Body topology

#2521

Can I get some feedback on the lines? The intended outcome is a still render, so I think this should be fine, but I would like to get opinions.

#2522

It’s a nice model, but if you truly need forty blend shapes to deform well and it’s not a case of being excessively nitpicky then I have to think that’s a giant red flag that your workflow has problems. If you have a good mesh and good skeleton you should be able to get it to do just about any motion humanly possible without killing yourself skinning it just right.

You can’t avoid poles completely without having your vert count balloon excessively but even 5 edge poles cause me big problems in any areas that needs to deform well or need to receive heavy zbrush sculpting. There’s nothing worse than adding a new animation and realizing you have to fix your skinning or topology again for the ten thousandth time

Ngons are even worse than poles, and you seem to have quite a few of those. The problem there is zbrush can’t take an ngon and mudbox must be quad. So it’s going to change your mesh. And your vertex order. Now your skinning info and your uv maps are all complete trash. So you have to redo these or try to transfer the old maps etc.

So yeah you can do your detail in maya but it is probably going to make your life a lot harder, especially with animated characters.

#2523

Hi! I just got back to work on an old model, and I´m also testing comet´s pose deformer to correct deformations. I though I would post some wires and the test, as the topology has been working pretty good so far, I though it would be useful to post. I thing it can be improved a lot, but it is kinda ok for now. First spot I tested is the shoulder as it is to me the most complicated deformation going on in a humanoid character.

Check!


And the videos!

http://www.tmk3d.com/shoulderTest01.mov

http://www.tmk3d.com/elbowt01.mov

Jonnhy

#2524

It’s a nice model but the edges are getting messy in places:

avoid the random terminating edges (i know Stahlberg does it in the first posts an’ all but it’s 2010 now … it’s just not done anymore). You’ve got a totally pointless edge running up the spine.

When running an edge, imagine that it’s a muscle it should in general follow the muscle at the surface.

#2525

Nice feedback.

IMHO the 4 sided polygon struggle is a pointless one, lots of those ending edges gives me details I need to disappear smoothly as I apply a mesh smooth and they usually are far from articulations. The geometry in the wrist gives me some “bone pushing” detail plus the edges that will come in handy for when I need to sculpt my blendshapes for problematic deformation or just to cheat some muscle.

I’ll keep your image for reference in case I need an instance to remodel some spots (which I guess will happen actually). So thanks again for the feedback!

#2526

Hi all I am from Armenia, sorry for my english :). I created the human body. often i start from the beginning, and discovered a new technique, modeling the incorrect topology model, I realized that subdividing still all the converts in the quadrangles and it helped me to add significant and minor details on the human body.
and here is an example
LOW.jpg
That’s how I made required form
HIGH.jpg
Do you think is correct model ?

Your comments will help me a lot, thanks in advance

#2527

Do you think is correct model ?
Your comments will help me a lot, thanks in advance

#2528

Looks Great! You’re going to do well in the CG community if you maintain that progression.

Unless I’m mistaken, the technique is not new- in fact, Stahlberhg starts the thread with the same concept.

And, I believe that the general consensus is that sticking to the grid system is best for a basemesh; high frequency detail is done through sculpting and textures, and geometry deformation should occur through quads and triangle faces.

I think there’s a lot more to the subject, but topology should generally be kept as simple as possible.

#2529

Not at all, the technique is far from new. I just got to try it through the comet’s plugin which makes it all a lot easier. If you work it out with maya’s blendshape there are some extra connections needed for the model not to translate the corrected deformation. So for a long time it was not a “production ready” method, it has some unpredictable results here and there.

About the topology, yes, quads and tris are far more recommended for deformation, and I like to have in polys all the detail I can so it is posible to handle those details in blendshapes and deformations. I made this model for the cybergirl workshop some time ago, and after a long while I got my hands on the plug in, topology is backing me up well so far, we’ll see what happens with other areas (wrists, knees, etc)

#2530

Im having a real difficult time to get the proper edgeflow going. Im very new to organic modeling and I can’t for the life of me seem to get a good working topology for the upper body Im currently working on.

Watching some of your wireframes that you guys are posting is helpful but I don’t seem to get the same results. It takes me an age to figure it all out, so Im wondering if there’s a screencapture of someone modeling a body out there? Stahlbergs work is mostly the references Im using right now and Im trying to mimic the edgeflow but I seem to fail :slight_smile:

I’ve been checking out Digital Tutors body modeling guide but they seem to be clueless when it comes to the type of edgeflow and topology we’re discussing in this forum. Anyone participating in the debate that got any homemade vids? =)

#2531

You could benefit from drawing the lines on reference photos prior to modeling. Also, terminate the edges from the head which are currently running down the body, they may be making the process more trouble.

And what sort of criteria do you have for “good working topology”? Have you been running animation tests? Or rendering the mesh with a proxy smooth? Or importing the mesh into other packages?

#2532

I’m actually trying to get approval to use a few images of our models for the Wrapit gallery, as soon as they’re out I’m going to make a post here, also trying to explain the reasons behind them. Hopefully it won’t take too long…

#2533

The criteria is nothing more than wanting to be able to model a human body with an edgeflow that allows good deformation for possible animation. The model Im currently working on, I would perhaps pose in different ways, so topology for ‘animation’ wouldn’t really be required I think. However, when people look through my reel, I don’t want them to think “oh, he modelled it but it won’t hold for crap when animating” you know =) The mesh I showed in my previous post I’ve proxy and rendertested some to see that the mesh is ok, which it is.
I guess I’m too anxious to model organic stuff like Stahlberg and other people on this forum, forgetting that this is my second human I’ve ever done. Practice makes perfect :stuck_out_tongue:

#2534

I thought you all might be interested to know there’s an active discussion going on on our blog at TurboSquid.com that deals with topology, what constitutes “good topology,” and how it fits into a the bigger issue of what makes a high-quality 3D model.

We’ve been talking to some of the biggest buyers at TurboSquid, who come from all kinds of industries, and we asked them to tell us about their own standards. As you’d expect, it varies depending on the industry. But we did get some consistent feedback across industries, and we share that in our blog post.

We are hoping to engage 3D modelers in this discussion about professional or quality standards. So check out the blog and feel free to put in your 2 cents.

#2535

I am a student and I have always been told and read, work with quads. Now I am reading that tri/ngons are good? What changed? How do you know?

#2536

The Stahlberg topology is an example of how you can get the most of out your edge flow and use the nature of modern subdivision algorithms to enhance organic inconsistencies. A word of caution with this style, skipping to creative topology solutions before you are comfortable with textbook methods will probably cheat your understanding of both ends of the spectrum.

Current displacement/normal mapping techniques allow for bypassing the creative topology methods, although I wouldn’t discount the value of it.

If the final output is a still, 2D image, then your topology can be less rigid. If you are modeling for animation, you’re going to stick to mostly quads. Tris can be used in places where they are less visible or suit the mesh’s deformation needs. In any case, with animation, you should be open to testing various configurations with 3 and 4 sided polygons for the best results. If your toolset includes a sculpting application, stick with quads and tris only.

Check out some behind the scenes Final Fantasy stuff:
http://www.youtube.com/watch?v=DEybd106bBU&feature=channel

Their models are animating on tris, which were probably procedurally derived from quad/tri meshes.

Go with what works for your needs, as a student- my belief is just get it done. There is so much to learn, and it’s usually not the details as much as it is the process. The nature of the industry is automation and so change is expected.

#2537

In general i’d say stick to quads and nice flowing loops it’s easier in the long run and makes it very easy to increase or decrease the level of detail in your model as you need. A model for sculpting really should be all quads, tris can be handled but may cause unwanted bumps as you subdivide; ngons should be turned into quads. After UVing and you can cut down the poly count by using triangles but get uvs from a clean flowing quad mesh.

#2538

hi there
Actually I’m quite involved into organic modeling and topology. I made some heads\faces recently and now I began to study the body anatomy. Right now I’m not familiar with rigging\animation and I’m doing my best to build the correct topology for deformations. I have some problems with approach I must take. The male body in most cases has pronounced muscule structure that help deformations AND defining the shape of the body. The female body instead have less muscles, so the question is: do I have to define the muscles with edgeloops instead of shaping the overall masses and flesh??? (it seems to me to have another flow PIC).

Thanks.

#2539

well, you can. as you said, the male muscle edge loops help with deformations. not JUST cuz he has muscles, but also because that’s the correct movement-deformation. why wouldn’t the same thing work for better deformations on the female body? :wink:

#2540

The female body has less muscles? The human bodies are the same really, only deviate where it’s absolutely necessary. Even then, it probably doesn’t need to happen at the topological level.