Citywatch rig

[squill]

Hrm. I compared the joints on the new and old skeleton and found another discrepancy.

 

The new skeleton has a new joint: "RightToe_end". It is also missing a joint that was on the old skeleton: "RightForeArmRoll". So the total joints are the same, but the names are not. That's probably causing the export errors.

 

edit: As I suspected, I'm now getting a "joint names don't match" error.

 

jep i see what you mean. The current imported md5 skeleton seems pretty much f#$% up compared to ascottk's original skeleton. I would rather not remove RightToe_end because it's definitely useful for animating the feet although the ForeArmRoll could be left out but i see your not going to get this working in-game without re-exporting the animations right?

[Springheel]

I would rather not remove RightToe_end because it's definitely useful for animating the feet although the ForeArmRoll could be removed but i see your not going to get this working in-game without re-exporting the animations right?

 

 

No, the skeleton won't work with any of the current animations unless it has exactly the same joint names.

 

I'm not sure why there was a LeftToe_end but not a right one, but if it makes you feel any better the Left one wasn't weighted to anything anyway. It had about 15% influence over a single vert--just enough to keep it from being culled.

 

The current imported md5mesh seems pretty much f#$% up compared to ascottk's original skeleton.

 

It does seem odd that it has asymmetrical joints. Are you saying it didn't used to be like that? No one has worked on it since Ascottk as far as I know.

[squill]

It does seem odd that it has asymmetrical joints. Are you saying it didn't used to be like that? No one has worked on it since Ascottk as far as I know.

 

yeah the original base skeleton was symmetrical having a total of 75 joints instead of 67.

I think somehow those joints were left out using the md5 importer and no one have noticed this, shit happens

 

although that skeleton was mostly build so it could be used for FBIK in Maya and MotionBuilder, it also uses a lot of unnecessary joints. Besides this, have you also already tried to import the fbx file?

 

I think this also shows we do need to replace the skeleton soon or later because we don't want the inconsistency in joints.

[Springheel]

yeah the original base skeleton was symmetrical having a total of 75 joints instead of 67.

I think somehow those joints were left out using the md5 importer and no one have noticed this, shit happens

 

I suspect you're right. Wouldn't it be possible to re-export the original skeleton and animations with all the joints?

 

I think this also shows we do need to replace the skeleton soon or later because we don't want the inconsistency in joints.

 

There's no such thing as "sooner or later" at this point. We're supposed to release this mod in a year. It's already "later". There's no question that it would be nice to have a good, clean base skeleton. The question is whether or not it's worth redoing all kinds of work to redo the skeleton *at this point* in development, when we're already behind schedule.

 

I find it hard to accept the idea that there will suddenly be a glut of animators who appear from nowhere when the mod is released, ready to create new, wonderful animations, but unable to do so because the skeleton has a few extra joints.

 

The current skeleton may not be pretty, but it works. Your animations are testament to that.

 

Besides this, have you also already tried to import the fbx file?

 

Not yet. I'll have to update and try tonight.

[squill]

I suspect you're right. Wouldn't it be possible to re-export the original skeleton and animations with all the joints?

 

i guess if you take the original skeleton and replace it within your model file, I could re-export the animations. I'd have to check the def files to be certain all the joints will be exported.

 

I find it hard to accept the idea that there will suddenly be a glut of animators who appear from nowhere when the mod is released, ready to create new, wonderful animations, but unable to do so because the skeleton has a few extra joints.

 

the extra joints are not the problem, I've even disabled them(leg and arm Roll joints) in the animation rig from being able to rotate.

 

Like i said before the biggest issues are the twisted joints and their placing and alignment. This new skeleton your testing fixes a lot of the issues we have when animating the characters.

 

I don't see (apart from your mesh tweaks) the big issues of exporting and tweaking the animation again. I love animating (it's my job ) and most animations i did were done in a few hours or half a day.

 

The transfer of animations would take some days not weeks or months. I really do think it's a big advantage for the animators if we replace the skeleton. To be honest it motivates me a lot knowing we're testing this new skeleton and improving the rig

[Springheel]

I don't see (apart from your mesh tweaks) the big issues of exporting and tweaking the animation again. I love animating (it's my job laugh.gif) and most animations i did were done in a few hours or half a day.

 

I can only go by the progress I've seen thus far, but for the past six months, there have been less than half a dozen new animations (not counting placeholder ones) added. And we have another 40+ on our list.

 

Like i said before the biggest issues are the twisted joints and their placing and alignment. This new skeleton your testing fixes a lot of the issues we have when animating the characters.

 

You're far more qualified to argue for replacing the skeleton than I am, but the above leaves me wondering why, if that's true, replacing is necessary.

 

Tell me ultimately what you'd like to do, and we can work out how it can get done with the least amount of fuss. If we're not talking about adding/removing too many joints, it might still be reasonable from a mesh perspective. Maybe we should make a new "base reference" character, which you could rig and weight entirely in Maya, without having to worry about any LW version, and I could then take our character meshes and modify them to match.

 

And again, if we go this route we'll have to work in a separate branch until everything is fixed.

[Springheel]

Maybe we should make a new "base reference" character, which you could rig and weight entirely in Maya, without having to worry about any LW version, and I could then take our character meshes and modify them to match.

 

Just to expand on this idea further...

 

There's nothing that says the citywatch has to be the default base character for all our animations. When you think about it, the proguard is probably the character that will be used most often in traditional missions.

 

What if you took the Maya version of the proguard and turned it into your ideal animation rig? Make a new skeleton, weight everything the way you want it (you'd have to import the full-fingered hands but I could upload those as a separate mesh for you), and tweak and export the animations to match. Since none of your changes would affect the citywatch animations in either the trunk or the models_src folder, you wouldn't even need a separate SVN branch.

 

The benefit is that the new skeleton is a totally separate project. None of the current meshes will be broken during the process, regardless of how long it takes me to modify them. All the current meshes I've worked on can continue to use the citywatch skeleton and animations even after you've finished the new one, but once that's done I can slowly start to transfer the meshes over to the new skeleton, using the proguard as a weight reference.

 

The more different the new skeleton is from the current one, the longer it will take me to transfer meshes over, but if I've got an updated reference to follow it shouldn't be too bad. And all of the characters will continue to work in the meantime.

 

How does that sound to you? You and Domarius could even coordinate your efforts if you like.

[Domarius]

Hm so you're saying we should go ahead and make a new rig for the pro guard, and make all our new anims on him.

 

That's fine but I think I'm the only one who can convert the old anims to the new, very quickly, using layers, so I'd have to do that anyway, like I proposed earlier.

 

 

But as far as I can tell, there's nothing wrong with the current skeleton (the one in the game) in terms of what joints are available. Only the positions of the joints in relation to each other, and the subsequent weighting. It just needs tweaking in this manner and re-exported to the game, which I can do. I'll just continue with that for now till I know better about what you need.

[squill]

But as far as I can tell, there's nothing wrong with the current skeleton (the one in the game) in terms of what joints are available. Only the positions of the joints in relation to each other, and the subsequent weighting. It just needs tweaking in this manner and re-exported to the game, which I can do. I'll just continue with that for now till I know better about what you need.

 

 

if we are going to re-rig a base character we should start from scratch. Not only are the joints off position in relation to each other also the orientation and way you rotate them are far too twisted to fix manually. I strongly suggest to replace the skeleton with a new one like i build last week. The extra Rolll joints in the legs are unnecessary and could be removed.

 

That's fine but I think I'm the only one who can convert the old anims to the new, very quickly, using layers, so I'd have to do that anyway, like I proposed earlier.

 

i don't know exactly what you mean with using layers to quickly convert the animations because i haven't checked the layer plugin yet. But with a character set which you've used already it's pretty easy to transfer the animations done with the last animation rig and edit the keyframes.

 

the way i convert the older animations done in MB or with the fbik rig would be to bake the keys. Although some of the smaller placeholder animations could be redone by us.

 

I'm first going to finish the sheathe animation for the current release. I'll also upload the animation rig with IK hands.

[Domarius]

are you using the current rig to fix these bendings? The new skeleton Springheel and i are testing fixes the bending problems on the fingers, arms and legs because i've set the positions and rotations of the joints straight in the new skeleton (you can import the skeleton.fbx into max/maya to see what we're testing).

I'm posting these screenshots to confirm we are looking at the same version

 

The knee is better but it still collapses at only 90 degrees

 

Same with the elbow

 

The "3 fingers" move a lot better, but the rest of the hand still has some incorrectly weighted verts, and the pointer finger bone is still not aligned with the mesh, making it collapse when it curls.

[Domarius]

if we are going to re-rig a base character we should start from scratch. Not only are the joints off position in relation to each other also the orientation and way you rotate them are far too twisted to fix manually. I strongly suggest to replace the skeleton with a new one like i build last week. The extra Rolll joints in the legs are unnecessary and could be removed.

No worries - you should just build a new one then, cause I don't know what the requirements are - I know we need special joints for the head attachment, sheathing etc. but not how to make sure that all works.

 

But outside of that I'm eager to take on any other work. You can do a quick job rigging it, and I can go over it and spend the time fine tuning everything, that could save you time.

 

 

i don't know exactly what you mean with using layers to quickly convert the animations because i haven't checked the layer plugin yet. But with a character set which you've used already it's pretty easy to transfer the animations done with the last animation rig and edit the keyframes.

Well you'll just have to trust me - it's easier with layers. You don't touch any existing keyframes. You just add a new layer with ONE keyframe at the very start, for each joint you want to offset. And you offset them in that layer, and the whole animation is fixed. Done. I do it all the time for my old anims, whenever people said "He's not leaning over far enough" or whatever - did I go through and tweak each keyframe for the back throughout the entire animation? No, I always made a new layer called "pose tweaks" and any changes to the overall pose for the entire animation was done there, on the first keyframe.

[squill]

Why did you guys make a new rig? I probably missed something, but just so you know - the old one is fine, as far as the heirachy goes. The only problem is the positions of the joints relative to each other, and the subsequent weighting. I can fix all of these problems.

 

the new skeleton was a test to see if it was possible to replace the old one without too much trouble which it didn't turned out to be.

 

i think your missing the importance of joint orientation. Translations on joints are not bad because it's easily to adjust without effecting animation. Rotation however causes a lot of problems if they are twisted and bend in weird angles. This is the case with our current skeleton.

[Domarius]

Cool - can you show me examples though? I've found the actual structure of the skeleton to be fine for animating in the past.

 

We can now weigh up the twistedness vs the work getting a new skeleton in the game, cause at the moment I could happily resurrect the old skeleton and just fix the positions and weighting.

 

I'm going to be on standby on this forum for the rest of the day, just doing other stuff, untill we sort this out. Maybe you'd like to chat on MSN to speed things up?

 

domarius@hotmail.com

[squill]

No worries - you should just build a new one then, cause I don't know what the requirements are - I know we need special joints for the head attachment, sheathing etc. but not how to make sure that all works.

 

But outside of that I'm eager to take on any other work. You can do a quick job rigging it, and I can go over it and spend the time fine tuning everything, that could save you time.

 

i'm just mentioning this all because i though you were going to adjust the old skeleton. But i'm glad we can do this together and improve our methods.

 

Well you'll just have to trust me - it's easier with layers. You don't touch any existing keyframes. You just add a new layer with ONE keyframe at the very start, for each joint you want to offset. And you offset them in that layer, and the whole animation is fixed. Done. I do it all the time for my old anims, whenever people said "He's not leaning over far enough" or whatever - did I go through and tweak each keyframe for the back throughout the entire animation? No, I always made a new layer called "pose tweaks" and any changes to the overall pose for the entire animation was done there, on the first keyframe.

 

i trust you I'm just not used to layers because i've never used that kind of tool in Maya. The way i would do this would be to select the curves and keyframes in the grap editor controlling the leaning, and move them up or down to in- or decrease the bend. Works fine and i know exactly what's going on. So i can only say also learn to use the graph editor so you'll have even more control over your animations.

[squill]

I'm going to be on standby on this forum for the rest of the day, just doing other stuff, untill we sort this out. Maybe you'd like to chat on MSN to speed things up?

 

domarius@hotmail.com

 

i'll just having to do some of my daily things before i will continue to work on this. but sounds like a plan

[Domarius]

It's not a way of ignoring the graph editor - it's another tool altogther that makes huge sweeping changes very very simple. I guess you have to really try it out to appreciate what I mean by my examples. I don't think oDDity ever did - just got some (outdated) 2nd hand opinions and never actually used it himself.

 

An akward analogy I just came up with, is basically using one thing to do many things can be more complicated. Sticking with one timeline for everything could be compared to trying to design a helecopter with a rotor that moves it in all directions at once. Layers can be compared to adding a rotor on the tail, which is another layer of movement. Now one can move the helecopter independantly of the other, and get a more complex result with a simpler interaction.

[Domarius]

Okay I'll be refreshing the forums intermittently

[Springheel]

Ok, so what is the plan from here?

[squill]

Ok, so what is the plan from here?

 

Domarius and I are creating a new skeleton, still based on the current skeleton and last new skeleton. We'll use the proguard as our base character mesh. Also because the mesh is already better centered around the current skeleton.

 

For the next release i think it's smart to stay with the current skeleton, because having to rush things will not make our work any better. Would be nice if we did get it to work in time but i can't guarantee that.

 

I'll finish the animations needed. So in the mean time we're working on getting the skeleton up and running, converting the animations and tweaking them.

[Springheel]

For the next release i think it's smart to stay with the current skeleton, because having to rush things will not make our work any better.

 

Oh yes, I certainly agree there.

 

Domarius and I are creating a new skeleton, still based on the current skeleton and last new skeleton. We'll use the proguard as our base character mesh. Also because the mesh is already better centered around the current skeleton.

 

Ok, sounds good. My only requests are that you keep the new skeleton as close to the current one as you can to minimize changes I'll have to make to the mesh--use the same names for joints where possible, etc. (also don't change the uv-map of the proguard, though there shouldn't be any need to do that). Do you need the full hand meshes from me?

[squill]

Ok, sounds good. My only requests are that you keep the new skeleton as close to the current one as you can to minimize changes I'll have to make to the mesh--use the same names for joints where possible, etc. (also don't change the uv-map of the proguard, though there shouldn't be any need to do that). Do you need the full hand meshes from me?

 

we try to keep it close, although it's now better fitting the mesh. I did remove some extra joints in the legs and positioned them to fit the mesh. Do you also know what more is needed to make this character work like the af_pose etc. ?

 

Domarius will do the weighting in Maya and i'll setup the animation rig. I've taken the hands from one of your new characters which Domarius is modifying.

 

It's save to test this with the proguard right? otherwise we could always revert, but this character would be the base character to test on.

[Domarius]

It's important that the models be adjusted to best suit the skeleton, and not to make the animations look right. The animations can be easily tweaked using animation layers, but a poorly weighted mesh fundamentally flaws any animation produced with it.

[Springheel]

It's save to test this with the proguard right? otherwise we could always revert, but this character would be the base character to test on.

 

Yes. As far as I know no maps are using the proguard at moment.

[Springheel]

BTW, Domarius, if you enjoy rigging the meshes, I have plenty of character models that need to be rigged before they can be added to the game. I can do it, but it's not something I'm very good at, nor do I enjoy it much.

[Springheel]

@Domarius: I see you're going to weight the shadowmesh as well--if you could remove the head part that would save me having to do it later.