Where to Start?

Hello everyone!

Let´s talk subdivision. To be more specific, have you ever had any of these questions?

How do we know when to add more geometry?
How do I know when to add more edges?
How do I know when to subdivide or smoothen the mesh?
Or, what is subdivision 3D modeling?

If so, in this article, we´ll dive in and explain how to use subdivisions efficiently.

First, we need to define what is subdivision modeling.
Subdivision is the process of creating multiple polygons from a single polygon, while maintaining the model forms.
We will use subdivisions to make our models appear smoother or to advance them further for details.
We will also call it the subdivision modeling technique or Sub-D.

Non Subivision vs Subdivision

Now that we know what subdivision is, here are a few examples of how to use subdivision efficiently.

In our first example, we can see a simple object and a few possible results.

1. Non-Subdivision approach.

If we were to add edges with a cut tool, eventually, the object would keep its structure in a smooth preview. The original shape remains unchanged.

We can also notice that with larger edge count, we do get the object’s initial state back.
Furthermore, the object will have more geometry and become more stable.

We have few examples. As we add more edges the edges get tighter, and we get our shape back.

  1. One subdivision level
  2. Two subdivision levels
  3. Three subdivision levels
2. Subdivision approach.

However, instead of adding edges manually, let´s use Smooth operation.
In short, we will subdivide the model and double its edge count.

This time around, the object will keep the curvature from our Smooth preview. As a result, we keep the curvature, and get more edges to hold the potential details.

The question is, why do we want to subdivide it that way?

Let´s take a look at another example.
What if we need tiny details on a curvature?
How would we achieve this result?

If we add edges manually the object will tighten up to the point that we get the original shape back. We do not want that.
What we need is the curvature from the smooth preview.
In short, we would like to keep the curvature but add more topology.

So if we subdivide it a few times, we will have enough geometry to sustain details on a curved surface.

In that case, our result looks like this:
Small tip here. Smaller the details, more subdivisions we need.

Subdivision and Details

Where else can this approach be useful?

The same approach also can be used to establish primary forms.

Let’s say we would like to keep the smooth corner, but we would like to have more geometry for the details.

In that case, we would subdivide the current state of the object. Our edges will remain curved, but we will also have enough topology to create stable details.

Should we model with Smooth preview on or off?

It depends on what type of surface we are creating.

If the smooth result is all that matters, then we can model with subdivision.

However, if the non-smooth preview version of the object is the final shape we´re aiming for, in that case, we will model without a smooth preview.

Finally, we can answer the questions.

How do I know when to add more edges?

If the topology is stable even with lower polycount, we do not need to add extra subdivisions.

When we notice that we struggle to maintain the shape with the current polycount, we will increase the level of topology. That way our mesh will be stable and we will have enough room for details.
First example is topology level for mid type details and smooth corners, where second is topology level for higher details and sharper corners.

Common mistakes

Here I would like to mention common mistakes to avoid when it comes to adding a subdivision.

If we have established edges that cause pinching, adding more subdivisions will not solve our problem. 

Instead, we would make sure our object has enough subdivision from the start before we create our detail. That way we avoid pinching, and maintain the level of geometry.
Level of subdivision we choose, will also depend on the purpose of the object.

As a general rule, objects that are closer to the camera, ideally will have denser topology. Objects that are far away from the camera, part of the background, or simply not clearly presented, can have lower polycount. In that case minor pinching will not be issue.