Just looking for some general advice/guidlines when making clothing and how to handle armatures with it.

For reference, I'm using blender 2.72.

So, when I create a model, then create clothing for it, then try to rig an armature to it, I'm having a couple of issues. The larger issue is the armature binds to the model, and more or less ignores the clothing, as it's above the other layer, even if I join the body and clothing into one object.

I am using automatic weights when binding the armature. I can go through and hand weight paint everything, but it's incredibly time consuming, a little awkward as there are multiple layers with the model and clothing, and the process is prone to error (I always seem to accidentally get some wrong vertices in there, say from the opposite side of the model, and it throws off everything). I much prefer to use the automatic weights, and just fix whatever little issues that are created, but with the clothing, it's pretty much the entirety of the cloth that doesn't get bound to the armature.

Offhand, I imagine I can dissolve the vertices of the model that are under the clothing, but that creates issues as well.

Anyhow, I was curious how some of the more experienced of you folk go about this. Do you just buckle down and hand weight-paint everything, or is there a method I'm unaware of? Thanks for any help! :)

There are a few things you should consider.

1. Many games which uses different armor/clothing etc. do it by using completly different, fully armored models. There are several issues if you use layers of meshes or in general disconnected meshes. When animated they will most likely penetrate each other and to ease these glitches , you need to tweak the skin-weights alot.

2. There are clothes or armor pieces which can be attached to a single bone. Eg. a arm bracer or helm. In this case don't use auto-weights. Assign all vertices of these models to a single bone (use vertex groups).

3. Auto-weighting works really good in blender and I use it all the time (no manual weight painting). But you need to know how it works to utilize it to your benefit.

Rough concept:

Step 1: For each vertex the algorithm checks the visibility of bones, whereas surface will block its visibility.

Step 2: If it can't determine one or more visible bones, it will try to move along neighbor edges and inherits its bone visibility. It will travel along the mesh until it founds some visible bones.

This is the reason, that layered models or models with lot of penetrating surfaces will result in a mess. Best to have a clean, connected mesh.

Here are some tips to utilze it.

I. Integrate meshes into the model instead of layer it. Eg if you have a belt, dont just put it on-top of the base model, integrate it into the base model itself.

II. If you need to animate some pieces by single bones, move them away from the main model. Eg the eyes, move it 3 units away, move the eye bones accordingly away and put them back in place when posing your model.

III. Utilize the blocking ability of surfaces. Eg if you have an avatar with wings and you want to avoid that the wing-bones influence the avatar-back, then use a not rendered surface (use different materials) to create a hull which blocks the wing bones from the back.

One sugguestion: When you are new to blender/modelling, just don't try to animated layered models. Keep them simple and clean

Thanks for the advice :)

I had tried joining the clothing to the model, but apparently that isn't sufficient. I'll need to look up in the docs how to fully integrate the two. For my current needs, I don't actually need them separated at all (there won't be clothing changes or anything), I just find the cloth physics tools to be incredibly fast in creating good clothing, rather than by modeling by hand.

By integrating the two, do you suggest deleting the model vertices under the cloth, and then joining them together where they would meet, or is there a better way to go about that?

Thanks for the tips (such as moving things away until posing and blocking the objects)!

By integrating the two, do you suggest deleting the model vertices under the cloth, and then joining them together where they would meet

... and eventually add new faces to close the mesh.

I personally like clean meshes and this is the way I do it. With time you will get really fast at cleaning up the models. There are more benefits when doing it this way:

- texturing gets easier (no wasted texture space)

- animation gets more controlled (no penetration)

- less vertices (better performance)

Thanks again for the tips, Ashaman73,

But, ugh. Closing the model back up manually is hard. I used the boolean tool to erase the original model vertices where they intersect with the clothing, ended up with bad faces, that got messy to clean up, so I started over. I tried reversing the boolean order to delete the clothing vertices where they intersect, but still have all the original model vertices to delete.

I started doing it by hand, and it quickly became an effort in futility. Perhaps it just takes practice and experience. I'll keep at it, but properly clothed models has slowed down my workflow considerably. I'll keep looking around for advice and tutorials, and play around with various methods on my own, but I'd still be open to a little more advice regarding this, if anyone is still reading this thread.

My kingdom for a tool that will just close the mesh automatically.

But, ugh. Closing the model back up manually is hard. I used the boolean tool to erase the original model vertices where they intersect with the clothing,

Avoid using the boolean tools, it creates a pretty mess. Try to do it manually.

If you are not an artists, you will often be baited by some pretty tools which produce some good looking effects with a few click. But eventually, creating art is a lot of work with a complex workflow with many things to consider. The few clicks which produces some pretty art at the start of your workflow, will eventually result in a mountain of work in the later workflow stages. An other pitfall is, that you get/buy/create some pretty character models and suddently you need to level all the environment stuff to an equal level of quality.

If you want to create art for your game yourself, you should consider the necessary workflow, because a complex workflow will pretty easily outshine the coding work.

Great advice. I'm currently trying to feel out my workflow for this, and trying to settle into a simple-ish style that isn't overly time consuming to produce, but still looks clean and professional. I'm not going for photorealism or anything, just clean working, animated models, neatly normal/specular mapped. Aiming for slightly cartoony (mostly to simplify the geometry a bit).

I had started going down the road of various programs to speed things up, but ultimately felt a bit unsatisfied and decided the effort would be better spent becoming better acquainted with blender/Gimp and perhaps simplifying my goals. I wouldn't buy the models, as I think things would just clash, ultimately.

I was mostly wanting a tool to close the model, as I increase the vertex count considerably for the cloth physics (it's wonky with low polygon meshes), and then lower the count afterward. But, it got difficult to close the two meshes at that point, as that process would set vertices in pretty randomish places that were more difficult to reconcile.

I may have to forgo the cloth physics and just sculpt the clothing by hand. It seems it may be the simpler option.

neatly normal/specular mapped

Normalmaps are difficult. The standard way to create normal maps is to sculpt a high-model and bake it to the low-poly model. This is a very costly workflow. There are other ways, like generating the normal map from a texture, but the result is often sub-optimal . Trying to directly paint the normal map or to paint a heightmap is really cumbersome.

I've done some research in art styles with a strong focus on an affordable workflow and in my opinion one of the best looking, non-aging and cost efficient styles is ...tata... TF2.

The basic benefits of such a style is, that

1. It uses no (or really seldomly) normal maps.

2. It uses really low model details.

3. It uses a AO map of the model as basic shading.

4. It uses a low-detailed, handpainted texture.

5. It uses exaggeration in modelling and animation, which makes the animation stuff a lot easier.

6. For environment they uses mix of photosourced and handpainted textures.

When you are interested in more informations, take a look at valves publications over here.

Yeah, TF2 is roughly the style I'd be aiming for, with perhaps less complex models. I also hand-paint the models (I find I hit a sort of uncanny valley if I try to use textures).

For normal mapping, I do create them from a diffuse texture, and discovered like you mentioned, it can be sub-optimal. However, I find if I go easy and restrained with my use of it, I can hide most of the shortcomings of that method. I use it for some light hinting of texture so things aren't toally smooth/flat looking, but don't aim for a totally photorealistic normal map. As of recently, I've just been using shadow mapping for the bulk of the shading, though have run into some shortcomings. My other shaders could use a lot of attention still.

I have the day off, so I think I'm going to spend it toying with different methods for getting "clothing" right, and just generally getting more acquainted with blender.

Edit**Update***

After trying loads of things, modeling by hand and skipping the clothing physics works tremendously better for me. Rather than create another layer of clothes, I found it far simpler to just sculpt and edit the existing model instead (workflow is much quicker as well). The cloth physics seem awesome, and I'm sure work wonders in certain situations, but for clothing for the models, at least for my purposes, excessively complicate things.