And while you're at it, you might want to compare it to the old ICE-SPH addon Thiago Costa wrote (the one that involved its own closed testing Google group). Comparing this PhysX stuff to Lagoa almost seems unfair, as Lagoa tries very hard at being/doing too much IMHO. In that light the old ICE-SPH addon (still available @ rray.de) seems to be a fairer comparison...

>>Is that Nvidia's responsibility?
Not Nvidia, more like that it is simage fault, because physx can handle videomemory overflow and automatically switch to the CPU mode.
>> cannot seem to add or delete particles or use age etc without the solver failing.
It is my limitation, I will finish these features after implosia's release and it must be the same stable as built-in particle simulation node:)

Also I have notices that crash is happens in case when we add a huge number of particles at once, if we add the same number during some time - all work fine, seems like that crash is related to some kind of overlapping. In this way I was able to add about 2 millions on my gtx480 1.5gb.

Very nice test btw!

ok think I reached 2 million now by introducing slowly. . rendered with Fury2 so thats GPU rendering ontop of the GPU simulation. works out about 3 seconds per frame. 500 frames rendered in 10 minutes.



Interesting fact discovered to turn on CPU rendering mode is to set the max allowed particles to more than your card can cope, it then does give a n ice message to successfully kick in CPU mode for allowing more particles in the simulation.

Tekano,
In this last video the scale of the liquid tank seems to be several meters in length and the simulation looks believable. Can you please do a test at lower scales? Like a glass of water or a bucket. I'm curios how's holding up on this practical example.

here you go, its just a matter of setting the gravity x10 so instead of meters its cm and fiddling with the stiffness & viscosity until it moves ok.

First of all thanks for the simulation!

Now are you sure that modifying the gravity is a good idea? I know that in 3d you tweak the values until it looks right even though the values are not by far the physical correct ones...
Anyway, it looks alright but not that good when compared to the larger one. The liquid pouring from the glass should go straight down IMO. Just an idea.
Too bad I can't play with this right now, but it's good to be on the forums most of the time.

Hi Mcnistor , am positive that changing the gravity in this instance is the best thing to do, the units in Softimage are arbitrary so 1 unit can equal 1 metre or 1 cm so this is all I am changing. the first splashy sim they were at 1 meter per unit the 2nd glass one is 1cm per unit. simples. Yes, in the last example they really should not come out so far as you say but then it was even more boring I think the viscosity or stiffness is too high so there is maybe some surface tension or other force making them push out that way. Often its easier to have a live action example to copy as pouring water can be fairly subjective after a few different sims

playing with this is much more fun than waiting up to 100 times longer for the lagoa or SPH on cpu simulation thats for sure

hi, i got a bit of time to play with this today, but i'm getting this on the script editor


// INFO : Cleared PhysX World!
// INFO : Get CUDA support failed! Created PhysX World!
// INFO : Succefully setted Fluid Attributes
// INFO : ADD STATIC OBSTACLE: STAT OBS ADDED SUCEFULLY!
// INFO : ADD NEW PT: New Particles Added Succefully!
// INFO : Performing Step Success!

some ideas? Dont think its using the gpu! i got a 560 card
Any suggestion where i should look to fix it?

Thanks, Max



edit: fixed by copyin the dlls in main softimage root folder!

I'll post an update later (most likely after kepler releasing), I am currently sitting on the 7300GT so it's a quite difficult to debug physx on it

This really looks fantastic! Unfortunately I get the same issue as mamali132002 has. Scene loads, particles apear, but dont get any simulation going on. Under plugin manager, the .dll is marked as invalid as well :/

In any case though. Great work as it seems from above!