Clothing/swimsuit modeling questions
aaráribel caađo
Posts: 690
I coudn't find exactly the swimsuit I wanted, so decided to model one. My start was fine—I used an downward-poinging pentagon, shoved verts around, and then extruded down from teh bottom to create the crotch and back. In Modo, this is fine, but I think it's causing problems with simulation because it has so much geometry in the crotch, where the swimsuit narrows almost to nothing. After simulation, it looks like I get poor Genesis 8 a wedgie. Also, even with mesh smoothing, its kind of jenky. That doesn't show up in renders, but it annoys me.
Experienced clothes makers, is there a better approach to the geometry? I've only made pants, shirts and dreses before, usually starting with a cylander, so this put up a plane and model from there is a new approach for me?
Related question: is there a good technique or tutorial for adding dimension to things that will get dForced? Modo has an easy tool for adding thickness (the Thickness tool!), but I assume I'd run into problems with two closely spaced layers even with autofit and definitely with dForce. But maybe, hopefully, I'm wrong.
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
Just out of curiosity, what does the base mesh look like before dForce simulation? You might be able to use the "paintable" tools aspect to control the amount of dForce application (I just forgot what this is called in D|S, but pinning verts groups in Blender). Also, looking at the UV map, you might also want to relax the UV's, especially in the crotch area, as this will really warp any textures applied to the swimsuit.
Thanks,
-David
The crotch area, especially at its narrowest has far too much geometry; geometry should ideall be uniform in size or change size in a gradual way. There are times where you don't want this, and it will give results you're looking for. You only need enough for the result you're trying to achieve.
In my simple example with a plane, you'll see how 3 geometry lines are merged together to give only one (I could also have merged the three centre lines, but I was using mirrors so ignored it; this raised a point though, it isn't necessary to merge them all but miss some to reduce the amount of geometry just enough for one's needs.). Now where you do this can be varied and comes with experience - and just what the shape will allow: or suggests.
My final comment: does the swimsuit behave as you wish it too, and would expect from a real-world piece of cloth? If so, leave it the hell alone. Job done - unless you have plans to sell/give away or fit to multiple models with various shapes.
RainbowLight's One-Piece Thongs at Renderosity look a lot like that. You might want to compare.
This is a conforming, rather than dForce garment, but it comes with a jillion JCMs and MCMs.
In my experriende dForce doesn't like too dense mesh. at least not if you have self collision enabled.
So in my view I would go for more evenly distributed vertexes.
Thanks everyone. I think I'm going to watch a video or two on the Retopology pen in Modo, then retopo it to radically reduce the geometry in the crotch. Then I'll probably make a weight map for dForce to ignore the groin, since that shouldn't change with movement anyway.
I'd like to model some thickness to the edges and the straps, like the Godess Bikini in the attached image, but I'm not sure how that will dForce. Does anyone know of something that has similar geometry that works with dForce? Other than the crotch/crack issues, this looks okay, but I have a feeling close ups won't be great.
Hello Aararibel,
I have some examples from Hexagon shown below. The first picture is an example neckline similar to your model, note it is 46 polys. The second picture shows the polys in which I will extrude to create the "thickness" of the straps/hems. The third picture shows the extruded surfaces, creating the depth of the neckline/straps. The fourth picture shows 2 levels of subdivision- note that it is 1600 polys! The fifth picture shows the mesh with smooth shading... not bad, but we can do better. I'll post some more shots below using a different technique.
-David
The problem with dForce is it is rather bad at keeping shape.
What I do is I add an extra row of polygons at the edge which I angle 90 degrees, and then make the 2 rows of polygons (the edge and the bend) its own material zone and in that I set buckling stiffnes to 100 and buckling ratio to 0. And add smoothing.
That can usually make it look like thickness.
Use the dynamic surface add-on.
You need to create a bridge between two sets of verts.
Think of the U-shape example below as a cross-section at the 'base' of your garment; it then curves back, and then up. If a simulation is performed then the geometry would just drop, achieving a slight lengthening of the garment.
The cross-section is named cube; the interior part is cube.001, and this is what the dynamic surface add-on would be applied to. It is created by selected two edges and filling them / bridging them - however your chosen software does it. The vertices need to occupy the same space as the corresponding vertices in the garment.
What this would do, is hold the shape of the curve; you might have to experiment a bit to find the best blace to connect the two parts together, but it works fine. I've also done the same as Starkdog and created morphs to add depth.
For me, infinitely thin cloth kills believability.
Hello Aararibel,
Here is another example... First, I take the base mesh and apply 2 levels of subdivisions. The second picture, I highlight extra edges to remove in the neckband. Now we can see the simplified mesh in the 3rd picture. In the fourth picture, I extruded out the neckband. I then selected all the outer edges, and added 1 level of bevels to "round-over" the edges. In the next 2 pictures, you can see the poly count of 932, which is way less than 1600 in the first example. One could add additional bevel steps and do some welding to eliminate tris and poles, but this should be sufficient for a dForce simulation.
-David
Thanks, everyone, that was helpful. I'm not sure it's a fully robust solution, but I was able to get this to work without blowing up (in my limited testing) by using your modeling technique plus a dForce weight map that removes simulation from the bottom strip where it's not needed. Unfortunately, I'm not gettng the main effect I wanted—defeating the autofit pulling fabric under the bust into the body, but that might come with more tinkering. I haven't played with keeping just the two outter edge loops as a separate material zone wth buckling dialed out yet. I'll try that next.
For the curious, I'm attaching the new UVs, model and a quick render of it in motion with dForce.
When I first learned modeling, over a decade ago, "non-manifold geometry" (aka, where you have a T-intersection) was considered absolutely forbidden and guaranteed to make everything break. So when I'm modeling, I stick rigidly to quads and manifold geometry (and try for regular geometry—I look aghast as a lot of Marvellous Designer to Daz geometry, even though it works well). But it looks like whatever technical problems non-manifold geometry caused have been solved. Do you use that bridge technique regularly?
Yup.
Non-manifold will affect geometry; the main reason it's best to avoid if possible, is where the poles have more than four, and definitely more than five lines joining on to them; it is best to have them in flatter areas, which will allow the mesh to deal with them more. It is impossible to model effectively without them.
I've been modelling a long time; I've never seen anyone say don't use them, but avoiding high pole-count is something I've seen a lot.
BTW, the waist area of your swimsuit, I'd consider removing an edge-loop or two; they're a bit close. Unless of course you want the effect it will give.
It still has problems, but the thing with a dynamic dForce add-on is that it is actually a separate object, and only treated as the same object for the purposes of the dForce.
As far as getting depth and detail to dForce geometry without forcing those areas to be static, I've been working on that for some time, and think I'm finally starting to crack the formula - the underwear set in this image is entirely dynamic, including the seams, underwire, and those bloody rings that need to simulate as part of the straps rather than just following them...
The fundamental trick is adding the details after simulation, rather than before.
Yup, I create morphs to give depth, and dial them in after the sim
Are you leaving details flat in the mesh before simming? I'm not sure what the morphs do, at least since I thought morphs have to have the exact same polys.
In annoying news, I made some small changes to my mesh, including dividing the outer edge loop into 2 rows, but simplying some geo elsewhere, and now it blows up immediately. Grrr…
Wow. That's amazing. Whenever I've tried simming something that goes through a hoop, it's prone to slipping out. I made some very basic curtains recently and only about half of the holes actually stayed on the rod. The rest simmed loose. I ended up changing my camera angle to hide the tops.
Don't dial in morphs that add depth to seams before symming as they will just flatten, particularly in places where there is pressure, such as the tops of shoulders.
Once the sym has run, dial them in to get the effect you're after.
Wow that looks fantastic... also looks like a bra most people under the age of 40 wear. A lot of 3d bras seem oddy fussy to me, Its all full cups and garters and meanwhile out in the real world we're in the bralette renaissance
Real Bras have a number of demographics with various groups of folks interested in them.
3D bras - mostly guys!
It's hard to express how glad I am to hear that, because that's basically the main reason I started on the project - I just wanted my characters to have more choice of normal everyday underwear. So, actually hearing people complementing not just the quality but the actual design is exceptionally reassuring.
As is, I'm not yet finished, but I do hope to try to submit it to Daz as a possible product once I've got it refined.
I've been getting a decent amount of feedback and interest from women on this project, as it happens. Which has been very much appreciated - aside from just boosting my confidence, it's also helped me a lot with bringing up the quality and accuracy.
Well, I don't expect to hear from women who want normal, everyday swimwear instead of skimpwear, but I did finish my project, which can be found here as a freebie. In the end, I failed miserably at depth. I tried a number of things and all of them ended up with blow-ups or DS stalling before starting the simulation However, close fit and texturing hides the worst of it. I'll be really excited to see Matt's project when he finishes it!