I am practicing hard-surface modeling, mainly for machine kind of stuff. I read somewhere that if you are not going to deform a mesh, it is fine to have Ngons in it, which suits me well at the moment. But I know I will eventually get into character modeling. I should learn to make good topology as well.
As I want a clear picture of the workflow I want to develop, I need advice on whether this dream workflow of mine is good or not, and what improvements I can make. I know every workflow has its pros and cons, I just want to know if my intended workflow suits for general purpose (enough) or not, and would appreciate some feedback and advice to help me make sure I am on the right track
According to my current understanding, a good way to model hard-surface stuff is a hybrid of sub-surf and boolean modeling. And my understanding of a series of good steps of modeling are as follow:
Block out the shape using basic geometry (all quads)
APPLY different levels of subsurf on objects according to its needed detail so I have enough edges to work with (still all quads)
Start deforming the shape, add, drag, move, scale the edges etc (still all quads)
Use boolean to cut shapes, and try to adjust the structure to remain all quads
Does this work? What improvements can I make to it? Is there a much more general and better way to do it? Or a better way to think about hard surface modeling?
This is a step to failure. Ideally your mesh should have the minimum amount of geometry to represent the shape you want and to be controllable. Its easier to edit 3 vertices than to tweak 20 of them.
There are many workflows for hard surface modeling. The one I use include:
*Clean subD modeling. Modeling with clean edge flow and minimum geometry to get details.
*Boolean+bevel workflow which works especially for blocky objects.
*Boolean+zbrush dynamesh and polish, which is ultra-fast for generating high poly models for baking
To iterate, sub-d should be applied as a last step to preview your mesh. Not to generate more polies for editing. The only time you collapse your Sub-D modifier is when you dont have enough geometry on a complex surface to handle small to medium cuts.
Minimal geometry and bevel is a good approach IMHO, at least for mostly flat surfaces. Using crease and/or hold loops can solve subdiv issues where rounded and flat parts connect without having to worry about topology.
Where things get tricky is when you need to cut into curved surfaces. Then you need to have clean topology for subdiv to behave nicely. The issue there is less whether it is all quads or not, but more that the faces are not stretched / of different size, which is not easy when you e.g. cut holes much smaller than the mesh you cut into.
There are some tricks for this situation using a shrinkwrap modifier. One is to keep a copy of the uncut mesh and shrinkwrap the cut mesh onto it (after the subdiv modifier) to smooth out the irregularities the cut introduced.
If there are better methods, though, I would also like to learn about it.
So ideally for a subd workflow, I should use modeling tools to get the basic shape, add subd but don’t apply it, crease, create holding edges or bevel for rounded corner, adjust the structure the way it is needed for detailed boolean cut to try to remain clean topology.
But if that’s the case, how do I adjust the structure for boolean cut? Does that require using a lot of knife and fill tools? I feel like I would frequently mess up the part if not the whole mesh to add enough loopcuts for boolean cut, could you tell me how you normally achieve it?
Final question, related to the second workflow you mentioned, is it possible to create rounded corners with a heavy boolean workflow without sub-d and retopologizing ? My current horrible way of doing this is bevel and add segments till I feel that is enough, which can’t be the right way.
“how do I adjust the structure for boolean cut”:
On a flat surface, you only need to make sure your geometry is even and minimal, after the boolean cut, you clean up the messy area using merge and knife tools and then check smoothing. For curved surfaces with complex details, sometimes you either have to collapse the sub-D modifier or use the method mentioned by @omgold which works well.
“Final question, related to the second workflow you mentioned, is it possible to create rounded corners with a heavy boolean workflow without sub-d and retopologizing ?”
Retopologizing what exactly? Why do you need to retopologize your hard surface models? That’s just extra unneeded work. Build your models carefully and you dont need to retopologize anything. For a non-subD workflow, you create large fillets and bevels by hand and boolean any details needed. Weighted normals modifier can be added to smoothen the mesh normals and make it look better. For the boolean details, I make everything non-destructive, boolean modifiers with bevel modifier to add bevels. Sometimes this doesn’t work and you just have to collapse the booleans and add the bevels yourself.
I should have said it more clearly. I meant retopologizing after crazy booleaning so I can use sub-d. But I guess, as you mentioned, it is unneeded if I build my model carefully.
I think I get it now.
Thanks for your answers, they really help.
