The renderer is the most critical part of any 3D engine. It is responsible to do all the computations required to produce pictures. Its typical inputs are a description of the 3D model to be rendered and the observer’s data like viewpoint & view direction. Typical renderers found in 3D game engines or other realtime 3D graphics applications are made part of software runing on classical processors and part of algorithms frozen in special purpose hardware known as « graphics processors ». We think that such a shaky architecture will not resist a long time the ever increased complexity of 3D models and features set. We strongly believe that the only way to not crash soon into a complexity wall is to bet on pure software renderers. We are commited to do our best to be a key player among these next generation renderers with Kribi.

The Kribi Renderer was designed from the start, more than 10 years ago, for best possible scalability with increased 3D models complexity. It has evolved continuously until now through steady software optimization & numerous refactoring of its program code for capitalizing on many new CPU features. Pictures that were taking hours to compute in early days of the Kribi Renderer are now generated at realtime frame rates, though with exactly the same field-tested core architecture.

Our main goals for the Kribi Renderer since project inception are :

• State of the art support for advanced description of the 3D models database. So-called scene graphs are an example.

• Fast and realistic rendering of scenes of an ultra wide range of complexity. We want it best in class for models made of hundreds of polygons to multi billions of polygons.

• Perfect portability among target configurations. This include 100 % 3D-APIs and OS independence. The key point here is to avoid entirely the infamous «drivers issues».

• Best possible scalability with increased 3D models complexity. This include increased polygons count, increased light sources count and more & more features-rich model database schema.

• Best possible CPU Scalability with increased clock rate and number of processors in a system.