The following contains currently open thesis topics in our areas of research. If you are interested in a topic or have a thesis idea of your own that might fit our research interests, please contact us directly.
Open Streetmap. We have population statistics, density, number of cars. Streets have capacity. Create a massive parallel simulation in a Vulkan Computer Shader. An alien attack where people jump into their cars and try to leave Vienna. What happens?
If you are interested in this topic or for more…
Learning game. A group of adolescents is planning something really stupid and dangerous. You play one insecure member who has the feeling that this is not a good idea. In the game you explore several ways of expressing your opposition or leave the group.
If you are interested in this topic or for…
Story driven serious game about the experiences of a dead son that died in a road side car accident because he was driving too fast and tried to overtake in a curve. Show the perspectives of the parents, siblings and friends talking about their loss. Show what happened in flashbacks. Enable…
There are several tricks to counter depth fighting. Normal depth values vs inverse z, or making infinite depth possible. See http://entertain.univie.ac.at/~hlavacs/Topics_EC.pdf
You can also make the near plane as far away as possible, and maybe zoom in with a small FOV. This might work if long…
What does C++26 reflection and other goodies mean for game engine development?
If you are interested in this topic or for more information please contact Prof. Helmut Hlavacs (helmut.hlavacs@univie.ac.at).
Create a simple path tracer (offline). Implement a parameter n that determines the number of rays that are sent out per pixel. This n then determines also the noisiness of the result. Use a Deep Learning approach to denoise the pictures.
If you are interested in this topic or for more information…
The idea is to create a 360 degree video from some cool environment, e.g. a canyon, mountains, etc. Then render a photorealistic car into it. This could be used for production of car commercials, where the actual cars are added to the videos in post production. Film a 360 video with drone and render…
First idea: train a DL GAN to create image sequences (motion compensated image differences) taken from a video. Use the last N frames to create the next frame. So instead of having a video file, you would actually store the GAN and the first frame to kick off the video sequence, the GAN would then…
E.g balancing something. Use small perturbations in friction, mass distribution to make the ML model robust. Use domain randomization (OpenAI grasping models).
If you are interested in this topic or for more information please contact Prof. Helmut Hlavacs (helmut.hlavacs@univie.ac.at).
