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 1001000, 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.