To find out how Ipo driven shape keys can be used to create a muscular simulation for a model, I have been working on this female model, which I have been doing for a long time. Currently I am working in the shoulders, which seem to be one of the most difficult areas.
This is an animated wireframe. These low frame rate movies are a great resource to quickly find errors (in fact I see many in this one).
(edit: SORRY! removed, GIF file was too big and used much bandwidth.)
Here are some quick animated movies with textures. This is the first time I use Mencoder (thanks to this thread) I hope you will have no problem playing them. Mplayer should of course work:
My prevoius model didn’t have good edgeloops for the shoulder and back areas, causing a lot of problems when creating the deformation keys, so I discarded my old shape keys and started looking at muscle diagrams, photographs and actual people, and after a lot of reshaping, these are my results.
There are almost no vertical or horizontal edgeloops. This new configuration follows the flow which the older model already had for the rib area.
Here are some color coded images to show how the topology relates to the inner structures:
Unfortunately I have also found the limitations and problems of this approach. It doesn’t seem to be usable for animation, because rotations of bones on various axes cause wrong values for the driven keys (I am not sure if this is a bug). This system can work for simple posing if you manually adjust the key values.
I will put some images and animations looking from the front as soon as I correct some errors.
Does it look natural or exaggerated, is she too muscular, what do you think?
I think its a great animation and model actually.
she does’nt seem too muscular but like a slim women who takes care of her body. although her ribs may show just a tad too much but I am no expert on the human body so I will just stop talking.
Tolobán, this is a great model you have here! Looking at my model now, makes me wonder if I should change the flow of her back and the ribs to…because this one looks so good!
It doesn’t look exagurated to me. It looks rather natural. Only the ribs on the fornt are a little to hard.
Can you show us the problems you are having? I’m just wondering… would the approach you are following be the best one to create the most natural human model? I was planning to use the technique you are showing on your website (bones + shape keys to correct the deformations). If you discover weird things and possible bugs, it’s maybe a good idea to post them in the bug tracker, so the next version could have a fix?
Wow, that is really really impressive(is that blender ). How is the deformation at a 45 degree angle or an angle that uses multiple shape drivers simultaneously? I would be very curious to know how you pull that off.
Very nice poly flow. I may use some thing similar in a future model, but I want to learn the muscles in depth before than. How exactly does this script work?
i have also played with the ipo driven shape keys for deformation, its is great so far, it can really help out a person who didn’t do a greatjob modeling in the first place. it also beats this method http://www.sonofpat.com/shoulders.html
Thanks to you all for your replies! It is very useful for me to elaborate to answer your comments.
env: I need to displace some vertices a bit more.
Warder and Wim: Good point about the ribs, it is easy to exaggerate a bit.
(Kaeru:) Yes, I am using a standard Blender armature, combined with driven shape keys (tutorial), plus a setup of constraints.
Now I want to show you how my shoulder bone setup plays a big role. The upper arm bone doesn’t pivot in a static point. I am not sure if this is anatomically correct, but in a very empiric, direct observation, looks good. Anybody can observe that the shoulder has a wide motion when the arm is moved, so I added a few constraints to try to mimic this. This little movies show what I mean.
The collar bone (going from the neck to the shoulder) has two “Track to” constraints: one goes to the forearm bone and the other to the floating bone (child of the chest bone) named shldup (it keeps the shoulder up or straight). The beauty of it is that the same setup works for all movements. It is very difficult though to find the correct values for the constraints.
The degree of deformation depends on the value the driver curve has for a certain bone rotation value, so to get and mix the degree of deformations you want, it is all in editing the curves. It is actually easier than it sounds (but read below for other problems).
Script? I don’t think there is a script mentioned. For this you don’t really need scripts.
Learning the muscles will of course give you a good edge, but don’t obsess on a purely anatomical approach. You have to be open to make big compromises for good mesh modeling.
I think the flow following the ribs does work better than a cylindrical mesh using horizontal rings, I owe some inspiration to the new Makehuman mesh.
About the upper back, the interesting contradiction is that the muscles of the back seem to have this strange flow, but the scapula bones show their shape from underneath them in a way that doesn’t go with this flow. Especially they make these two clear vertical lines which would suggest long vertical edgeloops, but you don’t figure out these lines in the muscle diagrams.
Here are a couple pictures which would suggest different approaches. I like the one with the two women as a base to my model. I like especially how you can figure out the muscles in the lady on the right.
A few problems arised on which I was not counting. For one thing, the X, Y and Z rotations don’t work as nicely and predictably as I assumed. All goes generally well if I rotate on the X or Z axes, or even a comination of both, but if I twist a bone (Y axis) the shape keys go crazy. If I query the current value of a driver by a little Python script, the values which I would expect for the X and Z drivers are not what they should be, and thus the deformation is not good.
I guess the only way to find out whether there is a bug is to go to the tracker. But that would not solve everything. I do hope the new Python drivers can help solve these problems and let users create a great rigging sytem.
Thanks… A book? I might consider it, when I learn enough…
I remember that tutorial was mentioned in this forums, but I think it doesn’t even work for Blender because the “Copy Rotation” constraints twist the bone at some point, just like “Track To”, which make both types very limited in usefulness.
Those animations are sooo good. You set a high bar to match. There’s such a great sense of motion of the shoulder blade under the skin. The deformation is very natural.
A very nice model well animated. I look forward to more.
can u sent the .blend file so that i can practice on it?
I want to do a very realistic model with every details.
I am an artist. and want to share serious works for a better project.
This model is superb. The topology on the back is perfect and the animations look excellent. You really should consider writing a detailed tutorial to help the little people -or as nyrathwiz suggested, a book.
Nice muscles deformation I tend to agree with the ribcage area though. For a female to be that fit around torso + arms area (with muscle tones) her ribcage looks a bit anorexic. Balance it out a bit maybe? (less flesh on arms or more flesh on ribcage)
Nonetheless I love the experiments with muscles. Keep it up!
thats awsom, very well done, only crit i have is that the perhaps it deforms a bit to much around the shoulder blade, i know its muscle simulation and not bone but it seems like the shoulder blade bens rith the muscle instead of actully moving. but all that said still incredibley well done.
peace
I am doing something similar on a Horse model I am working on, but I have found that I can not use just shape keys. I am useing a combination of shape keys and stretch bones. You will be able to get proper animation results if you use more than one bone to trigger the shape keys for the shoulder. Have one bone copy the rotation on the x axis, one on the z and one on the y axis, binding them to they’re respective keys.
This way when you move the shoulder forward and back you will be able to get the same animation as when you rotate her arm at the shoulder and move her arm forward and back.
I hope this works for you as I have not tested it on a human model yet
Good work. I think you are doing a good exploration. I would not be able to do the same.
There are still many things to tune well before you obtain a really good musclar/body animation.
As you know I am not an expert of 3D animation but I well know body from my drawing skills.
The biggest “errors” for your setup and animation can be seen particularly when you look at the rise-up movement of the arm.
The armpit should become more concave when arm rises. Becouse Pectorals, Latissimus Dorsi and Terse minor and major muscles relax to help the Deltoid to rise the arm.
Deltoid heavly contracts its self, while Trapezius stress help Deltoid and to hold and rotate scapula.
Realaxing Pectorals affects also breast that lengthens.
Also in the back to front movement of the arm there are some more things to take in consideration.
The arm is rised, so beyond the other muscles involved we should see a contraction of the Deltoid and the Trapezius.
Also we should see scapulas rotate a little around vertical axis, so they become a little more distant, when arm move forward, and a little more nearest, looking at the back of the character.
). How is the deformation at a 45 degree angle or an angle that uses multiple shape drivers simultaneously? I would be very curious to know how you pull that off.
