Wednesday, January 13, 2010


The author of this week’s post: research scientist, physicist and looker extraordinaire, Ian Hopkinson, Ph.D, who works in the UK. 

Ian  Hopkinson, Ph.D

My background is in science and recently I've been working on the science of appearance.  The science of computer graphics forms a part of the appearance science area.

There's a lot of science that goes into computer animation and movie special effects. One part of the problem comes down to making a model of light and it's interaction with the world, including difficult materials like hair, fabric and skin. More directly involved in the animation process is the question of how to help the artist bring a character to life: given a character in one pose, how do I get it to walk, to run, to jump with the minimum, easiest intervention from the artist? How do I make it as intuitive as possible for an artist to get the effect they want?

In this area my main interest has been in the first of these areas: a model of light. My desire is for this to be physically correct, rather than artistically appropriate. So if you wish to put your character in a pool of light, to my mind that pool of light must obey physical laws, it must be made by a physical light fitting that you could make in the real world, it must cast shadows and it must be reflected appropriately into the rest of the scene. Modern computer graphics software has many knobs for the artist to turn in order to get the effect they want, but these knobs don't have to obey a physical world. Lights can be made that can cast dark, lights can be made that only illuminate some objects in a scene. Similar problems occur in the physical description of materials. To me this is terribly wrong!

The computer scientists interested in new algorithms for the modelling of light aren't generally artistically inclined, so rather than creating new and interesting computer models each time they wish to try out a new algorithm for representing a material or modelling light they turn to a limited number of old favourites, such as the Utah teapot, the Stanford Bunny or the Happy Buddha. Learn these objects and you will be able to nod knowingly the next time you see them.

In some ways computer art packages are a new sort of tool (like a brush for a painter, or a chisel for a sculptor) but their controls are an abstraction. When a painter paints, they must use physical objects in the real world to create their effect. A brush may be designed to give a particular effect, but it is constrained by what the physical world can do. By looking at a brush you can get some idea of what effect it will have, and what you might have to do with it to get different effects. Computer software, on the other hand, is different: the designer of the program must chose how to link user input to on-screen effect.

The physics department I worked in had an artist in residence for a while, we respected her because she'd written a successful grant application to join us. It was very illuminating talking to her, because whilst we both had the same interest in how things looked I wanted to know what something really was whilst she was more interested in what range of things it could mean. This could lead to some frustration, I remember clearly her bringing a beautiful photo to lunch which could have been a galaxy, or it could have been a pattern of powder on the surface of a bucket of water, or it could have something else - the frustrating thing was she wouldn't tell me which it was!

I did some "research" for this post: I watched "Simpsons the Movie" and "The Incredibles". I enjoyed them both, and for broadly the same reasons: They are cartoonish in their content (impossible, exaggerated, colourful), and they have a sense humour. From a technical point of view, the lighting in "The Incredibles" is beautiful, and the rich characterisation of real-ish human figures is impressive too. However, in the end, the technical means by which they come to be on the screen is less important than the story, even for someone with an interest in the area.


Nora Lumiere said...

Thank you for this lovely guest post and for doing your research. I'm sure you're a better person for having seen THE INCREDIBLES if not THE SIMPSONS, THE MOVIE - The SIMPSONS are much better on TV, not something one can say about many films.

SomeBeans said...

@nora - it's a pleasure. It's interesting reading your blog on the artistic side of animation following my recent experiences with the computer graphics people.

Nora Lumiere said...

BTW - No need to put "research" in quotes as though it wasn't really research just because it was animation. Since it was the basis for your reviews of those 2 films, it was legitimate research, not pretend cartoon research. :-)

FleaCircusDirector said...

I agree that the disconnect that exists between the computer animators controls and the resultant animation can cause a problem and lead to animation that does not look good.

However pushing or even breaking the physical boundaries that constrain the real world has to be one of the good points about animation and is done in film as well. One of my favourite effects is depth of fields which allows the camera to draw the viewer's attention, this effect rarely occurs in real life with the exeption of people with bi-focal glasses. Increased contrast, improbably positioned spot lights, changes to colouring to affect mood are other examples of optical effects. Squash and stretch, anti gravity, cameras moving through walls are physical changes we would use. And of course then there are the dancing hippos and singing tea cups to consider.

Animation is an art and yes all art has an input from science and real life but being more than just real is part of the magic.

Nora Lumiere said...

I agree that art can mess with science and physics to create effects and beauty.
Scientists must be logical and linear to do what they do, but artists must break scientific rules to make magic.

Thank you for your wonderful comment.