| 摘要 |
A computer-based system automates the production of so-called pass--images (or just "passes") from data defining 3-D scenes. For example, a beauty pass of a selected object the normal surface appearance without any backgrou nd or other objects. A shadow pass shows only selected shadows and a highlight pas s, only selected highlights. The passes allow pixel-processing to be done efficiently. The processed passes can then be combined ("composited") to form the finishe d image. In the past, these passes were produced by editing a copy of the 3-D scene for each pass and rendering the edited copy. The invention automates the production of passes by filtering the 3-D scene through pre-specified pass definitions that override properties of the 3-D scenes. The results of filtering are rendered (rendering largely comprises the process of ray-tracing) to form th e passes. The system stores numerous pass definitions. Each time the 3-D scene is edited, the passes can be produced automatically from the pass definitions. This automation of pass production also allows the passes to be used in the authoring environment by allowing a pass preview of the 3-D scene rendering. The automation of pass-production provides a tool that goes beyond merely reduci ng the labor involved in creating passes, it provides the author the option of invoking a new perspective, a different kind of immediate feedback, right in the authoring environment. This feedback enhances the author's ability to focus on specifi c features of a scene as the author edits the 3-D scene itself.
|
