Here we go! I've been itching to write up an article on this kind of stuff for awhile, but.. like everything else, I don't have the time. But I can take a small break to answer some of these for you. :)
So first question, when are quads, tris, or n-gons better? Um, first off, n-gons should ALWAYS be your last resort. Hell, if you end up with a 5-sided poly, It's best to split it into a quad/tri than leave it as an n-gon. I have heard of some people saying that pentagons actually work when subdividing, but I've never come across a situation where that was acceptable.
When modeling anything that you intend to drop into a NURBS or want to subdivide, you will want to model strictly in quads. The occasional tri is acceptable, though. How do you know if a tri works or not? Good way to tell is drop your base model into a NURBS, and if the normals look good (all smoothed and no funky artifacts), then you should have no problems sub-diving further. I have yet to come into a situation where an all-quad model returned surface errors when smoothed. Even if the base model looks wacky, it should work.
As a note to NURBS and SubD: Use supporting edges! when making supporting edges, make sure the loops go ALL the way around the model. making a cut on a single poly will produce n-gons. And NURBS hates n-gons. Yes, it will produce a lot of extra geometry. Better that than funky normals.
I guess this eases into the knifing question. When making cuts, you can honestly cut wherever, as long as it suits the form that you're trying to produce. If you create n-gons, be prepared to fix it, or make loops from the get-go. I will admit, getting out of the habit of using n-gons took me awhile. They are easy and quick, but they produce bad results. You just kind of have to force yourself out of it. I will try to do a write up on anti-n-gons in the future.
This is all the time I have today. Will answer the rest when I get home. :)
No problem :)
So, normals. Are you asking about smoothing groups? Like when to break the normals and when to unify (based on smoothing angle)? Normals are interesting. Currently the workflow I take with normals in video games are probably only slightly different than film/mograph. Probably won't be using many textured normal maps in mograph than as you would in games.
For important, or high visual objects, I put all of the normals into one unified smoothing group. Reason for that is, if you have a textured normal map and you apply it to a model with more than one smoothing group (groups of normals smoothed at different angle thresholds), they conflict and screw things up. That isn't to say you can't or shouldn't. You just need to plan and/or use best judgement when creating your smoothing groups.
Honestly, for mograph, I rarely had to think about normals much. I wasn't doing any fancy tricks with them, or using normal maps. I mostly just applied a 45 degree smoothing angle to everything to give it a generally accurate look. :P If I wanted a lowpoly-look, I'd set the smoothing angle to 0.
Right, so Phong, Blinn, Ward, Oren, Lambert, etc..
(when I used C4D r11-12, I could never select Ward, Oren, Lambert, and others. I guess they weren't available?)
I'm honestly not sure of the technical aspects of these spec types. I just know that they are different methods for calculating specularity. Phong and Blinn are good for plastics and some glass, while Ward is anisotropic and good for metals. Lambert is your general matte shader with a high roughness, low reflectivity. Oren is usually used for rough surfaces. Also, Phong and Blinn have pretty good highlights (or rather, controllable highlights).
Let me know if there is anything here that you might have questions about (not sure If I made them more complex for you than they already were). Or if you have any more questions about anything else. Cheers!
Do mean mean how to go about cleaning them up? That's always a pain in the ass. I've been forced to get better at that ever since we've gotten interns in the office here. They'd hand their "finished" models off to me and I'd spend an hour or so cleaning up their mess, and then a few mins after lecturing them about the need to make clean models to begin with. I guess the same goes for stock models =/
Erm.. I think it first takes a good understanding of modeling on your own in order to become comfortable with cleaning up someone else's model. Because you don't really have any documentation available to you on how they modeled it, and why they did certain things. Just like uncommented code. Though with modeling, if you understand the methods, you can eventually pick it up with ease. Though there are times, I see a model and think.. "what the FUCK did they do to get this?!"
I understand that people use stock models because they are short on time and/or don't know how to do it themselves. Hell, I used to do that. But I got fed up with wasting money when I knew I could learn and get the model for free. It just meant that I had to spend some serious time to work through learning how to model well.
The trick? Practice, practice, practice... and more practice. Just pick a ton of really easy objects, and model them until you get into a routine. Then pick a little toucher objects. And so on. After modeling ~80 or so simple objects (restricting myself to quad-only), I was moving on to complex car parts with no trouble at all. Hell, the routine mentality works for just about anything. The same went for UV unwrapping for me. Was, by far, the most annoying aspect of 3D to me thus far. And now I can UV at ease.
Sorry for the long-winded post, and I know you didn't ask about some of that stuff, I just figured might as well share as much info as I can, regardless. :)
It looks like you're new here. If you want to get involved, click one of these buttons!