CollisionPOP and oversampling.

   4028   8   0
User Avatar
Member
509 posts
Joined: July 2005
Offline
Hi there,

I'm trying to make particles get stucks on my deforming object after they are emitted … but since ther's a oversampling if using “traslating geometry”… (which doesnt really need in my case anyway) how do I oversample the collision computation with deforming geometry?.. (its triangulated and I tried to subdivide it a lot as a chance to get more accuracy) … i got many particles “loose” the surface ..

thanks in advance for any help

cheers.
JcN
VisualCortexLab Ltd :: www.visualcortexlab.com
User Avatar
Staff
1072 posts
Joined: July 2005
Offline
Unfortunately, the deforming geometry hint does not support oversampling of the geometry in the collision POP itself. You'll have to oversample the entire simulation.

The collision tolerance parameter can also be of significant help in eliminating leaks.
User Avatar
Member
509 posts
Joined: July 2005
Offline
Ondrej
Unfortunately, the deforming geometry hint does not support oversampling of the geometry in the collision POP itself. You'll have to oversample the entire simulation.

The collision tolerance parameter can also be of significant help in eliminating leaks.

forgot to mention.. i tried oversampling the entire simulation too (paramenters outside pop network SOP right?).. .and no luck .. oh well… maybe I need to know what “oversampling” means… i used something around 5/8 as value… maybe i'd try much higher value?

tollarnce also is something I've tweaked… but didnt helped…

cheers
JcN
VisualCortexLab Ltd :: www.visualcortexlab.com
User Avatar
Member
639 posts
Joined: July 2005
Offline
Hello, I could be wrong, but oversampling is the number of cook happens in between frames to get a much better accuracy. If you set it to 5/8, you have much greater chance of “miss” than hit.

HOpe this helped.
User Avatar
Member
509 posts
Joined: July 2005
Offline
well.. talking about chances… you're right.. but I'm also .. I'd have more chances to “hit” it…

anyway .. seriously.. so what's the way to have such collisions better sampled?… everything i wanted to do .. was emitting particles from an object..which I want stay stick until a certain thing happens.. but unforntunately.. most of ‘em get unstuck cause of bad collisions…

I’m openmind for every alternative if this cant be solved with collisions accuracy..

cheers.
JcN
VisualCortexLab Ltd :: www.visualcortexlab.com
User Avatar
Member
639 posts
Joined: July 2005
Offline
Hmm… well, I guess trying to sample particle collision on any arbitrary shape object can be somewhat of a tricky to deal with at times other than, well, tweaking the sampling and collision tolerance . When a particle is traveling on a certain path toward an arbitrary shaped object, the particle has to determine “Am I still outside this object or am I already inside?”, so between the current frame and the next frame, there can be “infinite” amount of frames in between. Of course, this is without getting into the idea of voxelising or level set stuff. But thank goodness that we don't need such microscopic accuracy for every collision, so we can cheat a bit by turning up the oversampling collision tolerance – which for the most part will solve most of your collision issue unless you have very high velocity particles.

So, the faster your particles are traveling – chances are, you may need higher oversampling.

As far as the particles goes “unstuck”, did you set your “stuck” setting in your collision POP? i could be wrong, but once a particle is “stuck”, it's will remain stuck until you set an event to tell it to unstuck, no?


Anyways, hope that helped.

Take care.
User Avatar
Member
509 posts
Joined: July 2005
Offline
your explaination helped a lot.. really .. but my own scene here.. have particles that births from the object… stays stuck… and the object moves.. once the movement is faster than a threshold i decide with an attribute.. i give the stuck=0 “statePOP” … that's the final idea actually…

so the particles arent moving and then colliding.. they're born on the object.. and when the movement is too fast (which actually aint to my eyes.. i mean.. it aint that fast i think)… they get like stuck=0 all alone.


anyway … i guess is time to properly ask help about my goal on this scene.. ehehe… what I wish.. is some particles emitted when the object stops from a fast movement (or shape animation.. that's why I used trail until now and collitionPOP with “deformation”)… let's say like a wet tennis ball that loose drops every bounce on the floor.. any advice is really welcome.

cheers
JcN
VisualCortexLab Ltd :: www.visualcortexlab.com
User Avatar
Member
639 posts
Joined: July 2005
Offline
sum][one
your explaination helped a lot.. really .. but my own scene here.. have particles that births from the object… stays stuck… and the object moves.. once the movement is faster than a threshold i decide with an attribute.. i give the stuck=0 “statePOP” … that's the final idea actually…

so the particles arent moving and then colliding.. they're born on the object.. and when the movement is too fast (which actually aint to my eyes.. i mean.. it aint that fast i think)… they get like stuck=0 all alone.


anyway … i guess is time to properly ask help about my goal on this scene.. ehehe… what I wish.. is some particles emitted when the object stops from a fast movement (or shape animation.. that's why I used trail until now and collitionPOP with “deformation”)… let's say like a wet tennis ball that loose drops every bounce on the floor.. any advice is really welcome.

cheers

Hmm… I am not sure I fully follow you, can you post a simple hip file of the problem you're describing?

If you want your particles to only emit from the surface once it stops, you calculate the magnitude of velocity of your object, length(v), to determine if an object has stopped. If length(v) is 0, then the emitter object has stopped. So then you can ramp up the amount of particles to be birthed on your birth parameter, would this work?

Often time, I actually prefer to animate my birth amount rather than the activation – unless it's just simple on-off activation that I need, of course. : )

Unfortunately (or fortunately), how fast is considered “fast” is pretty arbitrary without any proper frame of reference. However, when you start dealing with dynamic simulation stuff, this unit scale can make a difference. I think most Houdini artists tend to look at 1 Houdini unit as 1 meter (which I believe is the default unit). So if you have a particle that travels a 1 unit/frame (depending on your frame of reference, this might look pretty slow or fast), this translate to about 50 mile/hour (or about 86 km/hr) – assuming 24fps.
User Avatar
Member
509 posts
Joined: July 2005
Offline
true .. tha'ts quite the speed I'm having… well.. i'm moving around 3 units (I also consider 1 unit as 1 meter in Houdini) in around 5 frames…

I'll post a scene later in few minutes… but yeah that's almost the result you wrote… with the differnce that the object starts still.. then moves.. then stops… so the “speed=0” expression wouldnt work maybe…

right now I'm actualy using the lenght() expression.. and I'm having almost the result I need exepct I loose particles from the surface cause of collision lost..

scene to come soon…

cheers
JcN
VisualCortexLab Ltd :: www.visualcortexlab.com
  • Quick Links