Render calls and geometry buffers – how do they work?

Hi c4cc,

Verge3D/WebGL renders a mesh by processing its geometry information (coordinates, normals, texture coords, vertex colors etc) stored in long arrays. Those are called “geometry buffers”. Big meshes imply large geometry buffers and require more video memory.

When rendering meshes, Verge3D/WebGL must mark a mesh’s geometry buffers “active” and also select a shader (a GPU program), and may be do some other stuff. Only after that, WebGL would issue a command to render this mesh. The whole procedure is called a “render call”. The more different geometry buffers and different shaders (in the 3D software, “materials”) are contained in the scene, the more render calls will be needed.

Preparing for a render call is a CPU-intensive operation that affects the FPS. Thus, you want to have as less of render calls as possible. This can be achieved by batching similar meshes into one, and by using less different materials. Hope this helps!

You can get the technical info from this page.

The more vertices, the more video memory consumption. One mesh – one geometry buffer. One mesh, visible on the screen – one render/draw call (provided post-processing is not used). Meshes that are out of the camera’s field of view are not rendered.

Any light source casting shadows will double the amount of render calls. Most of post-processing passes will also double the amount of render calls.

c4cc wrote:

What do you mean, post-processing passes?

Brightness/Contrast, Bloom, Afterimage, Greyscale, Outline effect, etc.

c4cc wrote:

what about bones and armature?

Do you see them in Verge3D?