This has been shown before but I thought I'd chime in since some people seem to be making it much harder than in needs to be by working on a solid Sub-D model.
The benefit of using a reference with a stack means you can work on a really simple model and get instant feedback. Don't worry about quads or tris, in fact, adding verts without edges is usually better. Even though the high-res is pretty sloppy all you really need to worry about is how it looks shaded.
Another added benefit to this is that you can go in at any time using the spline and fatten up your edges if they're not turning out the way you want in a render or bake. And you have a low res without a bunch of extra edges that will be easy to optimize for a normal map render.
http://www.valent.us/joakim/HolosightModel.maxhttp://www.valent.us/joakim/HolosightModel.obj
Replies
Definitely the smart way to work.
I'm not sure if this is ever the approach I would take. But that's just me personally. Maybe if I was doing movies or CGI that didn't involve as strict polycount limitations I would do it this way, but for games, no.
It sounds to me like it's a trade off of speed vs. quality. Not to say your result doesn't provide quality when shaded, it just has a lot of unnecessary poly's.
I'm interested in hearing people's thoughts on this.
It really depends on what you're talking about here.
If its a lowpoly mesh, sure you have to take the extra care to make sure it is constructed well, optimized, etc.
Highpoly for baking nm? Do whatever works, optimizing a hp mesh, or doing any other time wasting thing that doesnt actually affect the visual end result of the mesh is a total waste of time, there is absolutely no reason to do it.