Corporate

How Duygu Ceylan is using 3D and AR to Recreate Memories

June 6, 2018

Contributed by Meredith Alexander Kunz, Adobe Research

Duygu Ceylan, a research scientist at Adobe Research in San Jose, works at the cutting edge of 3D. A PhD in computer graphics and former intern, she is finding new ways to capture the three-dimensional shape and mechanics of human motion, empowering creativity in 3D imagery and augmented reality.

How did you decide to focus on computer graphics?

Growing up in Turkey, I really liked math. I started studying computer science because it was related to math.

The way I discovered computer graphics was through my passion for art. I had taken art classes, but I knew I wouldn’t have a career as an artist. Computer graphics gave me a way of participating in the arts because it is so visual. I thought, “This is my way of doing art.”

Could you tell us about your work on 3D imagery?

My work explores how we can gather 3D information—whether that is from a set of images, a single image, or from video—and how can we use that info to enable new kinds of creativity.

My PhD research was about how to get 3D information from 2D images. If you have multiple images of the same object, I studied how you can generate a 3D model of the object—in that case, a building. At a high level I still work on similar problems, but now I’m focused more on humans.

One of my published papers focused on capturing moving people and animals. You have a few depth sensors, and you record someone doing a performance or an animal in action. Then you try to reconstruct the motion sequence in 3D. It is a challenging problem if the subjects are moving. But it’s worth pursuing—with an accurate 3D reconstruction, we can capture our memories in 3D and recreate them in virtual environments!

What excites me most is that sometimes, a 3D image may not be the final output a user sees, but it is an intermediate representation to enable other applications like image and video editing. Working with a 3D environment allows you to add objects in a scene or to augment a person, like in augmented reality.

Much of this work is still in progress. I did have one piece become a feature in Photoshop Express in 2018. It’s 3D face tracking and painting. Users can track the 3D shape of a face in the camera view, and then can place a colorful artist-created tattoo on the face as it moves around.

What’s it like to be a researcher in this field? Is the pace of discovery accelerating?

From the technical perspective, the field is advancing very, very fast. Papers that appear at computer graphics research conferences now may already seem old, because research is being released prior to the event at research-sharing sites such as ArXiv. There is increased pressure to get work out quickly.

At conference presentations, I sometimes hear authors say: “At the time of the submission, our method was state-of-the-art. After the submission, there were these two or three other papers that report better numbers—but we also improved our algorithm, so we have better numbers, too.”

Have you noticed other shifts happening in computer science?

From a cultural perspective, people are talking more about diversity. I see that changing everywhere, not only in computer science. People are more vocal, and we should keep talking about it.

Adobe as a company is working to advance diversity. I’ve been part of Women Unlimited, both of Adobe’s women’s leadership initiatives called the Voice and Influence program and Leadership Circles, and the Adobe and Women Summit. I appreciate all these opportunities. 

How does being at Adobe Research help you advance your work?

In terms of industry research labs, Adobe Research is a great balance between publishing and product.

As a graduate student, I knew Adobe Research from all the papers they were publishing. I was really excited to get an internship with the group in San Francisco. I enjoyed the environment, people helping each other—both scientists and interns. Adobe Research was at the top of my list when I graduated. Then I was lucky enough to come back as a full-time researcher.

Now, I love being able to work with so many amazing interns each year. This summer, I will have two primary interns and will co-mentor a few other projects. We have great resources here—we can reach out to any student or professor around the world that we think would be a good fit for our project, and start collaborating with them.

What I really like the most is being surrounded by so many smart people. I can easily start exploring new areas, for example in computer vision or deep learning. The environment encourages you to learn.

Interested in learning more about Adobe Research? Drop by their site and apply for opportunities on our career site!