If I knew more about rigging i would try to make something more "game friendly" but i only started rigging and animating a few weeks ago, I'm not real sure what FK is but ill definately reasearch it a bit. I wouldnt mind trying to put something together and sharing it with everyone so we have a base rig/model to base our main toons off of.(then people can alter it to suit their needs). Good topic!

Also, do you do rigging in blender at all? i noticed in the original post you said "if someone wants to make it", you seem to be pretty knowledable on the topic, have you made something to submit so we kinda get what your looking for?

The rigs I use for creatures/heros in OSARE are amateurish.  No weights, just vertex groups.  I'm not even sure how to set up kinematics (I barely know what kinematics means).

One very nice thing for a "unified blender game rig" would be easily sharing poses and animation.  I would probably convert my models to all use a standard rig if one was available.

I dont see why "helper bones" are a problem, for example the 2 knee bones i have placed in front of the armature are just parented to the foot bone, so if you rotate the foot the knee rotation follows it, just makes it a bit more realistic. wasnt a pain to set them up, it was as easy as extruding a bone from the knee joint, pulling it out a bit and parenting it to the foot. Other than all the hand ik's and rotation setting the rest of the rig i submitted is basic stuff. If you want to simplify the hand rig just delete the hand all together make 1 single bone for the hand and add some fingers :)

Maybe you'll like to make something based on the makehuman.org effords:

http://www.youtube.com/watch?v=zeLM7EZTcqY&feature=related

If someone wants to learn more about rigging (not just the "dumb" follow-me tutorials, but also the explanation behind it) you can take a look at the Mancandy FAQ DVD by the Blender foundation. It's all about rigging and also a bit about animating. http://www.blender3d.org/e-shop/product_info.php?products_id=99

Having said that, you don't have to buy the DVD, as it's released under some form of CC, so you can legally torrent it or similar. If you do buy the DVD, the earnings go to support future Blender cause. All I can say is a proper rig can make animating things so much easier, I really recommend going through the videos (especially since they are also released under CC).

For the more specific things:

FK = Forward kinematics. Here the control goes from the highest bone in the hierarchy to the lowest bone. You rotate the tigh bone, calf bone follows, foot bone follow, toe bones follow. You rotate the calf bone, foot bone follow, toe bone follow, but the tight bone is unaffected. It's like posing these lego technic figures http://www.1000steine.com/brickset/images/8712-1.jpg

IK = Inverse kinematics. Here the whole chain of bones is controled by the end (target) bone. For example, you position the foot bone and the whole chain of bones in the leg will automaticly adjust to follow the target bone as best at it can. Instead of manually rotating each bone in the leg, you simply position the foot. And by controlling the foot bone, it stays firmly on the ground right where you want it.

Both methods have their uses. IK allows you to control limbs with very few control bones, while FK is superior when it comes to arcs (like swinging arms in a walkcycle).

Regards,

Lamoot

@ adrix: Hes got like 6 controller bones on a foot with 2 deform bones on it, lol. Not to mention hes using Bone shapes to simplifiy that clutter that is a foot rig.

Not an expert either, but from these free rigs (for Blender 2.5):

http://blenderartists.org/forum/showthread.php?t=187051

For sure a few nice tricks could be taken. I really like for example the idea used in the spider rig of having an IK controller that can be rotated and moved as usually, but when you scale it it rotates the knee (thus you don't have to swich selection when animating).

Otherwise, well here is an old 2.49 rig I build a long time ago:

http://opengameart.org/content/blender-humanoid-rig

I guess it's not really good, but it's down to the basics, and the proportions are pretty good too I think.

I could post up one of my generic rigs... Athough it has no IK.

http://www.blendercookie.com/2010/07/30/using-the-human-meta-rig/

Could be usefull ;)

Ok I made a first base. Still pretty basic, but usable:

http://opengameart.org/content/blender25-humanoid-rig

@adrix89 what do you think we have to do to seperate the mesh to allow for better animation blending?

Bump... check the discussion in the linked rig from me too ;)

Here is a nice way to rig a mouth that would also lend itself well to a game model (few bones), so we could also make a optional face rig:

http://blenderartists.org/forum/showthread.php?t=201712

Still planning to improve this some time but my PC with an actual GPU broke, thus working in Blender is extremly slow on this machine :(

This is awesome, thanks for the link! Would be great if it worked nearly well for midpoly and lowpoly models too. I will try that out soon!

I might try to put this (human) rig onto pfunked's human base model, and add a mouth to it, will be interesting!

Actually rigging seems to work surprisingly well on my slow PC (well at 10fps or so)... tonight I strongly improved the foot rig (according to the tutorial above, that with the action target, not the python script) and once I have cleaned up the rig a bit I will publish a version 2.0. Stay tuned ;)

Updated the rig with a better foot:

http://opengameart.org/content/blender25-humanoid-rig

I left out the widgets and the foot tumble control, since that seemed unnecessary and also slowed down my pc too much ;)

Should be now much more walk cycle friendly.

Na I wouldn't call this even close to done... this isn't supposed to be a quick rig, but a all bells and whisles first choice for humanoid game models ;)

Concerning the palm bones... well they offer a nice way to strech the hands and if you just don't skin them they are simple helper bones that never show up in the exported model. But I am planning to make that more clear by putting all the helper bones on a seperate layer. Maybe some widgets would be nice after all, so that one can hide all the helper bones...

Concerning the shoulders... well make a suggestion. I am really unsure how to set them up. Right now there is this "upper shoulder bone", which I think prevents the shoulder from deforming strangely. But I am not sure if that is the best solution.

The pelvis bone is there for better "hip-swing" animation, which I think is a pretty usefull feature, especially for female models ;) Could you elaborate in more detail what you mean with your comment regarding the disconnected pelvis bone? I currently don't see what the problem is there.

Regarding the foot tumble... well maybe I will still include this... but it's a really rarely used motion that can be done FK also. Besides... modern engines have a foot snapping system that does align the foot with the ground automatically thus doing most of the work where a tumble might be used.

What is further needed to rig the hand? It is already working fine... just some wigets and maybe some "action constraits" for typical movements would be nice... a FK switch would be indeed nice... I have seen that in a nother rig... but how do you actually make one? Got a tutorial on that?

Other remarks:

I am really unsure if a seperated copy rotation rig is really needed... the way the rig is set up, the main hiracy is exactly what you would want for skinning. So unless we significantly change that I don't really see the need for that (and it might just confuse the users). I think one of the main reasons such a setup is used in professional environments is that thus the modlers and skinners can already work on their parts while the riggers and animators might still be refining things on the main rig. Thus this way no time is lost waiting for the final version of the rig. In this case however the rig will be final before used as an asset, so I don't really see the need. But I would be happy if someone else would give some additional ideas why it might be worthwhile to make a copy-rot rig.

I am also thinking about creating a custom bone psydo mesh, like the did with the tarsal bone in the tutorial... there are several very nice rigs doing that (Alex's horse rig on this website for example), but I fear my current pc might be to slow to work with that than... and other that looking nice, I can't see the actual benefit right now. But maybe there is some other reason why it would be nice to have something like that?

But once this rig has reached a "final" state we really need to start making a animation preset. Is anyone around who has some previous animation experience? Mine is very limited to be honest.

Ok updated the rig again with a hand rig according to the tutorial you posted. It's still a bit uncomfortable to use without the widgets, but I will add those in the end.

Otherwise... I removed the lowest spine bone (seemed sufficient to have the pelvis), added a central "body" bone (not sure if needed) to move the entire rig, and added some "shoulder blade" bones... I hope that is what you ment with your shoulder remark.

I think the hand has a target... the bigger bone around the hand bone... maybe I should rename that.

You might have convinced me regarding the copy-rot. rig... (but again will add this in the end). However I am a bit confused about the GPU skinning. I am well aware that (especially on older hardware) you have only a limited number of bones that are "for free/hardware accelerated", but does that really concern us much in this case? AFAIK the exporter from Blender just exports the skinned bones, so if we have helpers as parents or childs those would never show up in the binary file and thus would also never be rendered in the GPU right? So as long as the number of skinned bones is low enough, how should the rest of the rig be of any concern?

Any more ideas or improvments?

http://blenderartists.org/forum/showthread.php?t=202260&p=1737552#post17...

Maybe something comes out of that too ;)

Ok, then we definitly need a copy-rot. rig.

Concerning the soulder... well I will think about it. What I like about the current setup is that it might be very usefull to conteract shoulder deformation if skinned to the mesh. But I agree it is a strange setup... however it is the one implemented in Blender's meta-rig.

Alternatives would be a sort of targetless IK setup like explained in the Mancandy FAQ videos... but that is not really functional IK and thus can not be used to animate holding a weapon well. Of course we could also set it up as the legs, but even though that sounds somewhat logical, I have never seen a rig doing it like that (for what ever reason).

Any hints on setting up the constrains you mentioned? As you can tell this is my first experience with an advance rig, before a simple IK chain was already rocket science for my rigs ;)

Ok I found out what the reason for that strange shoulder bone is:

http://wiki.blender.org/index.php/Doc:Tutorials/Animation/Armatures/BSoD...

Given that I think will switch to an elbow target design.

http://opengameart.org/content/blender25-humanoid-rig

Updated it again.

Changed to pole target IK both the fingers (simplified but same functionality) and the arms (still a bit strange in movement, not exactly sure why).

Also tried to have a shoulder that follows the arm a bit... tried to find any good infomation on how to set up a shoulder, but either there was simply a FK setup or a hugly complex 50 bones containing super rig... seemed impossible to find a nice but simple setup. This way it somewhat works, but I really don't like that the IK snapping between edit and pose mode is that big.

Also made it possible to move the hand a bit by FK, by lowering the copy-rot. contraint to .8 influence.

There is this set of scripts being developed to massively simplify rigging in Blender and it plans to include all of these more advanced approaches to rigging legs, arms and fingers etc. It's being developed by the main rigger for Big Buck Bunny and Sintel open movies. Here's a thread at Blenderartists . And some interesting youtube WIP videos:

http://www.youtube.com/watch?v=6lgwJWQ9_JE

http://www.youtube.com/watch?v=eUSyoO2BmaQ

I don't want to undermine your effort, but I'm wondering how usable this project will be with rigify in development.

@adrix89: I really want to have an automatic setup that moves the shoulder according to the arm movement... that really adds a lot to character animation to have some realistic shoulder movement. First I tried it with a follow target constrain, but the IK does the job much better. But I am open for other suggestions.

@Lamoot: I am aware of that, but unless they make it really fexible so that all sort of strange multipedal rigs can be made I don't really see the point. Once you have a nice rig for a regular human it is very easy and quick to adjust it to the proportions of the individual models.

In any case making this rig is really just a mean to an end, what I really want to have/create is a "plug-in" human animations set for all sorts of games (many FPS for now though). That is what is really lacking in open-source game development (and one of the main reasons people stick to modding games like Half-Life, where such a thing exists). That riggify system is really of no need or use regarding that.

Can't say how flexible or unflexible the rigify system will be. In one video he showed how it's possible to create a biped with 2 sets of arms. Not sure about the legs, especially those alien knees that bend forward. I'm keeping my hopes high for it though.

So the real goal is to provide a flexible and robust rig and also animations? In that case it does make much more sense to have an actual rig compared to the rigify approach.

@Lamoot: Yes, animations. But who knows when this will actually happen. I am still hopeing that a skilled animator shows up and just quickly does it... since I have never done walkcycles etc before... so it will be a slow progress (but hey great learning experience :-/ ).

@adrix89: You think so? If I move my arm around the shoulder always moves with it a bit. Especially when lifting the arm above the shoulder joint all the movement comes from the shoulderblade. I tried to limit the movement if arm movement is below the joint however... but somehow the IK and Target contraints override the limit rotation ones :( But with the low influence (.2) it has right now it seems to be ok... maybe the shoulder movement is a bit exagertated now... but real actors also do that to give their performance more life ;)

Concering the FK-IK switch... it seems to me that that is a lot of effort to set up (additional set of FK bones etc) while giving very little benefit (in fact all the movement can be done with IK, it might just be a bit more comfortable to do certain movements with FK). But maybe I will lower the influence of a few IK controlers a bit, like I did with the hand to allow for some fine tuning with FK. But feel free to correct me on that stance... I might be overlooking something where FK is actually needed.

Hmm... it's not like the rig doesn't have FK... it's just that I don't see a really good reason why it should have complete FK control via a switch. But you are right that the copy-rot. rig we are planning to do anyways could be used for that (unless the copy-rot. contrain interferes). But I will have a look at the tuts for sure.

Oh and I made a small update to the rig... limit rot. is working after all on the shoulder bone, I was just stupid and tested it with 100% IK weight. Now it limits the movement of the shoulder downwards a bit and also minimized the edit/pose mode snapping.

I think the current shoulder setup is pretty neat... but of course I know little about animating. It just seems too simple of a solution... and I wonder why it was absolutly impossible to find a good reference rig with an automatic shoulder. Maybe I am just a genius :p

He? I can't aprehend what you just wrote... With the current setup the shoulder does not move much unless the arm moves above the head. But small subtile movement can make a big difference.

I don't see the need for a rot-limt of the upper arm at all, could you explain further?

The last bit doesn't make sense to me at all :-/

I like the IK and the shoulder as it is ;)

But you are right the elbow target was messed up... I think by parenting it to the shoulder instead I pretty much fixed that. Have a look at v4.2 I just uploaded.

I also experimented with a IK controller for the spine in that version. I think it might be quite handy to position the head, and since it is targetless you can still FK controll the individual bones.

The main question I have now is though... will it work for animation too? I read a lot of times that certain IK setups work fine for posing a model but break apart when actually animating it. Does anyone have further insight into that?

http://imgur.com/lBsjN.jpg

Looks pretty good to me. Maybe I will add a small farward and backward rotation constrain to the shoulder so that is doesn't move quite as much forward... but it is really only a little too much I think.

About the spine IK, well you might be right, it does act a bit flakey. It is really nice that the torso follows the neck this way however, turned out to be useful in the pose above (which took only a few seconds to make).

Maybe I am a bit stupid right now, but I don't get your point. Currently the shoulders are in that position still as wide as the hips (maybe the head is a bit oversized and thus give the impression of having a very think neck), but I agree that they might be better a bit further apart, but really only a tiny bit.

If I lift my arms to such a position in real life, the shoulders also move inwards as they move upwards to level the gun with the eyes.

Last but not least... yes such extreme yoga movements might be a bit difficult to achive with the current rig (for that FK is needed), but seriously... who is going to make a yoga game ;)

but seriously... who is going to make a yoga game ;)

Nintendo? :)

I think this project is at the point where having revision control is going to help it by allowing a easier merging and collabaration.

Some place like http://gitorious.org or http://gitorious.org will work fine. Then do periodic releases here and elsewhere.

I dont know much about rigs, but currently one thing that is way off is hand movement.

There is this bone hand_target.R and when i move it everything before it follows (arm) but hand does not.

For example if i move that bone in front of character hand will still point to side, also there is no bone to control whole hand (all fingers).