DOAX3 approach to simulating breast movement

Firstly please don’t be a jerk just because the topic is about breasts. I get it. Some people find discussion of this stuff juvenile and distasteful. Having said that I’m curious about Dead or Alive Xtreme 3’s approach to breast physics and was wondering if anybody had any insights.

My assumption was that they’d use bones with spring physics - I’ve seen tutorials that take this approach. However I discovered that in DOAX2 they appear to have used something like shape keys. I was wondering if anybody had any ideas about how they might be creating breast movement for characters in their latest game?

Don’t know what game are you talking about but I can say using shapekeys for simulating breast movement in games (and not only in games) is barely usefull. I think such things like hair, cloth, breasts etc. in modern games are made with physic simulation in realtime inside game engine or baked to animation. Problem with shapekeys they have linear interpolation and requires lot of work and time to sculpt and set up while physic even with collisions can give good results in realtime.

Do you think that anyone will actually notice? (I mean, beyond the extent that every boy notices breasts.) :smiley: Most ladies wear bras … Be sure there’s going to be an actual return on your investment of time.

1 Like

I have experimented with this in blender. I found that a soft body simulation can work, if you have internal structure to your geometry to act as springs.

You do this by extruding edges to create internal geometry - then assign the attachment vertices (i.e. the verts that would attach to the body) to a vertex group and set them to the goal. This will lock them in place, allowing the rest of the geometry to hang off them and act according to the soft body physics settings:

There is more discussion around this approach in the following links

https://docs.blender.org/manual/en/latest/physics/soft_body/forces/interior.html

I use the same cage system but I use a Cloth sim as I find it works a lot better and is easier to get what you want, ( I have never been able to get Softbody to produce the desired results)
GIF 2-16-2022 12-35-31
Same Cage just viewing as Rendered instead of a wireframe.
physics_cages.blend (1.3 MB)

Yep - I have just tried that too and it works much better than soft body. You follow similar steps by pinning the vertex group where the booby would attach to the chest - then tweak the mass, internal pressure, stiffness, bending, damping parameters to get natural motion.

I have attached a blend file with a simple example.

Sula_nebouxii.blend (2.3 MB)

Usually game engine coders just do quick physics code to animate the bones that drives the breast geometry. Is not that hard, and in the past people have done this for older Blender versions.

This tutorial, and code is old, but if you know some python-fu maybe can adapt this to 3.xx

This video describes it in some detail (well… 30 seconds of detail):

Maybe this can give you an idea of how replicate this in Blender :slight_smile:

In Bullet physics engine, that Blender uses, there are softbody physics.

Back in the old 2.7 Blender, where it had the game engine, you would be able to mark a mesh as softbody and gain access to these features in a very easy and performant way.

Lots of people used the game engine to bake their physics since it was much more easy and faster to test.

Now with the current physics system of Blender, the workflows are not exactly robust or easy, perhaps other times is a big trouble to setup softbodies (due to lots of fine tuning needed or slow performance). So more or less people prefer working with jiggly bones instead.

However there are plans, that Blender developers will move the physics system into an interactive graph based mode in the following years, so there is hope for great improvement in that area.

Maybe they fixed it but i had problems with softbody and cloth sims on top. Fortunately they introduced internal springs and pressure at that time. Not easy to set up as softbody but it works great.
Only thing i missed was a simple dampening parameter. Its often an act of balance, back and forth between mass, tension and right amount of weight on the vertex groups and takes time but you get good results.

I also used it for fat simulation once.