marcello
02-06-2002, 05:18 AM
hi there, i've got some ideas on 3d modelling techniques, & after noticing all the great character modellers that are posting here, i though this'd be a good place to ask for opinions.
i've seen how efficiently renderman renderers like PRman & entropy (& BMRT even) handle subpixel displacemant, & thought that this could be used as a modelling technique in tandem with poly subdivs. An ideal workflow would be thus:
model general forms as a polygon subdiv control mesh > paint displacement alpha-map in a 3d paint app (deeppaint 3d...) > subdivide & displace at render time...
this means you'd have a light model to rig (with morph displacements for bone rotations ideally) & pose when animating. it could probably be little more than the kind of realtime models used in videogames. you could also bypass the 3d paint bit & use the lo-res nature of the control mesh to ease the unwrapping into a 'paint-shape' for uv-mapping (then just paint the disp. map in p/shop). you wouldn't have the fuss of wrestling with edge loops & other topological headaches when you build a detailed subdiv because you just paint on the detail! folds in cloth, muscle detail, geometry hair...
has anyone got any thoughts on this, perhaps some of you do this already. & if not, why isn't anyone doing this?!
also, i use 3dsmax's polys for modelling, & am trying to get some interconnection with the Entropy demo via Maxman1.7 (eval. licence) with no luck. has anyone tried this combo?
i've seen how efficiently renderman renderers like PRman & entropy (& BMRT even) handle subpixel displacemant, & thought that this could be used as a modelling technique in tandem with poly subdivs. An ideal workflow would be thus:
model general forms as a polygon subdiv control mesh > paint displacement alpha-map in a 3d paint app (deeppaint 3d...) > subdivide & displace at render time...
this means you'd have a light model to rig (with morph displacements for bone rotations ideally) & pose when animating. it could probably be little more than the kind of realtime models used in videogames. you could also bypass the 3d paint bit & use the lo-res nature of the control mesh to ease the unwrapping into a 'paint-shape' for uv-mapping (then just paint the disp. map in p/shop). you wouldn't have the fuss of wrestling with edge loops & other topological headaches when you build a detailed subdiv because you just paint on the detail! folds in cloth, muscle detail, geometry hair...
has anyone got any thoughts on this, perhaps some of you do this already. & if not, why isn't anyone doing this?!
also, i use 3dsmax's polys for modelling, & am trying to get some interconnection with the Entropy demo via Maxman1.7 (eval. licence) with no luck. has anyone tried this combo?
