RGL is looking to recruit a postdoc and a PhD student in the wider area of differentiable and inverse rendering / rendering systems / appearance modeling in the upcoming admissions cycle (Deadline for PhD applicants: December 15, 2020. Postdocs: no deadline, just drop me a message).
Are you excited about any of the following topics?
In addition to generating an image, a differentiable rendering algorithm propagates derivative information through the entire simulation (a bit like PyTorch/TensorFlow, but highly specialized to the physics of light). This enables fascinating and completely new uses of rendering algorithms to solve inverse problems. RGL is developing Mitsuba 2, a new rendering system that can be used to this end. Please take a look at the following three RGL projects to learn about our work in this area:   .
Modern rendering systems are large and complex. They involve advanced physical effects (spectra, polarization, wave-optical interference), GPU and CPU implementations, differentiation, and a variety of different rendering techniques (Bidirectional Path Tracing, Metropolis Light Transport, Vertex Connection Merging, Gradient Domain methods, etc.). Dealing with all of this complexity using standard programming languages (e.g. C++) is an increasing challenge. I'd to rethink how they are built, including the underlying programming languages. In this area, I'm looking for candidates who have a solid background in the area of programming languages and compiler construction.
Material acquisition and modeling
Materials found in the real world are beautiful and often cannot be described well by simple models used in computer graphics. I am interested in measuring materials to understand their properties, and to use those insights to improve the models that are currently used. My group has access to one of the world's most advanced motorized gonio-photometers for acquiring their optical properties, which is a unique opportunity for work in this area. Here are some examples of recent work in this area  .
What EPFL and RGL, and I can offer
EPFL provides a world-class, highly collaborative international research environment with competitive salaries and a beautiful setting on the shores of Lake Geneva. RGL is unusual in that in addition to conducting research on fundamental aspects of rendering, we also invest significant amounts of time to build practical systems (e.g. Mitsuba, PBRT) that constitute a widely used foundation of research projects around the world. Due to the small size of RGL (currently just 3 PhD candidates), I am able to provide extensive supervision to students–I enjoy reserving a large portion of my time to research and software development and often participate directly on some part of their projects.
RGL is part of EPFL's School of Computer and Communication Sciences (“IC”). Doctoral students in this school are part of a centralized doctoral program (EDIC), which is somewhat unusual compared to our European peer institutions (that said, such programs are common in top-10 schools in the United States).
One implication of this difference is that students do not directly apply to professors (e.g. me), but to a program. Admission is done by a committee of professors from different areas of computer sciences with the goal of recruiting the best candidates. The students selected by this committee receive a flexible fellowship that covers their first year. During this time, they can take classes and conduct small research projects (“EDIC projects” / “rotations”) in different labs, with the goal of choosing an advisor that will formally hire them into their lab at the end of the first year. This system is particularly valuable if you are interested in multiple research areas and would like to experiment and gain further experience before committing to a concrete path. Multiple rotations are not a requirement, however (in case you are only interested in a single lab).
The deadline of applications to EDIC is December 15 and consists of
- Statement of Purpose (ideally written so that it can be understood by a general CS audience)
- Grade Transcripts
- Curriculum Vitae
- 3 Recommendation letters
The recommendation letters are probably the most important part of your application—ideally, they should be from people with whom you have collaborated closely, and who can comment on your ability to do research. Recommendation letters from instructors who can comment on your grades but don't really know you well provide very little information relative to the grade transcript. For this reason, it's important to have at least 1-2 letters of the former category (ideally respected colleagues who can provide a detailed and positive assessement of your qualities as a researcher.)
Please reach out to me if you are interested in applying—I'd love to schedule a Skype/Zoom call to learn more about your background and interests before the EDIC deadline. The EDIC application form also provides a field where you can specify faculty of interest, with whom you have already communicated—please make sure to enter my name here.