Attach geo to alembic walking

[caskal 174]

Hey magicians,

Been fighting with this for 2 days, I have a person + suit I did on Marvelous, and I want to attach some tubes on the back.

Some people on facebook group suggested Rivet sop, as well as a tutorial with extracttransform + transformpieces, but I don't have luck. I also tried point deform but the geo distorts.

When I tried Rivet on the cloth itself, it wiggles, I guess because of the folds moving, so I tried directly with the person inside, and works better, but still not good.

I'm attaching some examples:

This was the best result so far:

T6a4mNQqS0.mp4

But gets fucked at the end:

2IwfUr3sMm.mp4

Any thoughts?

Thanks!

 

 

[caskal 174]

Here's an example of the wiggle with extracttransform + transform pieces

Y4aLvu4q2N.mp4

[anim 1,192]

- for rivet make sure you have proper up vector attribute to extract rotation data, you can use Poly Frame for that to extract it from topology or uvs...

you can also use group with more points to get more averaged results than just with one point, or also additionally having another point for the bottom and having lookat towards it to allow for better conform on the back (but the stable rotation still has to be established before)

- for pointdeform make sure you pack your canister before, also it may help if the pivot is closer to wherever it's gonna be attached to

- for extract transform you can just isolate back section for both rest and animated input, so that it doesn't consider arms and such when computing rigid morion

- you can also go with lower level approaches like Ray on rest, Attribute interpolate P and stable N/up from animated, or VEX based approaches xyzdist()/primuv(), uvsample() , or extracting points from the back and building stable transform from them ...

[caskal 174]

On 18/2/2021 at 10:24 PM, anim said:

- for rivet make sure you have proper up vector attribute to extract rotation data, you can use Poly Frame for that to extract it from topology or uvs...

you can also use group with more points to get more averaged results than just with one point, or also additionally having another point for the bottom and having lookat towards it to allow for better conform on the back (but the stable rotation still has to be established before)

- for pointdeform make sure you pack your canister before, also it may help if the pivot is closer to wherever it's gonna be attached to

- for extract transform you can just isolate back section for both rest and animated input, so that it doesn't consider arms and such when computing rigid morion

- you can also go with lower level approaches like Ray on rest, Attribute interpolate P and stable N/up from animated, or VEX based approaches xyzdist()/primuv(), uvsample() , or extracting points from the back and building stable transform from them ...

thanks for all the tips Anim, tried all of them, and the one that worked the best was an extract transform with some back points on the man itself instead of the cloth + a filter chop to smooth the walk, will post the short when I finish

Thank you!!