Good Topology and Wireframes

I’m a little unclear what constitutes good topology. Here I’ve made a mesh, and I spent time to reroute a lot of the loop cuts to be used by other parts of the mesh, so I don’t have to add more loop cuts for additional details on other parts of the mesh.

It’s all quads and subsurf friendly, and it seemed to me to be the best way to do it as I’m reducing geo, but when I look at the wireframe, it doesn’t look as nice as a lot of nice wireframes I’ve seen, and the wireframe renders are important as well for portfolio.

Is this acceptable? Should I run 2 separate meshes, one for a good wireframe render but more geo? Or should I throw caution to the wind and not care about poly count?

I’ve got a decent system but I’m already up to 10 million for the various parts including their subsurfs, and I’m looking to work more in VFX than games so I know I can run a bit heavier with geo. Just want to present the best mesh I can, maybe rerouting like this is attractive I don;'t know? What you guys think?

I’d consider the geometry passable, personally. You’re 3/4 of the way there, but you need to practice trying to keep your quads fairly even. You have a lot of squashed quads, then stretched quads, and in places some oddly shaped quads.

A good guideline I went by when I was newer with Blender was enable a Subsurf (albeit temporarily in some cases.) If the model still looked pleasant, then it was acceptable. It does get you into the habit of adding tons of support loops though which can make adjusting geometry difficult, so you should look at using modifiers.

On models like this, you don’t need to model everything as one big object. That needlessly complicates things.

Alas, on the the whole, your understanding of “clean geometry” is well founded and with some practice you’re on the way to making some amazing models!

3 Likes

A few thoughts.

Now off-line rendering albeit Cycles, Octane or Iray…etc, as long as your system can handle processing various tasks generating this object without an additional hassle of dealing with viewport performance hits, then pretty much dense cage topology shouldn’t be an issue.

However, worth noting from experience an ‘all quad’ rule of thumb may not always be an optimal choice, reason being plus you’ve encountered thus far, can be quite fiddly re-routing smooth polygonal transitions. So as mentioned above, look into separating the mesh into segments which can further mitigate resolution issues alongside shading errors from occurring in the first place.

Secondly, overall surfacing seems too be planar (flat) thereby inclusion of triangles, n-poles or (…god forbid) n-gons when thoughtfully positioned are generally acceptable, irrespective of proposed methodology toward finalization.

In addition I highly recommend trawling through a selection of linked info via the following artist’s cumulative work, he’s practically a veritable subd modeling guru. From whom I’ve personally learned on top of 15+yrs already as a hard surface modeler, extremely informative insight alongside many others who’ve also benefited as well:

FrankPolygon ArtStation

Frank Polygon Sketchbook

Polycount Technical Subforum - Long running Subdivision Modeling Thread

4 Likes

Me personally dont understand the necessity for perfectly square quads across the whole mesh especially when support loops and such are added because of a sub surf modifier. If this mesh has literally no shading issues and is all or mostly all quads than you should not worry. A lot of the times people will spend and waste a lot of extra time trying to clean up a mesh that doesnt need it for no reason other than saying “look what i can do” And in the industry all that extra time and energy is not going to change anything about how the mesh will look outside of a wireframe render. (Thats my personal opinion as someone who recently started working in the industry) if its quads and is shaded perfectly fine than everything else is a waste of time

3 Likes

There’s no such thing as “perfectly square” quads. You should be aiming to keep your quads roughly the same size if possible and following logical flows (keep poles to a minimum.) It’s considfered good practice for many reasons; it allows you to move the modifier stack, it allows you to create materials with relative ease, it allows you to create rigs with relative ease, and it allows you to customise your models with relative ease. It’s not just “showing off” or wasting time. There’s a reason for it.

You only add support loops if you are using a subsurf, and even then there’s often other tools in the modifier stack that you can use negating the necessity of said support loops.

You should understand all of this before going into the industry. You’d be a liability if you just throw crappy meshes at people to work with.

… what? The default cube is 6 perfectly square quads. The default plane is 1, the default grid is 100.

That’s a bit harsh, honestly. This mesh isn’t “crappy” - it might not be perfect, but it could be significantly worse

1 Like

I was talking theoretically and specifically to Benjamins post about modelling in general, not to the original poster’s specific mesh which I have already replied to. Apologies to the OP if that wasn’t clear. I have added a quote to clarify in my previous post.

Of course. But once you start creating models, it’s nigh on impossible to maintain perfectly square quads. I’m not thinking grids or cubes, I’m thinking objects like characters, vehicles, buildings, objects. Grids and cubes are simply the starting point.

I dont throw crappy meshes at people** you misunderstood. What i was getting at is in response to this mesh which is fine. he could indeed spend hours making them all look more pretty but it wouldnt change much. animations require specific, clean topology. I could show you plenty of wireframes of the stuff i make. its fine

1 Like

No need, I believe you. I may have misunderstood the argument or the perspective but I do get a bit irritable when I see people write things like “waste a lot of extra time trying to clean up a mesh that doesnt need it for no reason other than saying “look what i can do”” because I start imagining meshes full of triangles, nGons and tons of needless geometry that people can’t be bothered to clean up, try to pass it off as good work and not take the chance to learn from it.

The OP’s specific mesh is a perfectly fine, manageable object that is certainly texturable and customisable. There is scope to make his processes even better. Honestly, this post reminds me of where I was not too long ago with my modelling techniques so I’m genuinely excited for where he takes his models in the future.

2 Likes

Yes, thats why i mentioned all quads a few times. My thing was yes, make sure its all quads/deforms correctly, has no shading issues. if its animated than make sure its really clean. I need to rephrase it for directly talking about inanimate, secondary, background type objects that i was simply saying do not need to be the mona lisa of topology as long as everything that it should be is taken care of. Making it look clean when no one will notice is what i was meaning to say

2 Likes

And all this discussion really helps the OP about his mesh… IDK …

There are already some diagonal framing polygons like in the big blue area… i would suggest to do more like this… most of the polygons are coplanar… so why so much of them…:

also the left darkblue area maybe can be separated ?

1 Like

OK there seems to be some confusion maybe I didn’t explain myself well.

First I’ll put another screen grab here because there was a couple of separate panels visible that distort the view of the mesh in question.

I’m aware I could make it easier and make a few parts separate meshes (the circular door hinge for example), but my intention wasn’t necessarily to do it the most efficient way possible, I wanted it to demonstrate modelling ability, and therein offer a presentable wireframe. Point being, if I can do it the hard way, then I can probably do it the easy way as well. The extra work of rerouting doesn’t bother me (for portfolio) as it’s one of the most enjoyable parts of modelling for me, it’s like a puzzle.

I was more curious about the diamond quads for example, it’s a good way to channel 3 edge loops into 1, but it doesn’t look great in wireframe, ditto with rerouting round the bottom of the windows and doors. But that’s just my eye, maybe I’m wrong and it is expected?

The mesh as far as I can tell at this point is fully functional, with the subsurf (not applied) I’m able to twist and bend it without any visible problems.

From the comments I got a few things I also have questions about. What is the importance of even spacing of the support loops? Obviously when I’m reinforcing sharper edges there will be support loops close together, vs areas with no detail on flat surfaces…

It doesn’t matter what the wireframe looks like… all what counts is the minimum distortion in the normals so the model looks good if even viewed in a chrome like material… and if is textured later in a more dull material or even with dirt textures then this also doesn’t matter as much.

1 Like

I’ll be doing the texturing in Substance once I decide how to finish this mesh.

Blender Bob among others have stressed the importance of having good clean wireframes in your showreel.

Well yes indeed, this is why I’m asking. I don’t want to be perceived as “that guy” who passes on crappy meshes for other people to deal with. :smile:

I don’t know why you are using subdivision for something that is predominantly flat surfaces. You’d probably end up with a much lighter mesh just using bevels for edges. The problem with subdivision is that your support loops end up being subdivided too, and the whole thing ends up really heavy.

Another thing you could try is to apply the subdivision then use decimate (unsubdivide) and manual cleanup to keep it manageable. This should keep it in quads, unless you have some strange edge flow in places. 10 million verts seems pretty high for a single vehicle.

If you are using substance painter for texturing, then does it not work just as well with ngons and triangles anyway? I’m thinking of workflows like this.

1 Like

Hey Grant.

There are a lot of other objects and this one is certainly not the most complicated (this object is only about 30-40k if I remember correctly), the poly count isn’t really my biggest concern, I have a decent rig and this is only for my portfolio. This is only about presenting a good clean mesh and wireframe, not maximum file efficiency.

Sure I could have some tris and ngons, but if I don’t need to…

That’s an interesting tutorial in general though I’ll definitely check that out, thanks for that.

I’m just looking to make something nice and clean like this…

It’s a thing of beauty

1 Like

Does the use of any app for materials have anything to do with good topology for a specific purpose?

Ahh… so the you are up to make good models for VFX… you know that the meshes heavyly differ for games… so you may have mentioned that in you question…

In my original post…

“I’m looking to work more in VFX than games”

Ups… dummy me :blush: … first just followed this and then didn’t re-read first one… :face_with_peeking_eye: