Yesterday I did a post about the XBOX One, I included a statement about how Infinity Ward talked about getting better detail on their models they used a new process called polygon subdivision. I stated this was not a new technique, but was dependant on the processing power available. It was apparent to me that not everyone would know that this meant off the bat (though the words make sense their context doesn’t). So I thought I would explain what this is.
NOTE: Please excuse the crudity of the drawings I have done to show examples, some are proper wireframes but any drawing I have done you will know (the quality is terrible).
This is a simple wireframe model of a face. This shows the way the polygons are joined. This is also a very crude face (I did not do this one) and it helps highlight the point I am making. If you move your 3D camera closer to this model the polygons get bigger and so do the imperfections in this model.
So, the first image from a distance would show up OK, but as you get closer you would start to see the flat areas of the model, that the textures are applied to. This is where our nice friend Polygon subdivision comes in.
But through the process of either manipulating the model in real-time or, having different models that get loaded in as it gets closer and closer (I am not sure as I have not ever implemented any of this and it mostly theory).
The model starts splitting the polygons into smaller and smaller polygon. This process keeps happening until a limit is reached. As you an see, I have taken this model wireframe and subdivided the polygons up. Make them smaller and smaller. I didn’t change the shading to show the affect. But it is to highlight how this is done. The surface will appear smoother and smoother depending on the number of polygons the base wireframe has. We have gone from 5 polygons into 26 polygons.
Also something I should note, is not all polygons on models are triangles. They used to be, but due to increases in processing power the models can contain any shape polygon it requires to make the surface as realistic as possible. It is the number of polygons that is still a hindering factor. This is why things like shaders come into their own as they can help convey a depth to a flat texture without the need for polygons to actually create the depth.
I am hoping that this small little explanation helps make things a little clearer. I know it wasn’t needed, but I also thought for the sake of posting as well I would explain.