Optimizing your Mesh
As Mona works on WebGL/Browser it is important to get your asset polygon count as low as possible while keeping it looking the quality that you would like. There are many ways to do this depending on the 3D Tools you have access to and willing to learn.
This tutorial will look at two ways that go over a couple of the options available, Decimation/Remeshing is faster but less accurate, and Retopology takes longer to do but very accurate.
Both of these approaches will be looked at below.
Many 3D Creation tools have added these features in one way or another. Zbrush, Blender, even Rhino (since version 7.0 with the Quad Remeshing tool) have tools that allow you to quickly reduce the amount of polygons on a mesh. The main issue with this approach is that the result is not 'clean'. So in some cases texturing or modifying the asset may be more difficult, and the result is possibly not as optimised as it could be.
So it depends on how important and big is the asset (at least to start) and how much time you are willing to spend on it.
The Decimate and Remesh modifier can be applied in the Modifier panel in Blender after selecting the model in Object mode. Once applied you can play with the large number of parameters to get the best effect on your mesh.
Adding a Decimate or Remesh modifier to your asset.
Here are some examples of what Decimate modifier does to simple shapes.
Using the Decimate Modifier in Blender
And some more examples using the Remesh Modifier.
Using the Remesh Modifier in Blender
The Remesh tool is in the Object Data Properties panel after selecting the mesh you'd like to remesh. it is sometimes more accurate than using the Remesh modifier.
Using Blenders Remesh settings
Note the Sphere on the Remesh tool doesn't look any different as this asset did not work with the Remesh method. The state of the original asset will usually decide which method is the best to use.
These basic shapes are not the best examples of assets you would want to Decimate or Remesh, but they do show the results very clearly. And each asset requires it's own approach. There is rarely one method that suits all assets. The goal is the same however, your want the polygon count as low as possible, while keeping the shape of your asset.
Retopology is the approach of basically tracing the high polygon asset manually and creating a low polygon model. You have complete control where every polygon goes and you can make it 'neatly' with good edge loops so you can add/reduce polygons as needed much more easily. This approach also makes it much easier to unwrap for texturing.
This approach is basically required for characters as posing and animation usually plays a part, and the edge loops play an important role in how easy it is to rig, or apply to a skeleton, as well as how good it looks when posed or animated. So keep this technique in mind if you are looking to create VRM Avatars.
The simplest way to do this is by creating a separate model, a simple box or a single polygon plane for example, and then use the snap tools to place and extend polygons over the high polygon manually. This does take longer but the results are also much better and cleaner than using decimation or remesh tools.
As this is a more complicated approach, I will add two videos below. One is reducing a curved asset made in Rhino in Blender, and the other is one of the Workshops that cover optimizing Photogrammetry. This also covers 'baking' the high polygon asset onto the low polygon asset so that the textures can help pushing the detail to basically represent the original model exactly with a significant reduction in polycount.
Optimizing an architectural asset in Blender.
Optimising a complex Photogrammetry asset