mpm fibers not working correctly in new version

[PhilRadford]

Hi. Since the new update I have been trying various mpm fibers R&D tests and something seems up, sims not behaving as they should. I went back to some of my older scenes that I know worked perfectly before the update and they do not any more.

Has something happened here ? can someone test by animating maybe some rotation on a simple poly plane with a stands from normals to mpm fibers set up and let me know if its doing some crazy things and not behaving because form what I have seen and tested it seems very broken . I have the latest rebel pack and also the new MJCG compunds installed . There are no error messages just odd behavior in new sims and no behavior in old (pre new bifrost release) scenes. 

Phil

[Christoph_Schaedl]

Phil do you have an example scene or a video showing the issue?

[PhilRadford]

I have figured out what is wrong, it seems to be a bug !

If I open any scene with fibers made before the new release the 'mpm solver settings' node has been automatically disconnected from the 'simulate mpm' node . This has happened across all of my previously built scenes.

I have attached a screen shot of a graph that I have just opened. The knit is also disconnected and with a warning.

That's the first bug . I will move on to the problems with the new issues using new scenes in my next response with some examples.

[PhilRadford]

. . . and here is a video showing a new set up and a huge offset from the fibers trying to follow an animated sphere.
This offset didn't used to happen but now does . As you can see in the video I up res the detail a bit but still the issue happens. Try it our for yourselves.

[mjcg91]

Regarding first post, these are not bugs but compatibility breaks. 

"create_simple_knit" just hasn't been shipped with public 2.1.0.0, that's why it appears like this. I don't think it was intentional, but this compound has been moved to another namespace, and I think the devs just excluded the whole  category's compounds before releasing the public version of 2.1.0.0

the "mpm_solver_setting" output port as been renamed from "settings" to "solver_settings", therefore breaking the connection when you open scenes saved prior the update.

Now, the "constraint_mpm", it looks like a double transform issue. If your emission mesh remain static, and have only the container mesh then it will behave correctly. Maybe this one is a bug. (btw, if you use a "constraint_mpm", you don't need to keep the other source's connected. you can cut the connection as it is already passed inside the constraint node, otherwise you will simulate the same strands twice).

[PhilRadford]

---

@mjcg91 wrote:

Regarding first post, these are not bugs but compatibility breaks. 

"create_simple_knit" just hasn't been shipped with public 2.1.0.0, that's why it appears like this. I don't think it was intentional, but this compound has been moved to another namespace, and I think the devs just excluded the whole  category's compounds before releasing the public version of 2.1.0.0

the "mpm_solver_setting" output port as been renamed from "settings" to "solver_settings", therefore breaking the connection when you open scenes saved prior the update.

Now, the "constraint_mpm", it looks like a double transform issue. If your emission mesh remain static, and have only the container mesh then it will behave correctly. Maybe this one is a bug.

---

Understood re- the first issues, I am seeing a few nodes (not just knit) with this naming change. On the second issue regarding the new version and fresh scene, Yes if the source mesh is static everything is fine, but animated (even exported to alembic) does have the offset issue . I think the double transform sounds correct. I have tried multiple set ups (including alembic) and it always happens. Perhaps the devs could offer a workaround? 

[stamatk · Accepted solution]

Hi @PhilRadford,

Thanks for all the amazing work and for reporting your findings to us.  

Just a kind reminder that the "experimental" icon on the fiber/cloth/fluid/shell MPM nodes implies that these nodes are subject to some changes that may alter workflows and/or break connections occasionally.  

As for the constraint offset, I have noticed this issue on my side as well, and we'll schedule to have it fixed.

Thanks!
Kosta

[PhilRadford]

---

@stamatk wrote:

Hi @PhilRadford,

Thanks for all the amazing work and for reporting your findings to us.  

Just a kind reminder that the "experimental" icon on the fiber/cloth/fluid/shell MPM nodes implies that these nodes are subject to some changes that may alter workflows and/or break connections occasionally.  

As for the constraint offset, I have noticed this issue on my side as well, and we'll schedule to have it fixed.

Thanks!
Kosta

---

Hi Kosta, understood and happy to bring this to your attention . Is there a workaround ? only I am enjoying using fibers : ) 
many thanks

Phil

[stamatk]

Hi @PhilRadford 

I actually did find a workaround:

The animation has to be only on the constraint, not on the "emitter" geometry.  The emitter geometry is only responsible for creating the initial position of the fibers.  The constraint is the one that takes them with it when it moves.  So do not animate the initial geometry and do not group/parent/ the constraint geo and it works.  

I've attached a video, hope it helps.

Cheers,

Kosta

[PhilRadford]

---

@stamatk wrote:

Hi @PhilRadford 

I actually did find a workaround:

The animation has to be only on the constraint, not on the "emitter" geometry.  The emitter geometry is only responsible for creating the initial position of the fibers.  The constraint is the one that takes them with it when it moves.  So do not animate the initial geometry and do not group/parent/ the constraint geo and it works.  

I've attached a video, hope it helps.

Cheers,

Kosta

---

I will give this a go ! thank you !

[Dark__Night]

Hi folks does this mean we can no longer use 'create_simple_knit'? Could someone post the new one as a separate compound perhaps? I was enjoying my knitting 🙂