Mechanism Digital Develops Facial Replacement for Documentaries
June 30, 2010

Mechanism Digital Develops Facial Replacement for Documentaries

New York, NY - Mechanism Digital, a New York City visual effects and animation production company, has been ramping down from one of its most demanding projects of the year, and has also met many questions about how they accomplished the unique graphics in the recent MSNBC documentary, The McVeigh Tapes: Confessions of an American Terrorist.
The documentary is a two-hour special, narrated by Rachel Maddow, and aired on MSNBC to mark the 15th anniversary of the Oklahoma City bombing. It featured almost 50 hours of audio tape never before released to the public. The tapes were an interview of Timothy McVeigh, in which he describes the preparation and implementation of his bombing of the Alfred P. Murrah Federal Building. The tapes also offer biographical insight into the personal history of terrorist McVeigh.

Since very little film footage of McVeigh’s life and the attack itself exist, producer Peacock Productions went to long-time ally Mechanism Digital to develop the visual interpretation of the interview and flashbacks of the attack, ensuring that McVeigh’s recognizable face was depicted in the recreated events. In response, Mechanism conceived a new method of replacing an actor’s face with a computer-generated 3D model of McVeigh’s. A video showing an overview of the process is available for viewing on The Maddow Blog.
Traditionally, documentaries with limited footage from a historical event would employ conventional techniques like videotaping still photographs or shooting dramatized reenactments with blurred or hidden actors. The problems with these techniques are that they can’t depict emotions through facial expressions, and do not hold the audience’s attention for long. Peacock and Mechanism believed that this story deserved something much more unconventional. “For such an important chapter in American history, we really wanted to let the audience look into the eyes of the killer while listening to his voice,” recounted Lucien Harriot, VFX Supervisor and founder of Mechanism Digital.

Mechanism Digital’s involvement in the project began in pre-production. As Executive Producer, Lucien Harriot, spent considerable time planning and testing photographic layering techniques in the studio; working out all the kinks before going to set.

On the set, Peacock shot thousands of high resolution photographs of recreated scenes, visual elements and a stand-in actor depicting McVeigh in action. Mechanism then built a CG 3D model of Timothy McVeigh’s face and replaced every frame of the actor’s face with it.

Mechanism’s unique approach to facial replacement had never been used in documentary filmmaking before which gave the compositor every possible frame of final rendered CGI in advance. In developing and refining this technique, multiple 3D models of McVeigh’s head were pre-rendered every two degrees of rotation on two axes. Two sequences of frames were a combination of 11 expressions and 11 eye positions. The number of images was then doubled by rendering two camera focal lengths to match the lenses used on set; totaling over half a million images and layers. These images were brought into After Effects with Mechanism’s custom expressions allowing the compositing artists to rotate McVeigh’s photo realistic head in real time within the program.

Art Director Mark Palkoski comments on how critical today’s audiences have become: “You can’t get away with just 80% real and expect the viewer to get lost in the narrative – instead, the audience would be constantly taken out of the story due to face jarring. Our studio achieved significantly closer to 90% realism.”

In January, Harriot joined Peacock in Texas to supervise the first seven scenes with the shooting crew, who would continue for three more weeks. “Instead of traditional action, we invented a ’multi-plane shooting’ technique with the Canon 5D SLR camera, specifically for The McVeigh Tapes,” stated Harriot. Generally, the multi-plane effect is applied to archival images when no film exists; where the foreground is cut out and the background must be repainted. To reduce labor intensive post-retouching, Harriot directed the crew to shoot each element (foreground, mid-ground, background) separately so that hand-painting would not be necessary.

Back in the studio, Mechanism’s artists combined the CG layers, effects and photographic elements by compositing them in After Effects’ 3-D space. Dramatic camera motions were set to produce the highly stylized motion footage making up the 650 shots and exceeding all expectations in the finished product.