That shader is based on the facing ratio of the object (or the polygons). That means, for example, a part of the surface that points towards the camera will be white and a part that points away (to the side) will be black. And depending on how much a part is curved, it will be different shades of gray.
It's like a depth map, but instead of the depth driving the brightness it's the facing of the object.
You can find a tutorial that shows how it works here (just ignore the "ghost" part
But this is only the first step, since rendering the facing ratio will not create an outline. Instead it will be gradients of bright to dark. Later in the compositing process, you need to apply a gradient filter on the render to push the values around to make it into an outline.
When I made the turntable animations (putting all the render passes together in Adobe Aftereffects), this particular approach to make the outline created a few ugly artifacts in (mostly) the faces of the characters that had to be masked out to make them look nice.