Crashy heads

[Legacy_Mavrixio]

               http://dl.dropbox.co...crashy heads.7z

Those 2 heads are making players crash when logging or relogging, anyone can look at them?
               
               

               

                     Modifié par Mavrixio, 25 janvier 2013 - 05:07 .
                     
                  

            

[Legacy_dusty.lane]

               This one will take a look, Mav. Should've just asked this one!!

Buuut, aside from the hair defying all logic, this one can't see much that could be wrong with it. This one suspects it could be shadows, but doubts it. Could turn off shadows and see if those same players crash. If they do, this one is obviously wrong and somebody more experienced may be able to tell you. Or you're just SOL on the matter.
Maybe...

But what do I know? I am just a big stupid jellyfish.
               
               

               

                     Modifié par dusty.lane, 25 janvier 2013 - 06:25 .
                     
                  

            

[Legacy_Rolo Kipp]

               <scratching his...>

Hmmm... What is the part number directive in these heads?

beginmodelgeom pmh0_head125
node dummy pmh0_head125
  parent NULL
  #part-number 0
endnode
node trimesh pmh0_head125g
  parent pmh0_head125
  #part-number 1
  ambient 1 1 1
  diffuse 1 1 1
  specular 1 1 1

Other than that and the high poly count (1286) I don't see anything really out of place in the model itself. That hair, though. definitely a shadow-killer :-P

<...chin>
               
               

               
            

[Legacy_Failed.Bard]

               The scale and animation scales don't match, but I don't know if that would cause crashing.
               
               

               
            

[Legacy_Rolo Kipp]

               <does a double...>

Damn. You're right, FB.

Scale is 1 on all the Bioware heads I just looked at.
Scale is 1.15559995 on 124 and
Scale is 0.811886013 on 125.

<...take>
               
               

               
            

[Legacy_Rolo Kipp]

               <just putting...>

In the Bioware heads, what are these:

displacement 0.03
period 6
tightness 4
constraints 87
0
0
0
0
0
0
100
140
...

Is that danglymesh info?

Edit: NM, yes it is. ;-P

<...it out there>
               
               

               

                     Modifié par Rolo Kipp, 25 janvier 2013 - 02:49 .
                     
                  

            

[Legacy_NWN_baba yaga]

               complicated stuff to understand cuz in 3dsmax the preferences are different!
               
               

               
            

[Legacy_OldTimeRadio]

               In Max, these black triangles are the clue that face normals are flipped somewhere.  If you try deleting the bad faces, it appears in some places there are faces underneath which are pointing the right way.  Which means there are double faces.  This can happen for a number of reasons, such as sometimes happens if the model is ripped from draw calls. 

Going into the element subobject level and just clicking around, there are lots of subobjects when there shouldn't be.  Detaching all the elements using this handy tool reveals 470+ elements which is probably 100x too many.  Inspecting the elements, it looks like there are two copies of the hair present, one which is composed of about 7 big elements and the other which is composed of the other 400 disparate triangles.

These are probably "interior" and "exterior" hair.  Wound up removing what was the "exterior" hair and flipping the normals on the interior hair.  From what I saw they both were UV-mapped exactly the same.

There are the #Part indicators which Rolo mentioned, and that's a clue that the model was decompiled from a compiled state.  The #Part indicators are sort of an under the covers way of dealing with the model bits and AFAIK, they play the biggest role in supermodeling whcih is why, for instance, model compilers may ask the super model be present when compiling a model that uses them.  More information on how that works can be learned here and Ragnarok_mr4's 3rd message in the Omnibus thread "Change Phenotype Lag", among other places.

The scale disparity Failed.Bard mentioned is probably a result of someone wishing to enlarge the head.

Back to the poor model which has been broken into its (way too many) elements.  Face, ears are fine-ish.  Everything else is a steaming mess. 

* Removed all the disparate hair elements, leaving just a few big ones.  The overall look seems to be about 99% the same, afterward, though.
* Reset XForms on the model (had to do it old school, NWMax couldn't cut it)
* Welded Verts
* Welded TVerts
* Compiled w/Bioware's model compiler

I didn't really fix it so much as remove all the nasty-nasty I could and put makeup on what was left.  I only did pmh0_head124.mdl.  The other one was beyond hope.  There were no big groups among the elements and the hair style relied quite a bit more heavily on the inner hair showing through, something that didn't appear to be such a big deal on the head I did do.




               
               

               

                     Modifié par OldTimeRadio, 25 janvier 2013 - 07:35 .
                     
                  

            

[Legacy_Mavrixio]

               Thanks!
               
               

               
            

[Legacy_dusty.lane]

               Isn't OTR the best?
               
               

               
            

[Legacy_OldTimeRadio]

               LOL!  Pfft! Yer gonna jinx me!