FORTUNE — You don’t have to be Ansel Adams to take a cool picture with a smartphone. Part of that is because every time you snap a shot to Instagram, complex computer engineering is at work, behind the scenes.
Kari Pulli is one of the people on the other side of that curtain. Every day, he and his five-member team work on ways to make camera phones better. Pulli joined NVIDIA (NVDA) as its senior director of research in 2011. Before that, he worked at Nokia (NOK) for 12 years, two of which he spent as a visiting scientist at MIT.
Since 2000, Pulli has been working on concepts that have just started to roll out on mobile devices, such as augmented reality, which allows users to manipulate an image of the real world using apps on their phones. Pulli talks to Fortune about his transition from an individual researcher to a manager and how to chop an intellectual odyssey into manageable steps.
Q: How, in broad terms, do you make mobile cameras better?
A: We address the limitations of cameras. A human eye can see a lot of very bright and very dark things, almost at the same time, but cameras usually can’t. So outdoors, you take a picture of your friend, and there’s a bright sky in the background. Usually, either you only get a black silhouette, or you get your person and the sky’s completely burned-out white. We’re figuring out how to make those pictures better.
That seems like a broad goal. Is it daunting?
We have a palette of things we do. We have some projects that could take years, and then others where we hope to get some results within a few months.
The trick is that even if you have the long-term goal, you can break that into smaller goals that are worthwhile on their own. Then, rather than having this beacon in the horizon where you have to cross a couple of oceans and a few deserts to get there, and then possibly drown on the way, you say, “Okay, we are going to New York, so let’s first go to Las Vegas and have a good time there.” You break it into something that’s more manageable. Then, even if we don’t go all the way to that original target, we still reach something worthwhile.
There was one at Nokia. The first time I started looking at augmented reality was around 2000. That was a little bit too early to try to do augmented reality on cell phones. We made some demonstrations, and we could hardly get them running on a laptop. But I proposed a larger research project, and I had this end-goal in mind, but I hadn’t broken down the path.
So I decided that rather than fighting to get to the end, I would take one of the requirements to get there and work on that.
Cameras on phones are remarkably simple to use but the technology that runs them is so complicated. How do you communicate what you do to the rest of the world that uses these devices?
Well, I sometimes try to explain these things to my children, and that’s a good start.
Also, almost everybody on my team has a Ph.D., and part of that training is to describe what you’re planning to do so that somebody else gives you the resources and the time. You have to learn to sell your ideas even before you have done the research.
How do you manage a group of super smart Ph.D.s with different specialties?
I want to get people working on what they like. If I was the only one to set goals and push ideas, then they’re not going to work so hard, and they’re not going to innovate as much. But if they are part of the process of deciding what they want to do, then they buy into it and have more fun.
You also have been where your Ph.D.s are now. What did you learn from making the move from a researcher to a leader?
At that time that I became a leader, I was promoted because I was good technically. I was able to do lots of things by myself. I had a very young team, and it felt like I could do everything myself faster and better than everybody else. For a short time, I fell into that trap. But then I noticed that I was not training the guys working with me. So one thing is letting go and giving room for others to grow and learn.
Do you prefer working individually or leading a group?
I like both. Here at NVIDIA, I try to keep my fingers dirty with real work.
Especially in research management, I think you can’t get too far away from the real work because otherwise you don’t know what to expect from the guys.
You can’t lead them if you don’t understand what they do. You kind of have to earn the respect of the guys you’re leading.
What is the most exciting project you’re working on?
We just submitted this research for review: Last summer, we had a Ph.D. student from Stanford with us. The student … had this wacky idea that you should be able to edit a picture before you have taken it. I was a little bit dubious, but then we worked several months on it and we got it working, and it’s pretty exciting.
Photography is a creative experience, it’s not just recording the world. You choose the situation, and then you choose what story you want to tell. And now, we are giving you more tools to do that, and to do that while you still engage with the image-taking rather than postponing that process. Before you shoot, you can say, “What if I made the sky bluer? Will I even want to take this photo?” Or, “My friend’s face is in the shadow, I need to brighten that part up.”
That sounds amazing.
I hope that’s what the paper reviewers say too.