Jump to content

New Clothing Body Style Converter Beta v0.89f (10-26-2014)


Recommended Posts

Yeah I noticed it doesn't like DMRA to TGND but it does a great job on the armor/outfits so I just use nifskope and copy the TGND body part that I used for the conversion then load the converted one up and if it lets me I just remove the converted body branch out and paste mine into it if it doesn't let me then I just place it onto my player skeleton and do it that way instead. TGND also has a ton of verts on the ass cheeks as well I'm still using the b_50 with b_52_Upgrade I think my first trial of DRMA to TGND was the funniest after it was done being converted it looked like someone let the air out of the top of the breasts and then pushed the fronts flatish they looked like blocks.

Link to comment

Yeah I noticed it doesn't like DMRA to TGND...I'm still using the b_50 with b_52_Upgrade

In that case, you've got to try v.68.

 

It does truly amazing things with DMRA to TGND conversions.

 

This is a DMRA body that v.68 tool converted to TGND. Still the DMRA vertex count, but it is almost identical to a true TGND body. Check it out. DMRA_to_TGND.nif

 

Source Files For Reference:

 

The DMRA nif before conversion: DMRA_Template.nif

 

And the TGND nif used: tgnd.nif

Link to comment

Just wanted to share my latest conversion.

 

I've abandoned UNP Topmodel, as I could simply not get rid of the neck seam caused by switching to that body in BodyChange. Instead, I turned to BodySlide UNP. I composed a custom body there (based on UNP), used the converter to create lattices from UNP to my custom body, and then converted Northgirl armor.

 

The result...

post-368191-0-13290600-1396199420_thumb.pngpost-368191-0-39279200-1396199423_thumb.pngpost-368191-0-69858100-1396199414_thumb.png

 

Resounding success!

 

There may be some distortion at the lower breasts and the legstraps still float, but both are hardly noticeable. This conversion is sublime.

 

I'm *very* happy! Cheers for your work, you've got yourself a fan!!

Link to comment

That looks fantastic orgs1n.

I'm really happy with the results I'm seeing out of the v.0.68 version of my tool framework. The baseline conversion precision is now, essentially, 100%.

The added precision does have one drawback. Any flaws in the 'From' and 'To' nifs used to generate the conversion lattice will now be perfectly preserved by the conversion lattice. Similarly, if UV search is enabled, any bugs in the UV map used to match equivalent spaces between the 'From' and 'To' nifs can cause significant flaws in an otherwise perfect conversion.

Sample issue and resolution:

The Challenge was to convert the HGEC HA Upper Body to the HGEC A Upper body

post-20303-0-16051900-1396213702_thumb.png To post-20303-0-17014400-1396213699_thumb.png
          HGEC HA                                 HGEC A

 

And here was the result:

post-20303-0-93192700-1396210578_thumb.png

 

This was a very strange error.  Aside from the giant glaring hole, the conversion was flawless.  Something very strange was going on.

 

Comparing the UV maps in Blender demonstrated an obvious flaw in the UV Map for the HA nif.

 

post-20303-0-03375400-1396212184_thumb.png

The 'hole' in the mesh was caused by a nasty but easily missed UV flaw. Essentially, one of the Left side vertices was mapped onto the right side of the UV map.  The resulting uv distortion (the stretched and overlapping triangles), created a hole when the tool used the UV data to match equivalent spaces between the HGEC HA and HGEC A nifs.

 

Repairing the UV Map was fairly straightforward:

post-20303-0-15740000-1396212815_thumb.png

And finally the HGEC_HA to HGEC_A conversion using the repaired nif as the 'From' nif.

post-20303-0-31060800-1396213556_thumb.png

 

And here are the results of applying that high quality lattice to an extremely complex and challenging nif:

 

Before:  post-20303-0-26654800-1396225007_thumb.png

 

After: post-20303-0-87105800-1396225013_thumb.png

Link to comment

That was I think because I used a good body for the from and good body for the to but the actual body was bad as I found out it had no nipples so it gave it some. If you remove the top there's actually 5 inch pointy nipples underneath it.  :) Might be a good look for a monster breed. 

 

70 looks like it works really good no problems to report so far I just converted the sheva armor to TGND. 

post-25667-0-04172100-1396228690_thumb.jpgpost-25667-0-58373000-1396228706_thumb.jpg hands are getting pretty close to matching post-25667-0-07040500-1396228723_thumb.jpg I did see one thing though.

 

Right at the front of the neck there is a hole not sure why there was nothing wrong with that area. 

post-25667-0-06151400-1396228734_thumb.jpgpost-25667-0-38585600-1396228754_thumb.jpg

Link to comment

Thanks for the report.

 

I've recently begun noticing some intermittent and fairly unpredictable issues with wrist and neck seams.

 

I've got some theories about why it might be happening.  I'm currently testing a potential fix.

 

Basically...precisely matching the best match on a given face works great for transformations...except when the tool needs to match inter-mesh seams.  so, I'm modifying the Lattice Maker and Clothing Converter to treat non-manifold seam vertices for Skin type meshes as a unique class of vertex that need to be processed differently.

 

That is...wrist, neck, waist, and ankle seams should stop sucking with the next patch.

 

I'll post it if I manage to resolve the issue.

Link to comment

Good News.  My fix appears to have worked.
 
Clothing Converter V.0.71 v.072 is now live.
 
Change Log
 
v.70 to v.72
1. Wrist, Neck, Ankle, and Waist seams on 'skin' type meshes should now be converted properly (exact vertex match)
2. Mesh Names should once again appear correctly in relevant menus

3. Non-Manifold vertices that do not lie on primary seams will continue to be converted using the face location search.

 

Development Plan for future updates:

 

1. Improve detection of Neck, Wrist, Waist, and Ankle seam vertices.  The current method is a bit too hacky for my taste.

2. Simplify the generation of From and To nifs.  The goal is to allow you to select multiple nifs (say, foot, hand, body) and the tool will automatically merge them for the purposes of generating a conversion lattice.

3. Build a library that maps different UV standards.  The goal will be to allow you to generate high quality conversions even when the From and To nifs use different UV standards.  At the moment, the high quality results can really only be achieved when using From and To nifs with either compatible UV Maps or identical vertex indexes.

Link to comment

Looks like it's a success!!!! :)  Great work Gerra6!!! No problems I did have to make a new lattice as the old one still did it but after making a new one and overwriting the old one it worked like a charm.

 

Awesome.

 

It's always nice when a crazy idea actually works.

Link to comment

Oh...I also have a question for anyone using the latest version of Clothing Converter.

 

Clothing Converter has a bit of a design flaw...not a big one, but one that has been bothering me a bit for the last few days.

 

I designed the 'Search Distance' as a hard edge to the clothing conversion.  Up to that distance, the searching vertex will receive the full brunt of whatever transform was coded into the lattice surface at the nearest point to the searching vertex.  .00001 units past the 'Search Distance' the vertex receives no transformation at all.

 

This...makes me grumpy.  So...I'd really like to add some form of edge smoothing.  The idea is that, as vertices near the upper limit of the search distance, the percentage of the transformation that they receives gradually drops off until it reaches zero at the search distance limit.

 

Any thoughts?  Good idea, bad idea?  What sort of implementation would be both useful and user friendly?

 

One idea I had is to add a second distance slider, one whose value is added to current value of the Search Distance Slider.  So, anything less than Search Distance receives the full transformation coded into the lattice.  Anything between Search Distance and Smoothing Distance receives a reduces version of the transform, dropping off as the distance increases.

Link to comment

Errored out again on latest version, this time while attempting to generate a UNPBase->SevenbaseBombshell lattice. Error copied from the text dump below along with source templates. :/

Traceback (most recent call last):
  File "make_lattice.py", line 205, in <module>
    Main()
  File "make_lattice.py", line 190, in Main
    mainSearch(cs, to_mesh, from_mesh, side = side, ctx_verts = from_base, searc
h_resolution = s_res, vertFunction = setInfluences, face_search = True)
  File "C:\Users\Jonathan\Desktop\Skyrim Meshes Working Directory\Convert Clothi
ng RC_72_Light\kg\search_util.py", line 858, in mainSearch
    face_search = face_search
  File "C:\Users\Jonathan\Desktop\Skyrim Meshes Working Directory\Convert Clothi
ng RC_72_Light\kg\search_util.py", line 1816, in getVertexInfluence
    v1, v2, v3 = face_.verts
UnboundLocalError: local variable 'face_' referenced before assignment

C:\Users\Jonathan\Desktop\Skyrim Meshes Working Directory\Convert Clothing RC_72
_Light>pause
Press any key to continue . . .

unpToBombshellSamples.7z

Link to comment

Basically it required that the nearest point on a face(excluding edges) and the nearest point on an edge coincide *exactly* (within epsilon).

 

My *guess* is that an edge collided with a face somewhere and that collision location happened to exactly coincide with the calculated nearest point to to some vector that was out searching for a partner on the mesh.  Hence the *wow*.

 

You're the best kind of beta tester.  The kind that breaks shit.

Link to comment

OK, I think this should resolve that error.

 

Clothing Converter Release Candidate 0.73

 

Download:

 

Portable Package 73 Upgrade (upgrades Portable Package v.70, v.72 to v.73):  Removed

Light Download 73 (scripts only. Recommended if you intend to use 64 Bit version of Python): Removed

 

 

ChangeLog

 

v.72 to v.73

1. A rare crash bug has been identified and resolved

2. The tool performance has been improved.

Link to comment

Woohoo! Fixed one bug, made two new bugs.

 

This update fixes those two new bugs...so...

 

Clothing Converter Release Candidate 0.74

 

Download:

 

Portable Package 74 Upgrade (upgrades Portable Package v.70, v.72, v.73 to v.74):  Removed

Light Download 74 (scripts only. Recommended if you intend to use 64 Bit version of Python):  Removed

 

 

ChangeLog

 

v.73 to v.74

1. Eliminates potential source of distortion near UV seams.

2. Eliminates potential crash bug

Link to comment

I wasn't kidding.

 

Clothing Converter Release Candidate 0.75

 

Download:

 

Portable Package 75 Upgrade (upgrades Portable Package v.70, v.72, v.73, v.74 to v.75):  Removed

Light Download 75 (scripts only. Recommended if you intend to use 64 Bit version of Python):  Removed

 

 

ChangeLog

 

v.74 to v.75

1. Skyrim and Fallout Dismemberment Partition Maps will be properly preserved.

Link to comment

OK. This is getting silly

 

Clothing Converter Release Candidate 0.76

 

Download:

 

Portable Package 76 Upgrade (upgrades Portable Package v.70, v.72, v.73, v.74, v.75 to v.76):  Removed

Light Download 76 (scripts only. Recommended if you intend to use 64 Bit version of Python):  Removed

 

 

ChangeLog

 

v.75 to v.76

1. Target nifs with complete skeletons were having their skeletons improperly flattened and transformed. This bug has been resolved.

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. For more information, see our Privacy Policy & Terms of Use