Thinking Particles

#961

Not low-high poly replacement, but simulating simple, jointed geometry structure that the higher detail one follows, and then smaller local simulations where the high detail pieces get released and interact.

#962

Thank Hristo, but when i want similar cutted pieces from proxy and from high mesh, how i do same cutting on both? Or maybe i dont understand the workflow, can you please tell me more about it, like for a dummy :slight_smile: thanks a lot

#963

Hi 3ak!

Thank you always!! I could open it. I found it seems so fundamental,hehe. But it’s good opportunity to review basic stuff.Sometimes I messed up.

Kei

#964

hi,
may i know if there is a way to make the paper fall on the ground flat after it fall from the sky i have a problem with the SC node to achieve this effect.

#965

Hi, I think you have 2 options:
1- Play with the SC settings until it falls flat (I guess your issue is that they “stick” on the ground vertically?)
2- Do a “bring to” (just alignment) a few frames before they hit the ground.

could you post a sample scene with the issue?

#966

You’re welcome)

  1. SC need some thickness to operate on. you can’t use planes.
  2. If you want you paper sheets land only horizontally you can check for altitude (Intersect node or just paperZpos - groundZpos if you have plain ground) and after some threshold start to rotate sheets of paper. Obviously you need to have alignment info on them.
#967

Hey,

What actually Particle data is for? I saw lots of guy using particle data in his tutorial. But no one have defined it properly? I found two particle data in tp, one in helper and other in operator but what are the use of them? what is the main purpose of using particle data?

#968

Hi,

Currently I am doing a project, mainly splashes/ whitewater using Realflow Hybrido. I like the overall splashes, but we are missing some veery fine driplets that are created in nature when fast moving drops of water dissolve in the air into finer drops.

I don’t think this is something I shoud try to achieve inside Realflow by increasing the Resolution parameter until I get that amount of detail into my particles, because this might be too much in the sense of simulation time and diskspace.

I was thinking about creating a setup in TP by importing my Realflow particles. Inside TP, I wanted to build a system that checks if certain Realflow particles fulfill my conditions (age, speed…something else) and if they do, use them as emitters for tiny trail particles that serve as my dissolve drops.

Now, there’s my problem…How do I get Realflow BIN particles into TP? Has anyone of you guys ever tried that?

Thanks in advance for any tips!

#969

There is no public bin import tool available afaik. But pflow can help you with that)
Possible solution is to use fumefx to emit non-turbulent, heavy smoke from splashes and then advect particles by smoke vel field (or simply use fumefx birth/follow ops in pflow or TP). Then render them with krakatoa.

#970

I have a scene with several different particle instances that scale up as they drop to a deflector surface. To get the instances to not occupy the same space, Show Voxels was activated to increase the voxel level. After this was done, the viewport stopped responding to the time slider, but the scene rendered properly for the time scale. What’s happened? And can something be done to correct this? Thanks.

Couldn’t find where to start a new thread.

#971

Show voxels is not doing anything with the simulation, it just shows the vogel resolution in the viewport. In your case it probably was too much for it to handle - I’d turn it off.

#972

The helper is for reading particle data from particles besides the ones coming from a group node - like the ones from the outputs of SC, reference, psearch, etc. Connect that particle output to the helper’s input and you can get the particle channels like size, velocity, etc. from it’s output.
The particle data operator is for writing into these channels - so you can override the age or lifespan for example.

#973

Some “Thinking Particles”-Training over the weekend.

Thanks to Andrew Kramer for the original animation. :slight_smile:

Uli

https://vimeo.com/44619833

#974

hi Hristo
i was watching ur videos about the electric arc on your site
you r the master as usual
but i was wonder if i want to animate the target object how i make the particles follow it ??

#975

When I have a situation like this, when I need to operate on the position directly, I create another particle group that references the first one, and I use the position follow node to follow the reference. This way you can have both the direct manipulation of position, and smooth movement with velocity .

#976

Hey guys!

When it comes to dynamics simulation, I always end up iterating a lot and spending a lot of time waiting for a sim. So I’m currently looking into a hardware upgrade to improve my sim times for nCloth, thinking particles and the like.
Many of these solvers seem to not be multi-threaded, so I guess it’s all about raw GHz.

I’m currently on an Intel Core i7 920 @ 2.67GHz and caching 100 frames at 4 sub-steps of a brick-wall with SC collision takes almost 12 minutes.
The same sim takes about 6 minutes on an Intel Core i7 3930k @ 3.2GHz, but that’s still a lot when I do it 10-20 times a day.
Tightly stacked objects seem to be particularly slow with SC for some reason.
Both machines have 12gb of RAM or more and cache to an SSD.

What kind of hardware do you guys use in production for dynamics work?
Can you actually iterate, or do you have to cheat a lot with combining caches, multiple passes and key-massaging?

Thanks for the insight!

#977

In my humble opinion and experience mostly with pflow, ffx, rf, krakatoa/frost/xmesh but some TP, the fastest per core CPU is the way to go. It is far better to have 6 cores running at 4+ ghz than 12 cores running 2.0 ghz simply for the lack of full multi-threaded processes. Ram starvation can also an issue as soon as you have to cache memory to disk say goodbye to performance.

Caching to/reading from SSD SataIII is the best currently affordable option too, it can be up to 5x faster than SSD SataII.

I have no idea how everyone pulled off those huge building demos in 2012 :slight_smile: Can’t image that was quick to iterate.

#978

Thinking Particles 5 just released :slight_smile:

http://www.cebas.com/?pid=hot_news&nid=489

Some nice new features there

#979

It seems TP5 will have almost everything I need to have.

Only thing I still want is noise node.

#980

By not putting everything in one sim :slight_smile: