The environment was created using 3D matte paintings. "Our matte artists painted key views of the environment, and camera-projected them onto relatively light geometry in Maya, although the size and scope of the environment still made for a pretty heavy model," Hendershot observes. "Once the projection cameras and their associated images were set up, the projections themselves could be rendered in comp for each shot. Live Free or Die Hard was the first project at The Orphanage to use Nuke for compositing. Its powerful 3D capabilities were a true asset to our show, and allowed us to simplify the process. Instead of having a technical director project and render every environment painting, the paintings could be passed directly to the compositor -- along with a projection camera and the matchmove camera. By turning over images and cameras from painters to compositors, we were able to remove one person from the process. Ultimately, the size of the paintings and geometry became too cumbersome to keep all live all the time. Compositors frequently had to pre-render their own projections from Nuke to ease the process. Still, I think it was very helpful to use this approach."

Digital Dimension created some of the signature shots in Live Free or Die Hard. Photo credit: Digital Dimension.�

Combining 3D Packages
Two different 3D pipelines were employed to create the overpass sequence. Shots featuring the freeway in its regular condition were created within Maya and mental ray, using global illumination and final gather. Whenever the structure had to explode or collapse, the team used Houdini associated with Mantra. "Initially, we did many tests with rigid body dynamics to animate the destructions," Hendershot explains. "We obtained very nice results, but in the end, we found that this approach took too long. Instead, we opted to employ keyframe animation, using the rigid body dynamics tests as a reference to animate the main chunks of concrete.

"First, we broke the freeway geometry in Houdini, using cracking tools that were developed to create detailed broken geometry. Then, the geometry was rigged and hand-animated in Maya, mainly because our artists are more familiar with this package. Once the key animation was approved, we brought it back in Houdini to create secondary animation. We generated data that would indicate when two pieces of concrete would separate, when another piece would hit the ground -- basically, any kind of interaction between the hero elements. Houdini allowed us to use this huge amount of data to trigger many dust and debris explosions, and local changes to textures, all procedurally. For this type of work, we found that procedural techniques are much easier to employ in Houdini than they are in Maya. Houdini allows the artist to easily manipulate geometry directly in ways that most packages reserve for their particle tools." CG supervisor Dong Yeop Shin oversaw the development of most of the Houdini fx animation.

Led by lead CG artists Steve Cho, Josh Cole, Steve DeLuca and Kyle McCulloch, the compositing team faced tricky challenges with this sequence. Practical explosions and bullet impacts generated huge clouds of dust that had to be extracted from the plate, and brought back onto the CG environment. In many cases, the blue-screen didn't reach the top of the frame, which obliged the compositors to deal with dust clouds moving over a line separating the blue screen from a clear bright sky... The most complex collapsing freeway shots comprised of more than 100 layers.

Compared to the climax, the elevator sequence was more straightforward. During the course of action, two characters and their SUV crash into an elevator shaft. The sequence was shot in a five-story high set that was extended downward to create a sense of vertigo. Working from The Orphanage's Vancouver studio, CG supervisor Richard Sur and his team used Maya and mental ray to build a seamless extension. In addition, large rigs were removed from the SUV and the actors, and dangling elevator cables were extended. In some cases, original cables were entirely replaced with an animated CG cable.

A Shared Effort
Meanwhile, at Digital Dimension, a team led by compositing supervisor Erik Bruhwiler and 3D supervisor Andrew Roberts was just as busy crafting almost 200 shots. Digital Fusion was used for compositing and wire/rig removal, while the bulk of the 3D work was completed in 3ds Max.

In a major but low key visual effects sequence, a huge crowd is seen evacuating Washington D.C. "Given the restrictions of shooting at well-known government buildings, the production opted for a CG crowd," Roberts says. "We shot front, back and side photos of extras on set, and used those images to texture our Massive agents. Textures were cleaned up in PhotoShop and applied to a library of businessmen and women. Then, using SynthEyes, we 3D tracked the cameras for a number of crowd shots, creating low poly geometry matching the buildings, ground plane, stairs, and any obstacles seen in the plate. This geometry was brought into Massive and served as the terrain over which the crowd would move. To give our compositors full control to finesse our renders, we provided a number of passes and mattes. In some of the shots, we have close to 1,000 CG people moving through the scene!"