← Search

Tharindu Wickremasinghe

2 accepted papers

2026

NewtonGen: Physics-consistent and Controllable Text-to-Video Generation via Neural Newtonian Dynamics

ICLR 2026poster

A primary bottleneck in large-scale text-to-video generation today is physical consistency and controllability. Despite recent advances, state-of-the-art models often produce unrealistic motions, such as objects falling upward, or abrupt changes in velocity and direction. Moreover, these models lack…

Cited by 0SourcecodeScholar
2026

SeeU: Seeing the Unseen World via 4D Dynamics-aware Generation

CVPR 2026

Images and videos are discrete 2D projections of the 4D world (3D space + time). Most visual understanding, prediction, and generation operate directly on 2D observations, leading to suboptimal performance. We propose SeeU, a novel approach that learns the continuous 4D dynamics and generate the uns

Cited by 0SourcecodeScholar