Rendering Your Scene

Katana was developed from the ground up to address the problems of scalability and flexibility. Solving both problems was essential for dealing with the demands of highly complicated modern productions.

When it comes to rendering, Katana’s renderer-agnostic nature provides the flexibility to allow CG supervisors and pipeline engineers to select the appropriate renderer for the show or shot. The renderer connects to Katana through a renderer-specific plug-in. Currently, Katana ships with the 3Delight renderer (and accompanying renderer plugin) and an API that allows developers to support other renderers.

Scalability is also at the heart of rendering in Katana. Many renderers support procedurals that can be evaluated on demand (often called deferred evaluation) and are able to recursively call other procedurals. At render time, they are passed scene descriptions in the form of a procedural recipe to be run inside the renderer. Through this approach, very large scenes are easier to manage, and the resources needed to deal with them are reduced. In addition, deferred evaluation significantly simplifies pipelines by removing the need to write large scene data files (such as RenderMan .rib files and Arnold .ass files) for each frame before rendering starts. Renderers that don’t support these features are still usable with Katana, but they don’t leverage its full benefit.