duckydraws
patreon
Purple neko
🕑 Added 2022-07-15 05:00:03 +0000 UTCWhen I began updating the purple neko model, I wanted to address some deformation issues that were present in the original model. A crouching position, for example, had the model squash and crunch awkwardly in the butt and the area where the thigh begins.
There are a number of ways to address these, all with pros and cons, and I figured hey, let’s try them all.
One approach that I had heard of but haven’t actually seen used much was to use a specific rest-pose where the limbs were roughly at 50% angle between the expected min & max rotations. This is a stark contrast to the more common T-pose.
The logic behind this is that the model is basically “half-way” between the most common poses, which in theory should require less deformation to each direction.
To some extent this appeared to work as intended. Where there used to be odd distortions, the model had pretty nice deformation without any tweaking with extra bones or corrective shape keys.
The immediate down-side of this was the really odd angles I had to work around the arms and the legs. Managing topology at a constant 45 degree angle was an odd experience even for a model this simple. While the results were nice, I do not know if it is worth the extra hassle.
At this point there was still the question whether I wanted this model to be real-time engine compatible, or would I just use it in Blender for animations.
First option, while tempting, would require a lot more work with questionable returns: what would I do with it, and would anyone even bother downloading and running an executable for whatever it would be?
Second option would allow me to use all the tools at hand, which would allow creating content quicker and in an easy-to-share format. At this point, this sounds most viable.
This brings me to another point, namely…
To simulate, or not to simulate
In some of my later clips, I relied on various physics simulations to give the characters some boob jiggle and proper clothing that wrinkled as the character moved around. They look quite nice, but are a proper pain in the butt to create.
This clip had simulations to the wazoo. First there was a mass that controlled her tits. It was basically simulating a slightly inflated cloth that was attached to the spine. I had to be careful not to make too sudden movements because the “cloth” would then collapse into itself.
Having parts of the model controlled by a rig, and parts controlled by a simulation controlled by a rig, meant a lot of back and forth between different work states.
The fun did not stop there. Once those were done and simulated, then it was time to simulate the clothes. While I love Blender dearly, its cloth simulations aren’t quite on par with Marvelous Designer, which I’ve been using.
So this introduces a round-trip between two different software. What you do is that you essentially export the animation in a format called Alembic, which is like a stop-motion, frame by frame capture of the model in motion. These files can be gigabytes in size.
You send the animated model to Marvelous Designer, start simulating it frame by frame, and at around frame 512 you notice that, because of a small intersection in geometry, the characters shirt is now stuck inside her model. You despair a moment, go back to Blender and tweak the animation, and hope things work better next time.
But…
Even with all this hassle, it might still be worth it for that little extra it creates.
So, what does it mean for our purple boy? It’s time to explore ball physics!
The sack
I'm reusing a similar technique that I used with Emma's bust with slight alteration. I'll separate the bouncy bits into its own solid model and give it Cloth Physics with internal pressure. Then it's down to tweaking individual parameters until we get something that jiggles & retains its shape.
Bouncy boy! The parameters at this point are as follows. Don't take this like I have a clue what any of these does...
The plan is to use either this mesh as is, or to create some form of "vertex to bones to mesh" workflow. With Emma, the simulated mesh partially controlled the mesh of the character, but because I aim to simulate a layer of cloth on top of this character, I don't need to worry about keeping the jiggly bits attached to the main body.
It'll be interesting to see what issues comes up down the road. Unfortunately the only way is to try, and that'll take time.
Other things
- Tweaked the torso ratios and topology (again)
- Rebuilding fingers due to poor distortion
- Explored a "cage skinning method" to hopefully make the process quicker in other models.
That's it for the 3D updates!
Cheers!
-d
