velocirection
patreon
Kobold Jiggle WIP (and cubes)
🕑 Added 2021-12-09 08:22:00 +0000 UTCThere's a bunch of different soft body objects on the kobold! 2 for left/right [B]reast, 1 for lower tummy, 1 for upper, 2 for left/right upper arm, one for the butt (but I might one per cheek this time)
The imp has some for the butt, probably more for the rest of body later, but I'm already getting sick of messing with soft bodies >_>
I've also made a bunch of minor improvements to the character movement- I'll share that once I fix all the problems. I really wanted to see what the soft bodies would do though... I'd like to do a nice preview with the soft bodies and maybe some fluids.
So many things effect soft body simulation... Like the arm soft bodies kept bugging out and then I read that the scale effects the simulation, which... made me do some tests. (All cubes were baked with the same physics settings)
Here's three jello cubes that are all the same exact scale:
It seems like each one behaves almost exactly identically, it's hard to tell if any difference is due to how the viewport draws the wireframe.
Here are three cubes, but the left one is scaled at 1.00, the middle one is 3.00 and the right one is 0.5.
They all behave differently, which makes sense- a larger soft body will behave differently than a smaller one in the real world.
These cubes all have a scale of 1.00. Bigger/smaller cubes are using a normalized scale.
Once again, they behave differently, which makes sense.
Take a look at this gif:
The small cube is barely noticeably different. This is toggling between normalized scale cubes and cubes with varying scale. The medium and large cubes however are extremely similar (or exactly the same)
This is both sets of cubes, their simulations playing back at the same time.
Any flickering would mean that there is a difference between the sets of cubes. There's not any flickering on the medium cube (both have the same scale of 1.00). There's some flickering on the large cube (one has a scale of 1.00 and one has a scale of 3.00). And, on the smaller cube, there's quite a lot of a difference (one is scaled at 1.00 and the other is 0.5). But even then it's not terribly noticeable (as seen in the gif above this one).
This leads me to believe that scaling the object up or down doesn't impact the simulation much. But what has the greatest impact on the simulation is the distance of the vertices in the world, regardless of the object's scale... Which is extremely annoying.
It's nice to have soft body physics that would be accurate to the physical world, but I really think that someone (not me, I'm too dumb and I hate physics) needs to make a physics solver that doesn't depend on the scale of the object or the distance of the vertices in the world, but solves based on normalized vertex positions. Something intended for jiggling that doesn't need to be accurate to the real world... it just needs to bounce in a predictable manner. Like, have one setting for "bounce" one for "ripple" one for "stiffness" one for "volume retention" and what not.
I don't know though. That sounds like it would be difficult. But I don't get why there aren't like... stylized physics simulators. I just want to make some bouncy [t]iddy and butt, not scale-accurate jello cubes.
I haven't even told you about the "goal" vertex group and how drastically different weighting just 1 or 2 verts at "1.00" can make the behavior of the simulated object...
Or how checking "stiffness" to on (but setting the shear value to 0.00) isn't the same as checking it off. Or how different a shear of 0.1, 0.05, or 0.01 can be. It *does* make the object more stuff, but it's touchy.
Anyway...
This explains why tweaking all the freaking settings was so unintuitive in the past to me... It's still a pain the butt, but this gives me a clue as to why and I think now I can figure out values do what on larger/smaller objects.
