Hi, I’m Ben Kent and I’m a Lead Research Engineer at Foundry, as well as a screenwriter, producer and director.
I first joined Foundry in 2001 to work on the original Furnace suite of plug-ins, and I was one of the recipients of the 2007 Academy Award for Science and Engineering.
In 2011 I took a slight detour into the world of production and whilst making a number of shorts and adverts and doing a load of VFX work, I wrote, produced and directed my feature debut. Comedy horror Killer Weekend came out just this year, and so far has been licensed for distribution in over 15 countries.
At the beginning of 2018, with little left for me to do on the film, I returned to Foundry where I quickly remembered how much I loved both the company and the work. Now I lead the development of the NukeX research plugins and, more recently, the A.I.R (Artificial Intelligence Research) project.
Tell us the story of how you first got involved with the VFX industry.
I actually got involved in VFX by joining Foundry. It was my second job after University (I’d studied engineering) and a year in telecoms had taught me that that definitely wasn’t what I wanted to spend my life doing. I’d always been a massive movie buff, so when I saw an opening at Foundry, the potential to combine my hobby and all those computer vision modules seemed like a perfect fit. It’s a great way to mix my nerd side with my other nerd side.
You have been writing, producing and directing your own films since 2007 — what lessons have you learned from creating these films, and how have they influenced your life, and your research at Foundry?
One of the best lessons I have ever learned was if you’re about to embark on a tough project, make sure you tell every single person you can about it. Because then you have no option but to see it through without being really, really embarrassed.
I also learned that no Nuke release is as stressful as trying to direct an actor who didn’t get the lunch they wanted while a producer is telling you you’re going to blow the whole budget if you don’t get the next five setups done in half an hour.
When researching & developing a new plugin for Nuke, how do you approach solving the creative challenges artists face?
I’ve always done the VFX for my films, so I’ve gained a lot of perspective on what life is like on the other side of the fence and what takes up the majority of an artist’s time. As a team, we have a lot of great relationships with both VFX houses and Nuke gurus and we’re in pretty constant dialogue with them. Hopefully all that helps drive our research into tools that are actually useful and don’t just look clever!
We’ve also got an eye out for bigger industry trends and as you’d expect we’re looking into AI right now. We’re as excited as anyone by its potential, and I’ve already seen some incredible things that I’m not sure could be done another way. Our goal though is to use it to enhance artists’ productivity, rather than make an uncontrollable black box that really just shows off how advanced the state of machine learning is.
Generally when we design a tool we’re always trying to tread that fine line between making it easy to use and allowing artists all the control they might ever need for a particularly tricky shot. My friends in docs might curse me for saying this, but my goal is to always create a plug-in that someone can use without having to check the manual. But, big-up the content team, because in case we fail, there’s some awesome manuals there to fill in the gaps.
Speaking of developing Nuke plugins, you co-developed Furnace, which won an Academy Award in 2007. What sparked your interest in temporal coherence?
Temporal coherence and its evil twin, image segmentation, are the holy grails of computer vision. Once you can do these perfectly, almost everything else you might want to do just falls out in the wash. They’re really, really hard though!
Extending upon this work, you played a major part in developing SmartVectors for Nuke — one of the biggest game-changers we’ve had in Compositing in many years. How would you explain the underlying mechanics or thought-processes behind how it works to your Grandma?
It took me half an hour to explain to my mum how to delete a contact on her phone the other day (and I’m not sure I was successful), but I’ll give this a go…
Imagine you’ve got a 2D tracker centred on every pixel, and then you’re sort of smoothing the results together to get rid of any dodgy bits. Actually, that sounds quite simple. I might see if my mum understands it…
Do you think there are many applications for your research in technologies outside of VFX?
A lot of machine learning algorithms have potential applications in other areas of visual post, such as editorial, and that’s something we’re looking into.
I don’t see us going down the consumer route, but computer vision research is obviously all over that, whether in smartphones, driverless cars, security applications, etc. Possibly the most important example in recent memory is the use of cutting edge real-time face tracking to enable people to take pictures of themselves with cat ears.
What are the biggest challenges you see Compositors facing both now, and into the future? What tools are being researched or investigated to solve these?
Unfortunately, I think the biggest challenge might be the incessant need of TV manufacturers and broadcasters to keep doubling resolutions, which plays havoc with data and processing requirements. I’m really hoping Intel, AMD and NVidia are actively researching some good solutions!
Have you made any observations about how a developer vs. an artist might approach solving a problem in Nuke, or otherwise? Is there something artists could learn from adopting a developer’s mindset, and vice-versa?
My main observation is that whenever a developer makes a Nuke script, nodes are all over the place with arrows going every which way. No logical separation and design, no comments… ironically all the things you do have to do in code! From what I’ve seen a lot of comp artists do have a developer’s mindset, just with better dress sense.
Who inspires you, and what are the most valuable lessons you have learned from them?
Jerry Seinfeld. The man gets paid to have coffee with his friends. Genius.
Where can people find out more about you and your work?
For my general film stuff you can check out www.fidgetfilms.co.uk, and here’s the website for Killer Weekend: www.killerweekendfilm.com.
Thanks for having me, Ben!