BLASt3R code released: turn raw video into poses, intrinsics and dense depthmaps
kwangmoo_yi · x · 2026-09-09
Follow-up post linking the GitHub repo (naver/blast3r) for BLASt3R, a 3D reconstruction system for unposed, uncalibrated imagery:
- Feeds in an image collection or video and jointly outputs camera poses, intrinsics, and dense per-frame depthmaps via bundle adjustment
- Offline mode handles unordered image sets; online mode amounts to incremental SLAM
- No calibration, poses, or frame ordering required
The repo ships install scripts and requires a GPU with CUDA 12 (faiss-gpu needs Python 3.11-3.13); the project is at an early stage (11 stars).
More from Research
- ValsAI Launches Tax Agent Bench: 391 Expert Questions to Test LLMs on Corporate Tax — JenniferHli · 2026-09-09
- AI Intern reproduces ICML best paper "How much do LLMs memorize" for under $8 — _akhaliq · 2026-09-09
- Buckmaster & Alpöge publish Euler (112pp), Boussinesq (76pp), porous media papers with Lean proofs — xennygrimmato_ · 2026-09-09
- Boltz Lab is hiring as partners weekly solve problems traditional methods failed — GabriCorso · 2026-09-09
- Boltz creator Gabriele Corso named to MIT TR's 2026 Innovators Under 35 as models near the clinic — GabriCorso · 2026-09-09
- Boltz models keep improving as partners solve previously failed drug-discovery challenges weekly — GabriCorso · 2026-09-09