Leading mocap studios talk to Develop about leaping the Uncanny Valley, Pt 1

Talking Heads

Across the game development industry most disciplines are advancing at a great rate, all the time propelled Moore’s infamous Law. The momentum needed to satisfy public thirst for new tech magic takes a huge effort, but expectations are consistently met.
The exception to that trend is in the facial animation and motion capture sector, where the threat of falling into the Uncanny Valley looms large. New technology and techniques exist in ample quantities, but the closer the sector moves to delivering convincing human movement, the more potent flaws become.
Facial animation and motion capture is a field still in a fledgling stage, where numerous approaches thrive. Each has merits and failings, from audio-driven models and traditional marker-based performance capture to video capture and painstaking hand animation.
Regardless, the facial animation market alone is enormous, and growing all the time.
According to data provided by Image Metrics, in 2006 the sector was worth some $807 million worldwide. Thanks to the growth of the number of movies and games requiring the same tech, that market is expected to be worth a quite staggering $1.9 billion in 2011. While film accounts for much of that sum, in 2010 games promise to offer facial animation companies $150 million in business.
It is clear that there are plenty of developers keen to include high-end facial animation and mocapped performance in their projects, and a public hungry for the results. The real challenge, then, is not in building a market, but adopting the right technology and technique to guarantee the best results.
While some models offer memory cheap, expedient and low-cost solutions, others deliver expensive, headline-grabbing quality that tempts gushing reviews and the subsequent sales boost.
“In my opinion, entirely automated facial capture is still unrealistic on most medium scale jobs,” suggests Ian Jones of director RealtimeUK, which works with a number of motion capture firms to create pre-rendered game trailers and marketing movies.
“They accomplished it on Avatar, but it took literally years to get to the stage where raw data could be plugged straight into the facial rig with believable results.
“Although mocap data will get you fairly close the original performance, all the data that we have dealt with still requires a degree of artistic interpretation and those extra little tweaks that really bring a character to life.”
Spend any time with the staff at the studios leading developments in facial animation, and it’s clear that many others echo Jones’ sentiment that no single approach offers an ultimate and unbeatable solution. In short, choosing the right approach – or combination of approaches – for your budget and schedule is more important than plucking the technique that boasts the most technical muscle (see: ‘Guess Who’).

“When using motion capture verses hand animating, how the data is captured is secondary to providing a well planned film-like shoot and delivering a high quality result on time, and on budget. Today there are so many good methods of tech,” confirms Imagination Studios MD John Klepper
“The difference is in time spent, attention to detail, and skill in data handling post shoot, coupled with support during and after delivery. Most of the time our clients are battling with extremely tight deadlines, and a technology that won’t offer the fast turnaround they need just won’t do the job. For a recent triple-A title, for instance, we produced over 20,000 seconds of body and face animation, fully edited, in less than
two months.”
The demands of Klepper’s client base are typical of those facing facial animators, and are the driving force behind the evolution of the technology in the field. An apparent pioneer spirit is clear when talking to the likes of Image Metrics, Imagination Studios and Audiomotion, and its evident that energy is leading to a wealth of research and development and experimentation.
Oliver Bao, head of research at Depth Analysis, which is currently at work on Rockstar’s thriller LA Noire, has a driver’s eye view of where facial animation is heading, and is quick to point to the hurdles it faces.
“A real challenge facing facial animation right now is getting really high quality normal maps and them actually performing realistically,” he professes. “I know that my peers in the industry have been trying to do video framerate captures, with all the high-res normal maps, but keeping all that stable, and the stereo correspondence of the geometry in place, and the temporal correspondence; that’s quite difficult.”

The challenges facing the facial animation and mocap studios are certainly plentiful, and as service providers have to a cater for a continually more diverse and increasingly demanding range of customers, there’s an emerging need across the sector to greet clients with not only a barrage of information, but a simple honesty that hammers home a message that those who sign the cheques rarely like to hear.
“The real challenge is staying adaptive to the needs of the individual client and working within their constantly evolving pipeline, and ever-limiting timeline,” reveals Klepper. “That and convincing potential clients that, as clichéd as it may seem to say, with so many inexpensive solutions available today; you get what you pay for.”
Yet even the notion that more money equals better quality doesn’t quite cover every intricacy, and subsequently there’s still plenty of space for the more traditional creative disciplines. While that sounds like good news for those schooled in long standing techniques, such as hand animation, the constant pressure of budget means that like so many other skills, the work is being handled in markets where more generous economies mean less troublesome overheads.
“Aside from the costs involved, in either the infrastructure in terms of hardware, or buying services, the increasing desire for more believable characters as a device for potential emotion content is driving people to chose hand animation from cheaper labour markets,” suggests James Norman, lead mocap animator at Rocksteady.
Outsourcing will always court controversy and the odd sly glance, but in reality it is a means to an ends, and has provided developers with a vast and skilled new pool of staff. Such is the burden placed on those charged with bringing a humanity to modern games, that an international effort may be the only way to meet the market needs.
If that weren’t enough, there’s also a collective responsibility not just to perfect techniques, approaches and business models, but to push the ability of facial animators even further.

“Catching up with the expectations of the public in terms of graphics and visuals is something the game industry is still struggling with,” warns Klepper. “Games are judged nowadays in terms of presentation as much as content and tech, so ignoring the ever increasing demand for high quality graphics and animation is not something developers can afford to do anymore.”
It isn’t all negative though, as the capacity for innovation within facial animation and mocap far outweighs the challenges.
“I think that as game engines and the power of consoles increase in the next few years it can only go to allow, or necessitate extremely high resolution capture, where every muscle twitch and deformation can be remapped at run time,” muses Norman, who paints an exciting picture of the future.
“As this happens perhaps there may come a time where we should go back to the Valley and step right across. There’s certainly a trend to try to get players more emotionally involved with the gaming experience and one way to do this is to create situations or encounters that are believable enough to draw one in.”

Part 2 of this feature can be found here.

About MCV Staff

Check Also

Blockchain and NFT games are banned on Steam, but not the Epic Games Store

Valve and Epic are taking drastically different approaches to blockchain and NFT games on their storefronts, as Valve has moved to remove them from its platform.