I reduced this into single event flow and noticed, it worked fine until I added Particle physics. Max crashes when particle physics is utilized. Also it started crashing after I hange timestep from frame to 1/2 frames.
10-18-2023, 02:28 PM (This post was last modified: 10-18-2023, 02:29 PM by tyFlow.)
Anytime you run into a crash, please send the minidump. That, or the file itself, it the only way to begin to debug an issue like this. Your previous minidump showed the crash happening within Phoenix, which is why I recommended sending it to Chaos for further debugging. If you have a crash now related to Particle Physics, you'll need to send that minidump as well. Without sending the files/minidumps there's literally nothing that can be done to help further. Text descriptions or screenshots of flows provide no information that could possibly held to debug a code error.
Is there any development on this issue or any ideas what to try to get a little progress with this project? I would really need to move on to new projects, before my pockets are empty.
10-24-2023, 12:34 PM (This post was last modified: 10-24-2023, 12:37 PM by tyFlow.)
Well I was unable to reproduce the issue over here when you sent me the file, and your crash minidump showed the issue happening in the PhoenixFD dlr file...so your recourse would be to follow up with Chaos as I am unable to debug the plugins of other developers.
A potential workaround would be to change your PhoenixFD settings to export VDB files instead of AUR files, and then load the resulting VDB files into a tyFlow Birth VDB operator (file mode, with your PhoenixFD container assigned as the transform node), and use the VDB Force operator to drive your particles that way instead. That way you would avoid the use of the PhoenixFD loader to read the cache data into tyFlow and also potentially avoid whatever issue is causing the crash.
In the meanwhile I tried simplified version of the fluid simulation, but it didn't help. Max crashed as usual. I would really need to advance a little bit. I'm exhausted due to long working days trying to solve this puzzle and this is everything I got with tyFlow.
Now by following the earlier instruction I seem to have lost my tyFlow license permanently. Do I need to purchase a new?
In order to re-enable tyFlow PRO after this test you will either need to restore tyFlow.cfg to that folder and restart Max, or re-enter your license key.
I am somewhat busy with my own projects these days, but if I can do something to help you finish this, I am here (being many time in a pressure like this, I can fully understand the stress).
First of all, I might be VERY WRONG, but I think you might overthinking the whole project, and maybe it become too complicated, for no reason.
It's really hard to understand what exactly you want to achieve looking only at screenshots you gave, but if it's only a liquid that is sipping in the cyclone, and it needs to swirl on the way down, and you also need particles inside the liquid doing the same thing, that should be pretty easy bro.
I'd first do liquid with Phoenix FD until I am (or client) fully happy. Then I'd do just Birth fluid, fluid force operator with TyFlow to drive particles like liquid.
Of some particles are exiting the liquid, I'd use surface test to detect them (referencing the phoenix mesh) and delete them.
I am not sure why would you need Particle Physics for example.
So the particles can collide?
But if you in a camera shot that is seeing the whole cyclone, can anyone notice particles going to through each other (it will not be visible)?
But then again, I maybe misunderstood the whole thing, and maybe indeed is very complex.
Can you post a scene with just Cyclone (and other objects if they are involved in simulation), and explain clearly what needs to happen, and what your client wants to see exactly.
I can't promise anything, but I can give it a try tomorrow, or day after tomorrow.
And if you explain in very clear manner, maybe someone else will also be able to help you.
This system is for cleaning stones and sand out of the liquid. There are several stages, where those are being reduced and the fluid is cleaned stage by stage.
On a cyclone the flow goes downward near the walls picking up speed and the flow turns up in the bottom and goes up in the middle. Stones don't aren't following the liquid upwards. Instead they fall into a junk trap at the bottom, while the fluid mainly leaves the cyclone upwards. The object test sends stones below certain level to a new event, which don't have Fluid force. Without that, it would follow the cleaned fluid upwards instead of falling to the junk trap.
Particle physics is needed, because the particles are piling up to a certain level at the junk trap. Then the junk trap is closed and flushed.
I have managed to create somewhat convincing Phoenix flow earlier, but when simulating tyFlow Max crashes each time, when I increase steps per frame. Only one step per frame don't crash, but then the particles fly out of the cyclone through the walls. I think 4 steps per frame would do the trick.
I made another tryout with all new Phoenix configuration and I got this. It simulated normally with one frame framestep. As soon as I changed this to 1/2 frames, I got this.
I'm beginning to wonder if tyFlow and Phoenix are not quite compatible, when used together in same scene.
The problem with these software like tyFlow, PFlow, PFlow tools, Particle studio, Sandblaster, Schreiber Ipas plug-ins has always been, You can do very cool effects with those, but those are nowhere near what client wants You to do. That's why I have spent almost one year trying to do cyclone with no success at all and have customer jumping up and down with me.
Early 90's I made cyclone animations without particle systems. I just created lots of curves and lots of particles and made fancy cyclone animations by using follow path feature. No one complained about the quality. It was of course a tedious way to create particle animations - took weeks to do. Now thirty years later, I try to do the same with most sophisticated particle system there is, but haven't been able to produce acceptable results after spending almost a year.
Now I'm running out of money and I'm hoping this community could help me by providing ideas on how to finalize this project in order to be able to start new projects. I would have much more complex particle animations to create. Those too I have made by using the follow path method in the past decades.
If path follow works, and if your client will be happy with that, why don't you use it?
TyFlow have that option too.
Additionally, after the project is done, you can further test other options (unless you're sick of that project. )
What I meant, I didn't use any particle systems back then. I had object and animated it to path. Then I cloned it many many times while changing the parameters of each animation, path and object a tiny bit. They have path animation already, now they want to modernize the animation. That's why I can't go back to path animation this time.
10-27-2023, 01:49 PM (This post was last modified: 10-27-2023, 01:51 PM by tyFlow.)
By the way I did some more testing in the scene you sent me...I am actually able to reproduce the crash when I run it in 3ds Max 2023, but don't get the crash when I run it in 3ds Max 2024. I stepped through the problem bit of code and it seems as though Phoenix FD is crashing specifically when sampling its data at a subframe (which explains why you don't see the crash happening when timestep is set to "frame"). It definitely appears to be a bug within the Phoenix library itself. Have you reached out to Chaos for support on the issue?
Please also double-check that you have the latest Phoenix nightly installed. My Phoenix version for Max 2023 is 5.01.00, whereas my Phoenix version for Max 2024 is 5.20.02. It's possible the bug was already fixed between those versions.
I'm working with 3DS Max 2022 and Phoenix 5.20.02. Chaos knows this issue and they have the same scene as You have. They cannot replicate the crash. However what I've seen their screenshots, tyFlow seems to be working totally differently there that it does in my system. TyFlow particles fly to the opposite direction after birth.
)