Hey you know what’s better than one Maiko? just kidding. I got home late today and instead of working on ARIA, I spent some time tweaking the boob physics on the Maiko MMD model I have been editing for a while now. I am close to finishing but it won’t be perfect, simply because the physics don’t allow for big boob characters. There really no way I can see to do it where it doesn’t have tons of flaws. I will explain for those not familiar with MMD. Basically there are only 3 physics body shapes. as seen in image A.
A Square, a pill shape, and a sphere. Basically these make up all of the physics in mmd. the box can be made into a rectangle. The pill can the extended. The pill is usually used for arms and legs and the sphere is commonly used for the head and boobs (if present), and the box is usually used on hair and dresses. The pill is used for hair also it just depends on things not worth mentioning.
Basically the thing to take away from this is sphere is the tits. Two spheres to be specific, for obvious reasons. The problem will all the physics in general is that they are all rigid and don’t deform if they collide with one another and seen in Image B,
Basically the common way they are weighted to a bone is the with a smooth and even gradient. It causes the boobs to deform in such a way where they point out. I tried to illustrate below.
Since the only thing that is weighted the most is the tip where the nipple is, the only main thing that follows the physics sphere as is moves and bounces is the nipple.. The other parts of the boob does also but the pull is not as greats because the weight fades the closer it is to the torso.
Below I tried a different method where most of the boob is weighted. it has it own set of problems but the most of the boob remains in tack and follows the physics fairly well.
The problem here is that when the breast moves as it collides with other things it remains full and solid almost looking fake because they appear too solid and not deforming at all. even incorrectly.
So which is better? I think I’ll go with D because while it looks more solid then it needs to be it also look like you are going to get you eyes poked out. I made a video to show the differences. Like I said, both have problems, and you you can’t really do much about it because of the way the MMD physics are. Also keep in mind that for small boobs C works perfectly fine. And D works well in some cases. It’s when you starting getting to big ole’ phat titties that you run into deformation issues.
Update #1: Ok after doing a bit more study I think I have crafted something I like. Even did it with just the one bone. For purposes of description, I will call it “E”. It is not really a combination of C and D it is more like a modified version of D. I actually did something that at first (to me) makes no sense until you actually see the results then think about it and then starts to make sense. I put most of the weight at the top of the breasts. What it does is allow the boob to compress and expand while also rotating and moving along the axis. The drawing and video is below.
as you can see. no weight is below 50 for under the breast. It goes down to 10 just to fade out the edges.
right moder looks smoother i prefer this one it’s the C i guess
Yeah the right one is C the left one is D. I’m just not comfortable with the heaviest part of the breast being at the nipple. I’ll try a different way with two bones and see how it looks.
@Vortex00
The D version doesn’t bounce all over the place, while C is bouncing all over the place doing odd physics.
I can see why from now on you’ll go with D fixing the problem with collision.
The big question you should be, C works fine with small breast. But what is the limit of a bigger breast it’s the limit before it starts to go collision.
https://www.youtube.com/watch?v=9ynxQK_-zoU
here’s another example of C where you can see the pointy banana shape from 0:17 to 0:25
@Vortex00
I see, well the C can work with those that have a bra on dancing. I see what you mean about the banana shape.
I see why the D might be the best, it is less becoming a banana shape to it.
I guess you are still trying to perfect the MMD models.
Trying to get something like this which is closer to D. but still has a bit of deformation.
https://www.youtube.com/watch?v=Ltv9Hpclbdw
@Vortex00
Wow that is a perfect blend of C and D!
The deformation isn’t as crazy the way your video’s show the big breast models.
I think manage to make something like that. perhaps better idk.
will upload the video later today
Maiko’s boobs already amaze me every time, so I think there is not that much to fix.
I found that moving the boob joints further to the back, while leaving the rigid bodies in their original position looks good with some models. It reduces the rotation on the X axis, and adds more linear up/down motion.
The second idea that comes to mind is the setup in iei’s models. (ieinote on bowlroll) It might be too realistic looking for your girls (no excessive bouncing) and it’s a little more complicated to build. He uses a setup where the main boob bone is attached to the chest with 2 support bones, forming a parallelogram, so there are 4 joints. If the support bones are the same length it completely suppresses the upward rotation, resulting in a linear upward motion with relatively constant boob volume, but normally you would make the top support bone a little shorter to keep a little rotation. Additionally to the main rigid body it’s necessary to add 2 dummy rigid bodies for the support bones. I once re-built this setup from scratch for a model and it took me about an hour.
An hour is not too shabby. I been working on just boobs alone for the past 3 days off and on, experimenting and trial end error. I tried that method you mentioned yesterday by borrowing bone structure joints and physics from a arlvit model and yeah I noticed that I could achieve a “leaning forward” angle that is not possible by having a single horizontal bone controlling everything. However dealing with 4 joints was a pain plus I had to keep in mind how the axis is different for the vertical bone. And one of the joints would not even budge not matter what I did to it (the bottom forward joint). But those physics did have my original idea that the bulk of the weight should be at the bottom not right on the tip where the nipple is. So I was happy to see that.
However now I think it is actually backwards. In my current setup I have the bulk of the weight at the top with a gradient fade downwards at a angle. I’ll have to draw it later and add it to this post. Then a another fade towards the opposite angle. Then blend/fade and hard edges.
what is produces is a heavy weight paint at the top mid section of the boobs which I assume would only work on big boob characters. It allows the breasts to expand and contract and well as rotate without sagging too low because the bottom has the least amount of weight. I should be fairly noticeable in the updated videos I posted at the bottom of the post.
@Vortex00
Well the breast of both go there own ways, it isn’t like D where it’s all over the place. And it is not doing the banana effect. I’ll say this E is the best you have.
About the second video, I thought for a minute there Maiko was going to break dance. LOL
I notice that Maiko’s ass doesn’t bounce enough, than again it could be that she have no ass. Just a little meat at her back. Anyways since your practicing physics, careful than the ass chicks in future models don’t go all over the place. The way you are now dealing with the D physics. If anything if you plan to do more ass quake (LOL) I think testing in small ass, mid ass, big ass with chicks bouncing to see what would make the best effect of what your trying to do on that ass. LOL
Now the testing should be how is this new E physics work with small breast. To see if it’s all good on that. If so than C and D models will probably not be needed anymore.
Ass physics is a whole problem to itself. The main problem there (at least for me) is you can’t weight “paint” vertices that can be influenced by more then 2 bones.
Basically for the ass physics to work they way I want I would need to have the weight to be controlled by the hip bone, the ass bone for the jiggle, and also the leg bone. For right and left ass cheeks respectively.
There is a way to manually assign each vertices up to 4 bones, but you have to do each vertices individually and that is big hell no for me. I might try it to it in places where the edges don’t blend correctly.
Yeah Maiko does not have a huge ass but she has a little junk in the trunk back there by description. I’m sure it would be fine if I allowed more moment in the horizontal directions left and right, and also decrease the spring constant a bit. But I’m sure it’ll look like shit. Maybe I can try to allow more rotation. and decrease the spring only on certain axis.
“Maiko – What the hell” videos comes to mind to show the problem of only having two vertices, there is a issues where the ass kind of disconnects from the leg, you can see it start at about 0:44 and it happens around 0:47 or so.
http://spiralvortexplay.com/svp/2014/02/04/mmd-what-the-hell-maiko/
Then there is the Leeina model I was working on a while ago that comes to mind. Not much work put into Leeina just yet aside experimenting with ass physics.
The ass is a bit too big in the that model even for Leeina (but only by a bit lol)
basically the more “meat” there is to work with the easier it is because there is a less of a need to make it controlled by more than 2 bones.
http://spiralvortexplay.com/svp/2014/01/30/mmd-work-that-monkey/
Luma has a bigger ass than Maiko so it maybe I will have to do a bit of extra effort. But I plan to make everyone in UMCC with the same body as Maiko first. which is Zytra, Roise, and I think even Stacy, Luma’s sister, if my memory is correct, and Annie. I’ll just be bikini clad Annie, I’ll worry about her dress later. just need to make the necklace, it might give me problems because it sits between her boobs. And I think one or two npc characters have big boobs also.
@Vortex00
I see what you mean about the disconnects from the leg. I get the feeling ass physics is really going to be a challenge.
I almost forgot about the Annie white dress with the necklace! Holy shit man it’s like you can’t get a break out of the problems with MMD. But still it will be worth it when you start using them for animated series.
@justjoe2k
speaking of animated series I have a test motion I am using to kinda get a idea about how it would look if it was actually kinda good lol. I should be done once I get a few random guy student models done. I’ll work on that today. I also found some cooler looking hair for Kyle.
you can see various version of it here:
https://www.youtube.com/results?search_query=Requested%3A+Kobato
Of course I am using one of the the school areas I showed in the video a while back and not that brick town or whatever. and the little man is going to be a sentinel. I guess the red one Maiko has in her room in UMCC. And I guess the guy that saves the day will be Greel. Well that is how I have it so far.
And I’ll have a few things going on the background. Like Lynn and her bodyguard. and maybe a bunch of boys around Luma. and Kyle with his camera. things like that. And I think I found a butterfly. so it is all good.
Select all vertices (in the main window, not the editor window) and change to BDEF4. Then do your weight painting. Then normalize your vertices to downgrade the BDEF on every vertex you didn’t add an extra bone to.
Keep in mind that your transformation window is going to be useless laggy until you renormalize your vertices. You might want to consider separating the relevant vertices into a different file for this purpose, not sure if that would give you a usable transformation window.
Note that with this, you can do the breat you want, but it’s ridiculously complicated. You use multiple rigids for each breast: probably a tetrahedron of spheres, connected by joints strong enough to maintain the relative shape of the breast. But this is a tough weighting job in PME, you can’t get away with a linear gradient anymore.
Realistically, the trick to rigging different kinds of jiggles is less about the joints and more about the riging and weights. I think it’s probably unwise to aim for BDEF1 on any vertex.
One of the reasons collision is such a problem is because bone rigids are unstoppable juggernauts that will relentlessly power through any physical rigid. The overly complicated answer to that one, of course, is to create invisible physical proxies for your colliders to give them a bit of give.
Oh, I understand what you mean. Fixing it by editing the weightmap that way is pretty clever. I often played around with the rigid body setup, but never thought about changing the weight that way. Well, weight in this case is a bone strength scaling value not physical weight, so having more weight at the top makes perfect sense. I guess the terminology can be misleading xD
In the new triple comparison video the left two Maikos really look better. I think the middle one looks best most of the time. When viewing the extreme motion starting at 17 seconds frame by frame the left one almost has the boobs wrapped around her neck while the middle one’s look perfect most of the time. The middle one also better recovers from the downswing while the left one’s sink into her belly.
Yeah you noticed all the problems I saw with the left model physics. By having most of the vertices weighted at 100% deformation was basically impossible, I just had to kinda fake it. yeah by having them rotate into her stomach XD
yeah it made more and more sense as I thought about it. a bra or something supports mainly by holding them up from the bottom’ish region, a pushup bra for example, so there is less jiggly movement at the bottom and should be less weight because it doesn’t move much. so it makes sense to have the most weight at the top anyways. It could probably be better but I can check boobs of the list of things bothering me about the model.
Lupin is really a good motion to test this. those chest pumps are pretty extreme.
I tried to also fix her hair from slinging around all crazy at times, it was another thing kinda bothering me, but I’m just going to leave it as is. I have a few guess at what the problem is but even still I would leave as it is anyway even if that was the problem.
going to take a look as the ass physics then finish making the facials.
@Vortex00
I see those videos look interesting can’t wait when you finally start beginning to show a animated series in MMD form.
now I just gotta find the auido, or just extract it from the video might be faster then looking.
Physics are looking awesome in Streetz! Can’t wait to see Maiko in and out of some different outfits!
Thanks~
Yeah it should be interesting. I had planned to update a few more character first in this bikini. Mainly Zytra and Luma and other voluptuous characters. Then work my way down to smaller size characters.
updated this with the picture
Well that explains what is going on in what form the animation is pushing the breast.