I have a modeling question. In many timelapses I have watched that involved a lot of hard-surface modeling, the modellers have not extruded additional peices of geometry from the main section, but rather created a new object and let it intersect with the old one. Is that a good practice? Why or why not?
I assume it is because Chris Kuhn does it in the sci-fi hatch course.
I hope it is good practice. When I was working through the Mesh Modeling Bootcamp, a few times I separated meshes where the instructor didn't. My reason was that I was trying to get edge loops in all the right places to make editing easy. If everything is one large mesh, and different parts have different shapes, it's easy to end up with a confusing mess of edge loops. If some object seems to have a few big parts (cubes for example), and they need to be modified a lot, it seems easier to me to separate them. For making a still image, it shouldn't matter. You need to be careful with games and animation since it can have important effects. I hope some experts will comment.
This video is an example. The guy doing it makes a lot of other timelapses, all using that same technique.
It totally depends on the scenario and your workflow preferences, but a general rule of thumb that most believe to be good practice is "if the objects come apart in real life (even if you need a screwdriver), they should be separate objects." Of course by this rule, meshes shouldn't intersect and if they do they should be the same object, but that's not always practical or necessary. I'd say if the intersection is visible, make it a single mesh with no intersection. If the intersection is not visible and never will be, the meshes should be separate objects. You should avoid intersections wherever possible and definitely wherever it can be seen by the camera, just because it's physically impossible and looks unnatural.
I'm not an advanced modeler, but here's my personal experience. I think it's not always a bad thing to let your meshes intersect if it's not an important area. However, when it's possible to keep good topology, I try to make my things connected, so when I add a subdivision surface modifier, the transition becomes very smooth and you can edit it with edge loops if you want to have just a little bevel. An alternative is having just a little distance between two meshes and adding a back face for nice occlusion shadows, just like Kent explained in his "Wall of Mailboxes" course. If you really want a very nice and clean mesh, I would use on of the two options I mentioned.
I hope this answers your question!
I mean for mechanical objects that will be mechanically rigged.
I don't know anything about rigging, so unfortunately I can't give you any advice on that topic.
Same answer really. riggin really doesn't change things. and tbh, mechanical riggin is easier with separate objects.