Real-Time 3D Point Cloud Reconstruction on Mobile Phones
MonaJalal_ · x · 2026-07-20
This post introduces an open-source 3D scanning model that uses standard phones or videos to reconstruct rooms or caves into 3D point clouds in real-time while moving, without relying on LiDAR or depth sensors.
It highlights that the model runs at around 20 fps on a single GPU, works on raw indoor and outdoor videos, and showed no obvious drift or collapse during a 13-minute, 25,000-frame walkthrough. The author also turned the solution into a fully reproducible project and even created a low-cost alternative using a Raspberry Pi + rotating LiDAR.
Related event: SLAM3R: Real-Time 3D Scene Reconstruction from Monocular Video(4 posts)→
More from Embodied
- Skyfall AI's SKY-001 world model turns sketches into CAD builds in 3.9s, vs 199s for GPT-6 Astra — rohanpaul_ai · 2026-10-27
- Tesla FSD approved in Latvia, the 10th European country; Musk praises smaller nations for moving fastest — elonmusk · 2026-10-27
- WanGP v18.81 adds Tencent's Hy-Nomas world model, playable on 8GB VRAM — cocktailpeanut · 2026-10-27
- NVIDIA open-sources FLORA, a robotics data engine ~125x cheaper than GPT-6 Astra — nvidia · 2026-10-27
- Embodied AI startups need a different operational discipline, investor says — ___Mufasaa · 2026-10-27
- π0.5 study: raising demos from 10 to 438 lifts Place Ball in Bowl success from 7% to 83% — stingning · 2026-10-27