Wild things, part 7

That’s pretty much it for the new techniques developed for combining hand drawn and 3D animation for the Wild Things test. The only thing left to talk about is shadows. Here we cheated, in a really outrageous way – but it paid off.

When you create a computer graphic scene, you specify a number of light sources. The computer program calculates, for each pixel in the image, where is the visible 3D point at that pixel, and from that it calculates which of your light sources are able to illuminate that point, and which are in shadow. After all, not every light source can reach every point in the scene. Sometimes there are objects in the way that block the light from some light source or another – thereby creating shadows.

In order to make Max and his dog feel as though they were part of the 3D scene – even though they were really hand-drawn animated characters – it was very important that they cast shadows. Otherwise they would have appeared to look like they were just floating in front of the scene.

Of course, Max and his dog were not really 3D objects in the scene. So we couldn’t just throw some sort of algorithm at the problem of what shape their shadows should take – there is, quite literally, no mathematical solution to that problem. Fortunately, we had animators who were perfectly happy to draw the outline of a shadow. And here is where we cheated. Just as we had the animator draw the outline of a character, and then used a computer paint program to fill in that character’s colors, similarly, we asked the animator to draw the outline of the shadow that Max or his dog should cast onto the 3D scene.

In other words, we relied on the animator’s talent to figure out where the shadow should go. Once we knew the shape of the shadow in any given frame of the animation, we used that shape to suppress the lighting from the key light source in the 3D computer graphic lighting. The visual result was the same as if we’d had a magic computer graphic algorithm to cast true shadows onto the scene.

Note that we were not painting a shadow onto the scene. Rather, we were invoking the same computer graphics techniques that we used to light and shade the 3D background – except we were giving the animator a chance to add shadows to this 3D shaded scene.

On a philosophical level, this created a very interesting interaction between animator and scene. 2D hand drawings were being used to reach in and directly modify the physics of a 3D computer graphic simulation – in particular, blocking 3D light sources at selected pixels. Effectively, we were casting actual shadows from non-existent objects.

The results were spectacularly successful, as you can see from watching the Wild Things test.

The only final note – and it is an important one – is that we were very careful throughout the production to choose the colors for the 3D computer graphic background and the image-processed hand-drawn characters that would mesh together perfectly. I can’t overemphasize how important this is when making a film that combines work from two very different media.

Examined in hindsight, our little test for “Where the Wild Things Are” represented a new way to look at computer animation. It wasn’t the result a single technique, or even a single approach, but rather a mash-up of complementary techniques and approaches, a way of mixing the old and the new, of using the computer as a tool in a very different way. That little test floated around the industry in the following years, and ended up influencing many things that were to come after, from “Who Framed Roger Rabbit” to the “Toy Story” films and beyond. I would argue that the success of this test proved the point of what my friend Lance Williams used to point out, around the time we were first bringing Max and his dog to life: “Computer graphics,'” he would say, “is limited only by your imagination.”

2 Responses to “Wild things, part 7”

  1. davidmaas says:

    I’m inspired to hear of your approach on this bit of history!
    Generally, I find it odd how long NPR-interested artists been staring at 3D geometry, trying to get it to render nicely, instead of just drawing on it and having the 3D instruct the hand-drawn line how to move. The time is ripe to get back to this mindset – a marriage of media and toolsets.

  2. Michael says:

    Here is a little gem that I designed and worked the pipeline for (it was released in 2000).
    In this case using an off the shelf 3D application as there was no (affordable) 2D compositing software that could handle those images (A3 drawings scanned in at 600dpi, greyscale – thousands of them).
    Working in 3D with some custom plugins proved to be a lot more successful, even though there is very little “proper” 3D in the movie.

    2K, cinemascope, off the shelf dual CPU PCs with 512MB of RAM :)

    The movie (a 80MB FLV)
    Some stills from the movie.

    So much for war stories… I do find it interesting to see what happened within 16 years of CG though (from the Wild Things test to Ring Of Fire).

    Sorry for half-hijacking this, but I thought it fits well within the general theme.

Leave a Reply