The scene reconstruction results differ significantly from the provided examples.

Excellent work! However, I'd like to know why the results are so poor when reconstructing dynamic scenes. Yet, HUGS, the underlying project of HUGSIM, produces much more realistic results for dynamic scenes.

<img width="1817" height="806" alt="Image" src="https://github.com/user-attachments/assets/ffc432d6-6daf-48f8-a64b-7ab5b06bfdde" />