I am currently having a go at volume rendering using cubes. I have been extensively looking through resources on how to do this, and reading explanations on how to do it, and I understand the gist of it.

You send a 3d texture to the shader. Each voxel in this texture contains some data about a "virtual cube"(Could be colour or texture coords in a texture atlas). The shader then renders the cubes.

I read that you render a cube of a specified dimension(8x8x8) for example in the engine, which then gets sent to the shader with a corresponding 3d texture describing the location of the cubes. This cube is then ray casted through each of its pixels in the shader, and cubes rendered in this way. This is what confuses me.

What does this mean? How is it possible to render faces using this method?

Bump.