Zbrush2 and Silo

[Nemoid]

Cool i’m really interested into a similar workflow. usually i tend to model in Lightwave, with correct edgeloops and topology from the start, but not always i get excellent results cause this process is not so easy. Anyway, the idea of getting rid of real working topology at the start and then rework it later, amazes me, just because u can really project a full working topology even on sketches, and then apply your ideas to the model.

Also, i’m interested in to XSI tools for a similar process as u describe. do XSI is able to import a really high subdivided mesh from ZBrush as a template? i guess so, so maybe you can elaborate how many polys would be a limit for this operation in XSI?

ok LOL maybe i’m asking too much.

one observation though.

Traditional modelers with a lot of experience, tend to be organized well from the start, thinking to edgeloops in terms of animation and how the mesh will deform, just while modelling it. this requires alot of exercise, and knowledge of animation too coupled with modelling techniques. this seems to me so far a really good method to work, just because you will end up with a full working model with the exact amount of polys u want, placed where you want. the amount of polys determines their placement and design too, most of times.

traditional modelers doesn’t feel the need of a topology paint tool or something so far . obviously this will change in time, adding flexibility and the possibiliyìty to rebuild your topology with ease.

this being said, workin this traditional way its indeed a limitation for an artist especially if he becomes from traditional drawing and sculpting, and for this exact reason i like ZBrush so much.there was a time i literally hated polygons.

but after u have a low- mid res char modelled you can go in z brush and detail it with more ease i think.

another cool process could be just use z spheres in ZBrush and build your low poly model, in a very intuitive way ( it takes a few minutes instead of hours) position your points deforming low poly mesh as u want, then go and fix your edgeloops with XSI or othe apps for a full working low- mid poly cage import it in Z Brush and do all detail u want, subdividing your mesh etc. normal maps build detail for you for the final process. the first part of this workflow will make u work into less many hours IMO.

[Strike]

Hi Dave,

Each time you post, i’m always amazed by the quality you put on your details. Fabulous work ! You have a fan here…

[skycastle]

Thanks guys. And thanks Strike. I like seeing your post as well.

“Traditional modelers with a lot of experience, tend to be organized well from the start”

This is what I DONT find to be the case. Each time in the past where I worked
the traditional way ( cube up, then edge loops and so on… ) I would bring my
model into Zbrush for detailing. Almost as soon as I start I realise that the edge loops I imported were in the completely wrong place or wrong direction!
More surpriseing to me was that even very small, almost texture level details sugested changes to my original edgeloops in the base mesh. As Ive said in the
past working the “traditional” way is like creating ballon characters. Just way to smooth and not even close to what the character needs in terms of details
( this would be for more realisticly styled organic designs ). Thats one of the reasons why I feel that working out as many form related issues up front is the way to go.

[chadtheartist]

I’m intrigued by the concept of edge-loops, in light of this new discovery in the topology brush.

In the past, why were edge loops defined? Was it just for animation, or did it also help define the mesh in a more controlled fashion? I’m obviously not yet a professional, so I am merely going on what I’ve seen and read on the internet. But it looks like, at least to me, that the whole edge loop functionality is still very much a possibility and a necessity with the topology brush. In fact, I believe it could be very formulaic for certain instances, i.e. faces, torsos, etc… Therefore, instead of having to worry about achieving that formulaic edge loop form in the beginning, you can systematically do it later, and possibly faster.

From my studies, I’ve found that vertex pulling is not a good way to model organic shapes. Zbrush obviously eliminated that with it’s amazing toolsets, one of them being the 3/4 modeling view. However, it’s zsphere creations sometimes have problems in certain areas. And as for someone who is trying to get into this industry, the wire-frame that the zspheres makes is not ideal to show on a demo reel. Now, thankfully, in comes the topology brush, so there is an answer to that puzzle at least.

Now, allow me to brainstorm for a moment. There are few ways these two programs can work together. I guess it will be up to the individual user/or company, to decide what they will use it for. But here’s my idea of it’s use.

Let’s say I take a model, like this:

http://homepage.mac.com/chadtheartist/.Pictures/djinn/topotestlowpoly.jpg

I rough it out in Silo using the soft selection tools, like Dave did on his model in this thread. I took that model into Zbrush and smoothed it, to get rid of the inconsistencies due to the soft selections limitations in Silo

http://homepage.mac.com/chadtheartist/.Pictures/djinn/topotestfront.jpg

And then I start working on redoing the topology. I’ve found that doing hands in this fashion is not an easy task, in Silo or Zbrush. So I redid the hands the old fashioned way. I don’t know if the topology brush would be a help on doing hands anyway. Especially considering the many different fingers, etc…

http://homepage.mac.com/chadtheartist/.Pictures/djinn/updatedhands.jpg

Here is my idea now, since I won’t continue this until the topology brush is made available. I had originally wanted to make the mesh really, really detailed. I figured then I could just redo the topology, and that would be it. But then it donned on me, why redo all the detailing twice? So I’m going to be trying an idea of doing a medium res mesh, just to get the general shape and form, and then detail the model with better topology in Zbrush.

For instance, I have a pretty good place to start building the face using the topology brush. Now, I can focus on the topology without having to worry about focusing on volume too. IMO, this is going to be the biggest benefit of working in this fashion.

But in retrospect, I can see the topology brush being essential for future gaming models. The fact that you can redo the geometry on many different objects is going to really help with keeping the same silhouette of the model. Using the tools that Ariel pointed out to do normal maps, and even displacement maps, is going to be one heck of a time saver. Interesting stuff all around.

Of course this is my take on it, so I’m sure there are other uses for this that I haven’t thought of yet, aside from being able to add things to models like clothes, etc…

I wonder what they are going to think of next?

[Giantsun]

…a “make great model based on what I did” button ?

[lodenfactory]

Cool thread skycastle, your works are amazing
Anyway I was wondering if the UV process is better to do in Zbrush with something handy like GUVtiles or whatelse or outside Zbrush in your opinion. The unwrap process could be very tedius sometimes so is really cool import the model, press a buttom and start texturing but is there some cons in doing that? Since here we are talking about “workflow” I was just curoius…
Bye

[Ariel]

Also, i’m interested in to XSI tools for a similar process as u describe. do XSI is able to import a really high subdivided mesh from ZBrush as a template? i guess so, so maybe you can elaborate how many polys would be a limit for this operation in XSI?

XSI is not as efficient as ZB in handling high ammounts of polys on meshes. LEt’s say you make a 1 million poly model in zbrush. I would import this into xsi and run a polyreduce operation o bring it down to around 100,100. At this stage the structure and planes of the model will still be intact. What will be gone are the small details, which you don’t want in your mesh anyway.

The good thing that I’ve found about this workflow is that topology never precedes design and aesthetic. Topology becomes almost an afterthought, which you complete at the very end of the process, without compromising form. You could think of it as scanning a maquette and then importing it into a 3d app for re-parametrization only that here you would sculpt the ‘maquette’ in zb.

This process was described by Eric Sanford in a Siggraph03 presentation. you can watch a video of his demo here:

http://www.softimage.com/downloadsrv/process.asp?file=/education/seminars/eric.zip

[Ktaylor]

Well that is just beutiful,

I have a question for you skycastle. Lets say you have this awesome character rig and it has all the muscles on it and the slide around and colide with each other and just rock the house, now when you bind your character to this rig and start to do the muscle simulations what happens with all the muscle definition that is modeled into the mesh itself.

Seems like they would “betray” one another when the simulated muscles moved around under the muscles that dont dont move around, know what I mean?? I know you guys do that a lot there and I was wondering how you get it to work.

Thanks, and as always great inspiring work!!

[sumpm1]

I think that the edge draw function in Silo is only a first step in the process of topology automation. I dreamed of a tool for ZBrush too that would automate, or simplify the redraw process. Rather than redraw every single edge, you could simply have an optimize topology function that somehow finds the flow of the model, recreates the base topology, and then applies the proper displacement to keep all of your details. It is unlikely that this automatic step would get it right every time, or have what you had in mind for the base topology. For that there would refinement methods, such as drawing flow lines that would not be exact edges drawn onto the model, rather key edges that directed the directional flow of the topology, and the optimize tool would fill in the rest for you. For example, on muscle topology, you could draw in the valleys of the muscles, and in the creases of the armpits, and elbows, since these will need to animate.

In the case of the topolgy redraw tool in Silo, you have to redraw the entire mesh, I see this as a limitation of the tool. I would rather see it possible to redraw a portion an d have it updated to fit in the mesh. Let the cpu think about how to make quads and such. Being limited to redrawing an entire model , rather than the portions of the model that need it, seems like it could become an inconvenience

There could also be a topology slide tool. This would allow you to keep the current topology and lock the current shape of the model. You could the slide the mesh edges over the model like cloth and keep the shape that you have already spent time to work out.

I think it would be the hardest thing that Pixologic ever did, but if anyone would or could do it, it would have to be Pixologic. Dave, I am sure that Pixologic has a plate full. But I would love to see your influence in this area of 3d.

[pnoland]

I’d like to see this as a Zbrush feature as well but don’t underestimate the Nevercenter team. Once they unleash Silo 1.3 to us all I’m sure we’ll flood them with more feature requests like the ones you’ve stated…they’re awesome developers so I’m sure they’ll be game for some things like this. The tools is new as far as functionality goes so there’s probably room for them to play around with it. We’ll have to see when they release it publicly! …which I can not wait for much longer I’m gonna burst!? :lol:

[Abyssis]

hey-ahou skycastle, you and Strike are both c_r_a_z_y I like your work as well

---

Abyssis

[Nemoid]

Thanx Ariel for the infos and link

[Ludenhud]

As usally when I see such a great modell I would like to see an szcript of it Awsome work you got there dude, keep on going

[vivien]

argh.HOW do you do it?!?

[custom_junk_mail]

I personally haven’t had as much pleasure in using the post-topology brush approach. I still find it easier and faster to establish the edge loops at the beginning in the rough form then use zbrush for detail work and export displacement, normal maps, and/or bump maps (depending on whether I’m doing the animation for a realtime environment or not) to add the detail to the low level cage. Sometimes, if my rough level 1 cage is really sparse, I use the second level cage instead for my animation work.

When I’m doing high-res models, I also use ZBrush for corrective morphs (along with regular morphs for muscle animations) instead of fiddling with vertex weighting and bone placement all day long. I can make them really quickly in ZBrush.

I also don’t see why, in animation packages like Maya which support hierarchical subdivision surfaces, the entire hierarchy can’t be exported. I bet that’s a strong possibility for future updates and would make it so you can animate ultra-high detail zbrush models very easily without the need for displacement (unless you are going for speed and need to eliminate some resolutions at the top of the hierarchy).

I use Lightwave which doesn’t support hierarchical subdivision so I can’t look forward to this feature, but it’s something to think about.

I think that as computers become more powerful, topology in faces at least will become unimportant. Right now, the only big advantage to establishing proper edge loops in the face when using ZBrush for creating high-resolution morphs is that it’s easier to avoid the jagged stair-step look in creases and wrinkles which can form if the model’s edges don’t lie on those creases. This is simply a matter of resolution, and as computers can handle higher and higher resolutions, the effect won’t even be visible anymore.

[Nemoid]

well, maybe one day we will have enough resolution to do that and topology polyflow and other will be a thing of the past. that day will be a great day!!

seems that an app like ZBrush is able to handle those millions of polygons because of it’s way to work. incidentally i’d like to know better how it does this , but maybe it’s a factory patent…

other apps aren’t apparently so powerful, and that’s why the low-mid poly mesh has still a reason to exist for now. so a good process seems to me that of using zbrush to fine-detail and texture an existant mesh,with a good polyflow made in Lw , Silo or other package. a good polyflow seems for now to allow even less subdivisions in Zbrush so this is very good.

[nergal]

Hi Skycastle. The modelling and detailing of this model is quite spectacular. When utilizing
Silo and ZB2 workflow, How do you redraw topology? What is your recommendation for learning Silo? What helped you?

Thanks a great deal for sharing and inspiring…

[Jason_Belec]

This is a very interesting thread. With a lot of good points. I think people should make sure they differentiate between game and other areas to see if certain trends appear.

I come from the feature, television and vfx side of things. The most useful modeling tool for me prior to Zbrush was NURBS and Photoshop. I could create high detail models with little mass (polygons) until render. I used an old ILM approach (Cubic Mapping) for the details derived from Photoshop - almost exactly as TextureMaster/ProjectionMaster work. Then Zbrush came along and what took weeks or even a month was now a week or less. The major thing I noticed was not how complex a rig was, not how high quality weight maps were or even how many detailed morphs were used - BUT - how you modeled and placed elements. I found that in 80% of the models if you modeled it right it acted right and usually with the simplist rig. Yes, this is hard - especially if the design is by someone else who just made a cool doodle. I am alos taking into consideration working/fixing models for others in order to complete a project.

The reason I bring this up, is someone mentioned ‘more powerful computers’, other things like EdgeLoops, TopologyBrush, etc. were also mentioned. Well, the problems won’t go away, they are just pushed along the scale, and this discussion begins anew. You have to try a few things, it’s always easier to something that has been done, but you can apply what others do to your own models to prevent issues. Forums like this help, in the last year alone I have scene facial rig setups that reduce production time by several factors - as an example. Yet the ability to do these has been around alot longer.

I for one have never given topology a second thought. All I have cared about is - looks right - acts right. This is not to say it isn’t important, or that you shouldn’t consider it. However as with all things, perhaps you should look at the process and not necessarily the tools. I am big on pushing the creative approach (traditionally trained), and I am a process/technology freak (main focus of my career). I don’t like redoing anything, I don’t mind fixing it, but I should only have to do it once.

I will have Zbrush2 at months end - it did get a chance to put it through it’s paces recently and found it amazing - the time savings is fenominal.

I probably haven’t contributed anything useful, but…

[Nemoid]

I think we re discussing these things about ZBrush/Silo because sculpting in Zbrush is realy natutal, and brings a totally intuitive approach to modelling organic shapes shapes.
modelling that way directly in Z Brush implies millions of poligons but Zbrush handles them very well. i wouldn’t be so surprised if a topology brush or other clever tool will find its space into zBrush sooner or later, to let u model as u mwant , naturally, and then simply use your mesh as a tempate to draw on (IMO this isn’t redoing things, infact you’re only optimizing things a lot, to use way less polygons - good for animation):lol:

obviously, the best and more effective process is currently to build your model into another app, then fine detail it in z brush, for all things u cannot achieve into a normal app. but i think we are not so far to change this way to work, wich is indeed more technical than artistic.

[Tim_Leydecker]

ironed out some spelling mistakes:

One step towards being freed of legacy edgeloops may be the transfer tool
in Maya 6.01´s bonustools (remember it´s grandfather, the wrapmesh?)
in short the workflow could be described as follows:

*Model a neat lowrez blockout of your character, describing the pose.
Use loops and such to get a feel for the volume.

*Import into ZBrush. Detail it like crazy. Bend. Sculpt. Subdivide. Play.

*Import back into Maya. Adjust your lowrez cage to better fit the highrez one.

*Redo the UV´s on the Lowrez version. Do a transfer calculation from
highrez down to lowrez. (The highrez doesn´t need good UV´s or a
similar, subdivided topology to make this work)

*End up with a Normalmap describing the difference between the highrez
and the lowrez. Use that for displacement or just for rendering.

The same should work inside ZBrush, e.g. creating the Normalmap from
the two meshes, the updated lowrez and the highrez version at least
by using the MayaNormalmap to displace to higherrez and refining the
result in ZBrush before redoing the calculation in ZBrush itself.

The last sentence sounds like overkill, that´s due to the fact that I don´t
know if ZBrush can generate a Normalmap from two different objects or
if it is restricted to bake subdivision levels.

tim

P.S: Editing the Normals and creating a useful shadingnetwork is quite
a pain still. Be it due to 16bit limitations in Photoshop or slight buggers
in the creation process like AA killing transitions between UV borders.