Thor Vestergaard Christiansen

I am a PhD Student at DTU Compute in Copenhagen, Denmark supervised by Professor J. Andreas Bærentzen and Assistant Prof. Morten R. Hannemose. The title of my PhD project is Neural Form Representation and I work on learning based methods that can represent geometric shapes.

Email  /  Github  /  Google Scholar  /  CV  /  LinkedIn  /  Twitter

profile photo

Below you find a list of my published papers.

Neural Representation of Open Surfaces
Thor Vestergaard Christiansen, J. Andreas Bærentzen, Rasmus R. Paulsen, Morten R. Hannemose
Computer Graphics Forum (SGP), 2023
project page / paper / code

We use a simple MLP to learn both the Generalized Winding Number and the Unsigned Distance Field to represent open surfaces such as human facial expressions and different types of garments.

Teaching Activities

During my BSc., MSc. and PhD I have been a TA in the following courses.

DTU 02580 Geometric Data Analysis and Processing, F2024 - Spring Semester 2024

02504 Computer Vision, F2024 - Spring Semester 2024

02580 Geometric Data Analysis and Processing, F2023 - Spring Semester 2023

02504 Computer Vision, F2023 - Spring Semester 2023

02580 Geometric Data Analysis and Processing, F2022 - Spring Semester 2022

01037 Advanced Engineering Mathematics 2, Aug2021 - August 2021

01037 Advanced Engineering Mathematics 2, Aug2019 - August 2021
Space Experiment
Thor Vestergaard Christiansen

I designed an experiment, which was performed on the International Space Station ISS. The teaser video is from the European Space Agency, ESA, and provided by the Danish Broadcasting Corporation (DR), DR, P3 Essensen. Also, see the DR Article.

DTU Oticon Audio Explorers 2018
Thor Vestergaard Christiansen

Together with a team of engineering students, I won the Oticon Audio Explorers competition in 2018 Competition. The image is from Oticon © from the Oticon Audio Explorers Tour 2018 to New York City.

Talk at the Pioneer Centre for AI, Copenhagen
August 21st 2023

I gave a talk on the topic Neural Form Representation and addressed some of the challenges with regards to using learning based methods for 3D shapes.

This template is kindly borrowed from Jon Barron.