Page 9 of 11

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 07:09
by CybeREX
ikam wrote:1/ I fixed visual artefact bug.
2/ Changed the joint behavior, see example.
3/ No change for now for chains.

For me all example that builded in that file just fall down to the floor

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 07:26
by Skinnytorus
Is your build up to date? Watch this link to see if there were any commits before you download any examples:
https://github.com/wzman/werkkzeug4CE/commits/master

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 07:46
by ikam
all example files here are obsolete if you don't have exactly the same code version that I used to build them. It's really just for testing stuff, nothing stable yet in this thread
You need to update the code and recompile wz at each modification I send on github to get latest file exemple working, and after some update, previous exemple file may don't work as I changed some stuff.
better way is to get lastest code update (with git software and this command : git pull origin master), compile code, and build examples yourself from scratch, don't use previous operators and don't do copy/paste from old files. An old operator may be not compatible with new ones or produce indesired effects

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 08:26
by Skinnytorus
ikam
It might also a good idea to set default location for new joints at the top of the RBs (collider's?) local bounding box with some offset (like 0,1 or something like that). Saves some time...

What do you think of that?

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 08:50
by ikam
why top and not left or right ?

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 08:54
by Skinnytorus
For me it's natural: a Christmas ball hangs with its hook upwards ;)

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 09:44
by ikam
lol. yea why not but i'm not sure you will build a chistmas tree in all your demos :p

I made a version working with joint multiplication. I not guarantee all case, but it should works for basics. Anyway in concrete cases, multiple stacks of nested multiplied joints in tree, become a bit hard to design for the user and it seems uncommon for most of concrete cases.

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 10:07
by ikam
and same version with an alternative joint operator, that take only 1 input and search in all tree. maybe more clean to build joints

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 10:14
by Skinnytorus
lol. yea why not but i'm not sure you will build a chistmas tree in all your demos :p

You can add both as an option if you like ;))

physx3.wz4 Yep. Multiply seems to be working as expected. Good job :) I'll take a look into the latest file soon.

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 10:36
by Skinnytorus
I like the basic idea.
1) A drop-down list of all available rigid bodies in both joint operators wouldn't hurt.
2) Have you added any functionality for the situation when multiply is on input of the joint operator?

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 10:41
by ikam
2/ it works if you switch from first to all for instances param

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 10:56
by Skinnytorus
Both joint ops still don't work with chain properly.

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 11:23
by ikam
yea, chain need a complete review

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 12:55
by ikam
fixed

Re: new Physx

PostPosted: Tuesday, 29.July 2014, 15:23
by Skinnytorus
Yeah! Working! Thanks.
1) IMHO JointIndex in Chain must be consistent with RB terminology. Otherwise, it's a problem to understand how it works.
2) Could be cool to randomize the multiplied RB motion a bit in your last example. Any ideas?