Physical and neural methods for realistic image generation

For appointments please consult the ZEuS of the Universität Konstanz. The exercise can be found here.

This is a directed studies course (exercise) in which students have to read the literature and do the practical course work. We will meet from time to time and discuss the content of the books. Please send in addition to registering for the course an email to Prof. Deussen. 


Learning Content: 

This course teaches methods for generating photorealistic images of virtual scenes. To achieve this the light propagation within the scene has to be simulated using global illumination.

Lecture Content:

  1. Introduction to Raytracing 
    - Basic algorithm, extensions, data structures
  2. Light transport and global illumination 
    - Rendering equation, global light transport 
    - Sampling, Monte-Carlo Integration 
    - Path-Tracing, Particle Tracing 
    - Radiosity
  3. Radiosity 
    - light transport revisited 
    - creating the equation systems / Form Factors 
  4. General aspects 
    - Finite element methods 
    - plenoptic function / image-based rendering 
    - Holography
  5. Neural creation of images

Qualified Students: 

Information Engineering students, bachelor in-depth lectures or master program


Basics on computer graphics, programming skills (C++)

Credit requirements:

Oral examination (20min)

Registration in ZEuS,

A list with timeslots will be made available in the lecture and later in the office of Magdalena Lipska (Link).

Examination date:

First date: 30/07/24 (Room ZT809)

Second date: TBA


Successful participation yields 6 ECTS

Course Material

Log in to view the course material.