/3/ - 3DCG

>>593598
>Thanks, this is immensely helpful. I come from a Quake level design background, where you could get away with making stuff from a bunch of primitives because the compiler would optimize it all later.
Meshes aren't like BSPs, anything that works like a boolean in poly modeling generally leads to messy results if you don't have a good idea of what you're doing, or have a way with dealing with the resulting geometry.

>I am trying to wrap my mind around the right way to do things in 3D modeling. I don't yet have even the ability to execute my vision of, for example, making a curved surface cut into a simple cube. I need to learn the fundamentals, so any tutorials about basic stuff would be helpful.
There are a variety of techniques you can utilize to execute any kind of model, although some are more efficient than others depending on what your reference is. In my example I used edge-extrusion modeling, where you just extend new polygons from a base plane, shape the overall form of the object, and close up the holes right at the very end once you're happy with it.
Another possibility is drawing splines to conform to more complex shapes, and then extruding geometry from them, boolean operations are much easier to work with when using splines, so you can just combine and subtract shapes to get the right form you need.

For your next question, you can "cut" curved surfaces into angled objects just by chamfering them, that is, you start by simply extruding out the bounds of the curve, and then you subdivide just the edges along the planes where the curve is supposed to occur. Do note that the spherical cut in the cube require a "quad-chamfer", and not all programs have by default, so you may need a find script. It's also fairly safe to boolean a curve out of a surface if it's flat along at least one axis (such as these arches), but trying to boolean the sphere into the cube would be a disaster, so it's better to use the chamfer method.