ZBrushCentral

ZBrush, Modeling & Topology

ZBrush Artworks
#1

First, let me say that this is a very interesting tool you have here. I am very impressed with the robust tool set you’ve come up with in such an early iteration of your product. Though I’ve only got access to the demo as of this moment, the new ZShperes modeling features look very interesting to me. One concern I have about the product however has come to me while I’ve watched Ken Brilliant’s scripts of the alien modeling.

I am very impressed with the sculptural talent I’ve seen in these demos, but as attractive as the appearance of the model is, I really find the topology of the model to be a problem. I’ve been studying facial modeling for the last year or so on the DSF (Spiraloid) and other websites. What the ZScripts have shown me so far is a total disregard for the concept of edge loops and control of the resulting mesh. From a 2D imaging perspective, this is not a problem. However, if I was to adapt these high res mesh objects for animation, I suspect that the lack of structural planning would be a hinderance.

I’ve looked at some of the early ZSpheres examples with facial elements, and these appear promising. For those of you out there with access to the full version of the tool, I’d like to ask a favor. If you are working on characters to animate in whatever application is your preferred, would you be willing to post some wireframes for me to examine, and perhaps share your ideas for a workflow that would get around this problem. I’m specifically interested to see how ZSpheres could attack the issue. Thanks for any information you can supply.

<font color="#949494" size=“1”> November 05, 2002 Message edited by: nimajneb </font>

<font color="#949494" size=“1”> November 05, 2002 Message edited by: nimajneb </font>

#2

turning around that question

Heya,

thanks for bringing up this matter. I am in no way sophisticated in animating. I think i know what you mean with a facial setup which is supposedly good for animating.
This is most of the time a tedious task of pushing vertices into an eveloping position to recreate the facial features (e.g. let polys intersect where a crease in the skin is likely to appear when being animated).

Now i found this to be too time consuming for my amateurish attempt - what i did then was using morph targets for facial animation -> this can be done in ZB in no time with its very powerful modeling features. The results are pleasing aswell.

Now this method, although being used by many, seems to have disadvantages to the method mentioned above. I can understand that realtime renderers would prefer the lowest poly count as possible but other than that i have no clue why i should go the long way when i can have results in minutes.

Please if you have any additional info on that subject fill me in.

Thanks in advance

#3

The topology issue is a matter of control. You mimic the structure of the facial muscles to facilitate natural movement. It’s not so much about following wrinkles and folds, as following the movement and not just the form. Here’s an example of some good topology from AW and the DSF.

Meshes that don’t take this into account have two problems. The first is mesh density. Notice how simple the control cage is of the mesh in the picture. To control the corner of the mouth for the smile and frown, is was a simple mater of selecting and tweaking a small set of verticies. A dense mesh would result is hundreds of verticies that would have to be pushed and pulled on to get the same expression. And for anyone who’s had to do that, you know the hell of missing verticies, wrong selections, strange deformations and smoothing issues. Complex mesh makes control of the mesh complex.

This is not totally about economy of polygons however. The more important issue is the structure and how it relates to movement. Notice the lines of the mouth in the picture. The edgeloops circle the mouth, following the shape of the underlying muscle and through that its movemenet. If I want to open this characters mouth, it can be accomplished with a simple non-uniform scale and some minor fine adjustments. The radial nature of the meshes topology allows for this. With the dense mesh of Ken’s alien, it would not be so simple. It’s mouth is like a grid. If I moved it out of its created position, I would have quite a job. ZBrushes poly pushing tools would be a big help, but what they would not address is the nature of the movement. By incorporating the movement in the model in its structure, I avoid the problem of attempting to pull the mesh at odds to its grid.

All this being said, I’m very curious to see how ZSpheres would apply to the problem. Some of what I’ve seen of ZSphere heads show a very promising structure around the mouth and eyes. Show me wireframes folks! :slight_smile:

Thanks for the question Nemo!

<font color="#949494" size=“1”> November 06, 2002 Message edited by: nimajneb </font>

#4

Hi Ben,

Hey, Kalamazoo, I’ll be flying up there in a week or two.

Since ZBrush started more as a painting and illustration application, I don’t think the majority of those on this forum are very familiar with the constraints and advanced techniques of polygon modeling for animation. I am a little familiar with it. My main software is spline and patch based. It uses no actual polygons, but you build a ‘cage’ much like the one you show above. The nature of the splines is that they will actually lie on the surface, and are represented as smooth flowing lines, rather than the staight line-segments in the picture above. However, the modeling technique is very similar for the same reasons. Edgeloops are very desirable, and the lowest possible ‘point’ count is best, because that makes animation of the skin much easier than a high-count.

The low-poly modeling technique that is now being used by some ZBrushers is pretty new. Ken B’s alien was done before the current verion of ZBrush came out. Ironically, Ken B is now one of the more advanced Low-Poly modelers.

However, even those who start with a low-poly mesh usually increase the density as they go along beyond what you would probably desire. Since there are no constraints regarding polys for an illustration versus an animation, we go ahead and increase the density, and you may not see the models that you would like to see.

That said, check this out:

To get the highest detailed hand, I simply divided and smoothed the lower density version. The lower density version could be made to appear smooth in ZBrush by turning up the Smooth slider. It could also be used in another application where the mesh would become the control cage or the spline points.

One problem in ZBrush is that you do not have total control over the mesh, as far as deleting or adding points or edges. You can delete or hide points, but it is not real easy to select single points yet. However, by using Zspheres, you can get the major edgeloops in place around the eyes and mouth. You certainly could get them around the ears and other places, but for humans ears aren’t really animated, so I think it is much less critical there.

<font color="#949494" size=“1”> November 06, 2002 Message edited by: Jaycephus </font>

#5

Hi Nimajneb

As already mentioned, building 3D models with meshes optimized for animation and ones made for still images are different. The tools needed are different too.

But ZB has made advances to produce good low-poly cages with decent edge-loops–at least as a good starting point. This does require ZSpheres though.

Below is a starting point for a character face (a chimp) done in ZB

From here, I continued to increase detail and sculpt in ZB, but you could also export this to another program for further refinement.

I’ve also modeled a high poly mesh in ZB, exported it, and rebuild an animation friendly version in another program. Here is a tutorial on that process: Click Here

This method is fun because you can concentrate on the ease of 3D sculpting first, and deal with the technical aspects later.

Incindentally, I made that character example head you posted in this thread. (What is “AW” and “DSF”?

#6

Hey Ken,

I thought that image belonged to Martin Krol (AKA, Ambient-Whisper) of the Digital Sculptors Forum (Spiraloid). Didn’t notice the credit at the bottom of the post. Sorry about that. I am coming at ZBrush from the point of view of applying it as an animation tool. Would you be willing to detail the process of building a head using ZSpheres so I could see the tool applied? Or does something to that effect exist already?

and

Jay?

Why the hell would you be coming here to the land 3D forgot :wink:

-nimajneb

#7

Here are the steps I took with a zsphere to make a model:

to get the indentations for the mouth and eyes, zspheres were simply pushed into the head. When skinned, this gave nice “loop” geometry around those areas.

From there, the geometry was sculpted with ZB tools, pushing and pulling.

When the form was taken as far as it could go with the current geometry, the mesh, or areas of it, were divided for more faces to work with.

In this example, I did divided the mesh farther than if I were to use it for an animation model. You can stop at any point and export for further modification if that is your desire.

Even though I did make a higher resolution mesh, I like having the “flow” of faces set up as best I can early by starting low poly. I find it easier to model that way.

There are other examples of this method on the forum, I don’t recall where though.

<font color="#949494" size=“1”> November 06, 2002 Message edited by: Ken B </font>

#8

Hi nimajneb

Zbrush can do what you want and its pretty easy too.

Here I have posted a model I’m working on and as you can see I made the High mesh head in ZB added detail then exported it to a different program and cleaned up the polys/splines. As you can see the mesh is reduced a lot and for animating it in messiah the mesh works fine I can also rework it in a different programe to get it just as I want it and then import back into ZB for texturing or more modeling what ever. works pretty well.

<font color="#949494" size=“1”> November 06, 2002 Message edited by: dfaris </font>

#9

Ken,

Thanks for the process images. It looks to me like the ZSphere acts almost like a blobby or a metaball creating a surface parametrically. There’s very little information about ZSpheres on the website, and they are why I’ve become interested in ZBrush. Quick, intuitive layout of good topology. Thanks for taking the time out to post on this issue. Now that I’ve looked over your site, I’m remembering you and your methods. I believe the ZBrush tutorial you offered is a adaptation of the technique you apply with traditional modeling (drawing edgeloops onto real sculpted models). Its the total opposite of the way Bay attacks the problem. Your way really separates the creation of the model into two problems. First form is attacked, then good topology is laid on top. Very logical, especially from a sculptors point of view. I like it.

Thanks again,

and

dfaris,

Think you missed my point, looking over this nice head. Specifically around the eyes, nostrils, jawline. Notice how gridlike your mesh is. These grids are at odds with the way that we know the creature would move when flaring the nostril, gaping the jaw to strike, widening the eye. How would you rotate the jaw open on this character and maintain the jawlines shape and proportion?

In ZBrush you could just paint the form and detail. To animate it however, the structure has to be fixed. You can only reshape what you’ve got. So attention has to be given to what a feature in the model does as well as how it looks. What I’m saying is that modeling from a Sphere may give you a good looking model. It doesn’t however give you useful topology. It doesn’t take into account the function of the form, and for animation that is a major concern.

-nimajneb

#10

dfaris,

Think you missed my point, looking over this nice head. Specifically around the eyes, nostrils, jawline. Notice how gridlike your mesh is. These grids are at odds with the way that we know the creature would move when flaring the nostril, gaping the jaw to strike, widening the eye. How would you rotate the jaw open on this character and maintain the jawlines shape and proportion?

In ZBrush you could just paint the form and detail. To animate it however, the structure has to be fixed. You can only reshape what you’ve got. So attention has to be given to what a feature in the model does as well as how it looks. What I’m saying is that modeling from a Sphere may give you a good looking model. It doesn’t however give you useful topology. It doesn’t take into account the function of the form, and for animation that is a major concern.

I know what you mean about the mesh form but in order to get a low poly mesh you have to make some kind of change as compaired to the real thing or the mesh would be huge. This modle is not the finished model and I have already animated it in messiah and it works fine. I did do some more tweaking of the mesh after pic was taken. The model in the pic you posted above has the same grid look around the mouth and eyes. I’m not sure what you are trying to say if your trying to fit the mesh to the way the real muscules lay across the face then the mesh would have to be huge. If the bone setup and constraints are done with care you can get the same effect and look if thats what you are saying.

#11

dfaris,

It’s not that you want to mimic every muscle in the mesh. That’s not the point. You want to use the muscles as a guide to the direction of the edgesloops in the model because the contraction of those muscles is what pulls skin one way or another. The dino head you’ve shown, specifically around the nostril, eye, and mouth jaw doesn’t do this. I’m not saying you can’t animate it. You can animate any set of verticies you like. What I’m saying is to get a better and more controllable deformation there are some structural conventions that modelers attempt to follow. Look closely at Ken’s model in the pic above. Notice how the edges circle the mouth opening. Also around the eyes, notice how they radiate out of the socket. Look again at your model’s eye socket. See how the grid adapts itself to form, but doesn’t follow the contours of the socket. The edges of your mouth don’t circle the opening, but travel right down into the neck. All this being said, the comparison between these two is not ideal. You’re likely not doing complicated lipsync and facial expression with your dino. It’s design is very likely not impacting on your animation with it. A human face (or any expressive face) is much more plastic. What I’m attempting to get across is that paying attention to function of the mesh (topology) is as important to animators as the form of the mesh. I may not be explaining this the best. I’ll look for some links in my library that can give better examples.

-nimajneb

#12

Hey Nimajneb, I’m not sure if this is what your interested in, but I find thay the new adaptive skinning method for use with the Zspheres creates outstanding uniformly parameterized surfaces. (It takes a little work, but it beats the heck out of using Paraform.) This is a work in progress of a dragon I am creating. Let me know if this is close to what you were asking about. Just a quick question, what 3D APP are you using for animation?

#13

dfaris,

Found a very good resource for head modeling in my links:The Human Head. This site is a great primer on the importance of topolgy. Lots of visual examples.

sirquadalot,

This is what I want to see, which is ZSpheres applied to character development. I’m expecting the nostrils and eyes to be carving out though, and looks like you have them protruding. May just be that you’re early in it yet. I guess it’s really breaking down to how you intend to use the model. Not meaning animation or illustration, more facial expression vs. a more simple (though mesh dense) puppet. I do see things that concern me in this starting mesh, however. Lots of poles (non four sided verticies), and not in places that make sense. Example the jawline, notice the six sided pole and the corner of where the jaw would be. This makes it very hard for me to determine the shape of the jaw and or visualize its movement. There’s strangeness going on with the eyes as well now that I’m looking at it. Look at that grid on the eye, its almost like the skinning of the ZSpheres is doing something like a boolean there. These things are strange. Like I said, I have no idea how the ZSpheres skinning is implemented. I see you folks massing out forms with the ZSpheres, but the resulting topology is highly variable. Ken applies it, and gets something like a early cage I’m used to seeing. You apply it, and I see a group of pieces attached with poles. It’s weird. I wish they were in the demo, then I’d play with it myself and stop bothering you folks :slight_smile: Show me more on this one as you continue. I want to see it when you’ve got it closer to what you call final. Also show the ZSphere and perhaps the most simple low res skin of them. This one looks smoothed or at a higher density. Dense meshes hide topology problems. Low res are much easier to read, so to speak.

I use MAX as my primary application, to answer your question. Meshtools and Nendo made me a confirmed adherent to poly modeling. What I liked about what I’d seen with ZSphere is the quick, seemingly good lowres cages that it makes. What I was concerned about, and still am concerned about is how well the tool applies to facial structure as opposed to body structure.

Thanks again, both of you for your posts.

-nimajneb

<font color="#949494" size=“1”> November 06, 2002 Message edited by: nimajneb </font>

<font color="#949494" size=“1”> November 06, 2002 Message edited by: nimajneb </font>

#14

Thanks for the interesting link nim,

no news from me about ZSphere Skinning (sigh - no time) but i wanted to post this pic to illustrate what i think is what you seek in a ZB mesh.

Now i always like to look at things in most possible ways in order to understand the meaning and nescessity.
My question would be if high density models with no concern for topology, which could be created in ZB very fast with 3dSpheres, could be used to create morph targets to simulate facial movement. And wouldn’t the resulting render sequence look as good as as a model with very strict topology ?

Just the old … do i really have to do this if i can get similar results with that … question.

Pardon my english :wink:

#15

<font color="#949494" size=“1”> November 07, 2002 Message edited by: ÈZ </font>

#16

Hi Ben,

This is a really interesting thread because it is at the core of modeling for animation.

I think we all agree on the necessity of an organized mesh/topology to animate hires, organic creatures. As you state in one of your mails, there’s different ways to get there.

Computer modeling is unique in that it asks of the user to have the solutions ready before starting the creative process.
I think nendo/meshtools are great tools to implement those solutions.

Personally I love the way zbrush tackles technical issues and empowers the artist. It has a set of tools that mimic an artist’s way of thinking.

Adding zspheres to Z-brush 1.5 is a clear step towards more 3d tools and it makes Z-brush one of the few apps that allow “trial and error” modeling.

There’s a real challenge here : deal with the technical issues involved in making a organized, animation ready model without losing the intuitive feel of the software.

I would love to hear from other people how to approach this the z-way. What extra vertice level features would be essential to work with zspheres?

e1

<font color="#949494" size=“1”> November 07, 2002 Message edited by: e1 </font>

#17

Hmm im not sure about the differences in the models you speak of but I’ll give it a go ;)…

I think the difference in the meshes between Ken’s and Sirquadalot’s models are because Sirquadalot’s dragon is a “raw” Zsphere model at a pretty high density.

While Ken’s model is (as far as I can see) the low poly, raw Zsphere mesh modified by pushing and pulling the polygons (or points) into place. And then, as he said, he divides the mesh and repeats.

I know in the Zsphere models I’ve done the raw models can at times look pretty funky. But they are generated very fast and after you’ve made the mesh you just tweak it by moving point by point to where you want em.

So I don’t see why it shouldn’t be possible to get a mesh like you described, is it not just a matter of moving the polys?.

#18

Nemo,

I’d say okay to that statment with one hitch. You’d have to have modeled the inside of the mouth cavity (i.e. made the space) so that your could open the mouth without having to change the mesh. Likely you’d have to do something similar with the eyes as well. The real point is that you’d have to make yourself aware that areas of the mesh are going to need to change drastically in the morph targets and make sure that you had enough mesh to cover those changes. I’m not sure how much control over the mesh you’ve got using ZBrush, but none of your operations once you’re set on a mesh could change its topology. Seems from what I’ve seen of the demo, this would not be a problem, and that the operations that create geometry are separate from those that push geometry. I’m still not terribly confident that its a good solution, but what works is what you should use.

and

EZ,

? Nice fish. I assume that this is an example of simplest ZSphere skinning? :slight_smile:

e1,

That hits the nail on the head of why I’m looking at ZBrush in the first place. By passing the planning stage so that we can just be creative and worry about form first. That’s why I liked Ken’s approach for modeling that recreating topology on top of the model after. It separated the creatives from the analytics of modeling. I think that ZSpheres are going a long way to making something to start with that has a usable topology and working from there. What would be even more interesting to me would be a ZBrush interface linked to together with a heirarchial sub-d modeler. You could set up a simple control cage as a base level for morphing, and use the higher levels to add detail with a ZBrush like interface. Rodin (old plugin for MAX) had something like that, but not nearly so nicely implemented.

and

Karasuando,

It is just a matter of moving polys, but also the ease of moving those polys. With a high enough density mesh, I don’t doubt you could get any shape you wanted, provided you were experience with ZBrush and a good sculptor. Outside of ZBrush however good topology wins hands down. If you have to do something to a dense mesh without ZBrush’s tools at hand, you will know the meaning of pain :wink:

<font color="#949494" size=“1”> November 07, 2002 Message edited by: nimajneb </font>

<font color="#949494" size=“1”> November 07, 2002 Message edited by: nimajneb </font>

#19

Ah yes that is very true, I’ve done some modelling in Cinema4D and moving masses of polys sure is slower there.

The sculpting tools are very good for dense meshes, but I think they as good on a less dense mesh aswell. Only I pretty much just use the “move” option. But of course you can’t knife or create new points etc.

#20

Thanks for the reply,

this thread is very valuable already :slight_smile:

I can understand now why topolgy is important in regular 3D Software. I will give it a try sometime although i have no clue how to operate 3DMax or the likes.

<BLOCKQUOTE>quote:</font><HR>What would be even more interesting to me would be a ZBrush interface linked to together with a heirarchial sub-d modeler. You could set up a simple control cage as a base level for morphing, and use the higher levels to add detail with a ZBrush like interface. Rodin (old plugin for MAX) had something like that, but not nearly so nicely implemented. <HR></BLOCKQUOTE>

I believe this would come in very handy indeed - however i think Pixolator is going another way atm - take a look at this thread

As i see it - this technique will allow you to create a model with ZSpheres, then refine the model using attractor spheres, then applying a displacement map to get the final touches. To animate simply use the ZSphere base as your “bones”.

This certainly wont give you as much control as vertex pushing and might not be satisfactory for facial animation but it will produce fine looking, animated render sequences. But noone has seen the new features yet so i wont talk of restrictions there yet :wink:

I think it might be out of focus for this application to offer all the mechanisms available in “hardcore 3D packages” since the pixologic team always kept the artist in mind when creating ZB - it is far less mathematical than other software - but more “natural” in its behaviour.

On the other hand a software package lives from its features so your proposal should go onto the wishlist :slight_smile:

Again thank you for providing such useful information. :slight_smile: