Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Fast 3D Foundation Model Initialized Gaussian Splatting

About

This paper introduces a fast method for high-quality 3D Gaussian Splatting (3DGS) reconstruction without traditional Structure-from-Motion (SfM). The proposed approach leverages 3D Foundation Models (3DFMs) for camera pose and point-cloud initialization, then jointly optimizes both camera poses and Gaussian primitives using a depth-guided loss function. This enables fast convergence even from rough initialization with as few as 50-60 input views. To further improve reconstruction quality in sparse-view scenarios, an MLP-based pose refinement module is introduced alongside depth-guided supervision from the foundation model. Extensive experiments on Mip-NeRF 360, Tanks and Temples, and RobustNeRF demonstrate that the proposed method achieves competitive reconstruction quality (23.61 dB PSNR, 0.19 LPIPS) while reducing training time to approximately three minutes per scene. The proposed method produces ready-to-use 3DGS models at a fraction of the time required by existing pipelines, making it suitable for near real-time applications in robotics, VR, and autonomous navigation.

Anurag Dalal, Daniel Hagen, Kjell G. Robbersmyr, Kristian Muri Knausg{\aa}rd• 2026

Related benchmarks

TaskDatasetResultRank
Novel View SynthesisMip-NeRF 360 garden
SSIM0.6122
18
Novel View SynthesisMip-NeRF360 (room)
PSNR28.48
16
Novel View SynthesisMip-NeRF360 counter
PSNR26.16
11
Novel View SynthesisMip-NeRF 360 bicycle
PSNR24.22
4
Camera pose registrationMip-NeRF 360 bicycle
Rotation Error (°)0.4078
4
Camera pose registrationMip-NeRF 360 garden
Rotation Error0.3844
4
Camera pose registrationMip-NeRF 360 counter
Rotation Error (°)0.6045
4
Camera pose registrationMip-NeRF 360 room
Rotation Error (°)0.6652
3
Showing 8 of 8 rows

Other info

Follow for update