?
On a printed or PDF plan, draw the entry-to-room lines in one color and the rest-zone footprints in another. Crossings are friction. Parallel skim-paths are usually fine. A straight shot from the front door into a bedroom door or toilet door is a high-friction pattern in almost every residential plan I review—not because of superstition, but because arrival never gets a place to land.
Checklist: five spatial tests you can run tonight
1. First landing
From the entry photo, mark where the eye stops first. Ideal landings are a console, a plant niche, or a soft turn into living space. Bad landings are a blank wall at close range, a toilet door, or a direct sightline into a messy kitchen.
Fix without renovation: shift a low piece so the walk curves; add a rug edge that hints at the turn; close an always-open door that steals the view.
2. Sightline from rest
Sit (or lie) where you normally rest. Can you notice the room entrance without twisting your neck past 45 degrees? If not, rotate the seat or bed five to fifteen degrees and retest. Tiny angles beat new furniture.
3. Visual weight balance
Solid cabinets and bookcases read as heavy. Windows, open shelves, and mirrors read as light. If all the heavy mass sits on one wall, the room feels like a listing boat. Redistribute: a tall lamp opposite a tall cabinet, art opposite a doorway cluster, a plant opposite a media unit. Matching pairs are optional; distributed attention is not.
4. Work facing
Wherever someone stands for more than fifteen minutes—stove, desk, ironing board—note what is behind them. A wall-facing cook in an open plan often feels socially cut off even when the living room is three meters away. If you cannot flip the appliance, add a island stub, a pass-through shelf, or a mirror that returns a glance toward the social zone. The goal is partial awareness, not a full 180-degree stage.
5. Clutter as a routing bug
Piles form where circulation dies. Before buying bins, ask: why does traffic skip this strip? Widen the path, pull furniture six inches off the wall, or give the corner a single intentional object. Fixed routes starve junk of hiding places.
Where AI actually helps makers
Manual markup works, but it is slow when you have ten apartments or a client sending phone photos at midnight. Computer vision and layout models can:
- Detect doors, windows, and large furniture blobs from a single room photo
- Overlay approximate walk paths on a floor-plan image
- Diff two furniture placements so you see what changed, not just what “feels better”
That is the useful definition of “AI feng shui” for builders: geometry assistance, not mysticism. You still decide what to move. The model catches misses you ignore because you live there.
Useful workflow for a side project or consulting pass:
- Ingest photo + optional plan PNG.
- Label entry, rest, and work zones (even rough boxes help).
- Score against the five tests above as binary or low/medium/high friction.
- Propose one move only—ship the smallest change, then re-photo in three days.
Version the advice like product notes: “v1 moved sofa 20 cm off the door axis; sleep quality self-report improved.” Anecdotes become data when you keep the photos.
Privacy and honesty
Room photos are intimate. If you build or use a tool that stores them, prefer ephemeral processing, clear retention, and no training on user homes without consent. Spatial analysis does not need to become a surveillance product.
Also be honest about limits. A model cannot feel a draft under the door or the way afternoon light hits a pillow. It can tell you the bed backs onto a hallway door. Humans still walk the room.
One-evening experiment for builders
Pick one room. Shoot the three photos. Sketch the circulation graph on paper. Run the five tests. Change exactly one thing. Live with it for three days, then shoot the entry frame again.
Most “this place has weird energy” stories in rentals start as awkward geometry. Fix the geometry first. The stories often quiet down on their own.
If you want a structured second pass over photos and floor plans—entry flow, seating angles, room-by-room notes—I’ve been building Juhe (居合) at https://fengshuiroom.com/, an AI home feng shui consultant aimed at that workflow. Use it as another pair of eyes, then trust what your body reports when you walk the space again.
Top comments (0)