Interview: Eric Bourque, Autodesk (CGA2018)

EricEric lives in Montreal, Canada where he works in the Entertainment Creation Products Division of Autodesk as a Senior Software Development Manager for Arnold, Bifröst and Maya where he’s been for more than 10 years. Before joining Autodesk, he completed a post-doctoral fellowship with Pierre Poulin at the University of Montreal in the LIGUM graphics lab. Eric obtained his PhD in Computer Science from McGill University where he specialized in computer graphics at the Centre for Intelligent Machines under the supervision of Gregory Dudek. He has also completed a Master of Science (computer science) at McGill and a Bachelor of music at McGill where he majored in jazz performance (saxophone) and computer science.


Great to see you back again Eric! Does this mean you had a great time at CGA conference last year? 

Definitely! Last year was my first time in Belgrade, but I was spoiled since one of my close colleagues at Autodesk is from here (Nikola Milošević), and made sure that I got to see many of the special things Belgrade has to offer.
It was clear from the outset that VFX Serbia and CGA Belgrade take this event very seriously and the organization and content are top-notch. It was great to hang out with everybody, make new friends, and catch up with others during the event. The presentations were great, and I had some good meetings with customers as well. The student participation is high, which is always encouraging to see at these kinds of events.

CGADay1-335

Developing the CG community must be very important to Autodesk as one of the main supporters of CGA conference from day one? How do you see your users worldwide?

Yes, at Autodesk we try to support the VFX community as much as we can. We sponsor many conferences as well as events worldwide and also host some of our own community events in various cities to try to keep the local communities thriving. We understand that people are using many products to accomplish their ultimate job of making art, and are happy to do our part to try to keep the ecosystem healthy.

Judging by your research publications, your interest in computer graphics has started, for most of our audience, in the early days. How do you see the 90s compared to almost now 20s in the CG industry and the software development?

Yeah, I guess I’m old! Hehe 🙂 I can still remember seeing some of the earlier renders with global illumination for the first time and how amazing they were! They were using all kinds of approximation techniques just to make the rendering tractable on the Silicon Graphics machines at the time. Fast forward, and we have things like deep-learning-based denoisers for Monte Carlo renders that are performing remarkably well for static images. This would have looked like science fiction back then!

Violet_Image-courtesy-of-Enver-BK
Violet, Image courtesy of Enver BK

One of the things I find really cool looking back is that Marcos Fajardo bet his future on pure Monte Carlo path tracing when it wasn’t the popular thing to do at the time. As CPUs got faster and started having higher thread counts, and recently as GPUs are starting to exist with dedicated path tracing hardware, pure path tracing is proving to be the most widely adopted approach to final frame rendering for Film & TV. It’s really impressive that Marcos committed to that idea early on, and didn’t stray from it while others kept adding different approximation features which made many of the renderers more complicated to use. Arnold’s ease of use and artist friendliness stems from that commitment early on to Monte Carlo path tracing.
It’s also great to see that while rendering could be considered to be a mature research area now, there is still a lot of great research work going on in light transport, material representation, participating media, etc.

baby-cheetah_Image-courtesy-of-Yuriy-Dulich
Baby Cheetah, Image courtesy of Yuriy Dulich

Programming and jazz music, how do they play together?

“The finest dozen computer scientists I know are all musicians” – Steve Jobs.

There’s really some truth to this 🙂 Most people I knew when I was studying music were immensely talented at mathematics and computer science, and most computer scientists that I’ve met along the way have shown a serious interest in music. I think jazz musicians are even more adept at dealing with abstract problems, and reacting to the system around them since that’s what they’re always doing when they’re playing. I guess jazz musicians would make great pipeline engineers 🙂
While software development can be seen as being methodical, there is definitely art in programming. And even more creativity in debugging! I’ve always enjoyed pair debugging with someone to observe how they tackle a problem, and what “tools” they’ve built up over their career. The same way it’s really fun to watch a shot and figure out all the details of how the plate was filmed, lit, and how the CG elements were added to it. Sometimes on interviews I show candidates short selected clips from movies and ask them to tell me what they’re seeing – it’s a great way to see how people think and communicate about what they’re seeing 🙂

Moving from a programmer to a lead and manager role, what is the difference for you? Do you still have fun?

Yeah, I guess I’ve just continued to scale through our organization, first moving from an individual contributor, then to the lead of a small team, then a bigger team, then a few teams, and now I have an organization of around 50 people in my group. It’s really a lot of fun – I have people on my team all over Europe, the US and Canada, and a lot of very interesting people who have deep specializations in rendering, simulation, pure mathematics, systems programming, etc. As I’ve moved up the ranks, I’ve also been able to visit a lot more of our customers all over the world which is really inspiring since it’s great to see all the great work people are doing. While I manage a large team now, I do still get to have fun and find the right places to have an impact. Sometimes it’s through architectural work, sometimes it’s just about social engineering; putting the right people together so that the sum is greater than the parts.

This slideshow requires JavaScript.

Is your team all in one building or on multiple locations, how do you manage to support them on the distance, what are the tools to keep the whole team up-to-date?

I have larger teams in Montreal, Toronto, London, and Madrid, a few people scattered in Scandinavia, and then other individuals a little all over the place. We always try to use video for our meetings (currently using zoom for video conferencing which is great!) because that somehow bridges the divide, and lets you get all the non-verbal. We do try to bring people together at least once a year for bigger planning meetings and team building since it’s always easier to work with someone remotely after you’ve met them in person.
We use a bunch of online tools like trello, mural, jira, etc., but I have yet to find something ideal for whiteboard sharing. I think that’s what I miss the most when doing a remote meeting with people vs a local one – I really like to use a whiteboard.

At the last year CGA conference, we had the honour to get the first-hand update on Arnold development from Frederic Servant, Arnold celebrity. What can we expect this year?

Arnold celebrity … hehe. Love it! Fred’s awesome – I’m sure he has something exciting up his sleeve. He’s a French rendering nerd with a British sense of humour, so his presentations are always fun.

Frederic_Servant

What is the latest news in the Bifröst world of particles and liquids?

The work we’re doing on Bifröst is really exciting. Our adaptive data structures and new solvers that run on top of them are making for really impressive high-resolution volumetric simulations for a fraction of the memory (both during the solve and for the cached result). The same adaptivity applies to the other elements in the scene like colliders and emitters, so we’re able to get some great improvements in efficiency. Check out our SIGGRAPH paper presented by Michael Nielsen from this year for more details!
The main thing we’re working on right now is making really nice artist tools for people to be able to do arbitrary procedural work using the Bifröst visual programming language. Keep checking The Area (http://area.autodesk.com) for the latest updates, or join the Bifröst beta for a sneak peek.

Leading the team on both Arnold and Bifröst, can you tell us about their relationship? Any juicy gossip? 🙂

Bifröst and Arnold are definitely in a relationship, and it’s not “complicated” 😉 Expect a baby Arnold Bifröst procedural sometime soon …

The latest trend in deep learning is to go into the computer graphics. Has this buzzword the same interest in your team too? Where do you see deep learning implementation in the rendering technology in the future?

I think there are a lot of interesting possibilities using deep learning for rendering. But at the same time, I think it’s important to understand that it’s not a panacea: the main drawback with AI today is that while it can do things that are very impressive based on good training data, it’s not possible to fine-tune the result. In our industry, that’s a pretty big problem since artists are always looking to tweak things, create variations, and to creatively go from one variation to another.
AI-based denoising is a great example of how you can assist a look-dev artist by allowing them to iterate faster since they can render at lower sample counts, and therefore reduce their iterative render time while developing the look of the asset. It should also be possible to use AI in other ways as well to assist artists; for example to catch a mis-configured scene before spending time rendering it, or helping to light a scene from scratch based on a bunch of a priori scenes used for training. I think those areas are actually quite interesting and exciting. Another would be to use AI to help choose optimal paths for light integration in a Monte Carlo path tracer.
But my hesitation will remain around general use of AI until we develop ways to fine-tune deep networks and to understand what’s in the black box to a point where you can push it artistically in a given direction. Otherwise, there will be too many stressful moments when the director asks for that last minute change 😉

This slideshow requires JavaScript.

You’ve been to Belgrade for the first CGA Belgrade conference; how was your experience? What are your thoughts on our local talents and Serbian CGA scene?

I had a great time at the first CGA Belgrade for many reasons. I liked the intimacy of the event; it really felt like you could talk to everybody, and the events surrounding the conference were a lot of fun. I still recall an impromptu dinner where we had Mikhail from Mainroad, Szabolcs from Digic, a few guys from the Foundry, and a few of us from Autodesk. We had a bunch of great food, good laughs, great stories, and some fun competitive banter 😉
I also got a chance to visit Nordeus, where Ivan Stojisavljević treated us like royalty. It was a great experience, and some local artists like Igor Žanić and Dušan Ković attended and gave us feedback on the direction our products are taking. There’s obviously some very strong talent in Belgrade if Igor’s FX work and Dušan’s look-dev work are any indication.

Arnold_lego-render_Image-courtesy-of-Lee-Griggs
Arnold Lego Render, Image courtesy of Lee Griggs

Oh, and please, let us know – are there any places that you would love to visit in Belgrade again this time? What’s your impression of the city? How would you describe the flavour of Belgrade? Anything to recommend to our new guests?

I got to do a lot of fun extra-curricular stuff last time, and hope to again this year! We got to see a basketball game between Serbia and Austria last year, accidentally met Novak Djoković and his wife, walked all around the ruins, ate too much great food, found the best bartender in Belgrade, maybe even in Europe in my opinion, had some great espresso at a few different discerning coffee shops … and I might have had a bit too much rakia 😉
If I can do half as much this year as I did last year, it will be a success for me!
Now, to find that bartender, since I heard he left Bar Central …

 

CGA2018_Eventbrite_register
Once again, November becomes the month of computer graphics, as CGA Belgrade hosts its two-day journey through the latest news, trends and developments in the VFX industry. We have worked especially hard this year to expand our main program, which we are proud to announce will feature two separate tracks! For big-picture thinking and groundbreaking ideas, make sure to look for the Know It All sign. For hands-on training and insider tips & tricks, don’t miss the Know How stage.

 

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Discover more from VFX Serbia

Subscribe now to keep reading and get access to the full archive.

Continue reading