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

[Alwayon]

As I tried to understand your kind explanation above, I spent some more time observe these conversion result & I believe I've seen some pattern here...

 

It seems like the tool try to identify the neck seam and eventually find two "ring" of seams as you said.

 

 

 

More NIF and its corresponding log.rar

 

 

 

However, this seem to be quite different case comparing to the left and right wrists/ankle seams detection.

... I mean all of these are roughly cylinder-shaped meshes, each piece has cut-out on both ends ... but wrists and ankle detection seem never find two rings or non-manifold edges, but neck seam detection does.

 

Please forgive me for my lack of enough understanding/knowledge of edge-detection algorithm. But I am a bit wondering that  ... is there some way we can keep just the "higher ring" of neck seam and discard/ignore the lower one ? 

 

 

[emmietiie]

So, I'm using the latest version to try and convert DMRA clothing and armors to AGHGEC, and I think I got decent results. I don't have the texture paths set in nifscope, so it's hard to see. One issue I'm having is that some of the armors, like golden saint / dark seducer were made as a single outfit mesh. How would I go about combining the source and destination meshes as a single nif file to create the lattice of that?

I hope you will share once they are finished

[Monsto Brukes]

I haven't looked at this thread in "a while" and was drawn back to it recently.

 

So I have a question: From what source body to what target body worked well for you? And what outfit(s) did you convert?

 

I'm Just Curious.

[Alwayon]

I haven't looked at this thread in "a while" and was drawn back to it recently.

 

So I have a question: From what source body to what target body worked well for you? And what outfit(s) did you convert?

 

I'm Just Curious.

 

For me, I am testing UNPB to 7Base (all variations) / CBBE or CBBE Pushup to 7Base (all variations) / UNP to various bodies, Vanilla to my custom body, and a few more pairs.

 

My method is to wrap destination body mesh on source shape first, then make lattice out of that.

And I can say that Gerra's tools is working so well and quite perfect most of the time. Making mass armor automatic conversion a dream come true.

 

Of course, there are some drawback...  ie ... converting low res armor to quite different-sized body .. but this are also true when we are using Bodyslide to do the same thing.

 

What I am planing to convert (and may be ask for permission to share) are LSAR, and full set of vanilla armor + DLC, hopefully, after Gerra release some hotfix for this excellent tool.

 

[Kodiak]

Now I'm guessing this is a fairly safe assumption, but I just wanted to ask anyway. Does this tool only function for Skyrim models and meshes or can it be applied to Fallout armor meshes as well?

[gerra6]

The tool has been designed for and tested with Oblivion, Fallout, FNV, and Skyrim nifs.

[Kodiak]

Awesome! Thank you All I ever really see if Skyrim content, I'm glad to hear that it works for NV. That is why you never make assumptions

[Monsto Brukes]

 

My method is to wrap destination body mesh on source shape first, then make lattice out of that.

 

Do what now? If you would detail that a bit I'd appreciate it.

 

I tried to convert from CBBE to my Bodyslide mesh, and it was not good. I figured since it's as close to the same  as you could get on a technical level, that it'd come out well. It did not. some pretty bad distortions around underboob. Not horrible around waist and butt tho still unusable.

[Alwayon]

 

 

My method is to wrap destination body mesh on source shape first, then make lattice out of that.

 

Do what now? If you would detail that a bit I'd appreciate it.

 

I tried to convert from CBBE to my Bodyslide mesh, and it was not good. I figured since it's as close to the same  as you could get on a technical level, that it'd come out well. It did not. some pretty bad distortions around underboob. Not horrible around waist and butt tho still unusable.

 

 

I guess this is going to sound complicated but actually it doesn't.  Because I just learn all of this 1-2 months ago reading around. Before that, I don't know how to operate ANY applications mentioned here.

 

The key to good conversion seems to be letting the tools know how to morph the source body into targetbody_0 and 1 with BEST precision.

 

One way to achieve that is to have the target body mesh BUT in source shape.  Ie. UNPB body in exact UNP 0 or 1 shape.... and use it as the source body in lattice making

This will let the tools know exactly how to move each vertex in source body into another position to make it in target body shape. Because actually it is the same mesh with different vertex positions. Thus letting the tools have almost perfect morph-field (?) for this source->target body.

 

To have this, I normally feed the actual source body and the target body into a program called Wrap (by russian3dscanner.com, thanks to blabba's recommendation) .. let it wrap target body around source. ... get the result out then feed it into Mudbox to correct its surface ... then feed it into 3ds Max to use as a morph target, correct the vertices & neck/hand/feet seam ... and sometime use another 3ds Max script called Wrapit to polish it a bit ... then finally export it out as a NIF.

 

This output mesh will be my source mesh in lattice-making. I'll feed it into Gerra's make_lattice and use Vertex Index Search mode only.

 

Before I convert any armor, I will test my lattice out by trying to convert actual source body into target body.  It should produce almost exact same shape body which I always load into 3ds Max to compare & confirm.

 

(Of course, there will be some often-negligible distortion because the tool doesn't morph source into target mesh directly .. but it will be very-very close and quite perfect for armor conversion job)

 

If this body-only conversion turn out right, I am ready to convert any armor NIF.  And it turn out not OK, it is likely to to armor's fault.

 

However ... if body-only conversion turn out not-so-good. The lattice is not usable and there is no need to try converting any armors with it, the result will be almost certainly bad.

 

BTW, I still not have a chance to try UV Search mode yet. So I don't know about its quality/precision

---

 

In your case, I believe both of your mesh should already be Vertex Index Search mode compatible since both are just CBBE with different shape ?? So there should be not need for this wrap-it-around process.

 

There might be one exception when the breast area are too different.

 

In this case, the body should be converted perfectly but some low resolution armor mesh may show some clipping, I guess.

 

I still don't know anything well enough yet .. but I am willing to help if I can. So, if you don't mind, I can take a look at your target CBBE mesh to see whether there is anything wrong.

[Monsto Brukes]

 

 

 

 

 

 

My method is to wrap destination body mesh on source shape first, then make lattice out of that.

 

Do what now? If you would detail that a bit I'd appreciate it.

 

I tried to convert from CBBE to my Bodyslide mesh, and it was not good. I figured since it's as close to the same  as you could get on a technical level, that it'd come out well. It did not. some pretty bad distortions around underboob. Not horrible around waist and butt tho still unusable.

 

 

I guess this is going to sound complicated but actually it doesn't.  Because I just learn all of this 1-2 months ago reading around. Before that, I don't know how to operate ANY applications mentioned here.

 

The key to good conversion seems to be letting the tools know how to morph the source body into targetbody_0 and 1 with BEST precision.

 

One way to achieve that is to have the target body mesh BUT in source shape.  Ie. UNPB body in exact UNP 0 or 1 shape.... and use it as the source body in lattice making

This will let the tools know exactly how to move each vertex in source body into another position to make it in target body shape. Because actually it is the same mesh with different vertex positions. Thus letting the tools have almost perfect morph-field (?) for this source->target body.

 

To have this, I normally feed the actual source body and the target body into a program called Wrap (by russian3dscanner.com, thanks to blabba's recommendation) .. let it wrap target body around source. ... get the result out then feed it into Mudbox to correct its surface ... then feed it into 3ds Max to use as a morph target, correct the vertices & neck/hand/feet seam ... and sometime use another 3ds Max script called Wrapit to polish it a bit ... then finally export it out as a NIF.

 

This output mesh will be my source mesh in lattice-making. I'll feed it into Gerra's make_lattice and use Vertex Index Search mode only.

 

Before I convert any armor, I will test my lattice out by trying to convert actual source body into target body.  It should produce almost exact same shape body which I always load into 3ds Max to compare & confirm.

 

(Of course, there will be some often-negligible distortion because the tool doesn't morph source into target mesh directly .. but it will be very-very close and quite perfect for armor conversion job)

 

If this body-only conversion turn out right, I am ready to convert any armor NIF.  And it turn out not OK, it is likely to to armor's fault.

 

However ... if body-only conversion turn out not-so-good. The lattice is not usable and there is no need to try converting any armors with it, the result will be almost certainly bad.

 

BTW, I still not have a chance to try UV Search mode yet. So I don't know about its quality/precision

---

 

In your case, I believe both of your mesh should already be Vertex Index Search mode compatible since both are just CBBE with different shape ?? So there should be not need for this wrap-it-around process.

 

There might be one exception when the breast area are too different.

 

In this case, the body should be converted perfectly but some low resolution armor mesh may show some clipping, I guess.

 

I still don't know anything well enough yet .. but I am willing to help if I can. So, if you don't mind, I can take a look at your target CBBE mesh to see whether there is anything wrong.

 

 

 

 

 

so

1. start with the mesh that you want outfits for
2. use wrap
3. use mudbox to correct the surface
4. use 3dsmax for further correction
5. use 3dsmax > Wrapit for further correction
6. export from 3dsmax as nif
7. use the clothing converter suite to convert the bodymesh
8. use 3dsmax to visually verify conversion.

lol dude that's certifiable. How many hours do you think you put into that process? If you knew nothing of those programs before starting, I'm guessing 100.

 

I'm not dingin you. . . i'm just sayin... that takes a lot of motivation to do all that. congratulations on being a highly focused individual.

 

 

[Monsto Brukes]

so this is what happens with my conversion

 

source body (cbbe)  

 

target body (bodyslide variant I call FFW)

 

body converted using the above lattice 

 

outfit converted using the above lattice 

 

lattice was created using Vertex Index Search, without UV Search or Skin Only. sliders set at (top to bottom) 3, 10.0, 10

 

conversion was done using only Don't Cross X, sliders @ 3, 10.0, 1.0, female only.

 

Any suggestions? the body conversion is pretty close... breasts are too high, and point upwards more, but the big obvious problem is the  . . . hmmm would you call that "seams"? (=

 

[edit] oh yeah v85

[Alwayon]

 

 

 

so

1. start with the mesh that you want outfits for
2. use wrap
3. use mudbox to correct the surface
4. use 3dsmax for further correction
5. use 3dsmax > Wrapit for further correction
6. export from 3dsmax as nif
7. use the clothing converter suite to convert the bodymesh
8. use 3dsmax to visually verify conversion.

lol dude that's certifiable. How many hours do you think you put into that process? If you knew nothing of those programs before starting, I'm guessing 100.

 

I'm not dingin you. . . i'm just sayin... that takes a lot of motivation to do all that. congratulations on being a highly focused individual.

 

 

 

I think you have a very good point here.  May be staring at 3D nude female models all days long give me strong-enough motivation to learn these stuffs.

 

 

Ha..ha.. I agree with you.  Those are not-so-good neck, hand and feet seam. The whole body is shifted up. So the lattice in this example is not usable. My best guess now may be skeleton mismatched between source and target bodies ??

 

Have you already checked if both are rigged to the same skeleton (ie. XPMS 2.06) ?

 

Edit : BTW, I think the lattice parameter of 3,10.0,10 was a typo, right ?

 

[Monsto Brukes]

 

 

 

so

1. start with the mesh that you want outfits for
2. use wrap
3. use mudbox to correct the surface
4. use 3dsmax for further correction
5. use 3dsmax > Wrapit for further correction
6. export from 3dsmax as nif
7. use the clothing converter suite to convert the bodymesh
8. use 3dsmax to visually verify conversion.

lol dude that's certifiable. How many hours do you think you put into that process? If you knew nothing of those programs before starting, I'm guessing 100.

 

I'm not dingin you. . . i'm just sayin... that takes a lot of motivation to do all that. congratulations on being a highly focused individual.

 

 

 

 

I think you have a very good point here.  May be staring at 3D nude female models all days long give me strong-enough motivation to learn these stuffs.

 

 

Ha..ha.. I agree with you.  Those are not-so-good neck, hand and feet seam. The whole body is shifted up. So the lattice in this example is not usable. My best guess now may be skeleton mismatched between source and target bodies ??

 

Have you already checked if both are rigged to the same skeleton (ie. XPMS 2.06) ?

 

Edit : BTW, I think the lattice parameter of 3,10.0,10 was a typo, right ?

 

Not a typo. I threw those numbers in there to see if it made a difference. it did not.

 

I am going to check the skeleton.

[Monsto Brukes]

it was the skeleton.

 

I hadn't the slightest idea how to check or make cbbe have the matching skeleton, so I just made a bodyslide of vanilla cbbe which gave it the same rig/skele.

 

there is a VERY SLIGHT difference in underboob and stomach, but it's otherwise spot on.

 

Thanks.

[Alwayon]

it was the skeleton.

 

I hadn't the slightest idea how to check or make cbbe have the matching skeleton, so I just made a bodyslide of vanilla cbbe which gave it the same rig/skele.

 

there is a VERY SLIGHT difference in underboob and stomach, but it's otherwise spot on.

 

Thanks.

 

You're welcome.

 

Sometime, I also found that making lattice & converting with 1, 10.0, 0.0, especially with that Node target = 1 part, gives me a slightly more accurate result on 0.85a.

 

I still not sure why or how it work that way.

[myuhinny]

I have converted tons of oblivion armors some of the older bodies don't convert well. I have converted some outfits for fallout NV and have converted some outfits for skyrim I have even done some oblivion to skyrim body conversions where I used a skyrim body for the from file and oblivion for the to file and some came out pretty well it all depended on the skyrim body being used.

 

CBBE bodies without BBP/TBBP are the only ones I convert to 7BASE/UNP because even though you converted them to 7BASE/UNP you will still have those nasty hip marks that you get when using 7BASE/UNP body and then try to wear a CBBE armor. So after conversion I just delete the CBBE body out and replace it with the body that I converted it to.

[foreveraloneguy]

OK, been thinking about the question earlier of how to get tops to stop folding into cleavage. I don't know if this is workable, but what about comparing the angle between the vertices in the original mesh and the converted mesh, then, if the angle increases by more than $DEGREES, limit the angle change to something smaller. What I'm thinking of would work sort of like a curve in graphics software. I mean, after working with the tool a bit, it's easy to see how this happens, given that the conversion lattice is between bodies, but most armors/outfits are not skin tight so trying to follow the target body would almost have to cause artifacts like this unless something is in place to limit how much the angle of a vertice changes between the source and destination meshes. Or maybe it'll just be a limitation of the tool. I can make things better after doing the conversion by editing further in bodyslide and/or 3DS, but if I was any good with those tools, I probably wouldn't be using this tool. Which, by the way, is effing awesome.

[gerra6]

It's not a bad idea, but I don't think it will work.

 

I've been playing around with some similar ideas that would calculate the changes to the normal vectors of each face and then compare that to both its neighbors and the composite movement vectors from the lattice.

 

But I haven't yet come up with any reliable methods to filter out the false positives.

[foreveraloneguy]

I think the difference is that what I'm suggesting is to only look at the change in angle between 2 vertices in the source and destination mesh. Further, this could probably be limited to some distance from the center of the y axis and look for changes along the z axis, if the only thing you're trying to do is fix the problem with the areas around the chest collapsing inward. Something like if the original angle is 3 degrees in the source mesh and the same two vertices have an angle in the destination mesh of 80 degrees, something is probably not right there.

[kingkong]

it can be because at some regions the vertex are Inside :   in the brest region  some vertex/faces are crossing each other creating a confusing normals . and blender can not define  if the faces are Inside or outside .  some times ii do it manually inverting vertex/faces but in the brest region it is practicly impossible.

 

it might be a cause !!!

[Symon]

One suggestion I'd like to make that I suspect would be incredibly popular. A Real High Heels/De Real High Heels conversion option if such could be managed.

 

Some people adore 'Real High Heels' and some of us feel it ruins a perfectly good mesh. I'm assuming this would be the tool that might be able to do the job, if not,  I suspect another tool would be appreciated by almost everyone.

[AnnaFly]

One suggestion I'd like to make that I suspect would be incredibly popular. A Real High Heels/De Real High Heels conversion option if such could be managed.

 

Some people adore 'Real High Heels' and some of us feel it ruins a perfectly good mesh. I'm assuming this would be the tool that might be able to do the job, if not,  I suspect another tool would be appreciated by almost everyone.

Greetings.

 

 @ Symon.

 

 Not sure if i misunderstand your post , but Real HighHeels are already well converted by this tool ... so i 'm speaking about VipCxj 's Real HighHeels System ( for Oblivion , of course) , not tried with the Add1K1 ( not sure of his "name" ) system , nor with the Skyrim systyem ( never tried SKSE ) .

 

Cheers.

 

NB : ...  take a look at the picture from the post 509 , it's a real highheels model converted  ( with stockings) to this "picture-model" :

 

http://www.loverslab.com/topic/27557-new-clothing-body-style-converter-release-candidate-v072-3-30-2014/page-26

 

PS : the Windows 10 Preview OS is there now , i 'll tried this tool soon  with that new system ...

 

http://technet.microsoft.com/fr-FR/evalcenter/dn781239.aspx?ocid=wc-mscom-wol

 

[gerra6]

it can be because at some regions the vertex are Inside :   in the brest region  some vertex/faces are crossing each other creating a confusing normals . and blender can not define  if the faces are Inside or outside .  some times ii do it manually inverting vertex/faces but in the brest region it is practicly impossible.

 

it might be a cause !!!

Awhile back, I wrote a tool that can automatically identify and flip inverted face normals. I'll see if I can dig it up and post it.

 

Basically, the direction of a given face normal is determined entirely by applying the right hand rule to the vertex order of the face. When the tool detects an inverted face normal, it simply flips the order of the 2nd and 3rd vertex in the face.

 

The detection code probably needs a bit of work to make it more robust.

 

*Edit*

 

So I've given it a bit more thought and I think I've got an idea that might be an improvement to my flipped normal detection algorithm.

 

For the purposes of a right hand rule, a given triangle has two basic sequences: ABCAB and ACBAC

 

Or, to put it another way, ABC, BCA, and CAB will have identical face normals pointing one direction, while ACB, CBA, and BAC will have identical face normals pointing the opposite direction.

 

For a given pair of triangles that share an edge, say AB, I can simply compare the sequence to determine whether or not their face normals indicate the same side, or opposite sides. This eliminates the need for the fairly complex calculations that my current face normal fixing tool performs, and should be considerably more reliable.

 

Hmm. Gonna have to try that out.

[gerra6]

One suggestion I'd like to make that I suspect would be incredibly popular. A Real High Heels/De Real High Heels conversion option if such could be managed.

 

Some people adore 'Real High Heels' and some of us feel it ruins a perfectly good mesh. I'm assuming this would be the tool that might be able to do the job, if not,  I suspect another tool would be appreciated by almost everyone.

Hmm...

 

Well, the tool should be able to convert any mesh-based high heels meshes without too much trouble as is, so long as the pre and post templates use the same UV map. Note: I would not expect the default (nearest point on surface) search to provide satisfactory results, as it will map the heel to the ankle, etc.

 

With the exception, of course, of the foot. That's where things get ugly. Basically, the tool would need to somehow identify the high heel as a high heel and more or less remove it.

 

That...could be tricky.

 

It would probably require creating a template with the express purpose of morphing the shoe from a high heel version to a non-high heel version. Basically, go into a 3D modeling program and use sculpt/morph type manipulations to eliminate the heel by moving it into the shoe. Then use that and the original as pre/post templates to get clothing converter to automatically apply similar transforms to other models using the same high heel system.

[Symon]

Yes I agree, that's why I suggested it as a very nice idea if it could be done. (BTW, the default option was very unsatisfactory as you say.)

I've de-RHH meshes in Blender before and it can be a real pain, so much so that it isn't worth the time for an otherwise fine mesh.

 

Anyway, one for the long-term ideas pile.