Presentation
CAdam: Context-Adaptive Moment Estimation for 3D Gaussian Densification in Generative Distillation
Session3D Gaussian Splatting II
DescriptionCAdam resolves the mismatch between reconstruction-native densification and stochastic generative guidance in optimization-based 3DGS. By reinterpreting densification as statistical signal verification via momentum and context-adaptive selection, it reduces Gaussian primitives by 85%–97% across diverse backbones while preserving comparable perceptual quality, making generative 3DGS far more compact and practical.
Event Type
Technical Paper
TimeThursday, 23 July 20262:30pm - 2:40pm PDT
LocationRoom 408 A
Digital Library
PDF
Session Time & Location
Sunday, 19 July 20266:00pm - 8:45pm PDTHall K
Thursday, 23 July 20262:00pm - 3:30pm PDTRoom 408 A
Artificial Intelligence/Machine Learning
Modeling
Rendering
Full Conference Supporter
Full Conference
