Meshless Neural Rendering
master thesis
Status | in progress |
Student | Samuel Eiler |
Advisor | Changkun Ou |
Professor | Prof. Dr. Andreas Butz |
Task
Description
The crown in computer graphics is rendering. Rasterization has been well engineered years back, and ray tracing is already avaliable in consumer products.
Due to the recent success of deep neural networks in processing grid-based image data, there is a huge interests regarding processing 3D geometry data and render it in a 2D format.
Tasks
- Perform a literature review
- Explore and design an unified network architecture that be able to deal with (at least two) different 3D representation
- Tuning your network that plug rendering settings and produce desired rendering pictures
- Summarize your findings in a thesis and presenting them to audiences
Requirements
- Good communication skills in English
- Have experience or projects using Blender and Python
- General knowledge about machine learning and computer graphics