3D Graphics High Level Overview

More than 40 years of innovation in computing has granted us the ability to put 3D objects on a 2D screen. 3D computer graphics has moved from rotating a wireframe cube on a screen to Pixar animated movies telling stories like never before and video games looking more like real-life films with each new generation. With more realism, it really begs the question: how do 3D graphics work?

3D graphics work much like photo shoots or film shoots. Someone lays out a scene in a 3D environment that consists of lights, camera(s), and objects. The cameras then capture what they see for others to consume at a later time, or live!

In the physical world, odds are that the objects in the scene (say, a human or a chair) are "pre-made" and some light sources are naturally occurring: you don't need to make a lot of the components of a real-world scene from scratch. In computer graphics, we have none of that. We have to add any lights, cameras, and objects to our scene and define exactly where they are and how they act to achieve the desired look. This can be a blessing, because we can create worlds that wouldn't be possible in physical space, but it can also be a curse in that it can be difficult to achieve realism or set up the scene exactly how we imagine.

What does a 3D scene really look like? The scene hierarchy is as follows. We have a scene. The scene has a set of cameras, a set of lights, and a set of objects, which are called meshes. Every mesh consists of a material that defines how the mesh responds to light and looks to the camera. Every mesh also has a geometry, which defines its shape. These shapes can be simple shapes like cubes, spheres, and pyramids, or they could be complex shapes like a house or human.

Much like a physical world scene, rays of light project from a light source and bounce off of meshes. That light then reaches the camera so we can see everything.

In pre-rendered graphics like images and videos you can create in tools like Maya and Blender, this process of casting light, seeing how it affects objects in the scene, and using that information to draw the scene is often a slow process that produces relatively high quality results. In real-time graphics like video games, this process takes place between 30 and 60 times each second, but that speed comes at the cost of quality. The good thing is that the lines between these systems continue to blur as graphics processors get faster, though. This is why the graphics on gaming consoles and PCs look better with each new generation.

I hope this was helpful. If you want to see me making a really basic scene in Blender, read this post.