Content courtesy Dan Maas / Maas Digital LLC
1. Reference Material — I begin the production process by collecting
photos, diagrams, and blueprints of the hardware and environments to be
depicted. These references guide me as I build digital models of the robotic
probes and their surroundings. By constantly comparing my results to
real-world images, I ensure that my designs are accurate and realistic.
2. Modeling — Using a CAD system, I assemble primitive 3D shapes
(planes, boxes, spheres, and cylinders) into three-dimensional replicas
of the probes and their instruments. During this process, which typically
takes two to three weeks, I also assign material types to the surfaces
of the model: metal surfaces will later take on a metallic look, plastic
surfaces will appear plastic, and so on.
3. Setup — When modeling is complete, I assemble the separately-created
chassis, instruments, solar panels, and antennae into a complete digital space
probe, designating their relative positions and axes of motion. I also insert
"handles" to control each major moving part, much like rods on a puppet.
using these I can quickly and easily manipulate the model into any desired
position.
4. Animation — For each scene of the video, I position the digital
models and simulated camera within the 3D environment, just like a director
arranges a film set. To create motion, I indicate where the models should
appear at various key moments; the animation software then interpolates their
positions throughout the whole shot.
At this stage I work with a low-quality preview of the final image, as shown
in the illustration. These rough renderings give a good sense of composition
and timing, and can be generated quickly enough for interactive use.
5. Rendering — After the final animation sequences are approved, it's
time to generate high-quality renderings of the digital models and environments.
To achieve a realistic appearance, I coat the 3D surfaces with detailed
painted images or photos. The rendering software automatically adds shadows
and reflections using ray-tracing. These computations are quite time-consuming;
a single frame such as the one at the left can take minutes or even hours to
generate.
The entire Mars Rover animation was rendered on a network of eight PCs. Each
frame took anywhere from one minute to two hours to compute (there are 24
frames per second of footage).
6. Compositing — I often decide to render a scene in several different
layers or passes — the illustration shows a single frame consisting of three
separately-rendered elements. Each layer is color-corrected, blurred, and
moved individually, then combined into the final frame. This technique also
saves time if one element needs to be generated again.
Finally, to add an authentic touch to the computer-generated images, I
simulate several defects of real-world cameras, such as film grain (random
noise) and soft glows around bright highlights. These deliberate degradations
help to remove the unrealistically sterile appearance of raw computer
renderings.
Click here to watch the video.
