Are multiple shapes bad?

Hey folks,

I’m a coder by trade that is just starting to learn Blender for very basic models needed in some of my 3D android apps using libGdx.

I went through a tutorial today that showed a chair being made. To do this, a single cube was used and it was subdivided so that the back was raised and 4 legs were pulled down. Before starting the tutorial, my expectation was that 6 cubes would be used to construct the seat, legs and back.

I’ve noticed this practice of using a single object in most tutorials and I’m trying to understand why. Is this done for performance reasons or do people just find it’s the quickest way to make the models? Is it bad practice to use multiple objects? What if my object consisted of a square base and a cylinder branch?


It means you’ve watched bad tutorials if they didn’t specify why they modeled it as connected geometry.


Most objects in Blender consist of object data, with corresponding

  • object mode in the 3D view and object properties in the properties editor, and
  • edit mode in the 3D view and object data properties in the properties editor.

Mesh type objects can consist of one or multiple mesh parts. You have three options to make a model: As one object consisting of one connected geometry, as one object consisting of multiple parts without connecting the geometry (they could still intersect), or as multiple objects. Same type objects can be joined to one object (object mode, select, ctrl+J), and you can separate geometry to its own object (edit mode, select, P).

One asset doesn’t mean you make one model. Modeling stage can use multiple objects and models, and it can output multiple objects and models when those are used to produce the final usable asset.

In general it’s better to make a model either with separate geometry or as separate objects, whichever is better in the workflow. That’s because it simplifies the structure of any of the parts which in turn means less polygons, it can make UV mapping easier, and looks better when a real-life object consisting of multiple objects is modeled that way.

Multiple objects can also share object data which would make modeling and texturing faster, or they could use instancing which means the whole object or a group of objects is referenced multiple times which also saves memory. How much? This much

What can limit the options are the requirements for the model from the end use, or pipeline stages before that. Deforming geometry should be connected, texturing might be easier in some cases if some parts were actually connected, and the target application might need the geometry connected like in many 3D printing applications. But unless you know that you need connected geometry for something that is clearly separate, make them separate. Consult the documentation of the target application and read about requirements and optimizations.

Cubic base and cylinder branch could be made with any of the options. Connected or unconnected geometry, low poly, high poly, both. Whatever is needed.

Thank you for the great reply! Very helpful!