Skinning and unwrappping models

[thrasherstudios77]

Hey,

I am looking for some help with getting started with skinning a character and also with the method for unwrapping my character model so I can paint displacement and bump maps in Photoshop. I have looked for tutorials on these subjects and came up with nada.

I am not sure exactly what OPs I need to use and even where skinning and unwrapping take place, OBJ level, SOPs level? So I basically need to be pointed in the right direction. BTW how is skinning with Houdini compared to other packages. I have knowledge of skinning in Maya, which is a pain in the arse, hopeing Houdini has more powerful skinning tools. Does Houdini have smooth and rigid skinning methods, as well as influence objects and flexors, that is what they are called in Maya anyways.

Mahalo

[TheUsualAlex]

Hey there,

With unwrapping stuff, you can use stuff like UV unwrap SOP and such.

However, first thing you need to do is to make sure that you assign the UV coordinate to your model first. Best of all, it's 100% procedural!

So you may have something like:

Model

|

UV Texture SOP

|

UV Unwrap SOP

|

UV Edit SOP (to edit your UV and such)

|

UV Fuse SOP (to merge UVs)

Press Spacebar+5 in your viewport to go to the UV viewport windows.

As far as skinning character goes, I have only gotten as far as testing Houdini's character tool capability a long while back. So I haven't use it in a project yet. I find it just as good, if not better, compared to other popular high end software. The flexibility that Houdini allows you to have is just pure insane. Houdini uses pill style capture to capture weights. People from Softimage probably won't have any hard time understanding it. But from Maya base, I find this to be similar to Rigid Bind. The Paint Capture Weight is more like the Smooth Bind to me -- as you are painting smooth binds. I find Houdini's Paint Capture Weight to be faster than Maya's as well as much more predictable. When you use the pill-style to bind your character, this would mean that even after you bound your character, you can still make changes to your character even after you rigged and animated your character because of its procedural capability.

I personally prefer what I called conditional blendshape (set driven key blendshape), where I only use Capture SOP to the geometry only and then only using the Blendshape for any deformation. Although this is a lot of work and setup as you're basically blendshaping the points and model your character over and over again, this method, as I find, can preserve a much better topology. I do believe that this is the exact same method that Steven Stalhberg is using as well in Maya.

Personally, I say that Houdini 5.5 is pretty much capable of character animation. It's just a matter of production studios setting up their pipeline for it. And I just simply am very excited about what H6 will have!

Cheers,

Alex

[thrasherstudios77]

thanks you for the lengthy reply, I thought Houdini would allow you to go back into the model and tweak it even after skinning, that by itself is amazing. When you were talking about using blendshapes for deformations how would this be done. I already know how to make blendshapes, but do you mean after skinning your model and then see that at a certain pose the skin looks sunk in or messed up, that you just use a blendshape to reshape it for that particular deformation. Example:

Lets say you have the skin around the arm and shoulder socket not deforming right when you bring the arm back, would you just use a blendshape for the geometry around the arm socket to reshape it. If this is true how would you set up that blendshape, how do you trigger the blendshape to do its stuff only when the arm is brought back. Hope that makes sense. Never tried an approach like this and have a hard time visualizing the setup, sounds really cool though.

Does Houdini have flexors or influence objects for skinning? And if so how do you go about using them.

Thank in advance

[miguel m]

hi

if you right-click on a sop there's a command called "save texture image" that will save an image with your uv's stamped on it. that makes a nice reference to add as a layer for easier texture painting. btw, with this I often prefer to smudge the texture to fit into a nice laid out uv set than to move lots of uv points around to fit a texture.

the unwrap sop is great too. using the optinal second input for custom planes to unwrap to is great. Saves a lot of time. In other software i'd have to make several custom ortho projections, then battle them to fit and the overlapping. beware that unwrap makes one region of continuous uvs per plane, so you'll have seams. It does well with more or less homogeneous textures.

another option i use is to modify the source geometry then apply one simple projection. For example a head has lots of concave and convex surfaces that will give you overlapping with one single spherical or cylindrical proj. I use the smooth sop to even out a bit the surface, then the ray sop projecting the head halfway towards a bounding sphere turns out well. Then just need to copy this uv attrib to the original geometry.

for skeletal deformation there's just so many solutions (joint, arm, limb, etc) but these are a bit on the vintage side . i think the best start is using bones and the more standard tools. there are 2 capture methods and two weight editing methods, so there's plenty of flexibility there.

There are no "dedicated" deformers a la mirai or i guess maya, but no problem either, we got lattices, metaballs-bulge and more. Alex showed me a nice scene with custom bone-driven blends (still got it to show?). In my case I try to rely more on houdini's solution but I'd like it to have more than one behavior to choose from. In softimage I could choose from bycep, finger behavior and the like. I mean, those behaviours were built-in into the bone object and the distribution of the bone's "force field" changed with each.