April 18, 20045:49 am
"Spliting the render proccess into several steps has many advantages. You can obtain more flexibilty when composing the passes in post, and keep using them without rerender."
Yes, but it would be even better if we can use the Renderman native AOV rendering workflow to have everything rendered in 1 pass outputing all needed images separately (thus only saving time).
"The shader is mainly based on simple edge-detection algorithm, so I don't know how to change the width of line at an object-to-object base."
I wonder how do plugins like inkworks do it... They probabbly use a behind-the-scene-multi-pass-render-and-after-composite scenario methinks?
"For self-occluded objects, sometimes it is more efficient to use texture map to draw the ink-edge. "
I agree, but that wouldn't add up to being exactly a 'pipeline'.
"I will go on working and develop the whole pipeline for cell rendering soon."
I'm sure I'm not the only one who is extremely eager to see this!
P.S. Maybe it would be interesting for you ZJ to check out some MentalRay sources (which are provided with regular Maya5+ on all platforms). They have a 'countour shader' (which is really an inkline generating toonshading mechanism, not just contours), along with its sourcess.