Here’s a hypothetical example that I would appreciate some comments on.
If I wanted to model the head of this screw (see attached image), would I be better off using a cylinder with subsurf and subdivisions, or should I generate a curve to fit the head’s profile and spin it to generate the shape?
Are there situations when one option should be used over the other (complexity of shape, size etc.), or is it always entirely subjective?
Given that a screw is usually a tiny piece of an image (eg part of a much larger model), maintaining a low poly count is probably desirable, so in that situation I feel inclined to use a cylinder. Or am I wrong?
If you use a subsurf modifier, your poly count goes up. So I think it would depend on the situation. I think the key is: how closely will it be seen? If you are doing a close up, your render will look bad if the screws look like they have square sides. So you wind up using subsurf, or a lot of steps when you spin the profile. Either way, to get it to look good you use a lot of polys.
The further away the screw head is from the camera, the lower the poly count you can get away with. In extreme cases, you wouldn’t need to model them at all, just paint them onto a plane, and use a bump map.
If its a long way from the camera and you only want the head to show, spin the head profile with 8 to 12 steps, set smooth and normal map the screwdriver recess.
Closer to the camera it would be better to model with more detail, especially if the screwdriver recess is shown in profile. A normal map won’t show the correct silhouette.
I’ve found, with my own wonders over this, is that there is no 100% perfect way to do anything. It’s all about which route gives you what you want with the least amount of effort.
I would model the head, by starting with a cube with subsurf and do tha same with the shape beneeth, But i wouldn´t model the “screw” into it. I would make that by a curve.
