Low Poly Modeling Techniques
Date: 08-28-14 8:29AM | Category: Tutorials - 3D

I'm posting this as a follow up to my Summer/Fall 3D tip focusing on low polygon modeling techniques. It's somewhat confusing and I figured it could use some additional elaboration as well as a few graphics to help explain what the hell I'm talking about.

So first let's review that tip. Basically I'm suggesting that the best method to produce a uniform & usable low polygon model is to actually produce it as a high poly model, then manually delete edge loops to end up with your final low poly version. This sounds like extra work, but the high-poly subdivision features in today's 3D software can actually do alot of work for you in terms of smoothing out crude geometry, and producing more accurate non-linear forms. These techniques allow the artist to produce crude, low quality meshes (albeit with some tedious and specific edge loops) that are magically turned into brilliant, high res models. Especially for non-linear, man-made's really difficult to produce a natural curve if you're modeling polygon by polygon. (Anyone who has modeled a modern sports car will understand that frustration.) Like I said, the high poly method will require some additional edge loops (which can be frustrating to learn) but in the end those swooping, subtle curves will turn out perfect with little input from the artist.

Of course the other huge advantage that I haven't touched on is that now you'll have both high and low poly versions of your model, which will fit right into the normal mapping pipeline used in today's game engines.

One of the more confusing aspects is that you still have to build this "crude, low quality mesh" described above, and I know beginner would question why that isn't the same as a finished, usable low poly model. The reason is all about even and uniform edge loops, especially for animated characters. As mentioned above, there is a very specific method to building this initial geometry which is even lower detail, and more crude than the final result. In a nutshell, you can build a model even faster with lower geometry detail, and still end up with a better result.

Still confused? Well let's take a look at an example:

Look at the general edge uniformity on both models, the right model is cleaner, more accurate, and a much better final product. It may surprise you, but the model on the left has more polygons than the final (smoother) one: 2,146 vs. 1,728.