MrNeRF (@janusch_patas)
2024-12-25 | โค๏ธ 69 | ๐ 7
GauSim: Registering Elastic Objects into Digital World by Gaussian Simulator
Contributions:
-
We propose GauSim, a neural network-based simulator that integrates continuum mechanics to realistically simulate dynamics without idealized conditions, bridging the gap between simulated environments and reality.
-
We employ a hierarchical structure to efficiently simulate the dynamics of Gaussian kernels, reducing the number of predictions by around 90% while maintaining high fidelity.
-
We enforce explicit physics principles, such as mass and momentum conservation, in GauSim, resulting in an interpretable model and robust simulation.
-
We present READY, a multi-view video dataset capturing the dynamics of elastic deformations of real-world objects.
๋ฏธ๋์ด
