Good practice for Subdivision Surface modifier workflow?

[Myn]

Hi everyone,

In this lecture, we’re re-visiting the Subdivision Surface modifier to create our lamp base.

When adding the modifier, Mikey points out that the mesh shrinks as it smooths out the vertices. The 3D cursor below shows his base width before the modifier:

This is an easy fix; he scales up the mesh until it is back at the desired width. Nice and simple. He also edits his bezier curve to have a more pronounced lip, which the modifier smoothed out.

I have been questioning why we would use a Subdivision Surface modifier over just making an object high poly through techniques like bevelling or curve resolution. My understanding from reading around is that the Sub Surf modifier is “not real”, or “not destructive”. I.e. we can model something as low poly, use a Sub Surf modifier for the high poly look, and still have the option to remove the modifier later to retain our low poly mesh. This would be useful for exporting to a game engine, or Mikey hinted at “normal mapping” in a future lecture.

However, we cannot just remove the Sub Surf modifier in the example above. We have scaled the mesh up to make it look better with the modifier, and edited our curve. The mesh data has been modified, and we would have to go in and edit it to get our original look back.

A better example might be a chess piece from the last section:

If we remove the modifier, the object is out of proportion. We might have scaled out edges, or even added in more geometry like loop rings to control edges, in order to make the object look good with Sub Surf. If we want to remove the modifier, we would need to scale down some edges. If we want to add the modifier again, we would need to scale them up once more.

So I’m struggling to understand the Subdivision Surface modifier workflow. It strikes me that this workflow does seem to be destructive (even if the modifier is not), as we’re having to edit our mesh data to get the desired outcome.

I’m wondering if a good practice is to create two versions of every object; one low poly, and one duplicated for adding a Sub Surf modifier. This seems like the best way to keep a low poly version, but then creates problems if we want to start making substantial differences to our object. (E.g. if I wanted the pawn above to have a square head, I would have to edit both meshes).

My questions are:

1. Am I misunderstanding the Subdivision Surface modifier workflow?
2. Is there a good practice for using this workflow if we want to keep a low poly option?

I hope that makes sense, and thanks in advance.

[FedPete]

I think you are overthinking this, in this chess piece challenge. It’s just a way to teach you Blender stuff, and not the optimal way of working.

If you have a plan, depending on a lot of factors, then you decide on how you create and maintain (reuse) an object. Using the Blender modifier tools. This knowledge of how to do things is based on a lot of practice.

• If you create a game prop, then try to go low poly, where the details are still recognizable.
• If you go for realistic close-ups, then go for high poly and even a sculpted version.

If I see your pawn example, even for a high poly model, it has too many vertices. Lots of calculations. Especially when the pawn is presented at a distance, then details are less needed. Even your low poly model looks detailed from a distance. What I want to express, it only counts on how you are going to use the model.

If you say, I create something beautiful, to be used in another project. Then you are going to add a lot of effort, without knowing when or how to reuse it. This chessboard challenge is just a way of explaining things about blender behavior. Just follow the course, do the challenges, and practice a lot. Then after many, many hours of Blender, then you’ll understand better when to add modifiers, apply them or use different methods.
Blender is a versatile tool, many ways to solve the same problem.

Just have fun.

[NP5]

Not much I can add to FedPetes’s reply. ‘It depends’ there are no hard rules, You adapt workflow to the project and its intentions. You save early versions perhaps in case you may change course.

[Myn]

Thank you both for the replies.

I appreciate what you’re saying and completely agree, but I think I worded my question poorly so I’ll try to rephrase.

A question I’ve often had during the course is: “Why would we use the Subdivision Surface modifier instead of alternatives that add in geometry like bevel, curve resolution, etc.?”

I’ve searched that question a lot on the Udemy course Q&A, Stackexchange, YouTube, etc. which has really helped clarify the differences. But an answer that I kept seeing is that the Subdivision Surface modifier is not destructive, so the advantage is that we can just turn it on and off when we need to without affecting the base geometry.

It is that last part that I am struggling to understand. We cannot just turn the modifier on and off and have the high/low poly version ready to go. That implies there is no extra work needed. But when we add/remove the modifier we do need to manually make edits to the base geometry, such as scaling edges. That is what I was trying to illustrate with the examples above.

I think my question is: has this “just turn it on and off when we need to” been oversimplified? When we are told that we can just turn it off, is it implied that this involves editing the geometry afterwards?

Thanks again.

[bOBaN]

@Myn

I think my question is: has this “just turn it on and off when we need to” been oversimplified? When we are told that we can just turn it off, is it implied that this involves editing the geometry afterwards?

As always: it depends. if you are modeling just one thing / one object in isolation - small ‘volume’ difference doesn’t really matters.

For my hard surface work I usually can do that, because I start with bevel modifier which helps to keep the volume in check after subsurface.

Buuut… yeah… there are many ways in Blender… and almost all simple rules like "just turn it on and off when we need to” are oversimplifications.

(and overall my experience with subsurf is that you actually need to model your stuff with subsurf in mind to use it and have good results)

[Myn]

That’s really interesting, thank you bOBaN.

I’ve been avoiding using SubSurf in my own side projects, but I’m sure it will make more sense the more I use it (practice, right? ).

I understood why Mikey was adding and removing it in the chess set lectures, but when he started doing it in the animated lamp section it began to seem like a workflow, so I thought it was best to check if my understanding was off.

Thanks again all, I really appreciate the answers and advice!

[NP5]

Yes it is oversimplified as an expression. The key aspect is it is not destructive. It is possible to undo the shape alterations made to optimise the subdiv effects with relative ease in comparison with undoing multiple ‘destructive’ tools.
As Boban points out there are actually things to optimise the subdiv route that may well not be needed for destructive routes. Hence the fuzzy ‘it depends’.

I think a big problem is the (and it seems relatively recent to me) attitude of throw on a subdiv to make everything look less low poly. A quick easy way to bypass more high poly detailed working from the start.

[bOBaN]

@Myn

I’ve been avoiding using SubSurf in my own side projects, but I’m sure it will make more sense the more I use it (practice, right? ).

Yes it will!

@NP5

I think a big problem is the (and it seems relatively recent to me) attitude of throw on a subdiv to make everything look less low poly. A quick easy way to bypass more high poly detailed working from the start.

Yeah, now that current hardware and blender can handle more and more triangles - it is cheap to smooth out some imperfections or even “fix” some of the topology issues.

But! I have (almost) universal use of subsurf - I often use it to test my topology. Most of the time when topology has some issues - subsurf deforms the objects in really bad ways