Course Syllabus

Course Instructor and Office Hours:

Instructor: Matthias Zwicker, office hours by appointment
Teaching Assistants: Geng Lin, Saeed Hadadan, office hours Tuesday/Wednesday, 4-5pm, online: https://umd.zoom.us/j/97065812970

Time and Place:

Tuesday and Thursdays, 2:00pm-3:15pm, CSI1122

Summary:

This course covers advanced techniques in realistic rendering and modeling for computer graphics with a focus on recent techniques based on neural network. The first part of the course focuses on realistic rendering using physically based image synthesis algorithms. Students will learn how to represent light interaction with physical objects mathematically, such that these effects can be simulated with computer algorithms to create photorealistic images for a variety of real-world applications. In the second part, we will discuss how to create 3D models that can be used in graphics applications. We will focus on data-driven approaches, which is to leverage data captured from the physical world using cameras and 3D sensors to construct digital models. Throughout the course, we will highlight  and explore recent neural network and deep learning techniques to address these problems.

Content:

To model light transport for realistic rendering, we will introduce fundamental concepts like radiometry, the bidirectional reflectance distribution function, and the rendering equation. We will discuss techniques to solve the rendering equation such as Monte Carlo path tracing and importance sampling. We will also cover recent deep learning techniques that have been proposed both to accelerate Monte Carlo rendering and to solve the rendering equation in novel ways. In the second part of the course, we will cover 3D scanning and reconstruction, and texture and appearance acquisition. We will introduce 3D geometry representations such as meshes, and implicit and point-based representations, and fundamental geometry processing operations. Finally, we will discuss data-driven approaches to model 3D geometry and textures based on recent deep learning techniques.

The course includes programming assignments related to each major topic, and a self-directed final project.

Course Schedule, Materials, and Online Communication:

The course schedule, all materials, and online communication will be managed via this course page on UMD Canvas, the electronic learning management system of UMD. Access to these resources requires login using your campus ID.

Grading:

Grading will be based on the programming assignments (40%), the final project (40%), a final exam (15%), and class participation (5%).

Prerequisites:

An “Introduction to Computer Graphics“ (or an equivalent course) is recommended but not required. The course builds on concepts from calculus, linear algebra, and algorithms and data structures. Programming assignments rely on Python.

Academic Integrity:

We will follow the guidelines set forth by of the Department of Computer Science and the Office of Student Conduct.