new Physx

Re: new Physx

Postby ikam » Sunday, 27.July 2014, 11:42

1/ you can define a joint position (in rigidbody) with a bigger offset from mesh.
2/ can't see what you mean, any example ?
3/ I can do a chain operator that will be a seperate operator that take only one input (with multplied rigidbody above), it will link each piece to next one.
ikam
Operator
Operator
 
Posts: 911
Joined: Friday, 14.October 2011, 13:00
Location: France

Re: new Physx

Postby Skinnytorus » Sunday, 27.July 2014, 12:23

1) I see now. If I adjust joint transform controls in RB operator with anim stopped, actors return to their default positions. I have to start the simulation in order to position them. Would be better to be able to adjust them in real time without sim running i.e. like ordinary transforms.

2) I've got that sorted. Thanks. In your examples Multiply works only in respect of the free (unattached) actors which is good imho.

3) IMO there should be in fact 2 operators: for multiplied RBs and for custom RBs to form a chain (stacked above from left to right). It might be as well a single operator working in two modes - up to you.

In multiply mode, the operator should work in these sub-modes:
a) First mode - only Multiplied RB on input 1. The multiplied RB with index of 0 is parent. The others are links in the chain.
b) First-Last mode - only Multiplied RB on input 1. The first and the last indexed RBs in multiply are parents. The others are links in the chain.

In custom mode the operator should work in these sub-modes:
a) First mode - RB with index of 0 (from left to right) is parent. The others to the right are links in the chain.
b) First-Last mode - The first and the last indexed RBs (from left to right) are parents. The others in between are links in the chain.

So... What do you think?

Edit: In fact, it would be cool to make such chains multipliable (ChainMultiply operator? )(jelly fish tentacles, hair, etc.)...
Skinnytorus
Operator
Operator
 
Posts: 1300
Joined: Monday, 06.February 2012, 17:46

Re: new Physx

Postby ikam » Sunday, 27.July 2014, 12:57

1/ yes it need to play simulation to see joints in action. For now I don't see how to do without simul on.
3/ what are parents of chain ? extremities ?
ikam
Operator
Operator
 
Posts: 911
Joined: Friday, 14.October 2011, 13:00
Location: France

Re: new Physx

Postby Skinnytorus » Sunday, 27.July 2014, 13:06

3/ what are parents of chain ? extremities ?

Those are fixation points - the points the entire chain follows.
Skinnytorus
Operator
Operator
 
Posts: 1300
Joined: Monday, 06.February 2012, 17:46

Re: new Physx

Postby ikam » Sunday, 27.July 2014, 13:09

ok, but can you detail : "The first and the last indexed RBs in multiply are parents. The others are links in the chain." I don't see
ikam
Operator
Operator
 
Posts: 911
Joined: Friday, 14.October 2011, 13:00
Location: France

Re: new Physx

Postby Skinnytorus » Sunday, 27.July 2014, 13:10

Skinnytorus wrote:
3/ what are parents of chain ? extremities ?

Those are fixation points - the points the entire chain follows.


Edit: It's as if you were holding a chain. Your hand(s) are are parent(s) to the entire chain.
Skinnytorus
Operator
Operator
 
Posts: 1300
Joined: Monday, 06.February 2012, 17:46

Re: new Physx

Postby Skinnytorus » Sunday, 27.July 2014, 13:14

"The first and the last indexed RBs in multiply are parents. The others are links in the chain." I don't see

The first and the last indexed RBs in multiply are your hands in my example. The chain reacts if you move one of them or both in different directions.
Ok. Maybe then use 3 inputs:
1. Handle1
2. Multiplied RB (the chain)
3. Handle 2 (or no input if there is input1)
Skinnytorus
Operator
Operator
 
Posts: 1300
Joined: Monday, 06.February 2012, 17:46

Re: new Physx

Postby ikam » Sunday, 27.July 2014, 16:39

ah ok I see now what you mean.

But in fact there no need of parents and it's easier like this because of no need to identy parents or special treatment. all elements are links in the chain. And after you put another joint to first, last or anyone elements of this chain, it will be the rigidbody that will control the chain. for example : all links in the chain are dynamics rigidbody, and the another rigidbody linked to the chain could be a kinematic, static or another dynamic that will drive the head chain.
ikam
Operator
Operator
 
Posts: 911
Joined: Friday, 14.October 2011, 13:00
Location: France

Re: new Physx

Postby Skinnytorus » Sunday, 27.July 2014, 17:26

So, What's your plan? You want to make the 3 inputs as I described in my last message? If yes, then we can go even further: no need for multiply op on input for the chain - it's the chain operator that has to interpolate a certain number of RBs (i.e. clones of the RB in the second input) between handle1 and 2 positions. If there is no RB on input 3, then the chain's tail will hang loose. What do you think?
Skinnytorus
Operator
Operator
 
Posts: 1300
Joined: Monday, 06.February 2012, 17:46

Re: new Physx

Postby ikam » Monday, 28.July 2014, 06:06

Yes interesting but quite complex, it need to create 2 joints types (3 if three input) inside the same op (1 between handle and chain, 1 between chain links, 1 for chain tail and second handle) and manage the rigidbody multiplication.
I'll try to build it. In wait I put a simple chain op for testing overall behavior.

edit : another point, It become really hard with a complex tree with many multiplier when you want to attach specific elements... Maybe using a system of name to identify rigidbody could help
ikam
Operator
Operator
 
Posts: 911
Joined: Friday, 14.October 2011, 13:00
Location: France

Re: new Physx

Postby Skinnytorus » Monday, 28.July 2014, 07:22

I don't see how the simple chain works. Could you post an example please?

Maybe using a system of name to identify rigidbody could help

Why not? Give it a try. On the other hand, visual stacking could be more user friendly. Wz4 is no Houdini. No need for sophistication ;)
Skinnytorus
Operator
Operator
 
Posts: 1300
Joined: Monday, 06.February 2012, 17:46

Re: new Physx

Postby ikam » Monday, 28.July 2014, 07:37

a test.
You do not have the required permissions to view the files attached to this post.
ikam
Operator
Operator
 
Posts: 911
Joined: Friday, 14.October 2011, 13:00
Location: France

Re: new Physx

Postby Skinnytorus » Monday, 28.July 2014, 07:59

I see now. Thanks. It seems joints are creating artifacts. Take a look.

Edit: One more thing: if I stack RB Multiply under the Spherical Joint, joints don't seem to multiply.
You do not have the required permissions to view the files attached to this post.
Skinnytorus
Operator
Operator
 
Posts: 1300
Joined: Monday, 06.February 2012, 17:46

Re: new Physx

Postby Skinnytorus » Monday, 28.July 2014, 08:13

And what are jointindex0 and 1 in Chain operator?
Skinnytorus
Operator
Operator
 
Posts: 1300
Joined: Monday, 06.February 2012, 17:46

Re: new Physx

Postby ikam » Monday, 28.July 2014, 09:04

yes, I saw for artefact, it seems a matrix problem with meshes used to preview joint fixation poses.

Jointindex0 and 1 are used to identify which fixation pose to use to joint bodies. As a rigidbody has an array of fixation poses, you need to specify which one to use to joint. In chain you join each element to each other by two points.

But all that is not clear... need to find an easiest interface.
ikam
Operator
Operator
 
Posts: 911
Joined: Friday, 14.October 2011, 13:00
Location: France


Return to Development and contribution



Who is online

Users browsing this forum: No registered users and 13 guests

cron