Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Hinge Joints Alternative?
#11
(17-11-2020, 09:38 PM)tegleg Wrote: The stitcher seems the closest thing so far although my tests have been underwhelming to say the least. not a hinge but at least a way of joining 2 soft bodies.
lets hope the next update provides some kind of solution.

Thanks for your reply which was helpful. I totally missed the Obi.ObiStitcher Class! My project requires quite a few hinges per model - therefore I was looking for a solution that would not be too empirical. I guess it is time to go and do some testing with it.

I haven't found the stitcher being used as part of the examples / tutorials but maybe in a different Obi product?

If there isn't a way to restrict certain axis, then it's going to be hard to come close to emulate a hinge.
Reply
#12
(18-11-2020, 12:58 AM)Pheebau Wrote: Thanks for your reply which was helpful. I totally missed the Obi.ObiStitcher Class! My project requires quite a few hinges per model - therefore I was looking for a solution that would not be too empirical. I guess it is time to go and do some testing with it.

I haven't found the stitcher being used as part of the examples / tutorials but maybe in a different Obi product?

If there isn't a way to restrict certain axis, then it's going to be hard to come close to emulate a hinge.

Unfortunately my tests with ObiStitcher in trying to emulate some hinge of some sort have been underwhelming to say the least but I am not surprised as it was never made for this.

Even as a simple test, take the Barrels deformation demo scene and try and stitch 2 barrels together - no mater how I define the stich(es), the barrel will move and travel together without me even doing anything - similar to the dancing effect you can sometimes get with soft bodies in a scene Guiño

Also, I believe the only way to break a stitch is to disable it via the API. In my project, I need hinges to rotate on a single axis and break potentially - this is a physics game with soft bodies.

@josemendez Any recommendations / simulation config settings on making 2 soft bodies stitched together more still and not flickering / dancing?

@josemendez Last, how hard would it be to implement some hinge in Obi?
Reply
#13
(22-11-2020, 08:41 PM)Pheebau Wrote: Unfortunately my tests with ObiStitcher in trying to emulate some hinge of some sort have been underwhelming to say the least but I am not surprised as it was never made for this.

Even as a simple test, take the Barrels deformation demo scene and try and stitch 2 barrels together - no mater how I define the stich(es), the barrel will move and travel together without me even doing anything - similar to the dancing effect you can sometimes get with soft bodies in a scene Guiño

Did you try setting the stitched particles to the same phase value, so that they don't collide with each other? Stitching them and making them collide will likely cause feedback (dancing) effects.

(22-11-2020, 08:41 PM)Pheebau Wrote: Also, I believe the only way to break a stitch is to disable it via the API. In my project, I need hinges to rotate on a single axis and break potentially - this is a physics game with soft bodies.

Unfortunately stitches can't break depending on force. Triste

[quote='Pheebau' pid='8275' dateline='1606074097']
@josemendez Last, how hard would it be to implement some hinge in Obi?

Hinge constraints remove rotational DOFs (degrees of freedom) from a pair of bodies. This usually requires fine control of angular motion per-axis, as well as proper inertia tensors (which Obi only roughly approximates). While particles in Obi are augmented with rotation to some extent, they're not yet fully rigidbodyfied so to speak. Implementing hinge constraints would take quite some time as core parts of the engine need to be augmented/modified. We will consider adding this in the future.

An alternative could be using dynamically attaching particle groups to rigidbodies, then using hinge constraints on those rigidbodies. Less than ideal, but I think it's the best workaround for now.
Reply
#14
(23-11-2020, 09:02 AM)josemendez Wrote: Did you try setting the stitched particles to the same phase value, so that they don't collide with each other? Stitching them and making them collide will likely cause feedback (dancing) effects.

An alternative could be using dynamically attaching particle groups to rigidbodies, then using hinge constraints on those rigidbodies. Less than ideal, but I think it's the best workaround for now.

Yes I did set the same phase value for the 2 barrels being stitched but they are still dancing after a few seconds. It can be reproduced with the Obi barrels scene.

Funny you mentioned it because I actually thought about this alternative of using rigidbodies and hinges to connect the 2 soft bodies - I will need to create some new (helper) component to orchestrate the setup of a particular hinge (and the required placebo rigidbodies) between 2 softbodies and define the hinge transform - similar to ObiStitcher when creating stiches.

Thanks for considering adding native hinges to Obi soft bodies.
Reply
#15
(23-11-2020, 10:18 AM)Pheebau Wrote: Funny you mentioned it because I actually thought about this alternative of using rigidbodies and hinges to connect the 2 soft bodies - I will need to create some new (helper) component to orchestrate the setup of a particular hinge (and the required placebo rigidbodies) between 2 softbodies and define the hinge transform - similar to ObiStitcher when creating stiches.

Thanks for considering adding native hinges to Obi soft bodies.
hi did you get anywhere with this?
when i tried i get lots if drifting of the soft bodies.

Edit:
fixed by unticking constraint orientation on the particle attachment
Reply