-= Gametoast.com =-

Well, here we go. I've made a relatively simple walker mesh out of ATTE kitbashed parts, and now I'm trying to animate it. From what little I understand of it, you have to animate the bones (FK, is it?) instead of the effectors (IK?). The idle animation was easy, everything staying as it was from start to end, but now I'm trying to do idle_to_leftfoot. For a start, I'm not quite sure what that is - is it from the idle state to left foot planted in front, right in back, essentially a half-step? Or is it a full step, with both legs ending up in their starting positions?

Aside from that, I have a skeleton for the left leg with three bones. The top controls what might be considered the "thigh", the middle controls the "shin" and the lowest controls the foot. Between those, the ankle and knee are controlled by effectors (which move with the bones, but I'm animating the bones, not the effectors - is that a problem?).

I can rotate and set keyframes for the upper and middle bones fine, but for some reason, I can't rotate or set a keyframe for the lowest one. Whenever I press K or the set keyframe button, nothing happens (with the rotation transform selected for). Whenever I rotate it, it snaps back to its original stance. Why is it doing that?

You can use either IK or FK for the animation, IK is easier. You can actually export IK animations as well although some school of thought is that you should "plot" the animations first which converts the IK to FK. Which is good practice I guess but if you just want to do a quick check then export IK works just fine.

For the snap back problem it is because you are probably using both IK and FK and the blend mode is set to IK, for the bones you want to rotate FK wise you need to set the IK/FK blend to 0 for that bone chain. Select the top bone in the chain and in XSI explorer press enter, select the Kinematic Chain tab and move the IK/FK Blend slider to 0

For start positions for the different animations, if you have Ultimate Unwrap3D (purchased or demo) the skeleton position shown for the different mshs in the animation folders is the start position for that animation.

Did I make you go like this yet?

Also if you haven't yet have a look at some of these VAST XSI Tutorials they are pretty rudimentary in some cases but they do show pictures which is more descriptive than my typing.

Hmm - how could I go about getting it to use the stock AT-ST animations? That seems like it would cut out a lot of the work, considering how similar it is to the AT-ST.

FragMe! wrote:You can use either IK or FK for the animation, IK is easier. You can actually export IK animations as well although some school of thought is that you should "plot" the animations first which converts the IK to FK. Which is good practice I guess but if you just want to do a quick check then export IK works just fine.

IK is not supported by the exporter and will be discarded, this doesn't mean you can't animate in XSI with IK but the animation must be on the bones which = FK, animation on effectors is no good. If you grab the effector of a bone chain and move it then key the bone rotations that is still FK, if whatever you thought was IK and it exported properly then most likely it was FK or you had just enough FK to pick up the discarded IK. If you choose to key the effector instead of the bone rotations after moving the effector then you will need to plot the animation to the bones before exporting. Hope this helps.

Fiodis wrote:Hmm - how could I go about getting it to use the stock AT-ST animations? That seems like it would cut out a lot of the work, considering how similar it is to the AT-ST.

All you would need is the ATST skeleton and the process would be the same as if you were setting up a player model with Unit_Template_Beta. Have a look at the AT-ST in U3D to see what you might need for it to work properly.

Ace sorry to say but IK IS supported by the exporter I do it all the time. I have done walk sequences using strictly IK no FK at all. And if you want to be more detailed I am using constraint objects on the effectors to do the animation.
Pandemic may be doing the conversion to FK on the export I don't know because I am not plotting the FK prior to export or removing the constraints, really. And no the constaining objects are not included in the branch select for export.
If the debug command is in the bat, the resulting .zat file shows all (well the first 10) keyframes for all the proper bones.

Now I am not saying you shouldn't plot the animations I am saying if you are doing a lot of checking things, such as a walker, it is a lot easier to just export the IK rather than plotting, then reloading an unplotted version just to change something.

Well, FK doesn't seem to be much more trouble than IK, really, so I'll just animate the bones for now, to be on the safe side.

For the walker version that can use the AT-ST anims, I opened up the AT-ST's idle.msh animation file, and there was the skeleton layout, all straightforward and all. However, in the center of it where the "neck" and, um, "hips" would be, there are some primitive-looking things, as opposed to nulls.

FragMe! wrote:One thing that did happen when I was first building this up and testing it was the model was halfway through the ground, sitting right where the bone_root null was, but the pilot hp was still in the right place. Now what I did to fix this was change the bone_root from a null to a primitive (just a square in this case), moved it to position and froze transforms then dropped the skelton underneath that and re-eveloped (since doing that I have also found out about the set neutral postion (blast can't remember the exact command but it is under the transforms menu) which would probably do the same thing.
I don't know whether it was doing both things or just one of the two that fixed the problem and put it at the right location on the ground. (broke cardinal rule of trouble shooting changed two things instead of one) Also used the bone_root to transform the model slightly in the Z axis so that it would actually move forward and not stomp in place.
When I get time I was going to check that out again.

In the AT-ST's case (correct me if I'm wrong) the two primitives are bone_root and bone_pelvis (what I took for the neck and hips). Do they both have to be frozen? Or only one of the two? Or none, and their parents (DummyRoot and/or GroundDummy) being at 0,0,0 is enough?

You are on the right track with the skeleton, the geometry pieces are as you said bone_root and bone_pelvis. They can just be nulls you don't need the geometry it just looks cool. These as with any bone_ anything should not be frozen. grounddummy and dummyroot are at 0,0,0.

As for what you quoted I think I was just bashing around at that time using a msh to xsi imported skeleton, if you are building a skeleton from scratch don't freeze anything.

If you are looking at the skeleton in Unwrap don't be fooled by the way it shows the links between the bones. For example there isn't anything going on a diagonal from bone_pelvis to bone_l_thigh (or the right side either), the bone or rather root of the thigh starts at the end of the arrow.

You are trying to match your model to the skeleton correct? Not necessarily doing a whole new set of animations. Do you have a way to import the basepose skeleton into xsi? If you do and it is via Unwrap here is what you can get rid of, the unaasigned group, the geometry I think it will be called u3dscene that should leave you with just a set of green nulls. That is all you really need to envelope to.

Hint, if you branch select the bone_root null and then on the right Select Tab when you click on it choose "Select Child Nodes" it will highlight the whole skeleton, then going over to XSI Explorer press Enter this will call up the null display and size window. If you set the size to 0.25 if makes it a lot easier to see. Then hide all the root and effector nulls so you are just left with the bone_ nulls.

I am not sure if my ramble has help you at all but hopefully it wasn't just a lot of words.

FragMe! wrote:I am not sure if my ramble has help you at all but hopefully it wasn't just a lot of words.

Partly.

Actually, I have two walker meshes. One fits the AT-ST layout, the other I'm trying to make a new skeleton for.

FragMe! wrote: If you are looking at the skeleton in Unwrap don't be fooled by the way it shows the links between the bones. For example there isn't anything going on a diagonal from bone_pelvis to bone_l_thigh (or the right side either), the bone or rather root of the thigh starts at the end of the arrow.

You are trying to match your model to the skeleton correct? Not necessarily doing a whole new set of animations. Do you have a way to import the basepose skeleton into xsi? If you do and it is via Unwrap here is what you can get rid of, the unaasigned group, the geometry I think it will be called u3dscene that should leave you with just a set of green nulls. That is all you really need to envelope to.

Unforunately, I only have the demo of Unwrap, and so no way to save it in an XSI-readable format (Meshtool can't import it, can it?). Is there anything besides Unwrap that could do the job?