RK, Founder — Ritn3D
I run Ritn3D. I built the wall-detection AI, the end-to-end 3D rendering pipeline, and the native iOS, Android, and web clients. Every release ships from my machine; every support email is answered by me.
What I actually do
- Built the wall-detection pipeline. Trained the neural network that identifies walls, doors, windows, and room boundaries from a 2D floor plan (PDF, JPG, PNG). The benchmark numbers cited across ritn3d.com — 90–95% first-pass accuracy on CAD PDFs, ~80% on scanned blueprints, 75–85% on phone-camera photos — come from a June 2026 internal benchmark set I built and score against ground truth.
- Built the 3D rendering pipeline. Written in Blender with a custom procedural furnishing layer. Outputs interactive glTF for the viewer and manifold STL for 3D printing (Pro+).
- Ship the clients. Native iOS (Swift + Flutter), native Android (Flutter), web app (Next.js). One account, three surfaces.
- Answer support. Every email that comes into Ritn3D support is answered by me, personally. That's where most of the honest “what doesn't work” language on the site comes from — real bug reports from real users, not marketing.
Why I write these posts
Most of the AI floor-plan-to-3D writing on the internet is produced by content marketers who have never actually converted a plan. I built the AI, I run the benchmarks, and I ship the product — so when a post on this site says “90–95% accuracy on CAD PDFs” or “photo-realistic listing renders are the universal weak spot,” that's data I have first-hand, not something I read on a competitor's landing page.
Where Ritn3D genuinely loses to a competitor on a dimension (photorealism vs Matterport, design-from-scratch vs Planner 5D, full CAD control vs SketchUp) I say so. It's the only way this kind of writing is worth anything.
Contact
Support email is via the app's Support button or the website contact form. Both reach me directly.