Aller au contenu

Photo

Question about head weighting and bones


  • Veuillez vous connecter pour répondre
12 réponses à ce sujet

#1
Tchos

Tchos
  • Members
  • 5 072 messages

If a head and neck is in the same position as a stock head, and weighted to the same bones, with the same colours indicating weight strength, why would the neck, which is weighted to the neck bone and at its lower edge the ribcage, move instead with the head as if the neck and ribcage have no influence, with the bottom of the neck floating in and out of the shirt's neck hole?



#2
Eguintir Eligard

Eguintir Eligard
  • Members
  • 1 832 messages

If you built your own skeleton that's an easy answer. But since you didn't I don't know.



#3
Tchos

Tchos
  • Members
  • 5 072 messages

For clarity for those who may know, this is for one of the standard character skeletons.



#4
4760

4760
  • Members
  • 1 213 messages

One reason could be that the pivot is not at (0, 0, 0), if to reach " the same position as a stock head" you had to move the new head.

 

Another possibility is that "the same colours indicating weight strength" could still give different weighting. I'm not sure a 0.5/0.5 split between neck and head (orange) has a color visibly different from 0.48/0.52 for example. In the latter case, the vertex would be slightly more influenced by the head, meaning it will be closer to the head thus leaving a gap.

 

Did you try the SkinUtilities? You would still have some vertices to correct manually, but it helped me a lot to add new heads to the game (not released, they're part of new NPC).


  • rjshae aime ceci

#5
Tchos

Tchos
  • Members
  • 5 072 messages

I don't know about SkinUtilities, but I'll look into it.  However, the vertices at the bottom of the neck are entirely weighted to the ribcage, with no split at all, as I can see in the weight table.  I had to use the weight table to finish up each attempt after painting, since painting weights often results in more than two bones influencing a vertex, which it says is not allowed.  In each case, I zeroed out the bone of least influence, allowing it to normalise the other two.



#6
4760

4760
  • Members
  • 1 213 messages

since painting weights often results in more than two bones influencing a vertex, which it says is not allowed.

It is allowed, but the limit is 3. Did you check (under the bones list) how many influence each vertex? I think max sets it to 4 by default.

And does the message about "more than 2 bones" actually shows "more than 2 non-facial bones"?

If you don't want to use the f_ bones, untick the box "Head (cutscene)" in the Expotron utility page (or right-click on the mesh and remove the User Defined property "Head = 1"): that way, the head will be considered a normal body part, and three non-facial bones could influence each vertex.



#7
4760

4760
  • Members
  • 1 213 messages

However, the vertices at the bottom of the neck are entirely weighted to the ribcage, with no split at all, as I can see in the weight table.

That could then explain why "the bottom of the neck [is] floating in and out of the shirt's neck hole": if the neck hole vertices are not influenced only by the ribcage (in some cases, I noticed the neck and spine also affect these vertices), then obviously the vertex from the neck hole and the same from the head will move differently.



#8
Tchos

Tchos
  • Members
  • 5 072 messages

The floating in and out is the same behaviour that I saw when the entire neck was weighted to the neck bone, before I tried adding the ribcage weight based on what I saw was weighted in the original head.  Adding the ribcage, which mimicked the original head's weighting, did not change it. 

 

There is one thing from the original head's weighting that I did not mimic -- one side of the edge of the neck was weighted to the right collar bone.  The left side was not weighted to the left collar bone.  I'll have to try that, but would that really be so necessary just to make a neck stay in place?  I might as well just make whole-body creatures and not make interchangeable heads if it's going to be this unruly.

 

The only bones that have any influence at all in this mesh are the head, neck, and ribcage.  Those are the only ones that I added in the envelope list.  No facial bones.  (Yes, I believe the message was about more than 2 non-facial bones.)



#9
-Semper-

-Semper-
  • Members
  • 2 259 messages

before you do any manual weighting, play around with the skin wrap modifier. it should be available with 3dsmax8. it's a nifty little tool to transfer weighting data from one mesh to another. there's also a three part tutorial from autodesk to get you started.



#10
Tchos

Tchos
  • Members
  • 5 072 messages

Is that the process described in this tutorial?  Because that's how I started this endeavour.  Should I have just skipped the parts where it said to paint the weights after transferring the weight data and exported directly, even if the mesh has different vertices as in this tutorial?

http://neverwinterva...m-head-tutorial



#11
-Semper-

-Semper-
  • Members
  • 2 259 messages

no, it's not the same process - not even close. btw this tutorial is totally wrong. there's no sense in saving and loading weighting data which is restored per vertex id. if you import/create a custom model it will also have a completely new order of vertices, and you have to start the weighting from scratch.

 

a skin wrap modifier works similar to a texture baking process. it takes the underlaying geometry and fires a ray outwards. if it hits your new geometry the weights will be transfered to the closest vertex available. if your custom head ain't totally alien to the volume of the nwn2 head model, there's a huge chance that you will face a one click solution. even if there are weighting issues it's far easier and faster to manually reweight the affected parts.



#12
Tchos

Tchos
  • Members
  • 5 072 messages

Okay, that sounds like a winning procedure.  The new head is indeed a completely different model with a new order of vertices, but it's approximately the same size and shape as any other head.  I'll see how this works, and discount the tutorial.



#13
Tchos

Tchos
  • Members
  • 5 072 messages

Thanks, Semper.  The Skin Wrap modifier fixed the problem, and it was so much easier than the other method.