When you get in touch with me and want to work with me on a PhD, exchange visit, or the like, you should send me
When you write, please indicate in your message that you have read these instructions.
In your portfolio, you can include class and hobby projects alike, things you've done during internships, and, of course, research work. The portfolio should include, to the degree possible, pictures and videos (YouTube links are fine), as well as brief descriptions of what you did, what techniques you used and what major challenges you solved, what your role was in multi-person projects, and anything else that's relevant. For projects that have been written up in a paper, include links to the PDFs on arxiv or the final publication. Ask your artist friends for an example – nothing happens in the art world without portfolios, and challenging technical work is no different in that grades tell only a partial story about your technical talent and creativity. I will unfortunately not be able to entertain requests that are not accompanied by such material. For an example of a good portfolio, see here.
In addition to a portfolio, a strong recommendation from a faculty member doing related things at your home university can be very informative. Personal recommendations from faculty members you've done research with are the strongest, but also assessments of how strong a student you are in class compared to your peers ("top-N%") are ok.
I don't usually give out topics to the general B.Sc. thesis seminar. If you’re interested in a specific area of graphics, vision, or related machine learning, get in touch with me. I unfortunately only have time to supervise students who've mastered the basics of graphics and/or vision, as demonstrated by, e.g., doing well in my classes.
I only hire M.Sc. thesis workers for paid positions to work on research-grade problems that tie in with my scientific interests. I will consider supervising your project (without pay) if you have a topic in mind that's related and interesting, but not necessarily advancing the state of the art. I tend to have many such ideas lying around.
If you work in the industry, already have a topic related to my work, as well as an advisor at your company, get in touch with me.
I'm always on the lookout for driven, smart people to do research with. If you impress me with stuff you’ve done in the past (demo, game or other project, thesis, etc.), visibly display the ambition and excitement for doing research, and if we get along well when you show me all this, we’ll see if we can work something out.
For me, the goal for a PhD is pretty simple: making progress on unsolved problems that are relevant enough so at least some people care, and doing it better than anyone else.
I heartily recommend Bill Freeman’s notes, aimed at new graduate students, on how to do research. Find the link on my homepage.