Topology in Production, if it doesn't deform, does it matter?

Become a member of the CGSociety

Connect, Share, and Learn with our Large Growing CG Art Community. It's Free!

THREAD CLOSED
 
Thread Tools Search this Thread Display Modes
Old 05 May 2013   #1
Topology in Production, if it doesn't deform, does it matter?

I've just seen a fantastic video by Vitality Bulgarov of his Black Phoenix project.

Now, one thing I noticed is that there's a lot of n-sided polys, and topology seems to take a backseat over how it looks.

Forgive me if this is common knowledge, but is that generally ok for pre-rendered stuff?

I've been known to obsess over edge-flow, optimisation and topology, perhaps due to the realtime work that's been a part of my projects over the past few years, but when I think about it, I feel like I'm wasting a lot of time trying to get complex hard-surface models to smooth perfectly when it would look equally fine if I just added a bunch more geometry and bevelled an edge, and not worried about the resulting topology.

So, for you guys in production, regardless of industry, if you're working with pre-rendered stills or video, is it perfectly acceptable to forget about topology if it 'looks good'?
 
Old 05 May 2013   #2
Maybe. Does it make for extra work for someone else in the pipeline-say texturing and surfacing for example?
If somebody else doesn't like it and their reasons are legit then the answer is 'yes'. If not...
 
Old 05 May 2013   #3
Good point, I hadn't considered that.

I'm not even sure that there's a yes/no answer, I'm just curious to hear thoughts.
 
Old 05 May 2013   #4
When working in a pipeline, I would imagine that creating the best possible model is the way to go.

I think that if you are working on a still, you can get away with a few cheats here and there.

Generally my models look like bad stage backdrops-but I only have to produce the work for me-so its OK to cheat.
 
Old 05 May 2013   #5
YES it matters, because its going to be a headache for shading, UV setup, and texturing...sorry, qRemeshing something in zbrush does not qualify, it may look clean, but if you select a single edge loop, you get tornado selection. My suggestion, start clean, end clean, for production.

I get handed zb qremeshed or zb objects all the time, gets sent straight back. I miss the superbly clean sudD models I used to get in production from folks 3+yrs ago, things seem to be shifting to the sloppy hand off side (non senior/pre 2006 artists). If your just doing a personal project, do whatever, its your thing.
__________________
"SAVE KILLER BEES"

There are no "magic" programs that make you a better artist

Last edited by ambassador : 05 May 2013 at 05:05 PM.
 
Old 05 May 2013   #6
Originally Posted by ambassador: YES it matters, because its going to be a headache for shading, UV setup, and texturing...sorry, qRemeshing something in zbrush does not qualify, it may look clean, but if you select a single edge loop, you get tornado selection. My suggestion, start clean, end clean, for production.


That's a good post from someone who actually deals with this.
__________________
I like to learn.
 
Old 05 May 2013   #7
The answers you seek are in Vitaly Bulgarov's blog post-
http://vitalybulgarov.blogspot.co.u...ix-project.html


I haven't noticed his project until today but from what he says in the post and my understanding, his aim was to make his ideas quickly in 3D and rendered, and keeping the mesh clean was not priority but he kept the mesh light with very few subdivisions. His aim was to keep his creativity flowing instead of stopping to fix the topology.


The goal was not to practice in modeling but to practice in concept art utilizing 3d as just a tool for outputting images rather than creating a nice looking wireframe that fits production needs. The goal was to use 3d the way that would speed up the process of creating concept art rather than slowing it down. So I tried to minimize using such techniques as subdivision modeling since the clean-up of subd topology and edge-loop refinement takes time.


This doesn't mean there aren't issues when not modelling in quads, as he mentions there are artifacts in his renders(probably texture issues and smoothing problems) and he cleans it up in Photoshop.

Keyshot provides an automatic UVís solution for the box/cylindric/spherical shapes which worked most of the time. When I had artifacts I would just fix them in final render in Photoshop.


Also the aim of the work is a Book and not animation, so he can get away with a lot of things, as I think it would be difficult to animate these models but the intention was not for animation but concept art rendered at a high quality.

Like someone has said it all depends on what you are doing but working in quads is the general aim because of texturing and smoothing issues that can be caused by N-gons.
 
Old 05 May 2013   #8
Thanks for the replies guys.

Originally Posted by ambassador: YES it matters, because its going to be a headache for shading, UV setup, and texturing...sorry, qRemeshing something in zbrush does not qualify, it may look clean, but if you select a single edge loop, you get tornado selection. My suggestion, start clean, end clean, for production.

I get handed zb qremeshed or zb objects all the time, gets sent straight back. I miss the superbly clean sudD models I used to get in production from folks 3+yrs ago, things seem to be shifting to the sloppy hand off side (non senior/pre 2006 artists). If your just doing a personal project, do whatever, its your thing.


Gotcha. Guess I'll keep working with subD's then, I was just looking for clarification after a particularly frustrating afternoon working with them.
 
Old 05 May 2013   #9
Generally speaking, you have to master the trade before you can break the rules. Vitaly knows his craft, so he knows where topology is important and where it just does not justify the work involved.

Modelers can get tunnel vision and not see the big picture.
__________________
: my website :

Last edited by MasonDoran : 05 May 2013 at 08:05 AM.
 
Old 05 May 2013   #10
You might think it doesn't matter and then you try to UV unwrap it looks like a mess.

Quads... We want only quads... no tris... no n-gons... just Quads.

Thanks.
__________________
"Your most creative work is pre-production, once the film is in production, demands on time force you to produce rather than create."
My ArtStation
 
Old 05 May 2013   #11
I'll take a small handful of well placed tris and even nGons with a good, well distributed flow over some of the topological contorsionism I've seen people go through to keep it all quads.

The "it has to be all quads!" rule is this quasi-religious mantra that has long been dated, but since people didn't know in first place when it applied or how it came to be, still sticks around even when it has no real need to be applied.

Quads matter for nothing if your flow corkscrews around every limb and you dump 12 edge poles all over the place.
The odd tri or even 5sided here and there has long stopped being harmful provided that pipe downstream can deal with it.

Topology has one purpose, and one purpose only, serving the departments downstream.
IE: If it doesn't deform, it won't be textured, or will be textured with no need for readable UVs, it gets displaced at a fragment level (therefore density variance isn't very important), and the rendering engine doesn't complain, then it's superfluous to worry about it.

Rules of thumb and popular knowledge are a pain in the arse to deal with every time a new modeler comes on board.

If you want to take a genuine interest and be able to work out autonomously what topology to use when, then go and learn the basics of what rules apply and WHY, or work in some department downstream (rigging, texturing, technical shader or look-dev) so you understand the requirements. Otherwise, just practice all possible ways and rely on the HoDs downstream to your assets to give you or the modelling supe guidelines and stick to them.
__________________
Come, Join the Cult http://www.cultofrig.com - Rigging from First Principles
 
Old 05 May 2013   #12
IMO topology matters a lot for certain things and not so much for others. Anywhere you need to be able to make specific selections, it'll matter.

Otherwise, I don't see how it matters if it doesn't deform.

In the age of ptex, auto UV mapping tools, and advanced 3D paint tools, I'm not sure why anyone would have such major issues one way or the other unless they don't know the modern tools.

I use randomly decimated triangles on my organic geometry that doesn't deform or need a specific polygon flow. This way you don't see hints of polygon flow lines. My models are super high poly count and I think random polygons looks more organic than normal topology flows.




If you squint your eyes, you'll still see the topology of the quad version, but not the decimated triangle version.

Both spheres have the same number of polys, but the triangle version doesn't have uneven polygon sizes that stretch and converge around the model, doesn't have poles, or any topology artifacts where poles were patched but had to leave a few triangles that smooth weird since they're next to nothing but quads - like mini pinch points

People might say to just use SubD's or mental ray's approximation editor to smooth a mesh until the polys are small enough to be gone, but they make renders super long on heavy models since it has to recalculate massive smoothing for all objects every render job/frame. I'm at the point now that subD's/approximation editor literally makes the geometry disappear in the render as if it was hidden. Mental Ray apparently doesn't like 100+million polygon renders and just says no. Even on machines with 64 gigs of ram.


Anyway, what I do probably isn't the best idea for most people. I just can't work with a low res poly cage or major displacement maps at my job. I need near exactness right there in the viewport.

the other thing is I mostly use 3d volumetric procedural textures, so I don't give a damn about UV's other than models need to be in 0-1 UV space for 3d procedurals to render consistently.

Last edited by sentry66 : 05 May 2013 at 05:23 AM.
 
Old 05 May 2013   #13
The main reason for all quads is Zbrush which doesn't support n-gons and triangles can screw with it too.
Also, there was a time when not every renderer was able to subdivide non-quads.

Nowadays we still prefer to keep our organic stuff all quads for the first reason. Fortunately it's always possible to get rid of triangles by adding another edgeloop. Poly counts no longer matter that much.

I also do not tolerate too many poles, a modeler needs to have a pretty good reason to have more than 5 edges at a vertex.


On hard surface stuff that doesn't have to go to Zbrush, 3 and 5 sided polygons are completely OK. In fact sometimes they allow for much cleaner and lighter topology so they can be quite good.
__________________
Tamas Varga
 
Old 05 May 2013   #14
zbrush handles triangles just fine over here. Actually with my models I've had more predicable results by converting to 100% triangles before bringing a model into Zbrush.

I've had zbrush erase polygons on heavy models that were about 50/50 triangles and quads. But if I convert to all triangles first, then there's no issue.

Last edited by sentry66 : 05 May 2013 at 02:29 AM.
 
Old 05 May 2013   #15
Originally Posted by TeeJayEllis: So, for you guys in production, regardless of industry, if you're working with pre-rendered stills or video, is it perfectly acceptable to forget about topology if it 'looks good'?


Every industry will have its own requirements for topology. Game models with normal maps require one kind of topology for low-poly (tris on clothes lowpoly are fine (where a tri adds little geometry to define the silhouette, quad would add too much polys)), as it won't subdivide), but in deformable areas you'd better keep all quads for game models as well, like knees, fingers.
I'd say using quads is more predictalbe, and therefore, better. Unless you need to end this stupid loop, and killing it with a tri is better than making it going too far just to keep all quads. Such tris should be kept not in areas of deformation, especially long tris. But adding them is the last resort, and depends on experience, so practice will tell you if you should have it or not.
Tris are fine in areas of huge density models, especially hard-suaface subd's. For example for sub-d heavy-modeled clothing it can be safely used to end loops where polygon density is big and polygons are close, so it won't stretch neither textures nor deformation It would be very stupid making a huge triangle somewhere on the face.
N-gons may cause stretching of texturing for subd's. Also, n-gons and triangles cause shading artifacts for rendering - bulbs on geometry, which are very hard to get rid of, if you're modeling something like a sleek car. Stahlberg used them for characters to add bulbs etc, but there was no normal mapping, and displacement was too long (at least today it's much faster). To save several thousands polys for the sake of a "lighter" mesh makes no sense today. Just use normal mapping or displacement with clean quads.
One thing to keep in mind is you shouldn't have spirals, and loops should be closed whenever possible, so you can easily select them, delete, add another etc.

Last edited by mister3d : 05 May 2013 at 03:26 AM.
 
Thread Closed share thread



Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
CGSociety
Society of Digital Artists
www.cgsociety.org

Powered by vBulletin
Copyright ©2000 - 2006,
Jelsoft Enterprises Ltd.
Minimize Ads
Forum Jump
Miscellaneous

All times are GMT. The time now is 12:45 AM.


Powered by vBulletin
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.