Game Design Post 1
This week I began modeling. I decided to model a generic human, who will be my basis for all other models I make.
I began from the bottom up. I made the foot first, and starting from scratch is always the hardest part. This took me hours to do, because, literally, the foot was my starting point for my model. By this I mean the following: my style of modeling revolves heavily around extruding - either polygons or line segments. Polygons are essentially a combination of line segments, and line segments are essentially a line that connects two vertices. So, from my foot, I extruded, which in modeling means to extend, the leg.
The legs, while easier than the feet, are still complicated. The hardest part of modeling the legs is getting the curvature correct. If I were modeling for a 3D movie, I could simply add a hyperNURB, which is a mathematical ‘object’ that adds subdivisions to your existing model, to my legs. But, quite obviously, NURBS smooth things by subdividing polygons on an linear scale by doubling the amount of points. Meaning, if you take a cube, with 6 polygons, which has four points, and put the cube in a NURB object, with one subdivision, you will get 8 points, with ~ 30 polygons. As one can see, this more than tripled our polygon count, and in gaming, modelers achieve for the lowest polygon count, so that the user does not experience ‘lag’ from a bottlenecked GPU.
As it turns out, I did use a NURB, but after ending up with a few hundred polygons, I spent a lengthy amount of time reducing the number of polygons. (There is a reduce number of polygons command in my 3D animation program, Cinema 4D, but it does not work so well.)
After finally getting the curvature right on the leg, which is extremely hard when limited to a few polygons — in this case, my foot and my leg are only 80 polygons — I began modeling the torso and the upper body.
The torso was actually extremely easy to model. It was a simple extrusion and I shrank the radius of the ellipse that was currently the torso. This gives it a human-esque feel because, ideally, humans like to see a small waist, which is a lot easier to model than have in real life.
I have just began the chest about four hours ago, right before chest and is very complicated. Getting the curvature right, like the legs, is very difficult for this aspect. Not only does the upper portion of the body have to stick-out of the z-axis, but bottom-sides need to shrink on the x-axis, while the lower-back needs to be moved in on the z-axis and upper-back needs to slope-out, meaning move out on the z-axis. As one can see, that’s a lot to account for with only a few polygons, and it is crucial that the points are lined up symmetrically, for if they are not, the character will simply look bad. (Not to mention that my character needs to have perfect, and when I say perfect, I mean perfect, symmetry for when I am doing my character-rigging in the upcoming weeks.)
The hands were the hardest thing to model. The way I approached this, originally did not work. I made a cube, with five equal subdivisions (one for each finger), and tried to model a hand from that. After I finished, I realized the hands were useless because each finger was attached to another finger’s vertex. I had to start from scratch, this time with 10 subdivisions — 5 for the fingers itself, and the other 5 for the spaces between each finger. Originally, my group intended on making a first persons shooter, in which great detail of attention would need to go into the hands, as that would be the only seen thing, but we have recently decided to go with an isometric camera angle. Because of this, the hands have too many polygons, even after reducing the number of polygons in the hand from roughly 1200 to around 250. *Note that my entire body has about 200 polygons, and one hand has around the same*
Shortly, I will post a few screenshots of what I currently have at Friday, September 18, 2009. These renders are of very rough material, untextured, and will show the polygons on the model.
