Some thoughts on Modeling approaches

In CAD software, different approaches to modeling are used.

Most successful MCAD systems follow the feature-based approach, e.g. CATIA, Pro/Engineer, Inventor, SolidWorks, SolidEdge. In these systems, you define the complete modeling (design) history of a model or assembly of parts, by starting from a basic sketch (often 2D), which is constrained and made parametric and which is used for 3D model generation (extrusions, sweeps etc...). Further manipulations include blending, fusing and making holes. To change anything to the final product requires making changes in the history, but as long as the whole chain of events is maintained, it can be easily controlled with a chosen set of parameters.

From a CAD/modeling point-of-view, explicit modeling is an interesting approach. After the strong rise of the fully-parametric history-based modeling with features, people started to understand some of the limitations: to make minor adjustments, you have to dive into the modeling history tree, make adjustments and hope that the end result will come close to what was intended. But you have to work indirect.

An "Explicit" modeling approach removes the history tree and makes all modeling changes directly, on the end result.

Right now, the newer generation of the history-based systems introduce also some of the direct-modeling possibilities that "explicit" modeling uses, so you get a mixture of both.

These two approaches have their merits: generating parametric families of related parts, with the possibility of adapting to a chosen set of controls, seems the natural force of feature-based modeling, while direct manipulations of a model (possibly reverse-engineered) seems more natural with explicit modeling.

And what about Architectural CAD?

I guess that in the field of Architectural geometrical modeling, we are now in a transition period where many people embrace parametric geometric modeling, using systems such as Bentley Generative Components or Rhino+Grasshopper. And the whole core concepts of Revit are tied to this MCAD foundations (constraints, sketches, assemblies). Historically, the AEC CAD world has been one generation behind MCAD (at least in the approaches of CAD). We could thus assume that after the rise of these systems, some more "explicit" modeling approaches will be introduced.

In fact, the reason why SketchUp is so popular is the direct approach to modeling. You do not have to with the complexities that are inherent in BIM, but model how you want. That said, the ease-of-use of SketchUp is also its bottleneck: you can not make proper architectural model which carry full building information.

So where does that lead us? Be open-minded. Try to use different approaches. Try to apply them in your architectural design process. If the tool has at least some influence on the outcome, it might be good not to stick to a single tool. People working in SketchUp don't create the same designs as people using Rhino and certainly not when using Generative Components or Grasshopper. By trying CoCreate, or Inventor, or CATIA, you will learn at least a thing or two, which makes you more rich in how to tackle ideas and how to materialize them with CAD software.