Putting the character in animation

Putting the character in animation
Craig Chapple

By Craig Chapple

November 26th 2012 at 2:00PM

Develop discovers what's in store for the future of character animation

Character Animation has come on leaps and bounds during the current generation of consoles. Games such as L.A. Noire and Heavy Rain have shown developers and users the near-realism that can be achieved using advanced and impressive facial capture and animation techniques.

But despite making an impressive jump, advancements in this sector have slowed as we approach the seventh year of the longest console cycle in history, with developers tasked with getting the most out of hardware released in 2005.

Technology and memory has advanced greatly since the PS3 and Xbox 360 were launched, and although they are still impressive powerhouses, top range PCs easily outstretch these aging systems.

And whilst mobile and tablets have scaled up their capabilities at an impressive pace to offer console quality in terms of graphical fidelity, they are still now only matching limited console quality, not yet surpassing it.

A new era

But with the next generation of consoles seemingly close to hitting us – with Nintendo’s own contender the Wii U to launch at the end November – a new wave of technology is set to hit the game industry, with current technology set to receive a much needed boost to take character animation to the next level.

Epic is gearing up for the release of Unreal Engine 4, with its own title Fortnite harnessing the powerful development platform, and Square Enix has also signed a deal to utilise its features.

NaturalMotion is readying the release of Morpheme 4.0 in the near future, which is set to include a toolkit for building prediction models of an animation network, allowing a game at runtime to ask questions of its animation system.

Meanwhile, Unity is readying the release of Unity 4, which Unity animation technical director Robert Lanciault says will let developers create “out-of-the-box” interactive animations within minutes, as well as visual tools, live links and preview windows to give as smooth a user experience as possible through every step of the animation process.

One of the current trends directing the new breed of character animation technology, says The Creative Assembly’s motion capture manager Peter Clapperton, is pre-visualisation.

The man heading up the Total War developer’s brand new motion capture studio in Horsham – the largest developer-owned mo-cap centre in Europe – believes pre-visualisation is the way forward for the space, with an marked increased in demand for quantity and quality of animations from users.

“We felt there was a need to expand the mo-cap studio and services we provide in order to reduce clean-up time and provide us with the potential to pre-viz work through the real-time streaming process,” he says, of the motion capture studio which opened in August this year.

“Having a permanent facility is already proving to be extremely effective for prototyping a lot of work before hiring fight performers to come in so we can be more efficient and effective on the days of final capture.”

Another driving force in character animation technology is that of the relationship between AI systems with physics and animation. NaturalMotion CTO Simon Mack says there is a clamour for characters to appear smarter, requiring smoother and more realistic animations when obstructed by other NPCs and obstacles.

“One of the key trends we’re seeing at the moment is a desire for characters to appear more intelligent in games, from the way that key characters interact with the environment to the motion seen in crowds,” says Mack.

“As well as needing to create great animation networks, there’s a requirement to be able to more easily control these from AI systems. A few years ago there was a lot of talk about the interactions between animation, AI and physics.

"Physics integration has largely been solved, or at least understood, and it seems that now, studios are really calling out for the AI-animation interaction to be addressed.”

Counting the costs?

Havok behaviour and animation lead Jason Gorski and Epic Games lead animator Jay Hosfelt meanwhile state that iteration speed and the driving down of costs is another important aspect of the new era of character animation tech.

Hosfelt states that offering tools and methods for animators to iterate their work quickly is key, with the need to re-watch an animation multiple times to create the perfect look, making a speedier process vital.

“For gameplay animation, it’s very important to give animators the ability to tinker with blends and transitions in games and do it with a greater focus on iteration speed,” says Hosfelt.

“Animation involves watching something over and over until it looks and feels right. If you give the animator the ability to perfect these things and to iterate quickly, the quality of your games will only improve.

“There are numerous rendering solutions that will also showcase animation. Better skin shaders, lighting, and deformation options will enable game animation rigs to look more like their film counterparts.”

Gorski adds many studios are pushing more motion capture technology in an effort to reduce iteration costs, and looking for ways to automate the use of data to get up and running quickly, while others are taking on a more procedural animation approach.

It is this drive to reduce expenses, and with the likes of Unity and Mixamo driving sales costs down and targeting indie developers as well as big triple-A powerhouses that is democratising the character animation space.

Earlier this year Mixamo opened an animation store on Autodesk’s 3ds Max 2013 and Unity’s Asset Store. This month, the firm also made its library of hundreds of characters and thousands of animations available for an annual fee of $500, allowing developers unlimited access to all of its assets, while also not having to leave their favoured toolset and go to a separate site.

Character Animation for everyone?

Mixamo is also developing a brand new modelling tool named Fuse in collaboration with Stanford University, which, although still currently at an experimental stage, it is planning on releasing in January. The tool is designed help alleviate the problem of high-costs versus high quality in 3D character modelling. It will be free to use for developers, with users only paying for the assets they want.

It is efforts such as this, as well as the rise in power of relatively easy access platforms, such as iOS and Android, that means developers of all levels are able to harness previously expensive tools and animations for a fraction of the price – with the extra effect of increasing the quality of smaller titles in graphical terms.

“The overall plan is through technology enabling more people to create content,” explains Mixamo CEO Stefano Corazza.

“Instead of requiring specific technical skills and five years training in Maya or Max, these new tools and services allow anybody, even coders, to create beautiful characters and animations, and bring them to life in their game.

“We are massively broadening the potential userbase because there’s one level of obstruction that we created where people can use very high-level controls to create content without knowing even what a polygon is.

“So that’s the big shift, allowing the democratisation of character creation and animation where anybody will be able to create the content they like without needing any technical knowledge about it.”

Out of reach

NaturalMotion’s Mack says however that even though tools such as its own animation tech Morpheme is pushing forward democratisation in the sector, full state-of-the-art character animation is still out of reach for smaller studios in terms of making their own tools to use, but praises the wealth of options available from third-parties.

“Full featured character animation systems are large and complex things,” explains Mack.

“I don’t see smaller studios building them from the ground up. Fortunately, there are various mature middleware solutions available, including Morpheme, that are as accessible to smaller teams as they are to the big studios.

“Used properly, middleware engines give these teams a platform they can use to build their own innovative technology on without having to invest dozens of man-years in tools development. We certainly see smaller teams using and extending Morpheme in more inventive ways than many established studios today.”

Colin Urquhart, CEO of Scotland-based human body 3D and 4D surface image capture outfit Dimensional Imaging, which has done work on games such as FIFA 13, says that it is the large and spiralling costs of facial animation, particularly when creating your own solutions, that is one of the challenges facing the space.

He agrees with Mack that third-party middleware solutions and animations are perhaps the most cost-effective way of delivering quality character animation, particularly in the area of facial capture, but also stresses that with the rise in realism, the quality of acting must also rise.

“One of the key challenges that the industry is facing is the spiralling cost of continually improving the quality of facial animation,” says Urquhart.

“How can studios create hours of facial animation at ever higher levels of detail, quality and realism within reasonable budgets?

“I think the answer has to be to use technology like Dimensional Imaging’s to accurately capture the performance from real life talent. However, this then introduces the challenge that as the realism and fidelity of the animation improves, so the quality of
the performance must also improve to match.

"Therefore, I see the secondary challenge of obtaining good acting performances becoming increasingly important to the industry.”

Havok’s Gorski says another challenge facing the industry is combining automated processes with traditional animation. He says bringing these two aspects and techniques together could mean more personality and a sense of realism – even when given a certain amount of artistic flair to the game’s style – helping to draw in players clamouring for more personality in game characters.

“I think there will continue to be a tension between more automated processes – procedural animations, cloth simulation, IK solutions – and traditional animation as players seek more dynamic characters, but also desire strong artistic flare,” he says.

“Studios will need to find ways to leverage and combine both effectively to create really great characters. That’s where Havok Behavior’s modular design can really help out. Studios can customise the tech to suit their particular artistic vision, incorporating in-house tech when desired, while leveraging a powerful suite of out-of-the-box features.”

Breaking realism

Perhaps one of the greatest challenges that has faced character animation however is that of the uncanny valley, where facial animations that come close to conveying realistic human likeness, but still look slightly off, subsequently causing an uneasy feeling in the user.

Despite the likes of L.A. Noire coming close, games still struggle to convincingly break out of the uncanny valley.
Urquhart says that the issue is still as real now as it was in previous years, and with the rise in consumer expectations of graphical fidelity, particularly with the imminent arrival of next-gen consoles, the danger still exists of characters falling into the uncanny valley without realistic animations applied to them.

“I think that the uncanny valley is as meaningful in 2012 as it ever has been,” he says.

“In fact, as characters become ever more realistic, there is more of a danger of them dropping into the valley if insufficiently realistic animation is applied to them. This is one of the reasons why there is currently such a strong focus on achieving more realistic facial animation.”

Mack however believes that while there are a few examples of characters looking so realistic that their slight flaws were easily noticeable among players, the more concerning issue with achieving realism stems from unrealistic movements of character models across a game’s landscape and obstacles.

He goes on to explain that the gap between realistic appearance against motion has risen considerably, with characters often seemingly sliding across the when walking on flat surfaces and obstacles such as stairs, rather than conveying realistic movements.

“It’s surprisingly common to see beautifully modelled and rendered characters sliding across the ground and up stairs, walking and turning on the spot, failing to react to collisions with their environment and so on,” he states.

“There’s a huge gulf between the realism of character appearance and the credibility of their motion in most games. I find it interesting to watch somebody else playing a game and compare the performance of the characters on screen with those in an animated movie.”

Epic’s Hosfelt offers potential solutions for developers looking to avoid the traps and pitfalls of the uncanny valley, and advises there a couple of methods that can be successfully deployed to avoid its traps, highlighting the potential of uniquely stylised characters or animating on top of motion captured performances.

“There are two big ways to avoid the uncanny valley,” he explains.

“One way is to have stylised characters and animation, where you’re caricaturising reality. The second way, which pertains mostly to performance capture, is having animators go through and push it beyond what was originally captured.

“Even if you have the best capturing technology, you still need animators putting the final human touches on a captured performance. Facial capture will reproduce a smile, but it still takes an animator to make it a sarcastic smile or a bashful smile.

“It’s still very expensive to overcome the uncanny valley for video games, but I predict that one day that there will be better capture techniques, along with a faster way for animators to layer on top of a captured performance and put the final human touches on it.”

Next generation animation

It’s clear, then, that character animation still has its own hurdles to overcome before the uncanny valley can become a distant memory in our minds, but with the new wave of next-gen consoles, as well as fresh technology from the likes of Unity, NaturalMotion, Mixamo and Epic upon us, that future may not be too far away.

The new era of technology is also providing unheard of democratisation to the sector, meaning small studios and indie developers are also fast catching up with the ability to display realistic animations, although the truly real may still be out of their grasp economically.

As Hosfelt sums up: “As games use more complex state machines, more sophisticated animation trees and better AI, we will more readily believe that our characters are real, with real physics behind them. We’ll start seeing more procedural solutions like what we’re seeing with Euphoria and ultimately animators will have better tools to make all of this happen.”