Understanding the difference between CAD and surface modelers.

In a recent Blender Bros free newsletter by Josh Gambrell, key points about topology were discussed. He emphasized the use of different types of topology for various use cases. For example, in 3D printing, the primary concern is that the model is watertight or manifold and has sufficient polygon resolution for smooth surface transitions. In contrast, game development focuses on limiting face count due to resource constraints.

Reflecting on my experience in 3D since the early 1980s, I realize that some concepts are often taken for granted. One such concept is the difference between constructive solid geometry (CSG), which includes NURBS and CAD modelers, and surface modelers like Blender, SketchUp, 3D Max, and Maya. Additionally, there exist sculpting modelers like ZBrush and Blender's sculpting tools, as well as voxel modelers like 3D Coat, but are not part of this discussion.

It is important to understand the difference between these different types of modelers.

CAD modelers fall into two categories: parametric and non-parametric. CAD modelers use equations to define surfaces and objects, ensuring all objects are solid models—fully manifold and watertight. This is crucial for molding or 3D printing because it allows precise CNC tool cutting and ensures the object can be easily printed. CAD modelers offer high accuracy compared to surface modelers.

However, for 3D printing alone, a surface modeler like Blender can provide plenty of accuracy. CAD modelers tend to be expensive; software like SolidWorks, Fusion 360, or Onshape can cost thousands of dollars per year due to their advanced features like finite element analysis, parametrics, and simulations.

Parametric modeling in CAD allows adjustments to models long after creation. For instance, changing an axle's length or diameter will automatically update related parts. This feature requires advanced setup skills and significant computing power.

Non-parametric solid modelers like MOI 3D, Plasticity, and Rhino also perform CSG but lack extensive analysis tools or parametric capabilities. They require manual deconstruction for modifications.

Solid modelers excel at operations like filleting any surface, performing Boolean modeling (adding, subtracting, intersecting objects), and creating accurate offsets for hollowing objects—capabilities less robust in surface modelers.

Surface models are defined by a sequence of polygons all connected together to form a mesh. Using shading techniques, normals can be smoothed across the mesh to create the illusion of a smooth surface even when it is not so.

Surface modelers are typically better suited for creating organic shapes due to subdivision surfaces that ensure surface continuity. These subdivision surfaces are surprisingly good at generating smooth continuous surfaces, typically better than CAD modelers. For this reason, creating a car body using CAD is a difficult process which requires special CAD software to maintain the correct surface continuity.

Surface modlers can also facilitate easy deformations of surfaces using mesh displacement. So for instance you can add a true wood texture to a mesh that can be 3D printed. This is extremely difficult to do in CAD.

Surface modelers are not great at Boolean operations, or filleting between surfaces. Unlike their CAD counterparts, there are many instances where they will fail.

Surface modelers do not inherently maintain manifold objects and this can be a difficult process for those wanting to create printable objects in software like Blender. However, add-ons can help identify and fix issues. 3D print slicing software also has the ability to fix minor manifold issues.

It should be noted that the industry standard 3D printing file format called STL is in fact a surface modeling format. An STL file is constructed of many different triangle faces all connected together. For this reason, it is much easier to edit an STL file inside of Blender than to try and convert it to something a CAD program can understand. Many times that is not even possible in CAD.

Hollywood special effects rely on surface modelers for optimized scenes without concern for solids or 3D printing. Surface modelers are also superior for animation and rendering photorealistic scenes.

And of course gamimg depends on surface models.

For beginners in 3D printing, starting with a basic solids modeler like TinkerCAD ensures manifold and watertight exports. If one really wants to use CAD for engineering or even concept design projects, it might be beneficial to use CAD software like Plasticity or MOI 3D due to their one-time purchase models compared to the very expensive subscription-based CAD options.

CAD modelers can export surface meshes but surface modelers cannot export CAD data. It is possible for a CAD program to try and interpret mesh data but typically it's not an easy thing to do and you can end up with a bit of a mess.

When a CAD program exports to a surface modeler like Blender, it can generate difficult to edit meshes with extremely poor topology as it creates its own surface normals so that it will render correctly. Because of this, it is difficult to actually edit a imported CAD model inside of Blender-- unless it is a very simple model. That said, Blender will render perfectly an imported CAD model. Typically they come in as OBJ, STL, or other mesh format.

While not fully parametric, Blender's modifier stack offers a procedural approach similar to parametric modeling. Using this approach can provide a non-destructive model which enables users to quickly inefficiently make edits based on iterative feedback.

In conclusion, understanding the differences between CAD modelers (CSG) and surface modelers is important depending on your project needs—be it engineering precision or creative flexibility in creating your models.