Muscle System/Univ. Figures -- Monkey Biz! (NSFW)


Note: Skin shader breakdown starts @ post #27


Based on the work I’ve done rigging human figures for naturalistic posing & animation, I’m now working toward a full muscle simulation. I’ve seen this done a number of times so I’m not really breaking new ground except that this uses my already-tested rig with few adjustments. This has greatly simplified the first development stages:

Since I already have in place Constraint-driven helpers to deform a skin mesh to imitate muscle action, it hasn’t been a major task to apply the same principles to a modeled muscle layer. So far I’ve mostly been trouble-shooting the skin-wrapping methods, but I’ve gotten good results with early tests of the shoulder and biceps region:


Some of the shoulder controls in the rig


Early modeling of the shoulder & bicep region

Since the vertices of both the muscle layer volumes and the skin layer are being driven by the Armature, the problem of he skin “sliding” too much over the muscle layer is greatly lessened.

My next step is to divide the muscle layer into 5 segments (extremities and trunk) to solve the problem of having the skin “wrap” to the wrong muscle when the surfaces are closely adjacent, such as in the armpit. Each segment will be a separate target for multiple Shrinkwrap modifiers on the skin mesh. I also want to use this to add detailing to the hands and feet. The face is already under a separate armature that use muscle motion as its design basis.

One chief benefits so far has been a quantitative and qualitative leap in muscle form and definition, without increasing the base mesh density at all. Since I use 2 levels of Subsurf when rendering (and sometimes 3), there’s really no penalty in using these as a way to help implement the Shrinkwrap effectively. By placing one Subsurf before and one after Shrinkwrap on the modifier stack, I can eliminate a great deal of the “gap stretch-marks” that this technique can generate. I’ll of course be testing it under much heavier deformation & animation scenarios as well.

Great idea! I’m very interested in seeing the video! c’mon vimeo

It’s up now, sorry for the delay, not sure what went wrong the first time.

wow really nice stuff here… curious to see more!

very interesting approach
(definitely following this thred, i’m really curious of what can be accomplished using this method and how to implement it)

Me, too! :smiley: This is very much in the R&D stage right now, but it does look promising. My main goals are achieving better muscle definiton without resorting to Multires sculpting & normal mapping (except perhaps for the finest details), more natural-looking skin/muscle interaction when animating and posing, and keeping the rendering requirements within my current boundaries, i.e., no super-high Subsurf levels, no inordinate demands on the armature/meshes system. Working around some issues with Shrinkwrap will require a heavily-loaded modifier stack on the skin mesh, but I don’t think it will produce any huge performance issues. Only testing will tell.

I’m also thinking I can use this to improve definition of the knuckles and tendons of the hands and feet, but that could become very complex due to the individual digits, maybe more complicated than the results justify.


I’ve had some time lately to bring this project forward to the next step, using it in an animation setting.

Using my generic female figure I’ve adapted the rig and its underlying muscle forms to a basic walk cycle. This required some considerable changes to the use of the Shrinkwrap modifier, basically having to break the body into separate parts – limbs and torso – to keep the skin acting properly during motion. I’ve found I’ll have to subdivide the limbs as well for more extreme flexing at the knees and elbows, but that’s for stage three.

The Universal Figures project is intended to provide a malleable base mesh and a highly functional rigging system for naturalistic human motion in Blender. The muscle system variation is intended to provide a means of incorporating soft-body physics where appropriate, as well as adding higher definition to the physiology without using higher-resolution meshes or relying solely on static normal-mapping of musculature details.


The character-specific adaptation included using the Shrinkwrap parameters and underlying muscle mass modeling to depict a lightweight skeleton over a slim, well-muscled, but not heavily-muscled, frame, consistent with a fairly common notion of physiology for elves, and female elves in particular. The frames shown were rendered from the walk cycle in 29-frame steps, basically an alternate-foot stride – full walk cycle is 59 frames @ 30fps.

nice work… hope all is progressing well…

Re-opened at OP’s request in order to post updates.

Nice to see this project going: for those (me in particular) that cannot model anything, using a parametric workflow is always better than nothing.

Looking forward to that !

Thank you, Richard & the BA crew!

Been a long time between updates due to a work project (yes, with Blender!) that went way longer than expected, and a need to just get away from the biz for a time, a vacation of sorts. But the siren call was just too strong!


Working with a successful proof of concept (previous posts), I decided to improve the process in two areas – accuracy of anatomy and response to the Shrinkwrap “skinning” process, a well as improving the underlying forms’ response to the armature. So I built a skeleton, both as a useful file and because it has really helped in getting the musculature more accurate, the attachment points of the muscles at the joints acting as landmarks of a sort. The bones also provide shrinkwrap targets for portions of the skin mesh where the skin rides tightly over the bones themselves, such as at knees, elbows, and in women, portions of the pelvic girdle.


The rear view shows places where the musculature need refining, such as the triceps and the scapular regions.

My previous figure armatures used elaborately constrained “helper” bones to assist in getting proper deformation at the joints, and this system is intended to replace that to a large degree. It will, however, use similar constraint systems to activate the muscle forms for contraction on flexing, for a dynamic muscle response.

This figure is placed squarely in the “average” somatotype for a young woman, approximately 15% - 20% body fat, a parameter that can be adjusted via the shrinkwrap values as well as underlying muscle mass modeling. Once all the areas of concern in the system have been ironed out I’ll be experimenting with varying the general look of the figures (e.g., heavy and thin), adding soft-body response where appropriate, and of course working on the male model as well.

Since this is also a “universal figure” project, I spent some time reworking the modeling of the face, which uses my Othello face armature, hence no skull and facial muscles. I also wanted to spend some more time using the Cycles hair system, as it was quite new the last time I worked on the project. I’ve revamped my skin shader, mainly just tweaks to existing values such as the specular components and the SSS values, but I think its a lot more successful now. The “bump” component is proving troublesome, though, so I’ll save closeups on this model (code named “Sara”) for a later date.

Outstanding work, chipmasque! I’m sure I remember seeing mention of this somewhere before, and I’m very glad to see it continuing to be developed! Your animation videos look promising.

I have a question, do you think this approach could work for a very muscular character, like a Hulk? Or would something so extreme present too many issues compared to more traditional character rigs?

Thanks for the kind words, James. I am certainly aiming at being able to do highly muscular figures using this approach. However, I’ve found in some tests that a relatively dense mesh – needed to get the kind of definition to a muscular form that the Hulk represents – isn’t optimal, because the vertices tend to shift from point to point during animation, making the skin “shimmer” rather oddly. A nice “special effect” maybe, but not what I’m looking for. However, my tests also show that optimizing the mesh topology for shrinkwrap response is a possible solution to that issue. My current meshes have good general topology but are not ideal for a highly muscular look. I also anticipate that normal maps can help in this kind of instance, providing a measure of definition beyond what the system can provide, while the system provides for dynamic muscle mass response.

Wow… Excellent work. Very technical and neat.
The kind of thing I would love to have done for my 3D guy… But I’m not an expert in rigging and those kind of things…

But your results for now are really outstanding.

Curious to see an anmiation with a male character.
And yes, Topology is one of the key, I think.

Keep on the good work, Chipmasque. :slight_smile:

Alex

Thank you, Albron! I will be producing a male model as well once the “bugs” are sorted out, and also doing more animations, though these will likely be just OpenGL pre-viz renderings since Cycles takes so very long to render on my comp. I have to score a much newer machine to consider any fully-rendered sequences.

This is my latest work on the “Sara” facial modeling, refining skin, hair & eyeball shaders. One of my goals for the “Universal Figures” aspect of this project is being able to produce numerous looks from the same base mesh, so I make a “new” character for each stage, and “Sara” is the latest.


Very nice. Actually my brother and I are thinking about a similar system that also should generate the Muscle-Systems based on a skeleton.
Im very glad to see someone actually startet what we had as crasy Idea for years. It would be very nice if you could put up some tutorials and maybe even files about your MuscleSystemEvolution. A solid 10 of 10 stars.


A series of test poses to fine-tune deformation at various flex extremes. Not much more to do on the mechanics of the system, so the next step is to adapt the progress so far to a male model, which will also require refinements of the underlying skeletal and musculature models.

So far this experiment has been largely successful but to fully capitalize on its potential I’ll have to radically rework the figure topology, a daunting task to say the least.

@@chipmasque Awesome, I’m working on a similar personal project, I haven’t seen this project until now, truly brilliant work mate,

Cheers,

Jim Morren

Re-opened at OP request.

Many thanks, Fweeb.

Direct link to video, in case your browser blocks it: https://vimeo.com/173080625

The video pretty much says it all. For all intents and purposes the development phase is done; I’ll be posting some breakdown info for any interested in pursuing this approach as I adapt the system to my generic male figure. I’ll also be opening up the subject matter for the thread to cover more Universal Figure material such as skin shaders, ethnic variations, somatotype variations, etc.