So this is my study diary.
Hello, I apologize for not having a sketchbook.
Since I'm not very good at it, it didn't seem like the right thing to do.
I've recently started learning about
3D modeling topology.
To truly understand it, I've created a personal exercise Inspired by Frank's posts on this forum

,
The idea is to start from hardsurface shapes ruined by Booleans, and then recreate a valid topology that the subsurface can digest.
No shape changes are allowed. No retoping to a new mesh is allowed.fixing the topology of a cube "tortured" by
Boolean operations so it works with a
subdivision surface modifier without changing its shape or running a full
retopology.
I'm sharing my progress here and I'm open to tips or suggestions for new, challenging shapes.
I'm starting to work on a "simple" shape (not for me)


Replies
I start recreating what I believe to be the basic loops.
Also: i can not really see if these n-gons are really coplanar.. so its not completely clear how this really looks like.. at least for me
Hi.
I didn't say I'm just starting out. I'm very slow; it's like moving pieces of a puzzle. So, I'm trying to send better images, sorry
I hope these lines help you better understand the orientation of the faces.
My intention was to bring loops in a circle around the "hole" made by the UV sphere.
Then find a way to "kill" some of the loops, because if I brought all the loops present in the hole, the resolution would be too high for the entire mesh.
So I have to kill some loops in the hole area, I suppose, and then, with E-POLES, change the flow orientation as I move away from the hole.
Too much theory
But two things are giving me a headache:
1) A circular topology clashes with a straight-line quad topology; I don't know how to reconcile them.
2) I have areas that appear to have many vertices, and even knowing the tricks for reducing vertices, I don't know how to make some of them "die."
For example, this. On the circular part, I have many vertices, but if I looped them all, I'd have TOO MANY edge loops that extend to places they're not needed.
So I need to find a way to lower the resolution, killing some edge loops, but it's a puzzle I can't solve right now.
Well, why not ? Cleaning up an existing mesh involves resurfacing, and retopo is just one of the processes available to resurface geometry - by snapping to a reference surface.
Overall this looks like a rather simple shape to cleanup/rebuild in about 10 minutes or so, as It really is just a matter of reducing the density of the concave sphere by about half, connecting things from there, and then adding support loops or marking bevel/crease weights ...
Hi Pior ty for your post!
The fact is, I've been trying to learn topology for a long, long time.
. If I had remade a mesh from scratch, it would have been much easier, and MAYBE something better would have come out of it. Just to be precise, I didn't take 10 minutes, hehehe
It's precisely because I'm so hard to understand these things that I'm trying to tackle them head-on. For years, I've escaped modeling, inventing endless software pipelines to avoid retoping and UVs. By doing this, at least in my head, I think, I'm tackling the problem.
I let my head work, and I search for solutions online, I'm "inside the question" .
At least I think this is better. Maybe I'm wrong, I don't know.
But you gave me an idea. At the end of this exercise, I'll also create a mesh from scratch. was a good exercise i suppose.
To understand a few things. Meanwhile, and we're at, I think, 4 hours straight. The topology has improved, I assume because the performance with subsurf has improved. But I think there are a LOT of errors. I don't know how good it is.
I've been doing 3D for years now and I've never gotten any better. I can't get into the details, I can't do what I want. I've escaped this for years, but by mastering topology, I'll be free to create what I like, without having crappy UVs that make the model worse. That's the goal.
i think Maybe it's getting better. But damn it, it's like writing with the left hand, I find it incredibly difficult. I don't know why.
That'd be different from your intended exercise of course, but what I am getting at here is that there is nothing wrong with changing strategy midway if one approach just doesn't seem to work as well as one may have hoped.
The design of this one random part is also quite unusual, with all these strong changes of direction thus making it not necessarily representative of a real world scenario.
I understand. Thank you for your time and your kind words.
Unfortunately, I don't have the deep understanding to understand or choose one method over another for modeling. In the end, I've always thought that the only thing that existed was modeling something with primitives, adding a subsurf, and supporting loops. That's it.
I didn't think anything else existed. This is very frustrating because it further highlights that I've spent all these years on the wrong path, learning practically nothing.
But if not even these exercises help me grow in modeling and topology, what's the right path? What path should I take to learn and improve?
For all these years, I've modeled a lot, but it seems I haven't grown one iota in my awareness, if I don't know topology or the world of modeling.
I'm very confused and frustrated; I just don't know which path to take at this point.
I don't think there's a right way or a wrong way, and attempting to fully model this part in subdiv is certainly an interesting exercise in and of itself. I am sure you've learned a few things by doing so already, and it is a also a good test to see what is enjoyable (and practical) and what isn't. And I am sure the more experienced highpoly modelers around here will chime in with suggestions and solutions.
FWIW I personally always find it frustrating (and borderline absurd) when modeling a part ends up taking so much longer that designing it. Having to spend too much time and energy on redirecting loops or solving flowing quads is a bit nonsensical when compared to the much more straightforward sequence of actions one would perform when modeling the same part in CAD from a drawing for instance.
Now this is not a suggestion to use CAD for this of course. What I am trying to get at is that if there's just too much friction being introduced by this or that process, then perhaps the more tangential approaches become viable even if they may feel a bit hacky.
Also I think it's always good to remember that at the end of the day it's the lowpoly that really counts because that's what goes into the actual game. Hence building a clean lowpoly with hard edges first (when possible/appropriate) is IMHO never a bad idea. It's never wasted work, and it can be leveraged later to either build a "good-ish enough" pseudo-high (using bevels or a round edge shader), or be used as a base for a proper subdiv model after removing a few edges. Win-win. Whereas worrying too much about building a high first can mean spending too much time on superfluous things.
Now of course this doesn't really solve the actual exercise of finding a working subdiv-only solution for this part, and some projects would certainly require a high to be built first and foremost in order to be art-reviewed. Just sharing a slightly more tangential approach that I personally find more flexible and most of the time faster (but with some compromises).
Thanks for the response and for the image. I really appreciate you taking the time to share your point of view.
I figured that solving a complex topology would give me more creative freedom. What I'm looking for is the ability to work without the stress of thinking: "I can't add details here, otherwise the mesh won't have a clean UV," or "if I add details, retopo will be a nightmare."
I want to have "mastery" over a model, and I can only achieve that by learning to handle any kind of topology. That's the whole point of my exercise: if I can fix a topology problem, I'll be able to create one... :)The approach you showed is a great solution for a professional workflow and a very useful tip for production.
But I totally understand the professional, problem-solving approach, and you're absolutely right. From your smart point of view, I can see how my reasoning must seem a bit like driving against traffic and with little logic.|
Thanks again for your time and for the image, which was really clear and helpful. Best forum EVER Polycount!
eg. if you're modelling consumer electronics or cars designed in the 90s you likely want to be using a NURBs based modelling toolset.
More recent stuff is more likely to translate to subdivision modelling or modern CSG based CAD
if it's a casting - use booleans and zbrush
Yes, and that's exactly what I want to do.
My goal will never be low-poly production for games or animation. My only aim is the production of 3D hard-surface, and later, organic and artistic models.
An example would be a rendering of a spaceship flying in a valley. Or a steampunk boat sailing in a sea. My interest is to produce "artistic" renderings (I apologize for the exaggerated use of the word).
To do this, I need to be able to work on detailed models but also understand topology. I need to be able to combine multiple methods into a "certain" workflow. And I've never been able to find one.
My goal is to use CSG modeling for hard-surface models. But then I absolutely must know topology because I'll always be working on texturing and export.
and work on a model that's good for texturing.
I've been fighting with this for years. It's a loop that always brings me back to the starting point. I need to know topology to be able to model without limits on detail. But then I need to be able to do retopo easily while CONTINUING TO HAVE FUN
I want to have artistic limitations, and I have a ton—I suck. But at least I want to have a workflow that lets me do what I want.
I create my work without limits on detail or constraints.
Retopo everything that needs to be retopo'd.
Work on texturing in full "relaxation."
Materials, nodes, etc. Scene, lights, rendering.
For years, I've tried to master steps 3 and 4, but I can't produce if I don't have 1 and 2. In my work, I've tried to "escape" this, but everything always leads me back to 1 and 2.
I can't create if I don't know how to make a good topology or retopo a model.
I can't retopo if I don't know topology, which is necessary even for a simple retopo. Because you can make a TERRIBLE retopo.
This is what has blocked me for years.
I apologize for being long-winded, but I'm here among experts to find my direction and explain myself well.
Thank you all for your valuable help; your advice is always appreciated and carefully considered.
All that said I think that finihsing that one test part in subdiv is still a valuable thing to to do ... if only to get a feel for the (IMHO absurd) amout of time it takes to merley get to a point where you can start texturing it !
But at the end of the day I think it would be *even more* valuable to do so with a part that actually belongs to one of the scenes you want to make - like a simple prop that you already have a design sheet for, and that you know will be seen up close in the foreground of a shot.