Deepening the canvas

My continued research on painterly rendering techniques is beginning to bear fruits: PainterMan, a NPR RenderMan shader, has been rewritten from scratch and now works properly on any surface using a new and thoroughly optimized algorithm that provides much more control over the look.
Sublte displacement creates the illusion of paint strokes. Support for shadows and multiple light sources of any type has also been added. To my amazement it works best at high shading rates in ranges of 100-1000, whereby the time to render a frame has gone down a lot more than I could have achieved through code optimization alone.

As a matter of fact, this algorithm produces a look so different from the last version that I decided to keep the old shader alongside the updated one, simply renaming the new version to PainterMan II. Surely both can prove useful for different objects in my film.

Here’s a sample of what it looks like in a rough test scene:

That’s not all I’ve been up to lately… I have another NPR project cooking for use in Sleeping Dragon: Disney’s Deep Canvas technology looks mighty impressive, but there’s very little technical information or even documentation available on the net. A SIGGRAPH paper from 2003 proved somewhat insightful, but lacks detail here and there – not much of a surprise, of course Disney doesn’t want to give away its trade secrets just like that.

From what I gather, it should be possible to create a system similar in look to Deep Canvas using only standard programs and a little scripting. Alas, I don’t have the time nor desire to create a fully working implementation just for my short movie, so I’ll have to make do with a little hack, if at all.
As it stands, only few scenes in Sleeping Dragon would gain from a system like this, so it’s not my main focus to get this running at the moment. I’ll keep working on it alongside the film and post updates if anything noteworthy comes about.