Inconsistent Results - Same File, Same Machine

[mitchfx]

I have a fairly complex setup using several birth shapes, Physx Binds, breaks, splines, collisions, etc.  If I dial everything in to something that is exactly the results I want, then simply re-cache or save/load the file and preview again...the results are slightly different.  This is making it very hard to get an approval from the client or to repeat a setup and expect identical results.  Is this a known issue with the PhysX engine or other solvers used in TyFlow?  Are there any ways to avoid this?  Using the latest beta build 16022 and Max 2019.  Thanks.

[tyFlow]

http://docs.tyflow.com/tyflow_particles/...indSolver/

[mitchfx]

http://docs.tyflow.com/tyflow_particles/...indSolver/

Thanks for the reply and link.  Very interesting and useful information.  I triple checked my flow and I don't have any regular particle bind operators, only PhysX binds.  Does this setting impact other parts of TyFlow besides particle binds?

[tyFlow]

Hmm, if you're getting indeterminism without cloth or particle physics operators, that could be a bug. You can send the scene to support@tyflow.com for me to look at, but I'm out of the country for a week so can't check it for a while. You can also try setting max threads to 1 in the main tyFlow settings to disable multithreading in case it's a threading issue.

[mitchfx]

Hmm, if you're getting indeterminism without cloth or particle physics operators, that could be a bug. You can send the scene to support@tyflow.com for me to look at, but I'm out of the country for a week so can't check it for a while. You can also try setting max threads to 1 in the main tyFlow settings to disable multithreading in case it's a threading issue.

I will strip down this scene and send over a file within the next week.  In the meantime it sounds like a single threaded test is worth running as well. I'll let you know if that makes a difference.  Enjoy your much deserved vacation Tyson!

[mitchfx]

Hmm, if you're getting indeterminism without cloth or particle physics operators, that could be a bug. You can send the scene to support@tyflow.com for me to look at, but I'm out of the country for a week so can't check it for a while. You can also try setting max threads to 1 in the main tyFlow settings to disable multithreading in case it's a threading issue.

I will strip down this scene and send over a file within the next week.  In the meantime it sounds like a single threaded test is worth running as well. I'll let you know if that makes a difference.  Enjoy your much deserved vacation Tyson!

I just realized my flow does have one Object Bind (Surface) operator in it.  Is this also subject to the strict determinism behavior?

[tyFlow]

Object Bind shouldn't be subject to that, no. Try single-threaded mode and let me know your findings. Recently I discovered Find Target might have a threading issue in some circumstances so maybe that is it as well.