Sponsored By

10 Tips: Getting the Most from Motion and Performance Capture

What makes the difference between a successful motion capture shoot and a mountain of useless data? What techniques allow you to get the most out of your actors? This article offers 10 tips from Bungie, Rebellion, and more.

Tristan Donovan, Blogger

July 4, 2012

9 Min Read

[In the second of Gamasutra's discipline-specific series gathering important tips and techniques direct from developers, Tristan Donovan delves into motion and performance capture. You can read his prior article on audio, here.]

Motion capture technology has come a long way in the past decade. Cameras are cheaper, the software is better, and even the challenges of tracking eye movements are being overcome. At the same time, motion capture has also become more commonplace in games -- so much so that it's more common to see motion capture than hand animation in most big-name console titles.

That's all well and good, but what makes the difference between a successful motion capture shoot and a mountain of useless data? What techniques allow you to get the most out of your actors -- and are there times you shouldn't even use trained actors at all?

In this feature we ask the motion capture pros behind games such as Halo: Reach, Fable III, Total War: Shogun 2 and Sniper Elite V2 for their nuggets of mo-cap wisdom.

1. Prepare, prepare, prepare

If there's one single piece of advice that everyone involved in performance capture agrees on it's that preparation is everything. "Of all the advice I have for successful motion and performance capture, this is the most important," says Kurt Nellis, the technical cinematic lead at Bungie.

"Before you set foot in the room with your talent suited up and the crew looking at you expectantly, know what you are shooting. It saves a tremendous amount of time, as you won't be caught with your pants down because you don't know the answers to the questions everyone will undoubtedly have on set."

Bungie's Halo: Reach

Pete Clapperton, motion capture manager at Creative Assembly, the studio behind the Total War games, adds that it's good to set out the rules of communication in advance, too. "During the shoot there's normally a lot of people, and it's always a good idea to make sure everybody knows exactly what role they are playing on the day, because you don't want three people giving directions. The main roles are the person working the system, the person watching for any markers that fall off, and the director -- there are other people, but they need to know when to shut up."

2. Save time with animation

When making Sniper Elite V2, Rebellion Developments waited until the game reached first pass standard before turning to motion capture. "This allowed us to fully explore the style of the animations and react quickly to the evolving design and code requirements as the project progressed," says lead animator Zsolt Avery-Tierney. This meant time and money wasn't wasted on collecting redundant or incompatible motion capture data and that the team entered the shoot with a clear idea of what they wanted.

"Everything in Sniper Elite V2 that we gained from our actors was initially defined and designed by our animators. This not only served better communication on the day but also ensured our animators retained ownership and remained devoted to the final delivery," he explains.

Another benefit was that because the content was already well defined, Rebellion could tell its mo-cap studio its exact prop requirements and measurements in advance: "This maximized time usage on the day, drastically reduced asset waste, and relieved the third pass burden on the animation team."

3. Think about real-world solutions

"Many times in this industry, people try to come up with virtual solutions to real world problems," says Troy McFarland, motion capture lead at Bungie.

"Often, a simple change in a costume, prop or set will enable us to capture better data and make it easier for the talent to get into the scene and act. On the early shoots in Halo: Reach, I noticed that the hands in the mo-cap data weren't rotating as much as they were in the video reference of the actual talent.

"On closer examination, I found that our suits were causing the problem. They were one piece from shoulder to knuckle. I had our mo-cap suit vendor create a custom suit for us that had shorter sleeves and gloves with extra fabric where we could place the wrist markers. The result was dramatic and a nice side effect was that the suit could be worn by a wider range of talent."

4. Actors don't need previous mo-cap experience

Freelance performance director Kate Saxon, whose credits include Fable: The Journey and 007: Blood Stone, says many assume wrongly that acting for a game is different from film or stage performances.

"Actors are trained to adapt their skills to the medium they are working in and there's a misconception in the games industry that games are somehow very different and you can only use actors that have already done performance capture, which is utter nonsense," she says.

"Sometimes people assume that for motion capture they need a physical performer, but actors are inherently physical performers, who are used to doing a whole performance using their whole body, face and voice."

5. Don't dismiss full performance capture

Jimmy Corvan, business development manager at motion capture studio House of Moves, says that when it comes to capturing cinematics full performance capture is worth the extra effort and expense -- so you can capture the actors at their most human.

"The traditional way is you send a voice actor into the booth, and they recite the lines three or four times until they get the delivery perfect, and then you take another actor onto the motion capture stage to act it out until it's perfect," says Corvan.

"And it's then up to the animators to blend everything together so that the scene looks perfect," he says. "But humans don't operate on the perfect level, and what you get in full performance capture is the small nuances of imperfection that actually make us human, such as the draw on the words the actors say as they become more exhausted."

6. Make wireframe props

While props can help produce better performances by actors, they can also get in the way of the motion capture markers. "Early on in Halo: Reach, we were using realistically proportioned and weighted guns," says Bungie's McFarland. "These posed two problems. First they blocked the chest markers resulting in longer clean up and delayed delivery of the moves to the team. Second the hand positions relative to the shoulder were off from those of our purely fictional rifles."

To tackle the problem McFarland fabricated a wireframe rifle for the actors to use that was based on the screenshots of the in-game gun taken from Maya. "One thing to note with this prop is that I incorporated a toy gun for two reasons. One: the toy gun already had a comfortable grip and not needing to fabricate a handle saved me a lot of time. Two: it had a working trigger and, as silly as this seems, it helps an actor to have a moving, clicking trigger when they work on a scene."

7. Not all actors need be actors

Depending on your needs, actors might not be the performers you need. For Shogun 2: Total War, Creative Assembly looked beyond the acting world for people to play its samurai warriors. "We required people who were relevant to the martial arts for Shogun 2, so we had Britain's Olympic Kendo team come in to do some stuff for us," says Clapperton.

This did mean a bit more direction was necessary, though. "We found that the way they did their movement and performance looked processed because it was very accurate and correct, whereas we were looking for people fighting in the middle of a battle, so we had to ask them to step out of their comfort zone -- but they were happy to take direction."

8. Be specific when directing

It's important to be clear what you want from actors, says Saxon. "I've been working on a game recently that I can't name, where I've been very involved with the stunts," she says.

"The character has that attitude where their chin juts out, and they are always vying for a fight, and I have to explain those kind of physical choices to the stuntman -- and that the way this character would fight wouldn't be like a trained military person, but like a street fighter where it's more messy and muddy, more erratic.

"The stuntmen are great at taking this on board, but the danger is that without somebody pointing that out, you can get into a fight when playing the game and suddenly everyone's behaving like trained fighters."

9. Trust the people at the shoot

"Ideally, everyone in the room, including the actors, is there because they are good at what they do," says Bungie's Nellis. "When a mo-cap technician comes to you with a problem, listen. They may have bad news about the best take you've ever captured but it's better to address it now than find out you had unusable data later. If an actor says they didn't feel right on that last take and they think they can nail it on the next one, you'll often be amazed at how right they were. You won't always be able to take their advice, but the crew will definitely appreciate that their expertise is being utilized. An appreciative crew is a crew that will work hard for you."

Rebellion Developments' Sniper Elite V2

10. Establish naming conventions

Defining a clear system for naming your motion capture assets improves communication, says Rebellion's Avery-Tierney. "In Sniper Elite V2 I knew my stances, motions, and directions, so my naming followed suit. This gave us a good focus and precision in our takes, made reviewing footage and making selections much easier, and stopped us getting muddy with overly long or complex naming," he says. "For example 'walk_forward_jump_to_mantle_walk_jump_back_down'.

"Although the terminology conveyed a lot of content, the take also contained many fidgets that I would have missed if I had not gone back to review it. If I had named the asset I would have called it 'StepUp_JumpDown_Fidgets_01' and allowed the ambiguity of the last term to encourage further investigation.

"If I had conducted that shoot I would have captured, or at least chosen, each action as separate assets. This is because that's how I would have used them on project, saving valuable seconds of processing cost, and increasing the efficiency of the animation team to find and turn around assets from their schedule."

Read more about:


About the Author(s)

Tristan Donovan


Tristan Donovan is the UK-based author of the book Replay: The History of Video Games and a freelance games journalist who regularly writes for The Times and Stuff. He has also written about games for Game Developer, Edge, The Guardian, Kotaku, The Gadget Show, GamesTM and many others.

Daily news, dev blogs, and stories from Game Developer straight to your inbox

You May Also Like