a a a

Non-Photorealistic Rendering

Vol.32 No.1 February 1999

Why Non-Photorealistic Rendering?

Steph Greenberg
Director, The Physics of Cartoons, Part 1

Since computer graphics’ inception, programmers and artisans in computer animation have sought increasing realism in rendering, surface representation and movement. The “holy grail” of computer animation would appear, to an outside observer, to be an image indistinguishable from the reality we all experience.

For many years, I took this as my own quest, energized by the newness of the whole idea of computer animation, particularly approaching it from the low end world of the Amiga, where anyone with a few bucks in their pocket could participate in the 3D computer animation world.

The Quest Changes

Over the years, my goal has changed. I’ve joined with other artists who are not content with some duplication of reality, or making the non-real look real. Our inspirations come from the deliberately crafted world of traditionally rendered art, with pencils and paints. More specifically, they come from worlds created out of the imagination of artists — characters which have no real root in the real world, with traditions based in turning the pathological limitations of simplified graphic lines into worlds with characters that have endured for more than 70 years and rules that defy the laws of real world physics.

We know their names. All of us. Mickey Mouse. Felix the Cat. Betty Boop. Popeye. Bugs Bunny. Elmer Fudd. Donald Duck. Daffy Duck. Yosemite Sam. The Tasmanian Devil.

My earliest permanent memory is of a Bugs Bunny cartoon, Friz Freleng’s 1952 release, Hare Lift. I remember this because when I was two years old, and I was looking out the window of a 707, I asked my mother why I couldn’t see the entire world below us, like it was in Hare Lift. For some reason that memory has always stuck in my head, when I’ve forgotten so many important facts.

Since then, I don’t think a single week has passed in 38 years that I haven’t seen one of the shorts created by Warner Bros. animators so many years ago. And of course, there were other studios that contributed to the unreal alternative universe of theatrical cartoon shorts — MGM, Lantz, UPA, Famous Studios, Terrytoons.

Gas Planet Opens Eyes

In 1992 I saw Pacific Data Image’s (PDI) Gas Planet [3], and it changed the way I viewed computer animation forever. No longer were my dreams of doing a three-dimensional cartoon limited to the quasi-realistic world popularized by Pixar, a look related more to George Pal’s Puppetoons than the drawn classic shorts that I really loved. Animated characters in CGI could look more like drawn graphical characters, or like characters that were drawn but had a touch more dimensionality.

In 1994, I went to work for the Walt Disney Imagineering VR Lab, later renamed the VR Studio. The project that I went to work on was Aladdin’s VR Adventure, now an attraction at the DisneyQuest interactive entertainment complex. Editor’s note: See Real-Time Interactive Graphics column for a look at DisneyQuest. The concept was to create a completely immersive environment into which a guest could be in the animated world of Disney’s Aladdin. The characters have a drawn, painted look; and three dimensionality. This all occurs in real time while you fly through the environment on a virtual flying carpet. Aladdin’s VR Adventure was an amazing breakthrough both for real-time entertainment and non-photorealistic rendering.

But it wasn’t a cartoon. A cartoon is a subset of animated entertainment where the characters and their very form obey vastly different rules from those that we experience every day. The characters are forms that can experience extreme abuse, and show it in ways vastly different from things in the real world. They can be squashed like pancakes, stretched like taffy, can have the internal consistency of a water balloon, or the rigidity of steel. They can be malleable like clay, or shatter like glass. Their eyeballs and the gaze on which they are fixed can be exaggerated with projectile velocity, springing back with a force that can knock a character off its feet.

All these transmogrifications require is that they express an exaggeration of the conditions to which the characters are subjected. It helps if the transmogrifications are actually funny, but that is a subjective element that can only be verified by an impartial audience only after the cartoon is exhibited.

The Physics of Cartoons, Part 1

When we were working on the look for The Physics of Cartoons, Part 1, we were trying to strike that balance between the graphic look and three dimensionality. Conventional wisdom in the CGI community claims that the more real something looks the more it must behave like things in the real world behave. Attempts to do cartoony things with CGI that is rendered quasi-realistically have been claimed to have failed as a result of the audience’s unwillingness to suspend disbelief.

But with The Physics of Cartoons, Part 1, the very premise of which is to demonstrate the physics experienced in classic cartoon shorts and its effect on characters, we designed a look that would hopefully invoke the audience’s memory of classic cartoons. Doug Cooper, who created the techniques used for rendering, understood the form and set out with the goal of creating a new look compatible with the memory of classic cartoons. Editor’s note: you can read Cooper’s contribution elsewhere in this issue of Computer Graphics.

I decided that the backgrounds and foregrounds would be drawn, painted and layered. By creating the backgrounds two dimensionally, we could create oddly curved and angled buildings and mountains, with numerous scene changes, far more quickly than modeling and texturing them. By planning the camera angles and moves in advance, there was no need for fully three-dimensional backgrounds, because we knew what the final rendered scene should look like, and created the backgrounds accordingly.

The first stage in the background creation process was to take the representation of the scenes in the storyboards, and create backgrounds using the storyboards as inspiration. Some of the backgrounds were created first as pencil drawings, then scanned and colored in Adobe PhotoShop. Others were drawn and painted directly in PhotoShop or Fractal Painter. The backgrounds had to be wide enough so that the apparent camera could be moved around within the background.

The background elements were also created with separate elements, which could be moved as distinct layers to give a scene the illusion of three dimensionality. Elements close to the camera would move faster than elements that were distant where that effect was called for. Some background elements were created to be looped. That is, the left side was the same as the right side so that they would match up.

Once we had the characters animated and rendered, they were composited with the background elements. The backgrounds were created so that different elements could be moved around. In addition to straight compositing, SOFTIMAGE Toonz was used to animate the separate layers of elements, moving them in different planes in the same way multiple levels of background and foreground elements were created in the classic cartoons using transparent cels physically moved by a camera operator.

The character designs were drawn as if they were going to be animated by pencil. When the characters were modeled in 3D using SOFTIMAGE, they had to be modeled to reflect the drawn features and silhouettes. The internal setups on the characters had to reflect the ability to distort the characters as the line could be distorted in a drawing. And finally, the rendering technique had to present a graphic look that allowed the audience to accept extreme distortions of form and movement.

Even More Techniques Exist

The cartoon look, which ranges from that of The Physics of Cartoons, Part 1, to Gas Planet to various projects which actually look like hand drawn, inked and painted animation with outlines, is only one type of non-photorealistic sub-genre.

An example of recent techniques introduced at SIGGRAPH 98 is the paper “Painterly Rendering with Curved Brush Strokes,” by Aaron Hertzmann [2]. The apparent paint strokes created using this technique accentuate the curved lines of objects with varied brush strokes. The resulting image has the careful and deliberate appearance of a hand-crafted oil painting. Without rendering techniques such as this, creating animation in a hand-painted style with consistent brush strokes would be impractical because of the enormous amount of hand labor required to make each frame of animation, and the daunting task of attempting to make the brush strokes consistently enough for them not to crawl and flicker. Yet through non-photorealistic rendering, such a project becomes a possibility, should it fit the animator’s vision to pursue it.

Another style of rendering is represented by Cassidy Curtis’s Loose and Sketchy Animation [1], which takes a rendered scene, deconstructs the shadow detail and then uses a particle system to impart what appears to be drawn lines. The result is a simplified scene that seems edgy, because the edges move with the inconsistency of drawings that haven’t been cleaned up. The apparent drawn lines have direction, just as they would if they were drawn by hand. Added to the piece is a texture that even looks like paper, to enhance the illusion.

We aren’t limited to just the obvious drawn and painted mediums when looking to expand the palette of animation and rendering possibilities within CGI. We can look at anachronistic techniques, such as printing artifacts created by crosshatched steel engravings before halftones were developed. Or at woodblock printing, in which the artist scoops out the lines on wood creating strong, dark blocks of ink with light outlines when ink is applied to the wood and the wood is pressed to paper. These aren’t merely ancient historical printing processes. They are techniques employed by artists today when the artists decide they are the best medium for expression. Yet both of these techniques have been shunned or done on a very limited scale because of the amount of hand labor involved.

Steph Greenberg
Fax: +1-310-398-2774

The copyright of articles and images printed remains with the author unless otherwise indicated.

The Future is Wide Open

The possibilities for future work are endless. To have characters that can be three dimensional, and look as if they were brought to life by the brush stroke of a painter, for those very brush strokes themselves to be three dimensional, brings the formerly impossible to the world of the possible. The palette of tools and techniques which 3D animators can employ is expanding the range of expression available to artists who choose to work in the three-dimensional computer animation medium. Maybe it’s even shortsighted to draw an arbitrary distinction between other forms of art and the 3D computer animation medium.

We are reaching a point where the limitations of the images we are capable of creating are quickly being eviscerated, and the new limitations are imposed by our insufficiently expanded imaginations.

We are not bound to replicate something that looks like the world we live in.

I expect that during this transitional period, where technological barriers to creation are being shattered, there will be many odes to and parodies of classic art, as The Physics of Cartoons, Part 1 was to classic cartoon shorts. But as new techniques gain their own traditions, find their own equilibrium and their own voice through artists who are not yet old enough to recite the alphabet, these techniques will cease to be an emulative novelty, and will instead be freed for the expression of completely new ideas. They will be explored by people who never had the opportunity or perhaps the motivation to study classic paintings or classic animation, and who subsequently create entirely new artistic styles or genres of entertainment unbound by creations of the past. Or perhaps classically trained artists, unwilling to be shackled to past mediums but inspired by them, will be liberated by the infinitude of possibilities that CGI will offer.

With The Physics of Cartoons, Part 1, we found that many of the techniques and styles of animation used in classic cartoon shorts could be applied to CGI. But, even within the cartoon sub-genre of animation, there is so much to explore that we haven’t scratched the surface yet. CGI cartoons aren’t anywhere near their equilibrium, and the field is wide open to innovation and experimentation, in the look created by rendering techniques, the subject matter, character designs and the way the character and environmental forms are animated.


  1. Curtis, Cassidy. “Loose and Sketchy Animation,” SIGGRAPH 98 conference abstracts and applications, July 1998, p. 317; Website
  2.  Hertzmann, Aaron. “Painterly Rendering with Curved Brush Strokes of Multiple Sizes,” SIGGRAPH 98 conference proceedings, July 1998, pp 453-460; see Website.
  3.  Pacific Data Image. Gas Planet, Website.