Image courtesy of Deron Yamada. © 2004 DYA367.

Going back to its inception as American Dog, Bolt was designed to have a painterly CG look inspired by Edward Hopper and members of the Ashcan school of painting (urban American realism). In fact, the plan was to combine the urban, cinematic look of Hopper with the naturalistic exposure sensibilities of air/atmosphere in the cinematography from the films of the '70s (most notably, Vilmos Zsigmond's McCabe and Mrs. Miller). This included lens effects such as light scattering, blooming and lens flares.

Indeed, applying such a rigorous painterly approach to CG required lots of R&D on the part of the technology team (under the supervision of Hank Driskill). As a result, Disney filed four patents on the R&D done in 2006 and 2007:

1. "Multiple Artistic Look Rendering Methods and Apparatus," the massing/editing system.

2. "Method and Apparatus for Graphically Defining Surface Normal Maps," the normal painting technique.

3. Surface Shading of Computer-Generated Object Using Multiple Surfaces," the ray painting system.

4. "Rendering of Shadows with Hand-Painted Appearance," the painterly shadow work.

VFXWorld goes deeper into Bolt's technological accomplishments with Driskill (and you can learn more about them in The Art of Bolt by Mark Cotta Vaz from Chronicle Books).

Bill Desowitz: What was your role on Bolt as technical supervisor?

Hank Driskill: I helped shepherd the R&D on the film, helped supervise all the TDs, worked as the main liaison to the technology groups and, in production I'm kind of the "chief fire fighter."

When we first started our pre-production, the mandate from Paul Felix (our art director) was to be able to make CGI imagery that looked like a painting. That had been done here before with "Deep Canvas" back in the Tarzan and Treasure Planet days, but there the focus had been on brush stroke techniques. Adolph Lusinsky, our Look & Lighting director, was the first to make the observation that you could take a Hopper painting and shrink it to postage stamp size, way too small to see brush strokes, and yet you could still obviously tell it was a painting. This led us to a lot of research on the concepts of massing and editing, being able to control the amount of detail on an object. Look A/look B was our system for massing and editing.

We had our texture painters paint objects in layers: a base coat that's very loose, and levels of detail on top of that. We kept those levels separate, and gave controls to lighters to let them dial in where they wanted detail: they could put detail based on distance (things farther from the eye get looser and more impressionistic, or things farther from some center of interest in the scene), or they could put detail in well-lit areas and not in shadow areas (or the reverse, which creates this nice "blown out" feel to the lit regions).

It gave us a lot of ability to draw the eye where we wanted people to look, especially when combined with more traditional techniques like depth-of-field and such, but also let us work in stereo (where one typically dials down depth-of-field) and still be able to attract a viewer's interest to a particular part of an image.

BD: And how does ray painting fit into this plan?

HD: Ray painting was a technique for putting painterly edges on the silhouettes of objects. Ernie Petti, our lighting supervisor, came up with this idea. We didn't want to have to place thousands of brush strokes on our objects, like "Deep Canvas" had done, but wanted a similar painterly edge. Ernie came up with the notion of placing geometry around an object of interest that would have maps for things like opacity: a ray would be cast from the eye, pick up some properties from the first object and others from the second. We could, for instance, put painterly opacity maps on some surfaces around the main object, and render with the color of the main object and the opacity of the surfaces around it. This gave us a lot of flexibility to play around with the edges of objects at render time, without having to re-model or re-paint the objects. We built "flanges" on things like buildings that would hang off the buildings and give painterly edges, while more complex objects we'd just place cards around the object with opacity maps painted for various viewpoints. It's a little hard to explain, but it was a lot of fun to use and created some really neat silhouettes.