Noob question about smoothing techniques
vertex
Alright, my question may be a blasphemy to (some of) you, but forgive me - I'm total newbie and I feel like I'm missing something very important here... First of all, I googled, I searched forum a little, watched some vids and I still can't, how to say it,... it can't "click" in my head in right place. (maybe I'm just not made for 3d modeling...)
So, why smooth models using supporting loops and subdiv, which requires a very good topology (which is pain), when you can use scripts like Crease+, or Zbrush polish?
please, don't hate me
So, why smooth models using supporting loops and subdiv, which requires a very good topology (which is pain), when you can use scripts like Crease+, or Zbrush polish?
please, don't hate me
Replies
There's subdivision surface that needs pretty good topology, there are also applications that need good topology e.g animating the model.
Smoothing in the typical sense is averaging the surface, it doesn't necessarily require specific topology.
But seriously speaking, there is a good place and time for both approaches. If you are working with 3d (non concept), you need to know how to make decent topology anyways, so I don't think that having a good understanding of subdivision modeling hurts.
what I'm mean is smoothing edges like that:
But I've seen vids where people just use Dynamesh and polish in Zbrush - like here:
http://polycount.com/discussion/168610/proboolean-dynamesh-hardsurface-workflow-tutorial/p1
(and I'm not talking about booleans part, just about smoothing)
or ShapeShifer script like here:
in both the topology would not be good for subdiv
I get that topology is important for animations, but as I understand animations are made on low-polys (?), and I'm talking about high polys with many details.
Not sure how much related is this to your question, but if you would want to make a game ready hard surface asset, its just much easier to make the lowpoly, if you made your highpoly with subdivision, because you just remove that, and your support loops and you got your lowpoly.
I can also imagine a Zbrush game art hard surface workflow where you make a lowpoly, put it in Zbrush , you add your creases and stuff, and then you bake that to the already existing lowpoly as a normal map.
That explains a few things. Also I found a topic here on forum (quite old but still...) that help me. I would love to download some models with good topology to study, maybe I'll find something. (especially something like a cylinder with holes and spikes, that's a nightmare)
If it's static such as set dressing, non-interactive. Go for whatever workflow gets you the best result, quickest, with an optimised final asset. Good topology has a place here too though, it's easy to maintain and if you're in a setting where you're pushing models to different software for modifications then good topology usually results in a fairly painless transfer (n-gons will triangulate differently in different software). There's more reasons but I forget, but also it's just really sexy seeing those wireframes conform to shapes, they're really fun exercises to do!
If your object needs to deform, such as a character running, text bending, or cloth with dynamics. You need to pay much more close attention to your topology, else you'll get very undesirable results.
The very first question that you should ask should be, if this is for a static image or rendered video, or game, or how it would be used. Because this will determine how it should be approached.
Sadly, not even the best modeler on earth can make you understand by typing words. You'll have to seek out the theory behind different subjects and try most of it for yourself in order to learn it all. Then post your work, get criticism, take it in, and learn step by step. Challenge yourself.
TL;DR: Every workflow ever has their pros and cons. Learn them all if you can. You'll be a better artist for it. Use them for their benefits if you want. Some workflows have cons that make them incompatible with paid work for certain clients depending on their software and requirements.
These are just different ways to achieve a desired result. They all have their weaknesses and strengths, and I would suggest you learn how and why subdivision does what it does, so you'll always make an informed decision in whatever you choose to do. Part 1 and 2 of this is relevant:
Opensubdiv, Crease+, etc. are fast and easy to implement. That's pretty much the only positive.
I haven't heard of one such tool that isn't proprietary to either the parent software or a specific file format. If you're a pro and a client wants the highpoly model delivered in a modifiable manner and they use a different software than you, then your highpoly is useless to them.
The same goes for Zpolish. It's quick, proprietary and doesn't go well with an iterative process.
All of the methods I've heard of, save Zpolish are affected by the math of subdivision to achieve the desired smoothing effect. This means that even though Crease+ may be faster to do than adding your own supporting geometry, the underlying model will still be affected by bad topology.
Personally I prefer planning out my model so I'll get good topology and adding my own supporting geometry simply for the control and reliability it offers. If someone wants to buy a highpoly off me, then it doesn't matter what software they use as I can send any file format with all the supporting geo, and it'll work perfectly in any software. The con here is that this is probably the most time-consuming method of all. But it's not harder to do. Everyone doing highpoly work should understand subdivision, and so should be able to plan their model accordingly, making it easier than it looks.
As I understand how subdivision works, I don't have to work with subdivision. I add it to my model when I'm ready to bake.
Curiosity is a good thing. Keep it up!
Sorry if I come off as pretentious, I'm not good at being concise. I'm always up for helping out a newbie, so send me a PM if you want