#Stuttering Hair Physics

127 messages · Page 1 of 1 (latest)

idle kelp
#

When rigging in Live2D the stuttering is there a little but after tracking in VTube studio it's very pronounced. I used blendshapes to rig the ponytail and side pieces, I'll attach a photo of the parameters and deformer hierarchy. The physics are not keyed to anything other than the physics. Please help

#

Sidebang piece (1st, 2nd image)
Ponytail piece (3rd image)

errant jewel
idle kelp
#

Since at 60fps the stuttering was barely visible in Live2D but then was visible on VTube Studio, if that makes sense

errant jewel
#

let me copy your pendulum setup rq and see if i can figure out what's causing it

idle kelp
#

Let me know if you need any more information as well

errant jewel
#

okay i think it's really just due to the pendulum settings, i would adjust the values to allow for smoother movement when the head is moved slower.

idle kelp
#

I'm not the best when it comes to pendulum adjustments, do you have any pointers/resources that I can look at to correctly adjust it? Because I wouldn't really know where to start

errant jewel
#

i lowered the shaking influence and reaction time and that seemed to help

#

but i'm just on mouse tracking right now, see if adjusting those help on your end

#

the slower the movement the lower i usually set the reaction time

#

sometimes, increasing the minimum/maximum for the input normalization can make movement smoother as well, especially for pendulums where the inputs have lower effectiveness

#

since the pendulum will travel further even with a small input

idle kelp
#

it's still got stuttering, I tried increasing the normalisation for the angle but ill try doing pos x

#

kind of made it worse lol

errant jewel
#

angle normalization won't affect Angle X since the input is set to position X

#

hmm okay

idle kelp
#

cause I had the same issue with the face turn since I'm using aug angles. it had the stuttering. I changed it to copy the pendulum I had on another model I did and it fixed it

errant jewel
#

the input itself seems to be stuttering a bit too, what about turning up the smoothing for angle x is vtube studio?

#

what's the group for angle x look like on this vs the other model where it works?

idle kelp
idle kelp
#

(this model) Aug X group

#

(other model I used pendulum of) it has a higher normalisation since there's a greater angle turn (-60 to 60)

#

I have no stuttering issues on this other model

errant jewel
#

and if you copy the angle normalization as well the stuttering on the current model vanishes or does it stay?

idle kelp
#

It stays

errant jewel
#

okay and the stuttering is only present when you connect live2d to vtube studio or when you use mouse tracking as well?

idle kelp
#

It's present as well just not as prominent, so it must be an issue with how it's set up in Live2D

errant jewel
#

it's just strange that it stays even when you completely copy the physics settings from another model that doesn't have that issue

#

so i'm thinking it miiight be something else

idle kelp
#

I wonder if it's how I rigged the hair since the front bangs don't have the issue. I've tried rerigging the entire ponytail using a different method still using blendshapes but it persisted

#

The front bangs didn't use blendshapes

#

nvm they did me when I lie

errant jewel
#

how are the ponytails rigged?

idle kelp
#

So in the hierarchy for deformers:

[Category Deformer] - no keys, just organising

XY - only keyed to the aug XY

General Physics - only keys X1 and Y1 and it moves every piece using a temporary rotation deformer.
Then each section of hair has its own piece for X2, X3, etc.

#

I've tried removing the General Physic deformer where it moved all of them at once and had it just be individual deformers, but that didn't change anything. if that makes sense

errant jewel
#

i mean from the movement, could you record what it looks like?

idle kelp
#

I can send a video if needd

#

yep!

errant jewel
#

those look fine...

idle kelp
#

and I've triple checked that they're not keyed to anything other than the physics

idle kelp
#

Still stumped trying to fix this.. I've re-exported it, tried disabling different physics groups to see if it's affecting it, I can't figure it out :(

simple lance
#

can you tellme if in editting groups, do you have any blue or red?

idle kelp
#

I can take a video of it if you need

simple lance
#

ok

idle kelp
simple lance
#

ok, can you turn Physics OFF
and move angle parameter, to see how the hair moves without physics

idle kelp
#

So turn off the hair physics?

simple lance
#

all physics

idle kelp
#

I have aug angles so nothing will move without them so I'll keep those on?

simple lance
#

move the angle Aug X

idle kelp
simple lance
#

ok, by your other videos
Can you try to fix all childs objects to make sure they are always inside parent?

idle kelp
#

I'm a bit unsure by what you mean by that

simple lance
#

in your previous videos, youshowed the deformers

#

in those deformers, some of them, have pieces that stick out of the deformer and others that goes out when move on keyforms

idle kelp
#

So making sure the child is always inside this box on the keys? Just trying to make sure I understand

simple lance
#

yes, the Child object, should Always be inside the parent, in all its keyforms

idle kelp
#

ok! I'll try it on one of the ponytails and see how it goes

simple lance
#

this an example of child going out

idle kelp
#

out of the parent box and not the deformer correct? sorry for dumb questions

simple lance
#

the parent box and the deformer is the same thing

idle kelp
#

i meant the deformer currently selected sorry

simple lance
#

artmesh inside a deformer is also a child
and also you want it to not go out parent

idle kelp
#

gotcha. so if the physics deformer is going outside the box for the XY deformer, I should be making the XY deformer bigger?

simple lance
#

yes, that apply to everything

idle kelp
#

okay i understand you, i'll do some tinkering and let you know how it goes

idle kelp
#

does this look right? I redid the big physics one to make it bigger and then redid the xy to make it bigger as well

simple lance
#

that looks fine in default position

idle kelp
#

the first physic for the hair does a big rotation and always goes outside the XY parent deformer, do I need to make the XY even bigger?

simple lance
#

yes you need to make it bigger

idle kelp
#

alright i'll do that and let you know how it goes

simple lance
#

what happends if you move both of those parameters at max or min?

idle kelp
#

it goes out like this

#

both max

#

both min

simple lance
#

ok, in first picture it goes out
do you have the same as that but inverted? where X is min and Y is max?

idle kelp
simple lance
#

yeah ok, the parent deformer need to be bigger

idle kelp
#

are you sure the deformer isn't way too big to properly deform xy now?

simple lance
#

is not big if your child move that much
in live2D you always want to make sure child NEVER go outside of parent because it can do weird deformations

#

the more something move, the bigger the parent will have to be

idle kelp
#

okay ill give it a go

#

it definitely doesn't go out the XY deformer any more

simple lance
#

after you check that with all the deformers of the hair, we can check physics again how they moving

idle kelp
#

the individual hair pieces still go out so ill make the big physic one bigger and then increase the xy size again?

simple lance
#

yeah

#

btw are you following a tutorial or something? [i ask because some movement on X1 were similar in Y1]

idle kelp
#

for the hair not exactly. I'm just using the same standard movement I have for the hair. I've tried a few different methods but I've liked using blendshapes

simple lance
#

well using blendshapes is not the important part here
i was focusing more into the fact the two movements were too similar, that sometimes can give weird result when moving in diagonal
but thats something could be important after all deformers are fixed

idle kelp
#

The X1 and Y1 are inverted versions of each other's movement. Since I want the X1 to move left and right of course, and then for Y1 I want to move right on -1 to simulate going down and then inverse for 1

#

not really sure how else to rig the big movement of going up and down otherwise for hte ponytail

simple lance
#

that is not bad per se
but sometimes if the physics are not done correctly, they will neutralize eachother when moving diagonal

idle kelp
#

I understand, I'll probably do some practice just rigging hair after I finish this model

I'll send the updated movement in a little bit

#

checked for the smaller physics deformers too and they stay inside the big physic deformer

simple lance
#

ok, if all is good yu can go back to physics first test
if issue still hapening, desactivate all Y physics and test only X physics

idle kelp
#

it still persists

#

even with Y off

#

ill send a video

simple lance
#

ok can you record the same but with X physics? the group you showed is Y

idle kelp
#

yep

simple lance
#

ok do you see how it shakes more when you go slow
but doesnt when you go fast

idle kelp
#

yes

simple lance
#

thats because your pendulum

#

the settings you have right now in your pendulums, dont work with very slow movements [the shake is basicaly the physics stopping]

idle kelp
#

Ahh i see. Is there a way to find a happy medium between the two? Cause the stuttering is very noticeable when you do a small sudden turn left or right as well which would show in use

#

I do think it's helped for those small movements compared to the other ponytail I think

simple lance
#

try to change pendulum something like this
[i will say numbers in order of colums, ignoring colum called "no" and starting with duration]
20 / 0.89 / 1.8 / 0.2

#

Also change the inputs to be the aug parameters, not the original parameters

idle kelp
#

here's how it looked with the changes you reccomended, would increasing the scale in output numbers make the shaking come back again?

simple lance
#

you want output max to be 100 so you can use increase-decrease as you need
BUT you have to do that when you move strongly [or it will look weird as something we say is hitting limits]]

idle kelp
#

it's definitely fixed the stuttering problem but the output physics isn't exactly how i'd want it

should i tinker with the pendulum some more?

simple lance
#

you can do as much as you need

idle kelp
#

so some good news, i changed the y inputs to aug as well and there doesnt seem to be any jittering, ill test just doing that on the front bang pieces and see if that changes anything

#

so i changed it and I believe since its now aug input the stuttering has turned to a bounce rather than a noticable start and stop which is a definitely a bandaid fix, but what you suggested for the ponytail seems to have helped so i'll try it on the side bang

idle kelp