i3dgs: Official Implementation of SIGGRAPH'26 Paper on Immediate 3D Gaussian Splatting
rsasaki0109 · x · 2026-08-15
This repository is the official implementation of the SIGGRAPH'26 paper "Immediate 3D Gaussian Splat Reconstruction of Unordered Input with Global Consistency". It proposes a 3D Gaussian Splatting reconstruction method with immediate feedback capable of handling unordered image captures and very large scenes.
Key Features & Methodology:
- Utilizes fast place-recognition-driven matching for rapid processing.
- Employs cluster-based loop closure with graph-propagated correction for global consistency.
- Constructs a progressive Gaussian hierarchy.
- Performance: Achieves robust and scalable reconstruction in a fraction of the time required for offline processing.
More from Research
- Survey: 63% of US physicians use AI, 75% report reduced admin burden — iScienceLuvr · 2026-08-15
- Study Finds Vague Prompts Cause Coding Agents to Waste 7x Compute — rohanpaul_ai · 2026-08-15
- SPAR Project: Measuring the Verbalization Gap in Chain-of-Thought Reasoning — prajdabre · 2026-08-15
- AI Summer School: AI for Astronomy Lecture Recording — jamesmurdza · 2026-08-15
- MASt3R-SLAM: Real-time dense SLAM integrating 3D reconstruction priors — tom_doerr · 2026-08-15
- New Benchmark ActiveVision: GPT-5.5 Scores 10.6%, Humans 96.1% — jiqizhixin · 2026-08-15