Overview

MainLayer covers the whole screen. Within this layer, the user can add or remove a square grid of CC3Layers, by clicking buttons to increase or decrease the number of CC3Layers.

Each CC3Layer contains a separate 3D scene. Each 3D scene contains a separate camera and light, and so can be controlled independently. The light is positioned differently in each little scene.

Each scene contains a single main object. The object is chosen randomly, from a collection of templates, so each 3D scene contains a different object. The overall effect is a grid of tiles, with each tile displaying a single 3D object.

Within each little 3D scene, the user can touch the main object and then move a finger to rotate the object. When the user raises the finger, the 3D object briefly glows, demonstrating touch selection and control when multiple separate 3D scenes are displayed.

As demonstrated by the touch control, the object in each tile scene is a separate CC3MeshNode, and can be controlled independently of similar objects in other tiles. But since each of these objects is created from a template copy, there is only one copy of the underlying mesh data, thereby preserving memory.

When drawing this main layer, each 3D layer and scene must be visited to be drawn. There are several techniques that can be used to optimize performance under these conditions.

Of prime importance is reducing the number of times the color and depth buffers are cleared by each 3D scene. By default, the depth buffer is cleared on every transition between the 2D and 3D scenes, to ensure that the 2D and 3D artifacts draw in the order expected.

For many app configurations, this is not really needed. Here, we turn off depth testing in the 2D scene, so that any 2D nodes will be drawn over the 3D scene. We also tell the 3D scene not to clear the depth buffer between each transition between 2D and 3D. Except in the most complicated situations, this should be suitable for most apps.