Creating game cinematics

Once the studio and the client agree on the proxy work, artists begin modeling the high quality objects, environments and characters for the final render. These are the ‘assets’, the guts of the production. While game developers and CGI studios both churn out initially complex 3D objects reaching into the millions of polygons, only the CGI team has the luxury of keeping these for the final animation. Game developers use techniques like normal mapping (where detail is taken from a high poly model, put into a normal map, then applied to a much lower detail model like a standard texture) to give the appearance of quality in models while avoiding a pixel hernia in the game engine.

“A lot of in-game assets start off as high-poly source models,” explains Tom Drew, senior artist on Fury. “They’re used to help render out textures, and are baked down into multiple low poly textures.”

Gerard Roche, production manager at Act3, sees a tangible benefit in both industries working with detailed 3D models in the beginning;

“Normally for in-game characters a games company will create high resolution characters then lower the res until they have the optimal model for in-game. If we’re lucky we can get our hands on the original model, saving us a bit of time. Usually we’ll receive models between 8K-15K polygons. Our final models will end up anywhere from 100K-200K. With offline rendering we can have heavy poly count models, apply a wider range of effects and composite over the top that can’t be done real time as it’s too heavy on the hardware.”



While plenty of creative spark goes into the modeling and texturing, it’s only half of the visual workload. The other half is animation. As far as techniques go, there are two main methods for animating objects in 3D space, the more traditional of which is ‘Keyframing’.

Imagine you want to animate a ball rolling along a floor as part of a 1000 frame sequence. Rather than manually moving the ball a fraction in each single frame, Keyframing allows the artist to put the ball at the start of its journey in frame 1, the end of its journey at frame 1000, and the software will calculate all the movement in between.

While that’s tidy for some sequences, what about more complex stuff like the human form? That’s where Motion Capture (or ‘mocap’) comes into play. Using small reflective spheres placed at strategic points on a live actor’s body, mocap relies on special cameras placed around the actor that pick up the coordinates of each sphere as the subject moves, and translates them into animation data, which is then applied to the 3D model.



The power of mocap lies in the creation of a heightened sense of reality in animation, directly copying real movement down to the sub-millimeter level. For the Fury trailer, Act3 used motion capture for the major movements of two characters fighting, and blended in Keyframing for the facial expressions and hand gestures.