At the Mountains of Madness
So where we left off last time, we had the Rat fully modeled with a completely unusable number of polygons and I promised you that it would all make sense in the future. Well, today is not quite that future yet, however, a lot of what we will be doing here is going to directly affect our result. Our aim is to take the high-res mesh and make it useable in the game while keeping as much of the high poly look as we can where it matters.
The first thing we have to take into account when doing any kind of low poly work is that of silhouette, what do I mean by that? Silhouette is what it sounds like, it's the outline of the object relative to its background
The trick with this comes from the fact that this is an object that can be viewed from any angle at any time as it's a 3D model, so the silhouette is really a conglomeration of all of the convex shapes on the mesh. The thing we need to bare in mind with this is that we want to use as few vertices as possible while achieving this, as you can tell this can be quite a juggling act. Some of you might be asking, “but what about the polygons?” and you are correct in saying that polygons do in some manner matter however polygons are made up of vertices and there are a few things that can happen that can cause vertices to easily skyrocket compared to the number of polygons, for example, UV mapping seams and boundaries between smoothing groups both duplicate vertices. I’m not going to go into too much detail about this right now as it can get extremely complicated and extremely lengthy.
Another really useful thing to do before you start creating your low polygon model is working out what components can be reused, on the rat, for example, the entire mesh is fairly symmetrical and we can get away with modeling half and use a mirror command to duplicate the other side. Not only does this save us time, but eventually this can also be used as a way to save us a huge amount of texture space by having a lot of shared texture coordinates.
With all of this in mind we need to start building our low polygon mesh, this time I’ve recorded a video that's much shorter, however, it should give you a general idea about what the entire process was and I felt it would have been cruel to subject you all to the complete thing. Anyway, without further ado, here is the video.
The biggest mistake I made in this video was overestimating the number of polygons so the final “low polygon” mesh ended up around 40k tris in the end, which is around double what we are after. It's a fairly easy mistake to make when creating the first model for a project and in the end not a complete disaster as it's fairly trivial to go in and clean the mesh up afterward and reduce the number of tris.
After cleaning it up it ended up coming out like this:
So, what's next? We need to unwrap the model, bake the high res detail down onto this low polygon mesh, however I'm going to change gears a bit first as we need to create a shader capable of doing all the things we want to do to this mesh, so next time we’ll be going into a bit of detail on how that works.
That’s all for now, if you have any questions, please feel free to ask them below.