An Adobe Research internship isn’t your average summer job. Interns have the opportunity to dig deep into their own research interests and play a key role in developing innovative new technologies. Summer internships often turn into years-long research partnerships, help shape PhD dissertations, and lead to careers and professional collaborations with Adobe—and can even result in prestigious awards.
This year ACM SIGGRAPH, a community of computer graphics researchers and artists, recognized two former Adobe Research interns who did significant portions of their PhD research in collaboration with Adobe researchers. Minchen Li won the SIGGRAPH 2021 Outstanding Dissertation Award for his work simulating real-world contact between objects. Jonathan Ragan-Kelley received the SIGGRAPH 2021 Significant New Researcher Award for his contributions to systems and compilers for rendering and computational photography.
This isn’t the first time Adobe Research interns have made a strong showing in SIGGRAPH awards. Former intern Jun-Yan Zhu won the outstanding dissertation award in 2018, and several others have received honorable mentions.
Li’s internship led to simulations of objects colliding, squishing, and twisting—and a PhD
Li was working toward his master’s degree when he began his first Adobe Research internship, back in the summer of 2017.
“I decided to spend a summer with Adobe because I saw excellent research projects from Adobe researchers and interns at SIGGRAPH,” says Li. “I also saw connections between the research and Adobe products, and I wanted to be part of it.”
Danny Kaufman, principal scientist at Adobe Research, served as Li’s mentor. “What struck us right away about Minchen was his curiosity and ambition,” says Danny. “He immediately wanted to tackle new and very hard problems. To do so, over his internships he had to learn lots of new material which, each time, he picked up quickly. At the same time, he’s a fun collaborator and a creative problem solver.”
Li’s interests were broad, so Kaufman helped guide him toward contact mechanics—a critical area with many outstanding challenges, open theoretical questions, and practical industrial applications. Once he found his niche, Li returned to intern over the next three summers, through his master’s and PhD research. Li’s work produced first-author SIGGRAPH papers and his research during his last three summers formed the basis of his PhD dissertation, with Kaufman serving as the external member of Li’s PhD committee.
According to SIGGRAPH, “Minchen’s dissertation presents a breakthrough in the notoriously challenging and long-standing problem of robust frictional contact simulation nonlinear solid dynamics with guarantees of non-intersection.” Put more simply, Li deployed math and physics to build IPC, a tool that reliably creates previously impossible simulations of real-world objects colliding, squishing, and twisting.
“Minchen took on the problem of contact modeling and figured out how to make it work out of the box. Most technologies make it work sometimes, but designers just want it to work every time. Now we can put it into the hands of artists,” explains Kaufman. IPC is already being used in mechanical engineering, robotics, and animation studios.
Li recently joined the UCLA Department of Mathematics as an assistant adjunct professor, where he plans to keep tackling simulation challenges that matter for theory and industry.
Ragan-Kelley is changing computational photography and image processing — and bridging industry and academic research
When Ragan-Kelley came to Adobe as an intern in 2012, “he wanted to develop Halide, a technology he’d invented,” explains Sylvain Paris, Adobe Research fellow and Jonathan’s mentor. “So he spent his internship connecting the dots between his research and the places where it could be useful at Adobe. We worked together to identify the areas of biggest impact.”
Halide is an open-source language and compiler for computational photography and image processing. It allows users to write code and that works across all hardware—phones, servers, and desktops.
“Halide saves all of the effort that goes into re-writing and re-optimizing,” says Shoaib Kamil, Adobe Research principal scientist and one of Jonathan’s collaborators.
The SIGGRAPH Award recognizes Ragan-Kelley’s work on Halide, which it describes as “the industry standard for computation photography and image processing,” as well as his unique approach to research, “creating new abstractions that enable high performance by disentangling previously intertwined concerns.”
Since his internship, Ragan-Kelley completed his PhD and became a faculty member at MIT. He’s now also an Adobe Research consultant, and each summer he sends grad students to Adobe Research to continue working with Halide.
“Halide has become a powerful tool for practical development in products including Photoshop. It’s also a test bed for new ideas in the academic world,” Kamil adds. “It’s unusual to achieve both.”
“Many of my and my group’s best and most fruitful collaborations are with Adobe researchers, including relationships now spanning more than a decade,” says Jonathan. “On top of our academic productivity together, working with Adobe has provided one of the richest opportunities for our research to have direct impact on important products used all over the world.”
The “righteous feedback cycle” between Adobe and academia
The achievements of former interns, including Li and Ragan-Kelley, are part of a bigger commitment at Adobe Research.
“We do things differently,” explains Kamil. “We collaborate with universities and build long-term relationships with professors and their students. Our internships sometimes go beyond summer so the work gets into a publishable state.”
This approach gives interns a big boost with their graduate work: “My peers worked on ideas related to products at other companies,” says Li. “At Adobe you can choose your research, and it only needs to be related to the bigger research areas of the company. That’s why I was able to make all of my internship projects part of my dissertation research.”
Beyond internships, Adobe Research partners with academic researchers on mutually interesting endeavors. “We help academic researchers understand the real world, and we work with them to embrace speculative, exotic, nearly impossible projects,” says Kamil.
“Our relationship with academia is a righteous feedback cycle,” explains Kaufman. “We want to further what we know about the world and science.” Research collaboration allows Adobe Research to give back to the academic world while making innovative discoveries that may someday be incorporated into future Adobe technologies. It’s a cycle that creates benefits all along the way.
