Luxora — Lighting Design Studio

A browser-based lighting calculation workspace for room sizing, luminaire counts, 2D illuminance heatmaps, 3D room preview, design checks, and printable reports.

Included

• Zonal cavity method with CU taken from Room Index + ceiling/wall reflectance tables (3 luminaire types), LLF, target lux, power density (LPD, W/m²), and luminaire efficacy (lm/W).
• Input validation & sanity clamping — extreme or empty inputs never produce NaN/blank results.
• Room presets (office, classroom, retail, home, commercial kitchen, warehouse, clinic) that fill target lux, reflectance, CCT recommendation, and reference LPD / uniformity limits.
• Design checks tab: lux target, uniformity U₀ (Emin/Eav over the task area), spacing vs. mounting height, LPD vs. reference limit, luminaire efficacy.
• Point-by-point illuminance heatmap over the plan and uniformity estimate derived from it (not just an average).
• Interactive 2D plan with luminaire grid and heatmap; canvas 3D room preview with drag rotation.
• Exports: printable report (print CSS), PNG plan, JSON project, and DXF floor plan.
• Save/load project in the browser (localStorage).
• Optional server-side AI assistant; API keys stay on the server. The AI chat includes recent context and an offline fallback so the app stays usable with no AI configured.

Run locally

node server.js

Open http://localhost:8787.

Optional AI environment

Copy .env.example to .env and provide AI_API_KEY and AI_MODEL. The server exposes POST /api/ai as an OpenAI-compatible proxy. The frontend never receives the secret key.

The endpoint is guarded: per-IP rate limit, small request body cap, message length cap, only user/assistant roles accepted, and a server-side system prompt that keeps answers on lighting topics (override with AI_SYSTEM_PROMPT).

On Vercel, api/ai.js is deployed as a serverless function. Add AI_API_KEY, AI_MODEL, and optionally AI_API_URL under Project Settings → Environment Variables, then redeploy. Keep the variables enabled for the Production environment. (Rate limiting on Vercel is in-memory per function instance — best effort.)

The app remains fully usable without AI configuration.

Notes

Results are design estimates based on the entered assumptions and simplified photometry. The heatmap treats each luminaire as a Lambertian source; validate final designs against the applicable project standard and real photometric data.