Maya Cleanup question?

anonymous_user_bb0e176f1 · November 18, 2014, 2:54pm

Vaei;180763:

With the amount of issues that can arise from data stored in meshes I decided to write a script. Usually exporting as .obj can solve a lot of things but sometimes it’s easier to resort to rebuilding a mesh by hand especially if you don’t know the problem.

script does exactly that for you in a few seconds (depending on poly count and your processor), and it lets you keep your UVs, normals, and even skin/weights. It is a good way to isolate components as well such as geometry, UVs, normals, and rig.

I wrote it very quickly, I will probably do it properly later but for now with an interface, error checking, and hopefully better efficiency.

If you (after saving a copy…) select your object and run it, it will rebuild your mesh using only the vertex positions.

Note:
It is a draft and there is no error checking, so if you get an error then it was given incorrect information

To use:

Backup your scene, make a copy, don’t work in your originals

Copy/Paste the script below into your script editor (accessible via the button in the very bottom-right corner of the Maya interface)

By default, after rebuilding your mesh, it will copy over your UV’s, normals, and skin/weights. If you don’t want any of these change from ‘True’ to ‘False’ and note the capitalization of the first letter (it will error if you use ‘false’ instead of ‘False’) and don’t use quotation marks

If your mesh does not have a skin attached you must disable “transferBindAndWeight” by setting it from True to False.

Close UV editor. It will work with it open but take a VERY long time (maybe 5x longer).

Select your object (must be an object, not components such as faces or vertices or the script will break)

Run the script (Either press numpad enter, or click the single blue arrow on the menu at the top of the script editor) - Selecting the text prior to executing will keep it in the script editor, unselected text is consumed on successful execution

A 6k poly mesh took around 10-15 seconds on my i7 @ 3.4ghz, be wary of high density meshes.

If I re-write it I’ll make a thread for it, right now it’s very much a prototype but I whipped it up to see if it’d solve your mesh’s issues.

Here is the python code:
import maya.cmds as cmds
import maya.mel as mel

'''
PolyReconstruct v0.0.1 by  Taylor
- Will rebuild a mesh using only the vertex positions,
- Can transfer the UVs and skinning from original
- You MUST specify False for transferBindAndWeight if there is no skin

- Takes around 10-15 seconds for a 6k poly mesh on my i7 processor, be wary for high density meshes

-->CLOSE UV EDITOR BEFORE RUNNING  SCRIPT - IT WILL TAKE 5x LONGER IF IT IS OPEN<--

Anything that can update during the execution of a script will update on every operation, including
the UV editor making the running time significantly longer
'''

#--------------------------------
#CHANGE ONLY THESE OPTIONS
transferUVData = True#Keep UVs from original mesh
transferNormalData = True#Keep normals from original mesh
transferBindAndWeight = True#Transfer character skin and weight - DISABLE if no skin!

#--------------------------------
#DO NOT CHANGE BELOW
obj = cmds.ls(sl = 1)[0]
cmds.select(cmds.polyListComponentConversion(obj, tf = 1))
faces = cmds.ls(sl = 1, fl = 1)

newfaces = ]

for f in faces:
    cmds.select(cmds.polyListComponentConversion(f, fromFace = 1, toVertex = 1))
    vtx = cmds.ls(sl = 1, fl = 1)
    
    vtxPos = ]
    bgnV = vtx[0]
    finV = ]
    
    match = ]
            
    cmds.select(cmds.polyListComponentConversion(vtx[0], fromVertex = 1, toEdge = 1))
    mel.eval("PolySelectConvert 3;")
    vertEdge = cmds.ls(sl = 1, fl = 1)
    vertEdge.remove(vtx[0])
    for v in vtx:
        if not v in vertEdge:
            finV = v
        else: 
            match.append(v)
            
    if len(vtx) > 3:#poly
        order = bgnV, match[0], finV, match[1]
    else:#tri
        order = bgnV, match[0], match[1]

    for o in order:
        vtxPos.append(tuple(cmds.xform(o, q = 1, t = 1, ws = 1)))

    newfaces.append(cmds.polyCreateFacet(p = vtxPos, ch = 0)[0])
    
cmds.select(newfaces, r = 1)
newMesh = cmds.polyUnite(newfaces, n = "Generated_PolyReconstruct_Geo", ch = 0)
cmds.polyMergeVertex(newMesh, d = 0.0001, am = 1, ch = 0)
cmds.polyNormal(newMesh, normalMode = 2, userNormalMode = 0, ch = 0)

if transferUVData:
    cmds.transferAttributes(obj, newMesh, transferPositions = 0, transferNormals = transferNormalData, transferUVs = 2, transferColors = 2, sampleSpace = 0, searchMethod = 3, flipUVs = 0, colorBorders = 1)
    
if transferNormalData:
    cmds.transferAttributes(obj, newMesh, transferPositions = 0, transferNormals = 1, transferUVs = 0, transferColors = 2, sampleSpace = 0, searchMethod = 3, flipUVs = 0, colorBorders = 1)
    
cmds.DeleteHistory(newMesh)

if transferBindAndWeight:
    jnts = cmds.skinCluster(obj, inf = 1, q = 1)
    mif = cmds.skinCluster(obj, mi = 1, q = 1)
    bnm = cmds.skinCluster(obj, bm = 1, q = 1)
    skm = cmds.skinCluster(obj, sm = 1, q = 1)
    nmw = cmds.skinCluster(obj, nw = 1, q = 1)
    newSkin = cmds.skinCluster(newMesh, jnts, dr = 4, mi = mif, bm = bnm, sm = skm, nw = nmw)
    cmds.copySkinWeights(obj, newMesh, noMirror = 1, surfaceAssociation = "closestPoint", influenceAssociation = "oneToOne")
    
cmds.select(newMesh, r = 1)

Thanks for !

anonymous_user_bb0e176f1 · November 18, 2014, 3:27pm

There’s something I’m missing. It’s very frustrating…

Vaei · November 18, 2014, 8:43pm

Does the still persist with the mesh generated via the script?

anonymous_user_bb0e176f1 · November 18, 2014, 9:27pm

I’m trying it right now.

anonymous_user_bb0e176f1 · November 18, 2014, 9:46pm

Thanks for the script but it didn’t work. It generated a new mesh, but I still have the same with the tessellated spike with one of the materials.

Vaei · November 18, 2014, 11:37pm

Ah. In that case I can still take a look if you want to send me relevant assets. Otherwise I can’t think of anything off the top of my head that we haven’t already checked.

anonymous_user_bb0e176f1 · November 19, 2014, 12:30am

I sent you a PM. Thanks for your help!

Vaei · November 19, 2014, 10:34pm

Hi,

Sorry for the delay have been busy. I’ve found the problem.

Since we were able to rule out the mesh and UVs it was obviously the skeleton, so I made a quick script that checks each joint against anything I figured could cause an like .

script here checks for joints with scale values that are not 1.0:


import maya.cmds as cmds

cmds.select(hi = 1)
for i in cmds.ls(os = 1):
    for x in cmds.xform(i, q = 1, r = 1, s = 1):
        if x != 1.0:
            print i

And it output :


root|pelvis|spine_01|spine_02|spine_03|clavicle_l|upperarm_l
root|pelvis|spine_01|spine_02|spine_03|clavicle_r|upperarm_r
root|pelvis|spine_01|spine_02|spine_03|clavicle_r|upperarm_r
root|pelvis|spine_01|spine_02|spine_03|clavicle_r|upperarm_r|lowerarm_r

In the channel box Maya shows 3 floating points and will round if necessary. However, Maya is accurate to 16 floating point values so if you aren’t using python for error checking then it’s no wonder you didn’t find it.

These joints had these values: [1.0, 0.99999999999999989, 1.0]

Selecting the upper/lower arms for each side and setting to 1.0 again in the channel box then rebinding the mesh solved the error.

The small script I posted above, if you select your skeleton’s root and run it, should give you a list of every erroneous joint, when their scale is back to 1.0 the error should be gone.

anonymous_user_bb0e176f1 · November 19, 2014, 11:00pm

Hi,

Sorry for the delay have been busy. I’ve found the problem.

Since we were able to rule out the mesh and UVs it was obviously the skeleton, so I made a quick script that checks each joint against anything I figured could cause an like .

script here checks for joints with scale values that are not 1.0:
import maya.cmds as cmds

cmds.select(hi = 1)
for i in cmds.ls(os = 1):
    for x in cmds.xform(i, q = 1, r = 1, s = 1):
        if x != 1.0:
            print i
And it output :
root|pelvis|spine_01|spine_02|spine_03|clavicle_l|upperarm_l
root|pelvis|spine_01|spine_02|spine_03|clavicle_r|upperarm_r
root|pelvis|spine_01|spine_02|spine_03|clavicle_r|upperarm_r
root|pelvis|spine_01|spine_02|spine_03|clavicle_r|upperarm_r|lowerarm_r
In the channel box Maya shows 3 floating points and will round if necessary. However, Maya is accurate to 16 floating point values so if you aren’t using python for error checking then it’s no wonder you didn’t find it.

These joints had these values: [1.0, 0.99999999999999989, 1.0]

Selecting the upper/lower arms for each side and setting to 1.0 again in the channel box then rebinding the mesh solved the error.

The small script I posted above, if you select your skeleton’s root and run it, should give you a list of every erroneous joint, when their scale is back to 1.0 the error should be gone.

Thanks for looking into for me. I when I ran the Python script after selecting my root, is what it came up with. Not sure what I’m missing or if Im looking at the wrong thing…

Vaei · November 19, 2014, 11:16pm

You need to read the output in the script editor. All it does is tell you which joints have bad scale values.

anonymous_user_bb0e176f1 · November 20, 2014, 3:11pm

Where do I see the output? I’m assuming it’s above the the Mel/Python tabs. in the top section of the window. is what it outputted:

anonymous_user_bb0e176f1 · November 20, 2014, 3:13pm

The interesting part is that I also have that tessellation when I output my character from 3DS Max too, using their system.

anonymous_user_bb0e176f1 · November 20, 2014, 3:40pm

I think its an U4 since it only happens when importing a rigged mesh,

Here is another test mesh, time a basic model with a custom skeleton created in Maya with the join tools. I also included a non-rigged mesh. If you import them both into U4 and assign a tessellated material into the channels, the non-rigged mesh tessellates correctly and the rigged one has the spiking issues.