Back in the day, video game cut scenes were a patience-testing ordeal. If a character had to deliver exposition, you'd usually just sit there, staring at a soulless polygon character model, reading text which scrolled slower than an obese snail. Audio recordings of dialogue were literally unheard of, while having the character's mouth move was the stuff of your wildest malarial dreams.
Today, in 2014, we not only expect our video game characters to speak and look like humans, but to do it to an increasingly photorealistic degree.
Before the on-set of motion capture, much of these talking animations had to be animated from scratch. This was a time-consuming job, and it didn't always result in a natural looking result. Now, however, sophisticated #technology can deliver instantaneous results which almost defy belief.
Take, for example, the new generation of motion capture technology from 3lateral and Cubicmotion. These tech companies are the brains behind the facial animation of games such as Ryse - and the incredibly dialogue heavy - Grand Theft Auto V.
In a new video, a voice actor becomes a de facto actor as their face is transferred into a movie computer model nearly instantaneously. Check it out below:
How Does Motion Capture Work?
The explosion of behind the scenes footage has made tight-fitting spandex suits covered in golf balls a fairly familiar sight these days.
Traditional motion capture involves placing an actor in an open space which is surrounded by cameras. These cameras are designed to track the movements of special balls which are strategically placed on an actor's body - usually at their joints. When these balls are filmed, and the connections between balls established, a kind of computerized puppeteering skeleton results. Animators can then use this skeleton to provide movement to whatever wacky creation they can can apply to the skeleton.
The above method is perfectly fine for capturing broad movements of the body, but what about motion capturing the emotions shown on the human face?
That's much harder, as the display of anger, fear, joy or sadness often relies on subtle muscle movements that can be hard to imitate. This also adds a more complex technological element as high-resolution tracking is required to capture all the nuances of the human face.
In the past, markers - similar to those placed on the body in general motion capture - were used for facial motion capture. However, this could require over 350 markers to be placed on the actors face - which is a laborious and time consuming process.
Furthermore, it does not often result in natural looking emotions, with many characters' facial movements seeming overly-exaggerated or cumbersome. Examples of films which have utilized this method include The Polar Express and Beowulf - two title's criticized for the creepiness of their animation.
Now, the preferred method appears to be markerless technologies, which are both easier to use and more effective. Instead of using artificial markers, markerless tracking locks on to the natural features of the face - such as nostrils, the corners of eyes and lips and wrinkles - to create a customized tracking map for each face. This technology is less cumbersome and allows for a much wider range of emotion.
However, only about 90% of the recorded material is acceptable without editing, which means small tweaks are needed to fix up minor issues and make the expressions more realistic. Luckily, CubicMotion has got that covered, as their program also allows for manual manipulation. You can see some of this in action below:
Currently, CubicMotion offers their services to both the film and gaming industry, and claims they can even produce incredible renderings purely from audio samples.
Although these facial motion capture demonstrations are certainly impressive, I think there is still some way to go. I don't know about you, but I'm still getting a hint of the uncanny valley about these things.
These talking animations are...