Retopology is the process of taking a model with bad topology and making a copy of that model but this time with good topology. It is essentially, the process of fixing a model with bad topology. It is a very popular technique and that is growing rapidly. This is a fairly new technique and rarely ever existed a decade ago. Retopology didn’t exist as a technique because some crazy statistic said “3D modelers are getting worser at 3D modeling at an all-time high of 50% this year, something needs to be done before it’s too late.” Retopology is actually used to compliment the other modeling techniques where topology took a backseat. The process of retopologizing mainly involves the polygonal modeling technique (explained in this post).
Let’s take a look at how a 3D modeler may make use of retopology in his/her workflow:
- Create a 3D model using sculpt modeling (or 3D scanning).
- The result is a really good-looking 3D model (hopefully) but a model with bad, crowded, messy and unorganized topology.
- The 3D modeler now uses that awesome super-detailed 3D model with bad topology as a base, and creates a new mesh on top of that uses polygonal modeling to retopologize over the high-detailed model.
- The result is slightly less detailed but near accurate to the original 3D model.
- The 3D modeler may then bake the bump maps and such to preserve the details from the original model and apply this bump map image texture to the lower resolution retopologized version to fake the details.
Retopology in action
That’s basically the general workflow for retopology. You have the freedom to create high-detailed 3D models using the sculpt modeling technique without worrying/caring about topology and/or memory. Then later on, you “fix” the topology using retopology while being prepared to accept the fact that you may lose detail along the way.
In Blender, you would generally approach retopology by importing your high-resolution 3D model. You would then create a new mesh. Vertex by vertex, you would “stick” them on the detailed mesh. You can use the “Snap during transform” button which is shaped like a magnet and choose “Face” in the dropdown next to it. Vertices will now stick onto any closest mesh surface it could find. Below is what the tool looks like:
You would approach the same way as you would polygonal modeling (read about it here). That is, you would extrude a vertex, stick it to the detailed mesh, extrude another, stick that to the detailed mesh and keep it going on. All the while filling in faces where necessary (pressing F to do so) and taking note of good topology while placing your vertices. You shouldn’t guess where to add vertices or place them randomly wherever you feel like. If you know the right topology for your model, you will know what you are doing and where to place your vertices. With more practice, you should be getting quicker and quicker and vertex placement will become second nature to you.
