rit_sofa
07-06-2005, 01:40 AM
Hi everybody - This is my first time posting on CGTalk, so I hope this is the right place to be asking this. My question is whether or not it is possible to achieve a "Deep Canvas" effect in off the shelf software like Maya. If you are unfamiliar with Deep Canvas, it is the painterly texture technique used by Disney in the late 90's for 3D effects in Tarzan, Fantasia 2000, and Treasure Planet. Basically it makes it look like a handpainted Disney background that can be moved through in 3D space, as well as lit with different types of lighting. There is an excellent link here (http://cgw.pennnet.com/Articles/Article_Display.cfm?Section=Archives&Subsection=Display&ARTICLE_ID=50479) that explains the process in detail. (They also mention that the Disney process is based on Barb Meir's work from a '96 SIGGRAPH paper...) Unfortunately the technical aspects of this are a bit over my head, as I am more of a painter than a programmer.
http://cgw.pennnet.com/Articles/Article_Display.cfm?Section=Archives&Subsection=Display&ARTICLE_ID=50479
This article mentions that "As they paint, information about each brush stroke--its color, shape, the pressure used to create it, and so forth--is stored as data (not as pixels) so that the Deep Canvas software can apply the brush strokes to an appropriate position in 3D space."
(There's also a really great demonstration of the Deep Canvas process on the Special Edition Tarzan DVD.)
This is probably not possible without extensive programming. I am wondering if there is a way of creating a shader network that will come close to emulating this handpainted look. And I dont mean just painting handpainted textures and slapping them on the object. (Though I really dont know what the next step would be...I assume it has to do with lighting and the way the camera interprets the brushstrokes.)
Anyway, I'd be happy to hear of any suggestions! Thanks. :)
http://cgw.pennnet.com/Articles/Article_Display.cfm?Section=Archives&Subsection=Display&ARTICLE_ID=50479
This article mentions that "As they paint, information about each brush stroke--its color, shape, the pressure used to create it, and so forth--is stored as data (not as pixels) so that the Deep Canvas software can apply the brush strokes to an appropriate position in 3D space."
(There's also a really great demonstration of the Deep Canvas process on the Special Edition Tarzan DVD.)
This is probably not possible without extensive programming. I am wondering if there is a way of creating a shader network that will come close to emulating this handpainted look. And I dont mean just painting handpainted textures and slapping them on the object. (Though I really dont know what the next step would be...I assume it has to do with lighting and the way the camera interprets the brushstrokes.)
Anyway, I'd be happy to hear of any suggestions! Thanks. :)