Adobe Research Scientist Valentin Deschaintre has been recognized with the prestigious 2025 Eurographics Young Researcher Award for his outstanding work at the intersection of computer graphics and computer vision.
We talked with Deschaintre about why the Eurographics community is close to his heart, his work and how it impacts Adobe users, and what he loves most about working at Adobe Research.
Congratulations on your Eurographics Young Researcher Award! Can you tell us about the Eurographics community? And what it was like to discover that you’d received the Young Researcher Award?
Eurographics—the European association for computer graphics—is close to my heart. It’s a large conference and community, but it remains accessible. I contribute in various ways as a reviewer or a member of the technical committee. Most recently, I chaired the doctoral consortium.
I found out about the Award the week after the SIGGRAPH deadline, which is always a lot of work, so it was a wonderful surprise while I was recovering from that. When I look at the list of people who’ve received this award before me, it’s really inspiring, and I can only aspire to approach what they’ve done. This is a great recognition from a community I really care about.
Can you tell us a bit about your research?
I started my research journey working on digital material acquisition, which is understanding the appearance of surfaces from a few pictures and being able to recover the materials’ properties. For example, is the surface shiny? What kind of color is it? What’s the local surface variation? What is it made of? That sort of understanding.
As more and more technologies became available to generate images from images or text prompts, I started looking into generative materials. I also wanted to work on methods that provide control to users so they can edit and refine the images that generative methods produce. I believe this is really important for Adobe’s customers.
More recently, I have been looking into image understanding. Because of my early research, I often notice materials when I’m looking at things in the world. With any material, there can be variation, fading, maybe some slight damage. For humans, it’s trivial to recognize that it’s still the same material, but how can a computer understand that? So that is another part of research that I’ve been leading for a few years now both for 2D images and 3D representations.
I’ve also been working on intrinsic image decomposition and recomposition for generative models with colleagues in Adobe. The idea is that you can decompose a generated image into its intrinsic components, for example, the illumination of the scene. And then you can edit the components independently much more easily than you could when they are all joined together. After editing, you can recompose them and recover the original image with the changes you’ve made. This preserves a per-pixel editing capacity that I think users want—while providing really good-looking images and higher level control through generative models.
I want to note that this research has most often been done in collaboration with colleagues here at Adobe as well as academic partners and, most importantly, interns and student collaborators, and I want to recognize the quality of their work and how motivated they’ve been. The Adobe intern program is exceptional in that sense—it really brings in some of the greatest students in the world.
How does your research impact Adobe users?
Right now, I’m really thinking about control. What are the properties users want to manipulate and how can we facilitate editing so that Adobe users can get the results they want faster? Since I come from a graphics background, I like to leverage 3D tools and 3D understanding behind the scenes to facilitate workflows in both 2D and 3D.
When it comes to understanding what Adobe users need,there are two different approaches. First, we talk to the product teams, or to users at conferences. If there’s a capability that they think would be awesome to have, and if it falls into an area that I might be able to do something about, I may design a research project around it.
The other approach is to look at the state of the art. Research is a very community-driven endeavor, so we go to conferences and we engage with the big ideas everyone is thinking about so we can design our work around some of those topics. We think about the capabilities Adobe could provide with new research, and ways to enable even more control of those capabilities, or how we could find better data or adjust a technology in some way that would make it even better. In that mode, we go to the product team and suggest ideas that we think have good potential for products.
The best recent example is the decomposition and recomposition work I was just talking about. We shared an early version with the product team and we were blown away by what Adobe’s internal artists could do with it. For me, giving a tool to an artist and seeing their creativity—and how they use it for their own purposes—is one of the highlights of the job.
What’s your favorite thing about working at Adobe Research?
Adobe Research is at the forefront of the state of the art, and it’s full of brilliant researchers with a very cool mandate: to do good research that we believe is valuable to Adobe and then own that research all the way to the product. And it’s not just any products, but products that millions of people use.
As I mentioned earlier, research is a collaborative endeavor. Lots of the work I’ve gotten to do here at Adobe Research has had contributions from incredible interns who are passionate about what they do. I’ve also had mentors, both in and out of Adobe and in academia and industry, who’ve helped me along the way. I owe them a lot for helping shape the way I think about research and formulate problems. So this Eurographics Young Researcher Award is also thanks to them.
Wondering what else is happening inside Adobe Research? Check out our latest news here.
