Deep Facial

We created the "Earth Faces" animations for Deep Green, an environmental documentary by Matt Briggs. The film's message is that solutions to climate change begin with the individual, and choices that each one of us makes on a daily basis can have profound results in the big picture. From the beginning, Briggs sought a way to convey this message by visually linking symbols of the Earth and individual people.

Initially, Briggs and the Bent team explored a range of ideas that could help the film visually communicate this relationship. We eventually settled on the idea of creating human portraits out of clouds, as the faces could show the beauty and scale of what's at stake while connecting directly to individual people. The portraits are a metaphor for change that begins at home and create a direct visual connection with Earth's climate.

To achieve this goal, we created a series of animated segments showing different views of the Earth. Ranging from 10 seconds to half a minute, each view showed animated representations of human faces depicted as clouds. In the final film, these short segues serve as transitions between the film's stories, locations, and solutions.

We got our CG and compositing teams together for a little brainstorming and came up with a mixed approach that combined 3D CGI, live action, and still photography to deliver what Briggs was looking for. We did a bit of research, but could not find a precedent of anyone combining the elements in the way we intended to. The whole process was going to be one big, on-deadline experiment.

We started by shooting our live-action face actors with our Red Digital Cinema Red One camera in a darkened room. We used a simple lighting set-up of 1 tweenie light directly above the Red with some diffusion. This provided a nicely shaded highlight of the actor's face with a smooth fall off into black. Since the ultimate output was going to film, we shot this footage at the Red's full 4K resolution at 24fps. We needed some smooth, cloudlike motion as our end result, so we had the actors move very slowly and fluidly. We took the footage into Adobe After Effects and used the Time Warp plug-in to slow it down by 400 percent.

In line with Briggs' vision, these portraits needed to represent people from all walks of life. From age 1 to 80, from Korean to Iraqi, we sought out a diverse range of ethnicities, ages, and genders. We instructed our actors to look directly into the camera in order to achieve a direct connection with the audience. These details were vitally important to the film's message. After all, we are on this planet together.

We initially started out using the live action as reference for hyper-realistic CG clouds with the shape of the human face, but the expressions did not come through with the same emotion and vibrancy as the footage we shot. The faces were getting lost in a heavy dose of the clouds. So we took a step back and made the face the focus; the clouds had to mimic the live-action emotional subtlety. The CG clouds, while not secondary, had to support the original face expressiveness.

Using After Effects, we broke Red footage of each face into dozens of independently modifiable passes. Each pass contained different elements of the face itself and we added and subtracted textures from the luminance of each pass to break up the face in an organic way. Deconstructing the footage in this way allowed us to create depth and motion in the faces that mimicked real clouds in the atmosphere and still achieve the end product of a recognizable face. When the dozens of layers were combined, one on top of the other, they began to form an image of the whole face that was already starting to look very cloudlike.

The next element in the mix was satellite footage of the earth. We used these wonderful high-resolution images to create photo-realistic elements of the atmosphere. We sifted through thousands of photos and film footage clips of the Earth as seen from space, including photographs of different cloud formations and weather patterns.

The third ingredient in the mix was CG. In order to create a usable 3D Earth for the project, we modeled and textured a spherical planet in Autodesk 3ds Max and used the FumeFX fluid dynamics plug-in to create CGI clouds surrounding it.

To tie everything together, we imported the passes of the face footage and satellite cloud imagery into 3ds Max as individual mattes. This allowed us to tweak the characteristics of each matte, including its depth and movement, curvature and interaction with the CG clouds. Once all of the elements were interacting as realistically as possible in the 3D renders, we brought everything back to After Effects for final compositing.

Getting the look just right took lots of subtle manipulation of the individual AE layers at every step of the process. In the end, the human emotion of the original faces shows through.

Once we had the process settled, we were actually able to create cloud sequences more quickly than anticipated, allowing us to choose from a wide range of options. We were all quite pleased with results, but more than that, the experimenting we did throughout the process proved to be creatively rewarding in its own right.

Chel White, a 25-year fixture on the Portland, Ore., filmmaking scene, is a director and founding partner of Bent Image Lab along with David Daniels and Ray Di Carlo. The work of both a black humorist and a cinematic poet, his films paint indelible pictures of the human experience. His screenings include Sundance, Berlin, and the Museum of Modern Art.

Brian Kinkley has been with Portland-based Bent Image Lab for five years. Prior, he served as a motion graphics animator for GOOD Magazine, as well as a designer for Red Door Films.