Such a weird title deserves a weird project so here you go:
Originally it was just a fast head doodle in ZBrush but it looked quite cool to me so I decided to keep going on it for a bit and actually came up with a bodyscetch as well.
Here are two shots of the current ZBrush version(with partly not so cool anatomy…):
If you now wonder why I even post this here: I’m planning to finish the model or even give it a whole scenery, so it needs new topology since the basemeshes are rather simple (a cube - with quite a few subdivisions - for the head and just a few ZSpheres for the body).
Anyways, here’s the first step:
There is still some room for optimizations in various parts but that’s work which I’ll take care of a little bit later.
I’m going to be away for a few days from tomorrow on, so I’d appreciate if you just keep posting some C&C and I’ll answer any upcoming questions afterwards.
totally awesome! but i am wondering about your workflow. are you rebuilding the topology in blender? if so, i am wondering how, and if not, then yes, i am wondering why you are posting a zbrush model here. ; )
there are a number of interesting methods for down-rezzing models in blender, using the topology tools, or baking normal maps to apply to lower rez models. i would be interested to hear about your techniques, and experiences.
Well… sure I do, that’s the reason I actually post this here.
The usage of blender in production - of whatever exactly this project leads me to - will most likely be limited to retopologizing (with a bit of tweaking later on) and uv-mapping, but why not letting you take part since these are (or should possibly be) quite interesting steps for a lot of people here.
Ok, the technical stuff now:
The way I do it theoretically is quite simple:
I export a model from ZBrush at a resolution that reflects enough of the detail to get a good feeling of the overall shapes that need to be modelled (but not too much since blender can’t deal with very high density models all that well, and also because it’s simply not necessary) and then load it in Blender.
Afterweards I simply turn on the retopo mode and start modelling (poly by poly) on top of the design sculpt.
The normalmap thing…
Well, Blender still has issues with that but it’s getting better by the time.
Anyways, you still need to create a lowres model to apply a normalmap to in most of the cases and the retopo capabilities of Blender work quite well for that.
(You can find an example here if you’re interested.)
What kinda bothers me a bit is that the p.e.t. isn’t really working in the retopo mode, meaning it works as always but that’s it. The underlying model is completely ignored which it shouldn’t. A p.e.t. that behaves as you might have expected it to would improve the creation of a nice polyflow considerably because making that verty by vert is quite time consuming. Hopefully that will be the case soon…
That is one of the most hideous looking things I’ve ever seen. Freaking awesome.
As far as the re-topo goes, I’ve never tried it myself. I generally avoid making my meshes this dense, too. Everytime i see one this detailed, my inner rigger has a panic attack and passes out. It’s looking really good, tho… there’s 5-poles all over the place (my inner modeler has passed out as well) but you probably don’t mind them since it appears you know what you’re doing, and since it’s re-topo work, i guess modeling workflow is more about the early details than blocking it out… looks like fun… or something i’d break into a cold sweat doing.
I definately want to see this one finished (and animated if possible). The design is pretty bad-ass.
Haha, I can understand you just too well.
Sometimes it’s definately a pain but it’s worth it.
About the poles: Well, it’s not that I really want them to be there but poles are just something that exists and they are simply there when loops touch each other.
The bad part about it is that I need to model out a good amount of details and that results in a lot of loops if I don’t want the mesh too get too dense (which I really don’t want it to be) so it’s basically something that I have to come along with. Sometimes they’re annoying sometimes they aren’t noticeable, it’s like balancing on a knife’s edge.
Yep, you nailed it quite exactly: The design of the rough shapes was already done while sculpting but you totally loose any sort of decent topology if it ever was there at all, which is only the case very few times since you’re usually starting with a simple cube.
Going this way gives you a very nice opportunity though: You can concentrate on pretty much nothing but the design (and a couple of technical aspects of the used tools).
If you need to make some big changes you can simply go ahead and you’ll be finished after a very short time compared to the same actions done by traditional modelling. (And guess what happens if you don’t like your changes. )
In this stage of production (known as design sculpt creation) you give the model its shape and rough details but mostly not much more since it possibly won’t ever be seen again once you’re finished with the retopo work. (Exept the case of geometry projection.)
As soon as it’s finished you export (and also import) it and start retopologizing (what a daft word…):
Therefor you first have to get a feeling for the average density of the mesh that could fit your needs the best, meaning that you roughly decide how far each vertex should be apart from his neighbours to be able to capture enough of the underlying models shapes.
Then you start modelling while keeping that at the back of your mind; how you do it generally is up to you but here’s my way:
I focus on just a few details close/somehow connected to each other that need to be modelled out, and start drawing lines of verts (with the density I defined before) that follow the striking forms. Afterwards I simply connect those lines in a way that I think is accurate for the concerned part.
If you now think “SIMPLY?! This guy is insane!!!” and almost fall off your chair, jump up or whatever: I can understand you.
This needs quite a bit of practise in most of the cases but the golden rule for it is to keep the resolution of the model that you want to archieve in mind at any time.
If you follow this rule you will usually only face small problems even when trying to connect rows of verts that are far apart from each other, which f.e. sometimes was/is the case in this particular model.
If there are any special questions about that feel free to ask here or take a look at some poly by poly modelling tutorials.
Darn, I kinda wrote a whole lot, I should rather go back to modelling…
Thanks and I will definately keep going, I’m just not too sure about how fast…
Probably not, indeed. But the actual reason for that simply is, that I can’t catch enough detail in just a normalmap so it will more likely be a disp.map first and additionally some bump-/normalmapping on top.
Unfortunately that won’t include blender in any way since I’m not expecting the necessary features to come anytime soon. (Not even those to simply apply disp.maps for example, not to talk of the rest.) All that will be a ZB and XSI case, just as planned.
I hope I’ll get my website online till then because that stuff won’t really fit here I guess.
Anyways, I finished/refined the geometry/topology on the body and head for now, which means: onto the arms!