Philipp Erler
TU Wien
WED afternoon
In this workshop, you will learn about surface reconstruction from unoriented point clouds. Using the frameworks of Points2Surf and PPSurf, you will generate a dataset and train a neural network on it. The network produces a signed distance field, which is turned into a mesh using Marching Cubes.
Hardware:
You can use Google Colab. The free T4 runtime is sufficient. If you want to use your own hardware, you’ll need a GPU with at least 6GB memory and CUDA support. To not overload the CESCG network, please prepare the setup according to https://github.com/cg-tuwien/ppsurf_modeling.
|
Philipp Erler is a PhD Student in the Research Unit of Computer Graphics, at the TU Wien. His research interests lie in the areas of surface reconstruction and deep learning. He received his Bachelor’s degree in “Software-Engineering” at the HS Esslingen and his Master’s degree at TU Wien.