This interview with Technical Director and Project Lead Tony Barbieri is part of a three-part series that goes deep behind the scenes with Psyop regarding their four-spot campaign for UPS and agency Doner. For the other interviews, check this post.
From spot to spot in this campaign, how did the technical challenges change, and how far was the envelope pushed, technically speaking? Likewise, at any point in the production, did you or your artists have to reinvent the wheel?
From a technical standpoint, each spot was very similar in the challenges we faced, other than “Storm” which was a whole different beast in itself. The challenges differed in each spot more in the storytelling than anything else.
How to fit a story in 30 seconds, actually 25 seconds, introduce the scenario, play the scenario out, deliver the required information about the product, and finally reach a conclusion were probably the biggest challenges we faced when working on these first three spots.
We had a lot of action we needed to pack in in a very short time so we really took a lot of time in the beginning to flesh out our previz to make sure we had a well paced story. That was probably the longest process of the whole campaign for us.
As far as the machines, an interesting thing happened while working on these spots. We were just beginning to upgrade our machines to 64 bit, but we still didn’t quite have enough for the entire team. So we eventually reached a point where the artists that had been working on the 64 bit machines were creating these very heavy models and rigs, so heavy that the artists working on the 32 bit machines weren’t able to open and work on our scenes. We actually had to steal all of the 64 bit machines in the company so were able to open our scenes to work!
We had low rez versions of our assets but we got to the point where we had to work with the final geometry and it just wasn’t possible on 32-bit. Surprisingly, the render times really didn’t suffer. We had to be able to use more than 4 GB of memory to load our scenes up, luckily all of our render machines had already been upgraded to 64 bit, so in the end everything went pretty smooth.
I don’t think we ever had to reinvent the wheel per say—if anything, we worked very low level. I like to keep everything as simple as possible and not over complicate things. Sometimes it ends up taking more time when you try and come up with technical solutions, rather than just brute forcing through it. We modeled and textured what you see in the spots, there really weren’t any tricks involved.
We have been developing our pipeline for a little while now, and those developments are really starting to pay off in terms of efficiency and speed.
I understand that much of the corrugation we see in the models was built into the geometry itself, where texturing tricks relying on bump and displacement mapping were discouraged. At the end of the day, how expensive, in terms of time and processing power, was this decision, and was the determination to shy away from cheats one made by you or someone else?
The decision to not use cheats was mine. At first, I wasn’t sure I made the right decision, but in the end modeling something like that will always look better. Tom Cushwa and Soo Hee Han did an amazing job modeling all of those details in.
If we would have gone the route of texturing/displacing the corrugation, when we ran into situations where we needed to see the corrugation up close or get light to come through the holes in the corrugation, we would have had to mess around getting it all to look right.
The previz at the time we started modeling hadn’t even started either, so I had no idea what we were going to see up close and what we weren’t. I didn’t want to take the gamble early on and have to go back later and revisit the models. It really didn’t factor in that much more time getting the corrugation modeled, because we had such a great modeling team on these spots, and they plowed through the work.
Sure, the modeled corrugation added processing power needs, but I think the end results were worth it. We never had to scramble to get bad looking corrugation fixed because it was too close to camera or really deal with it at all once it left the modelers. That peace of mind was totally worth it.
How did you, as the chief foreman to your puppet-masters of rigging, arrive at the final character setup, and how did the overall design of the campaign affect the rigs themselves?
Well in the beginning of the project, when we were getting designs from the designers, I would go over them and find problem areas that I knew weren’t going to work when we brought the characters into rigging.
From the start Eben Mears, our director, wanted the spot to look like it was created with miniatures on a sound stage. With that in mind I wanted to rig the characters in a very believable manner, to really emphasis the possibility of them being “real”. To do that, we had to build the characters in a way that they would be built if were we were going to actually create real miniatures.
So in the end, the rigging more influenced the designs than the other way around. The character setup was really quite simple for these characters after that. The only atypical setup we ended up using was for the faces. Lee Wolland, our rigging lead on the project, came up with a system where he provided the animators with a rough facial setup for them to push around and give the characters emotion.
Then after a shot was approved, he would go back through and re-model the characters faces in the approved expressions, and animate on/off the visibility of the faces at the proper moments. Doing the faces this way gave us the more “realistic” stop motion look we were going for and fixed any UV stretching.
Despite everything appearing to be tailor-made out of cardboard, one would suspect that the shading and texturing pipeline, for something so visually synonymous, would be more streamlined. Is this true, and what types of considerations—technical and artistic—were made to suit this heavy-duty aesthetic?
It would be easy to think that when seeing these spots, but they were actually fairly difficult to texture and light. We needed to keep everything from looking too muddled together while at the same time giving the cardboard a very realistic look.When dealing with a world made entirely out one material, we have to take extra care about keeping our compositions interesting and textures varied enough to allow the proper elements to pop.
The ground plane was actually the hardest thing we had to deal with. None of us knew how it should look until very late in the game. When we did end up figuring it out, textures alone didn’t end up being able to hold up close to camera. We went through and modeled a lot of the ground planes by hand on a shot by shot basis.
Seeing that Psyop wanted to separate itself from fabricating anything deemed “too CG,” what aspects of each production was done by simulation? Moreover, was there any scripting that went into the execution of each spot?
They look like they are hand modeled because they were! Everything was hand modeled except for a very few elements. One of those being the crowds in “Gladiator” and “Circus”. The other being the ocean in “Storm”. The crowds were actually animated in Flash, brought into maya as a curve sequence, converted to a plane, and extruded.
I wrote a few custom nodes in C++ to facilitate this, and the process went smooth once we got a few of the kinks worked out in the beginning. The marbles in “Wild West” as well as the dust and debris in all the spots were created and simulated by our FX artist Miguel A. Salek.
“Storm” was the only spot that we really had to change up how we executed it. The ocean wasn’t modeled at all—or it wasn’t modeled traditionally. For the ocean we created flat planar geometry in XSI and using a great plugin written by Andreas Gebhardt, we turned these into ocean surfaces.
We then baked these surfaces out to point caches and brought them into Maya. From there, I wrote a few Python tools to extract edge loops and convert them to closed curve sequences for each frame. We then took those curve sequences, fed them through a couple of C++ nodes that I wrote, much like what we did for the crowds on the earlier spots, and they would output the cardboard geometry you see.
From one curve sequence piped through the nodes, it would output two planar cardboard sides with the corrugation piece in the middle. We would retain the history from the curve sequences and our animators could go back in after and push and pull the curves around to make sure the oceans were reacting properly to the current action.
Overall, what was the strategy for your lighting setups, and how much of it was done in 3D compared to a compositing package?
The lighting turned out really beautiful. Arisu Kashiwagi provided great style frames for reference on the direction we should go with our lighting. Using those reference frames, Andreas Berner, Helen Choi, Anthony Patti, and Shu-Chen Lin came up with some beautiful lighting setups to best portray the look Arisu and Eben established.
Our compositors and lighters work very closely in order to get that final look. It’s hard to really pinpoint how much comes from either team. I’ve heard that at some studios a lot of the times compositors just put together what 3D gives them. At Psyop, I feel like the compositors and the lighting artists really influence each other; it’s very much a team effort.
As the Lead Technical Director for the UPS campaign, was there any point when you where not able to fully reconcile the creative vision with the technical realities, and furthermore, how has this project matured your relationship with Creative Directors and Producers who may not fully grasp the technical demands of your job?
Honestly, I think we were able to really deliver the creative vision pretty dead on without having to sacrifice anything. In fact, “Storm” in my opinion ended up exceeding the initial style frames that were worked up.
Working with Eben is great, he comes from a technical background, so he really understands that sometimes what he or the client is asking for is going to be very complicated. He is always involved in the discussions when we are coming up with potential solutions to tackle a technical hurdle and very patient when getting results is taking a bit longer than expected.
Lydia Holness, our producer, also is very good at gauging how technical certain things will potentially be. I really didn’t feel any pressure from either of them on this job—if anything, I was given a lot of support in coming up with the solutions we ended up going with. “Storm” put quite a bit of stress on all of us before we figured out exactly how were going to create an ocean out of cardboard… It’s quite a challenge!
It really went very well though, and I can’t say there were any issues with Lydia or Eben not understanding the technical demands of these jobs.
We all know we’re not supposed to play favorites, but generally speaking, which spot from the entire UPS campaign is your odds-on-favorite, and why?
Honestly I love different things of all of them. If I have to pick an absolute favorite, for me it’s “Storm”. “Storm” and I had a very interesting relationship while we were creating it. It was very love/hate I think for all of us. It was a very big personal accomplishment technically so in the end it would be my favorite spot. I really love different things in each spot, it’s really hard to pick an absolute favorite.
In particular, what was the most rewarding aspect of this job for UPS?
Sharing the experience with the team and in the end having 4 really great spots to show for it. I’ve been working with most of the people on the team for almost 3 years now, and I really like to see what each person brings to the project each time.
Working on these spots was a lot of fun, It was a really great project to be part of.
In serving the dual role as Project Lead and Technical Director, is there ever a struggle in keeping all of your artists on the same page, while also projecting a sense of calm and leadership that the team can take comfort in?
Honestly, not really. I’ve been working with the core team for quite a while now and I think we all work really well together. Everyone knows what’s expected of them, and when we do get someone new on the team, everyone helps them and they typically get into the groove fast.
It can get hard juggling the responsibilities of a project lead and lead technical director sometimes on these big jobs—or even small jobs if they are highly technical. I see my role as someone who has to jump in and fill gaps. Sometimes that is rigging something, writing a tool, I’ve had to light shots, model elements, texture, it really depends on the situation.
“Storm” was a good example of that. Luckily it happened to be the last spot we delivered, but I really had to zone out and get through a lot of code/tests in order to come up with the system we ended up using. During that time, it was really hard to juggle everything, but thankfully the other spots were wrapped up, the team was fairly small and Lydia totally understood and helped me out.
On an extensive, taxing, four spot campaign, what do you do to stay creatively inspired, as well as the methods you employ to keep the artists that you are leading equally inspired? On top of that, do the technical demands of your TD role ever leave you with a creative deficit?
I really don’t think I need to keep the artists inspired, they are artists and I think they find their own ways to inspire themselves. Either reading Motionographer and seeing all of the amazing stuff that is coming out, or even seeing what other jobs are going on at Psyop will motivate us to really try and excel.
I think creating something we can be proud of is our main goal; that’s why we do this. There are plenty of other careers we could have chosen where we can make lots of money, or be highly respected or get whatever it is a person looks to get out of their career. I feel like people that get into this career and take it very seriously don’t need me to motivate or inspire them, they have that themselves. That’s why they are where they are, and they are going to work really hard to create something they can be proud of themselves.
As far as being in a TD role leaving a creative deficit, I think that being a TD is a very creative position. Creativity is not just being able to paint a pretty picture, or model the best looking orc character; it’s about coming up with interesting ways to solve a problem or convey an idea.
Being a TD on jobs like these can sometimes be fairly straight-forward and a bit tedious, but more times than not it’s about looking for a creative way to solve someone’s problem or deliver the director’s vision. It could be finding a faster way to do something that 6 months earlier took us 5 times longer to do, or figuring out how to make an entire ocean out of cardboard.
I find being a TD and solving problems is very satisfying and rewarding. Although, like I said above, I do get to still put my hands into other areas and get away from the technical side once in awhile. As a project lead I need to be a person that fills a gap in the team if I can’t find someone else to do it. So I do get to light, model, or texture sometimes if that is a role that is needed on my job. But I do think that being a TD is a highly creative position if the results you achieve being a TD is something that appeals to you.