📸 → 🧊 What happens when you point an open-source 3D model at a single photo? I spent time exploring open-source photo-to-3D generation — and ended up building Dioramic to put what I learned into something usable. How it works: 🔹 Upload a photo of an object 🔹 Hunyuan3D-2.1 (open-source) runs on a rented GPU — background removal → shape generation → texture painting 🔹 Get back a textured .glb mesh in about 4–5 minutes — view it in-browser and download it Why open source mattered here: Most "photo-to-3D" tools are closed, paywalled black boxes. Exploring an open model meant I could actually see the pipeline end to end — and Dioramic keeps that transparency: the Next.js frontend, the three.js viewer, and the GPU worker are all in the repo, so anyone can read it or self-host it. Tech stack: Next.js 16 · three.js · Clerk · Neon Postgres · Vercel Blob · Modal · Hunyuan3D-2.1 Every account gets 5 free generations — no credit card required. 🔗 Try it live: https://dioramic.vercel.app/ 💻 Source code: https://github.com/Hitesh-s0lank... If you're also exploring open-source 3D/generative AI models, or work with 3D pipelines, product visualization, or game assets — I'd love to compare notes. Feedback and stars on the repo welcome. #3DModeling #GenerativeAI #OpenSource #NextJS #ThreeJS #AI #WebDevelopment #ComputerVision #Photogrammetry #BuildInPublic

dioramic — turn a photo into a 3D model

Turn any photo into a 3D model open source, start to finish.

Turn any photo into a 3D model open source, start to finish.