Clothing Bodytype Converter V.5.0 (12.24.2013)

[rad87]

 

DMRA to anything smaller is always tricky. I have yet to crack it to my satisfaction. In fact' date=' if I do create a 3.4 release, it will probably be to deal with this specific issue, but I don't have any workable ideas at the moment.

 

[/quote']

 

And most of the cool things (at least the ones I found) are made for DMRA >.< I like big boob but DMRA is just huge, too much for my taste.

 

Here's a test update of the script that adds the option to use BB Butts weightpainting to aid in the generation of quality lower body transformations.

 

It is primarily designed to work with Bip01 Op3.R and Bip01 Op3.L bones, but any breast bone styles that contain a "3" should work. It also should be compatible with any full body meshes, or upper and lower body meshes that have been joined. (I've added code that should be able to tell the difference between lower body OP3 vertex groups and upper body OP3 groups)

 

The net effect should be to give you quite a bit of additional flexibility to target vertices for improved butt transformations.

 

As an example, I imported two lower bodies with BB Butts weightpainting. Then:

 

 

 

Which Gives

 

 

 

I will try it out today and share results later ^^

[rad87]

just got an idea. remember that cleavage problem? would it be possible to treat both halves (left and right) of the body separetly so that movement in one direction isn't canceled out by the movement in the other direction. so like making 2 separate latices and merging them together after they are calculated?

[gerra6]

just got an idea. remember that cleavage problem? would it be possible to treat both halves (left and right) of the body separetly so that movement in one direction isn't canceled out by the movement in the other direction. so like making 2 separate latices and merging them together after they are calculated?

 

Probably not. Any lattice will apply its changes to the entire mesh. There is no way that I know of to make the lattice apply its changes only to part of the mesh. ***Edit*** That's not entirely true. One option is to use vertex groups to define which vertices are affected by the lattice. If you type in the name of a vertex group in the VGROUP entry in the lattice modifier, it will only affect vertices that are members of that group.

 

The problem with that approach, unfortunately, is that you have to assign all of the vertices that you want modified together into the same group.

 

Could it be done? Maybe. Let me give it some thought. I have an idea that might work. It essentially would boil down to generating two lattices, one for the left side that is applied to an automatically generated vertex group on the left side of the mesh, and one for the right side that does the same.

 

We'd also need a mini script that would generate those vertex groups on any clothing prior to conversions.

 

Of course, this would wreak havoc on meshes that have breasts that cross the X axis.

 

I'll give it some thought and decide whether or not I want to tackle this.

 

***End edit***

 

 

Now, for meshes that are symmetrical, you can chop em in half, transform that half, duplicate, mirror (several steps here), and then merge it. That would probably work, but I'd generally recommend against it. It's a lot of work for very limited reward.

 

The key to keep in mind with automatically generated lattices is that they do about 95% (depending on the meshes) of the transformation work for you. The remaining 5% should just be minor detail tweaks to get the look that you're looking for.

 

You have two options. One is to manually edit the lattice. With careful use of proportional editing and node selection, you can sometimes improve on the generated lattice. But, and this is important, the same fundamental limitations to what a lattice can accomplish will still apply.

 

The other option is to manually tweak the clothing after it has been transformed by the lattice. This is a good practice in general.

 

Here's a quick primer on vertex editing.

 

1. To edit transformed clothing, first apply the lattice.

2. Once its changes have been made permanent, hit tab to enter edit mode.

3. Let's take a moment to set up a few necessary tools:

 

3.A. Transform Orientation:

* Click on View and Select "Transform Orientations"

* Note the four buttons (Global, Local, Normal, View)

* Each button defines a different axis of transformation. This is extremely useful.

 

3.B. Proportional Edit:

*Look for a target symbol, to the left of the magnet. (or hit o, or alt-o)

*When you select one of the two On options, a new drop down should appear.

*The options in the new drop down menu define different falloff algorithms when using proportional editing.

*When performing a transformation (like grab) while proportional edit is active, scrolling your mousewheel up or down increases or decreases the affected area

 

3.C. X-Axis Mirror

*Hit 'tab' to enter edit mode

*In the button pane, look for an entry "Mesh Tools More"

*Ensure that "X-Axis Mirror" is enabled

 

4. Select one of the problem vertices

5. Set Transform Orientation to View

6. Rotate your view until one of the axes (say, Z) is directly in line with the direction you want to move the vertex

7. Hit 'G' and then 'Z' to move the vertex along the Z axis.

8. Use your mousewheel to define the area you want affected by this movement.

 

That should get you started.

[rad87]

1. To edit transformed clothing' date=' first apply the lattice.

2. Once its changes have been made permanent, hit tab to enter edit mode.

3. Let's take a moment to set up a few necessary tools:

 

3.A. Transform Orientation:

* Click on View and Select "Transform Orientations"

* Note the four buttons (Global, Local, Normal, View)

* Each button defines a different axis of transformation. This is extremely useful.

 

3.B. Proportional Edit:

*Look for a target symbol, to the left of the magnet. (or hit o, or alt-o)

*When you select one of the two On options, a new drop down should appear.

*The options in the new drop down menu define different falloff algorithms when using proportional editing.

*When performing a transformation (like grab) while proportional edit is active, scrolling your mousewheel up or down increases or decreases the affected area

 

3.C. X-Axis Mirror

*Hit 'tab' to enter edit mode

*In the button pane, look for an entry "Mesh Tools More"

*Ensure that "X-Axis Mirror" is enabled

 

4. Select one of the problem vertices

5. Set Transform Orientation to View

6. Rotate your view until one of the axes (say, Z) is directly in line with the direction you want to move the vertex

7. Hit 'G' and then 'Z' to move the vertex along the Z axis.

8. Use your mousewheel to define the area you want affected by this movement.

 

That should get you started.

[/quote']

 

I'm already doing editing after applying the lattice. But the problem is that the part which is hardest for lattice is also hardest to adjust. If i had to chose I would prefer to have the brest area closest to perfect and the rest not so perfect. As far as assigning vertices to groups go I wouln't mind doing that myself if it would produce a better result, and that left-right approach could be optional in the script.

 

Thanks for the tips. I'm still a blender noob so most of them where new to me and I'm sure they will ease up the work.

[gerra6]

I'm already doing editing after applying the lattice. But the problem is that the part which is hardest for lattice is also hardest to adjust. If i had to chose I would prefer to have the brest area closest to perfect and the rest not so perfect. As far as assigning vertices to groups go I wouln't mind doing that myself if it would produce a better result' date=' and that left-right approach could be optional in the script.

 

Thanks for the tips. I'm still a blender noob so most of them where new to me and I'm sure they will ease up the work.

[/quote']

 

OK, this revision is *highly* experimental. I make no guarantees.

 

It adds two new options (Left Only, Right Only). If you enable one of them, it will generate a single lattice for that side. If you enable both, it will generate a lattice for the left side and another one for the right.

 

*edit* I'll upload a new version in a bit. There's a new bug with the breast bones that I need to track down.

 

**Edit 2** 3.4b Test updated

 

The X-Axis cross rule is refusing to behave. I'll have a 3.4.c Test version uploaded once I figure out why it's broken.

 

**Edit 3** 3.4c Test updated

 

The X-Axis cross rule should now be functional.

[rad87]

OK' date=' this revision is *highly* experimental. I make no guarantees.

 

It adds two new options (Left Only, Right Only). If you enable one of them, it will generate a single lattice for that side. If you enable both, it will generate a lattice for the left side and another one for the right.

[/quote']

 

if this script gets into the wrong hands we can expect some realy weird body modifications here

 

 

[gerra6]

OK' date=' this revision is *highly* experimental. I make no guarantees.

 

It adds two new options (Left Only, Right Only). If you enable one of them, it will generate a single lattice for that side. If you enable both, it will generate a lattice for the left side and another one for the right.

[/quote']

 

if this script gets into the wrong hands we can expect some realy weird body modifications here

 

 

 

OK, this revision should resolve a slight asymmetry problem.

 

*scratch that...something very irritating is going on*

 

OK, it should work properly now. However, in order to avoid evil along the X-axis, X-Axis vertices need to be added to both vertex groups with a weight of .5.

 

I've added a mini-script (create lattice groups) that creates the left and right vertex groups at the correct weighting with a minimal of fuss.

[rad87]

OK' date=' this revision should resolve a slight asymmetry problem.

 

*scratch that...something very irritating is going on*

 

OK, it should work properly now. However, in order to avoid evil along the X-axis, X-Axis vertices need to be added to both vertex groups with a weight of .5.

 

I've added a mini-script (create lattice groups) that creates the left and right vertex groups at the correct weighting with a minimal of fuss.

[/quote']

 

The asymmetry seems to be fixed now

 

 

Since now we can work on both sides separately do you have any thoughts how to deal with that cleavage issue?

[gerra6]

OK' date=' this revision should resolve a slight asymmetry problem.

 

*scratch that...something very irritating is going on*

 

OK, it should work properly now. However, in order to avoid evil along the X-axis, X-Axis vertices need to be added to both vertex groups with a weight of .5.

 

I've added a mini-script (create lattice groups) that creates the left and right vertex groups at the correct weighting with a minimal of fuss.

[/quote']

 

The asymmetry seems to be fixed now

 

 

Since now we can work on both sides separately do you have any thoughts how to deal with that cleavage issue?

 

*shrug*

 

It's hard to say. I'd probably start with different variations of lattice targets. From there, I'd maybe mess with different search weights.

 

The issue is, in part, that we've got roughly a 90 degree angle between the cleavage and the chest at that spot. I suspect that the sharp edge is related to the relevant vectors adding together and producing ugly.

 

The trick is probably going to be to find a sweet spot in the lattice node number (the number of vertices that influence each node).

 

I recommend setting the node multiplier to something relatively low (say 5) and experimenting with lots of different settings. Once you find a setting that looks pretty good, generate your high quality "proof" lattice with a nice high node multiplier. (anything above 10 or so generally gets you to the node cap).

 

In this case, one possible solution is to increase the X-Axis Search Weight.

 

Something like this:

 

 

 

Which produces this:

 

 

[gerra6]

Create Transformation Lattice V.3.5 is now live

[gerra6]

Create Transformation Lattice V.3.6 Beta is now live

[rad87]

Create Transformation Lattice V.3.5 is now live

 

do you sleep at all or just code for 24h a day? xD

[movomo]

Woot~ updated!

 

Comment first, install later!

[Nessa]

In the interest of helping things along, it might be a good idea compile a list of decent settings for different body types.

 

Most of the time I don't have much trouble as I don't go between vastly different bodies. However I had one conversion that went from LA sized girls to TGND and naturally me being me, I forgot to snap a pic of my previous settings. So now that I'm back to doing "normal" conversions I'm starting from scratch.

 

No bugs encountered on the new test one so far by the way.

[gerra6]

In the interest of helping things along' date=' it might be a good idea compile a list of decent settings for different body types.

 

Most of the time I don't have much trouble as I don't go between vastly different bodies. However I had one conversion that went from LA sized girls to TGND and naturally me being me, I forgot to snap a pic of my previous settings. So now that I'm back to doing "normal" conversions I'm starting from scratch.

 

No bugs encountered on the new test one so far by the way.

[/quote']

 

Are you saving the lattices once you create a conversion that you like? You should be able to reuse the lattice for any other conversions between those same body types (most of the time, at least).

 

Hopefully, the new "assign lattices to mesh" script should streamline that process a bit.

[movomo]

In the interest of helping things along' date=' it might be a good idea compile a list of decent settings for different body types.

 

Most of the time I don't have much trouble as I don't go between vastly different bodies. However I had one conversion that went from LA sized girls to TGND and naturally me being me, I forgot to snap a pic of my previous settings. So now that I'm back to doing "normal" conversions I'm starting from scratch.

 

No bugs encountered on the new test one so far by the way.

[/quote']

 

I think so, it's a good idea. We can just create some kind of library thread and share each other's .blend file. When someone gets a decent result -> then they upload it, and OP updated. hmm.. too many combinations??

 

btw I find this isn't very satisfactory for vanilla to preg type because it seems beth didn't care much about match between nude body and clothed body. I'm testing new version, hope it'll make better results this time.

[gerra6]

I think so' date=' it's a good idea. We can just create some kind of library thread and share each other's .blend file. When someone gets a decent result -> then they upload it, and OP updated. hmm.. too many combinations??

[/quote']

 

I think that's a great idea. But, as you've noted, the number of potential combinations is vast.

 

That said, I suspect that many folks are going to be interested in similar transformations, so there is definite value is saving, sharing, and improving any good lattices.

 

btw I find this isn't very satisfactory for vanilla to preg type because it seems beth didn't care much about match between nude body and clothed body. I'm testing new version' date=' hope it'll make better results this time.

[/quote']

 

One trick that might help is to set your lattice search distance to the maximum. With any luck at all, that *should* cause the lattice movements to propagate outwards in a manner that gets a better morph for even the vanilla clothes. This works best if you also increase the lattice scale (not much, just enough so that the base lattice is slightly larger than the clothes that you will be converting)

 

Another option is to get tricky with the source and target meshes that you use to generate the lattice.

 

One idea would be to generate a lattice using one of the sets of clothing that you want to convert as the source mesh and the pregnant body as the target mesh. Then, if necessary, get creative with the VGroup field and custom vertex groups.

[Nessa]

 

Are you saving the lattices once you create a conversion that you like? You should be able to reuse the lattice for any other conversions between those same body types (most of the time' date=' at least).

[/quote']

 

Normally yes. But in this case I've been going between what I'll call "wonky" bodies. They're strange variants of HGEC and are mostly the same but they all seem to be different in odd ways. Really it's entirely my fault for not getting a screenie of my previous settings (that essentially worked for -everything-). One thing I might do is actually dust off the printer and print out your descriptions of the settings again. If I've been working on something else for awhile and come back I can't remember a thing!

[gerra6]

 

Are you saving the lattices once you create a conversion that you like? You should be able to reuse the lattice for any other conversions between those same body types (most of the time' date=' at least).

[/quote']

 

Normally yes. But in this case I've been going between what I'll call "wonky" bodies. They're strange variants of HGEC and are mostly the same but they all seem to be different in odd ways. Really it's entirely my fault for not getting a screenie of my previous settings (that essentially worked for -everything-). One thing I might do is actually dust off the printer and print out your descriptions of the settings again. If I've been working on something else for awhile and come back I can't remember a thing!

 

I really have way too many damn settings. I wish I could figure out a way to get rid of most of them and still have the tool work for some of the wonkier meshes.

[rad87]

I really have way too many damn settings. I wish I could figure out a way to get rid of most of them and still have the tool work for some of the wonkier meshes.

 

how about the one not needed for most of conversions optional. like maybe not visible until you click on a button.

[gerra6]

I really have way too many damn settings. I wish I could figure out a way to get rid of most of them and still have the tool work for some of the wonkier meshes.

 

how about the one not needed for most of conversions optional. like maybe not visible until you click on a button.

 

I've thought about it. Blender seems to be very particular about the sorts of pop-up windows that you can draw in the 3d pane. I can replace the 3dPane with a full screen menu that I have full control over (like the nif import/export menus) but I haven't had much luck generating custom pop-up menus.

 

I can do sequential menus, like I did with the "Test" button, but that will probably just replace one confusing mess with another.

[movomo]

Sir, I'm getting hopeless results with this version. Especially arms and legs.

 

 

 

 

With 2.3b script,

 

but when I use 3.6a script

 

 

Test Setting - OP1 L/R

 

 

When I increase the arm ring or leg ring, it's totally screwed.

 

 

[gerra6]

Sir' date=' I'm getting hopeless results with this version. Especially arms and legs.

 

 

 

 

With 2.3b script,

 

but when I use 3.6a script

 

 

Test Setting - OP1 L/R

 

 

When I increase the arm ring or leg ring, it's totally screwed.

 

 

 

That's pretty ugly. Are those two separate meshes that have been joined, or a single fully merged mesh? The waist ring settings currently only work for meshes that have been fully joined.

 

For starters, though, try a higher lattice node targets number (~25 seems to be good most of the time). I'd also recommend enabling "Vertex Override" just to make sure that every vertex finds a target.

 

The Y search bias of .5 is pretty high, but reasonable with the belly there.

 

That said, I've definitely introduced a number of very ugly bugs with the 3.6s beta. I've been wrestling with them in my spare time all day. For the moment, the 3.5 version is the last relatively bug free version available.

[movomo]

 

 

Sir' date=' I'm getting hopeless results with this version. Especially arms and legs.

 

 

 

 

With 2.3b script,

 

but when I use 3.6a script

 

 

Test Setting - OP1 L/R

 

 

When I increase the arm ring or leg ring, it's totally screwed.

 

 

 

 

 

That's pretty ugly. Are those two separate meshes that have been joined, or a single fully merged mesh? The waist ring settings currently only work for meshes that have been fully joined.

 

For starters, though, try a higher lattice node targets number (~25 seems to be good most of the time). I'd also recommend enabling "Vertex Override" just to make sure that every vertex finds a target.

 

The Y search bias of .5 is pretty high, but reasonable with the belly there.

 

That said, I've definitely introduced a number of very ugly bugs with the 3.6s beta. I've been wrestling with them in my spare time all day. For the moment, the 3.5 version is the last relatively bug free version available.

 

It was joined and merged (via remove doubles command) object.

 

I got a little clearer result with 25 or higher Node Targets Setting. Thank you, but I find still 2.3b is better than 3.5 or 3.6b for pregs. I'll stick with 2.3b's outcome for now. My case is somewhat special, as I'm not very interested in breast or butt quality.

[gerra6]

 

It was joined and merged (via remove doubles command) object.

 

I got a little clearer result with 25 or higher Node Targets Setting. Thank you' date=' but I find still 2.3b is better than 3.5 or 3.6b for pregs. I'll stick with 2.3b's outcome for now. My case is somewhat special, as I'm not very interested in breast or butt quality.

[/quote']

 

With any luck at all, the 3.7 version should give cleaner results on pregnant meshes. That said, 2.3b shouldn't be giving better results.

 

I did some pretty major changes to the improve the efficiency of the back-end search algorithms. Some of those changes are currently active even when "Fast Search" is disabled.

 

I'll try re-enabling some of the less efficient (but possibly more accurate) code for disabled fast searches for the 3.7 release. It will be slower, but no slower than 2.3b.

 

*edit* So far it looks good. It's much slower that the new code, but the detail work is coming through much more cleanly. I think I'll need to take a good long look at my fast search algorithm to see if I can reduce the errors (probably save that for the 4.0 release).