Modeling with Quads or Triangles - What Should I Use?

courtesy of DT

Should I use quads or triangles on this model? The truth is, while it can come down to preference, there are some key advantages to using quads to create your models. It’s also not the end of the world – or your model – if you use both, but it is recommended to use as few triangles as possible to save you some major headaches down the line.

Understanding the different polygons.

For a modeler, there are only three polygons that matter.

Triangle

A triangle is the simplest polygon that is made up of three sides or edges connected by three vertices, making a three sided face. When modeling, triangles are typically a polygon type often avoided.

Triangles tend to pose a problem when subdividing geometry to increase resolution and when a mesh will be deformed or animated.

N-Gon

An n-gon is a polygon that is made up of five or more sides or edges connected by five or more vertices. It’s important to keep in mind a n-gon is typically related to a five sided polygon, but it’s not limited to just five sides.

A n-gon should always be avoided, they often pose problems at render time, when texturing and especially when deforming for animation.

Quadrilaterals

A quad is a polygon made up of four sides or edges that are connected by four vertices, making a four sided face. Quads are the polygon type that you’ll want to strive for when creating 3D models.

Quads will ensure your mesh has clean topology and that your model will deform properly when animated.

If it has been a while since you were in a 3D application or some of these terms seem new to you, take a look at the at this recent post covering key 3D modeling terminology.

So why choose quads?

When modeling with quads, the wireframe will have a much cleaner look and the model will be easier to navigate and edit. When you spend hours working on a project, you deserve to show it off. But, if the wireframe is messy you become very limited to what you can actually put out in the world for everyone to drool over. When you use quads you create clean, sleek lanes of polygons that are easy to follow through the model and provide beautiful edge flows that you can be easily modified by you or a team member. It is less distracting when you overlay the wireframe on a clay shaded model or even a textured asset.

It is also good to work in quads if you plan on passing the model off to someone else or a team. It is easier for someone to convert a model made up of quads to triangles than it is for someone to convert a model made up of triangles to quads.

Edge loops

Edge Loops typically are a continuous loop which no start or end point is determined. If you were to start to follow a loop from a highlighted vertex, you would eventually end up at that exact vertex. Edge loops are helpful to add detail such as wrinkles or folds, they can also be used to help define how sharp an edge is.

If an edge loop runs into a triangle, the loop has to end. This breaks the flow of the line and it’s no longer a loop.

Sculpting

So you want to add that extra little detail to your model, you better use quads. If you plan on taking your model into a sculpting application, such as ZBrush or Mudbox, It is best to avoid triangles as much as possible.

As stated previously, it is easier to predict how the geometry will be affected. Sometimes you may need to subdivide the geometry 4-5 times pushing your model over a million polygons. This is why you want to work with a predictable quad-based mesh. This also helps build a lower resolution version and accent the model using edge loops.

Subdividing

When subdividing quads, your results are fairly predictable. You have rows and columns made up of four sided polygons and it is easy to see where those polygons will be split in half once it is subdivided.

When you subdivide triangles, things tend to get messy. There really isn’t a visual flow to the model.

Smoothing

If you plan on smoothing your geometry or using a quick smooth preview feature, triangles will produce anomalies across the surface of the mesh. Because of the uneven amount of vertices, the triangle can cause blemishes or pinch the geometry. This similar thing can happen to geometry created with n-gons.

Animation

Quads produce cleaner deformations. Typically, artists will focus on areas where there will be a lot of bending and deforming, such as knees, elbows and wrists, and provide a little extra geometry that will benefit the rig and animation. With Quads, this is easily accomplished by adding or manipulating edge loops.

If you have a cluster of triangles in this area, it is harder to add or remove loops that will help benefit the animator. With Triangles, it is also harder to see a clean flow of geometry and they tend to produce sharp angles that can harm the mesh’s appeal. When it comes to animation appeal is important both to the model and the artist who provided the mesh for animation.

What about hard surface models?

When modeling hard surface models, the preference is still to use quads. The main reason: so you can quickly add and remove edge loops. When it comes to UV mapping your model, it is also easier when using Quads. There are less edges to clutter the texture editor when placing UVs.

Triangles are not a bad thing. They just have to be used strategically throughout your model. When using them on organic models, it is best to hide triangles where they will not be visible or in areas where very few deformations are happening.

A good thing to keep in mind when working in Quads is you can always convert it to triangles. There will be times when triangles will be preferred, an example of this is the final mesh of game assets or characters. In these instances most artists still prefer to work in quads and then convert their final model to a mesh built of triangles.

It’s always good to look at how other artists solve topology issues in their work.