...

Mingyang Song

PhD student
CAB G 65
mingyang.song@inf.ethz.ch

Basic Information

I am a joint Ph.D. student at ETH Zurich and DisneyResearch|Studios, starting in January 2024. My research is supervised by Prof. Siyu Tang and co-supervised by Dr. Tunc Aydin. I got my master's degree in RSC from ETH Zurich. My previous research focused on video restoration and compression. Currently, I am dedicated to advancing scene reconstruction and editing techniques.

Publications


Authors:Mingyang SongYang ZhangMarko MihajlovicSiyu TangMarkus GrossTunc Aydin

We combine splines, a classical tool from applied mathematics, with implicit Coordinate Neural Networks to model deformation fields, achieving strong performance across multiple datasets. The explicit regularization from spline interpolation enhances spatial coherency in challenging scenarios. We further introduce a metric based on Moran's I to quantitatively evaluate spatial coherence.

Authors:Mingyang SongYang ZhangSiyu TangTunc Ozan Aydin

We show that Dynamic Gaussian Splatting can be aggressively compressed by combining quantization-aware training with carefully structured motion vectors. The principle is borrowed from conventional video codecs: smoother content is cheaper to encode.