Loops vs Grid
polycounter lvl 8
Hopy this isn't too basic, havn't found a clear answer to this problem.
I recently trying to really get into 3d this time and i'm wondering why the general consensus seams to be to do the fencing with support loops instead of just making a grid.
I know that you, at times don't want to a grid-based mesh for reasons, but if you do, aren't support-loops unnecessary?
Example attached, thanks you!
![2022-06-19 15_43_47-Blender_ [C__Users_jbrem_Documents_Blender_Blender Files_DesignJammy.blend].png](https://us.v-cdn.net/5021068/uploads/NDJN1428Y6L4/2022-06-19-15-43-47-blender-5bc-users-jbrem-documents-blender-blender-files-designjammy-blend-5d.png)
![2022-06-19 15_43_03-Blender_ [C__Users_jbrem_Documents_Blender_Blender Files_DesignJammy.blend].png](https://us.v-cdn.net/5021068/uploads/SIPOAT73B3W7/2022-06-19-15-43-03-blender-5bc-users-jbrem-documents-blender-blender-files-designjammy-blend-5d.png)
Replies
Doesn't matter. Do what gets the job done.
In general, more geometry makes it harder to edit shapes, just because all those edges make any changes more complex. Simpler geometry for subdivision is easier to manage.
If you're modeling to bake normal maps, the topology mostly doesn't matter, as oglu says. Only matters how good of a bake it makes.
Really good advice here.
To further illustrate Eric's point: pure quad grid topology does tend to have a relatively consistent and dense geometry distribution. Which might not require support loops but can make it difficult to edit the basic shapes and control the edge sharpness. Most base meshes can be simplified by using support loops to concentrate geometry where it's needed and different topology layouts can be used to direct those support loops around the shapes.
Grid topology and support loops aren't mutually exclusive either. There's only so many times the base mesh can be subdivided, to sharpen the edges, before it becomes non-human readable. Which is why support loops are a fundamental part of subdivision modeling.
Support loops can also be used to control the sharpness of subdivided edges, without having to add any new geometry. Wider support loops produce softer edges and narrower support loops produce sharper edges. It's also possible to vary the width of a loop to produce an edge that goes from soft to sharp or vice versa. Something that would be a bit more difficult to do with a dense, quad grid mesh.
Catmull–Clark subdivision basically adds geometry and averages [smooths] the surfaces. So even with traditional subdivision modeling techniques, it's a lot easier to just create the basic shapes, add support loops and let the subdivision do the rest. While there still are a few situations where strict quad grid topology makes sense, most contemporary subdivision modeling workflows have evolved to become more efficient and user friendly.
It's way too easy to get bogged down trying to resolve complex loop flow before establishing all of the major forms. Which is why it can be helpful to focus on creating accurate shapes that flow together, before adding a bunch of support loops that increase the overall complexity of the model. A lot of the manual loop routing work can be automated by using bevel / chamfer modifiers and resolving the topology flow issues early in the block out.
As far as game art goes, it's also possible to bypass a lot of the high poly modeling process by using a boolean re-meshing and edge polish workflow in ZBrush. Though this workflow does have it's own drawbacks in terms of reduced manual editability and generating dense output meshes.
Streamlining a subdivision modeling workflow really comes down to learning the fundamentals then figuring out the tradeoffs and deciding which approach fits best with the specific goals of each project.
Frank! I think we need to allocate a sticky for all your gems of wisdom 😀
Jesus christ, thanks so much, especially @FrankPolygon!
My issue with support loops is that with lots of different details it becomes kind of a puzzle to resolve everything into another in the end, and i'm working on my portfolio right now and need some good looking topology, although i know that even ngons or tris could be considered to be "good topology". I just don't know what ppl who would hire me would look after. I always felt like grid-like topology looks clean af and gives the best opportunity to add details, but it's not suited for EVERYTHING, so my takeaway is to use the best out of both worlds where needed i guess.
Thanks!
Frank's sketchbook is a gold mine, been front-paged a few times I think
https://polycount.com/discussion/221392/sketchbook-frank-polygon/p1
Oh duh! yes of course....cheers Eric, put it down to a complete brain fade - must be getting old 🤥