FuseFX populates dozens of TV Shows with Golaem!

Check out how FuseFX populate dozens of TV shows with Golaem extras. Crowd TD Adam Broad shares some insights and breakdowns with us.

Can you present yourself and the studio?
My name is Adam Broad, and I'm currently a Crowd TD at FuseFX. I oversee all crowd work that comes into both the Los Angeles and New York offices and act in a supporting role for crowd work done in the Vancouver office.

On which projects did you use Golaem?
Working in Episodic Television allows us to utilize Golaem on a wide variety of shows, including "Veronica Mars," "Castle Rock," "Miracle Workers," "9-1-1," and "S.W.A.T." We've also provided crowds for Disney's new Epcot reveal and a host of shows that we're currently working on or haven't yet aired.

What is the size of the crowd team for a given project?
It really depends on the show. If we need to build new characters, we'll usually have between five to seven artists involved in the process of creating and prepping them. We have three body types and a male/female version of each so any clothing we make must be fitted to each body type, six times. Each character also has five different ethnicities included as different rendering types, so you can add less or more of specific representations to match the needs of a shot by simply adjusting a slider. As you can imagine, this takes a significant amount of prep work, especially when you're trying to pull off the illusion of diversity in a large crowd. For instance, our "Veronica Mars" characters had nine shoe types, fifteen pairs of clothes and swimwear, a vast number of hats and glasses, and around 25 different hairstyles. Each of these items was prepared with up to a dozen different texture variations as well. This only scratches the surface of prep work a crowd shot might have to go through to get a tremendous amount of visual complexity and uniqueness.

Can you give an overview of the number of shots you did and the type of shots (stadium, battle, city, any other…)?
Due to how quick the turnarounds for episodic and streaming television are, we've pulled off crowd work for just about any type of situation. Multiple stadium shots for “S.W.A.T.” including an evacuation sequence, a giant beach party with volleyball and football games taking place for “Veronica Mars”, mind-controlled townspeople for “Castle Rock,” charging medieval warriors for “Miracle Workers,” and a whole range of animal and insect crowd simulations for a few shows that have still yet to air. The number of crowd shots for any given show fluctuates but is usually relatively low unless the crowd in question is a story-driven piece. “Castle Rock's” mind-controlled townspeople were a big plot point, so we ended up doing 13 shots over two episodes for that.

Sorry but your browser does not support HTML Canvas Breakdown from Veronica Mars
Sorry but your browser does not support HTML Canvas Breakdown from SWAT

What is the typical number of characters you use Golaem for? 
The smallest number of crowd characters I've used on a show is about eight, for a big explosion outside a train station that needed a few extra people mixed in among the practical plate. They execute a ragdoll simulation and then get lost in the flames. As for the biggest one, we created around 13,500 for “S.W.A.T.,” filling up the Long Beach Auditorium for a boxing match. It's difficult to say what the typical count for a crowd is, as it really does depend on the show and the specific needs. Our capabilities allow us an unlimited crowd size, with the main factor being visual diversity and behavior.

How close could the Golaem characters get from the camera?
We've done shots with Golaem where we had characters as close as 20 feet from the camera, which is really a testament to how good the team at FuseFX can get these characters looking. That being said, we always prefer mixing crowd characters with real people in the plates as a buffer. It presents its own challenges because the digital characters have to match with the shot clothing, textural/color variation, behavior, hair, and accessories. Still, when we pull it off, it always looks better and is usually something I am quite vocal about in the run-up to a show during pre-production.

How do you feed Golaem with motions? Did you build a library of motions you reuse in multiple shows?
One of the best parts about building crowds at FuseFX is we own and utilize multiple XSens motion capture suits, giving us the flexibility to create our own motions in-house and on-demand to use on any show we want. When a show comes in, I will usually sit down with the Global Head of 3D David Blumenfeld and have a conversation about whether or not the suits need to be used. If we determine they do, we'll come up with a list of motions before we go into a mocap session and start shooting. We will also decide whether or not we need several variations of a specific motion and if that motion needs to be loopable. The suits give us a lot of freedom to capture just about anything we want, including data for props like weapons or tools. We have even gone as far as to implement a variety of custom rig integrations so that all of our in-house biped characters can be used not only with our mocap system but easily integrated into the crowd system as well. This provides a considerable amount of flexibility and speeds our turnaround significantly. We've used mocap on around half of our crowd shows, while we've been able to re-use motions on other projects, which has sped up production in many areas.

Could you give a bit of details about the typical crowd work involved on building a stadium shot for example?
In most situations where we build a stadium shot, we'll have an artist pre-model the seating for the stadium, depending on the camera angle. They'll construct the seats (which we can use later for mattes) as well as a polygon that sits just on top of it. From there, we can combine all the polygons, create a Navmesh, and then build a Poptool by component. In this way, we get one person per seat, allowing us to fill an entire stadium in under an hour. We will typically split a stadium up into sections via tiered seating for rendering as well as a foreground pass if the shot demands it. We'll also usually add a locator in at the stage level and register that in the Poptool "look-at" section, so right off the bat, the crowd actors are all facing the stage appropriately. Once the characters are in the scene, we tend to incorporate an 'adapt position' behavior, which allows us to move characters in XYZ space to fit them better to the seating.

Sorry but your browser does not support HTML Canvas Breakdown from Castle Rock
Sorry but your browser does not support HTML Canvas Breakdown from SWAT

How did the Layout help you make the shots faster?
The layout tool makes working in Golaem far more versatile. I've lost count of the amount of times I've had notes come in on shots for changes that have been able to be quickly addressed using these tools. Being able to quickly swap out clothing or remove and add people in certain areas, or even alter motions
including hand gestures using the rig options after a simulation has been cached is incredibly useful when it comes to getting shots out the door in a timely fashion. It also means you get more room to breathe while running an initial simulation; as a result, it isn't set in stone. The layout tools can carry a
lot of weight further down the line if they need to, making this a very useful addition to the toolkit.

Is there any part of a project you are most proud of or which prevented you from sleeping?
I recently completed a show with some complicated maggot and beetle sims airing in June, that came out looking really great. It was also our first foray into insect simulations using Golaem and was done using Golaem's built-in UV pin system on matchmoved geometry. On another project, we implemented Ornatrix hair and fur in Golaem for a host of upcoming creatures. As for shots that have kept me up at night, I think every show has its issues that occasionally crop up. Working in episodic television, time is always working against you. On many projects, it's not unusual to get less than two weeks to deliver an episode, so coming up with a firm plan in mind for a crowd ahead of time and sticking to it is something we always try to do to streamline the process as a show progresses.

Anything else you would like to add?
Golaem has been an excellent tool for us at FuseFX, and we've been able to develop a highly robust, flexible, custom crowd pipeline around it. Whether it's adding thousands of people to a stadium or just a handful of extras to fill out a scene quickly, we've used it for all sorts of purposes. The toolset has become tightly integrated into our pipeline thanks to our amazing team of riggers, character artists, and the XSENS motion capture suits and Manus gloves, giving us a tremendous amount of flexibility. We're able to go from nothing to a fully populated scene with thousands of characters, not in a matter of weeks, but a matter of a few days. It's, without a doubt, one of the main reasons our clients keep coming back to us for this type of work, bringing new and exciting shots of ever-increasing complexity. No crowd is too big, and no challenge too great for us to tackle!

Learn more