Single-Course English 5 ECTS

Physically Based Rendering and Material Appearance Modelling

Overall Course Objectives

This course takes its outset in the appearance of real world materials. The goal is to get as close as possible to replicating the appearance of real materials by computer graphical rendering based on mathematical/physical models. The participants will get an introduction into the physical models behind the photorealistic rendering of a digitally modelled scene. In addition to surface rendering techniques, this course covers techniques for simulating the light scattering that takes place under the surface of most real world materials. These volume rendering techniques enable rendering of participating media such as clouds, smoke, fire, and rainbows. They also enable rendering of translucent materials such as skin, marble, and most drinks and food objects. Material appearance is determined by the optical properties of materials. Computation of optical properties from the physical composition of a material is also introduced in this course.

In the entertainment industry, photorealistic rendering is used for computer games, animated films, and special effects in feature films. Moreover, the technology is applied for visualisation in material design and architecture. The most recent physically based rendering techniques can predict the result of taking a picture with a digital camera and, therefore, have relevance in quality control, spectroscopy, and biomedical optics.

During the course, we will mainly consider exact solution methods and only rarely work with fast real-time solutions. Graphics card (GPU) and multi-core processing may be exploited to speed up computations. If the rendering is sufficiently fast, we can use it for generating data for machine learning. This makes the course contents directly applicable towards creating inverse methods for image analysis and computer vision based on deep learning.

Implement rendering techniques for global illumination (for example path tracing or photon mapping for rendering both surfaces and volumes).
Apply Monte Carlo integration in rendering.
Simulate various types of light sources.
Use advanced camera and eye models.
Simulate examples of wavelength dependent phenomena such as dispersion, interference and diffraction.
Simulate subsurface scattering.
Combine shading and tracing techniques with theory for light-material interaction.
Create a tool for rendering realistic images augmented with reference data for machine learning.
Compute the optical properties of a material from a physical description of the material composition.
Analyse real-world light-material interaction and suggest ways of simulating it.

Visual effects: global illumination, Fresnel effects (metallic reflection as well as reflection, refraction, and dispersion in transparent objects such as water and glass), absorption, translucency, interference, diffraction, light scattering in volumes.
Methods: path tracing, photon mapping, microfacet modelling (BRDF/BTDF), subsurface scattering (BSSRDF), Lorenz-Mie theory.
Core elements: Monte Carlo integration, geometrical optics, radiative transfer, light scattering, electromagnetic radiation.

02562, or a similar course (some knowledge of heat transfer and Maxwell’s equations is an advantage)

Lectures and computer lab exercises.

Jeppe Eliot Revall Frisvad

See course in the course database.

Registration

Language	English
Duration	3 weeks
Institute	Compute
Place	DTU Lyngby Campus
Course code	02941
Course type	PhD
Semester start	Week 23
Semester end	Week 26
Days	Mon-fri 8:00-17:00