Meida Chen

[Yue Wang @yuewang314]

🎉 Thrilled to share that our paper “Deformable Beta Splatting” (led by the most awesome students Dylan Sun and Rong Liu, in collaboration with Andrew Feng and Meida Chen) has been accepted to SIGGRAPH 2025!

Project page: https://t.co/gI2Uu5U5Np
GitHub Code: https://t.co/1shEWBbhPb
Paper: https://t.co/kTvGFLHVnY

🧠 This work rethinks neural point-based rendering with three key changes:
• Deformable Beta Kernels — accurately model sharp edges, flat surfaces, and fine geometric details.
• Spherical Beta Color Encoding — disentangles diffuse and specular components with drastically fewer parameters than spherical harmonics.
• Kernel-Agnostic MCMC — a universal optimizer powered by opacity regularization, enabling scalable training across splatting kernels.

🚀 Highlights:
• SOTA visual fidelity across diverse scenes
• 45% fewer parameters than 3DGS
• 1.5× faster than 3DGS-MCMC
• Naturally supports frequency decomposition of geometry and lighting

[Meida Chen @MeidaC]

Gaussian Splatting for Large Scale Aerial Collection in Unreal Engine 5 https://t.co/45zDZIK1pt 来自 @YouTube

[Meida Chen @MeidaC]

Gaussian Splatting for Large Scale Aerial Collection (NEAR vs OFF camera... https://t.co/LRww85KEPs 来自 @YouTube

[USC ICT @USC_ICT]

The 2nd Challenge on Large-Scale Point Clouds Analysis for Urban Scenes Understanding at #ECCV2022 is underway. Join for a chance to win prizes and to present at the Urban3D Workshop! Visit: https://t.co/hwwNa9pSja #3DPOINTCLOUDSEGMENTATION #3DPOINTCLOUD #SEMANTICSEGMENTATION