View Full Version : What's max # of points in collision object?
07 July 2003, 05:18 PM
I am working on a very high poly collision object. I am getting an error that says: PFX: illegal polygon over points.
Does anyone know the maximum number of points allowed in a collision object?
Are there known workaround to this, like splitting the object into a bunch of individual collision objects? In which case, anyone know the maximum number of collision objects allowed?
I currently have about 20k points in this object. (thats just the top gutters)
Particles have to go into the holes in the decking, through the gutters on top, fall into the gutters around the sides and then into/through the pipes.
Did I mention thats after the helicopter crashes and skitters around on the deck... sigh. Any suggestions would be most appreciated.
A view from under the object.
07 July 2003, 09:16 PM
Whoa!!! First off, that is a very cool looking landing pad!!!!! Excellent!
Second, in VFX, always make the system do the least it has to for achieving the effect that you want. I would suggest making low poly proxy/collision representation of the objects that you need and make sure that you have only 3 and 4 point polygons in your collision object, higher point polys may not work well or could cause that error.
I would not actually even use the pipes where you can't see the particles moving inside them either. Setup a way for those particles to die off and create new ones at the ends of the pipes where they would come out. That extra processing is not something you should be worrying about or using your CPU cycles on.
Not sure you can do this, but perhaps even set a collision object with no polys and apply an image map for where you want the collision to occur with a grayscale image of the holes in the top of the object.
Good luck and be sure to let us see this work of art when you get it done!!!
07 July 2003, 11:51 PM
I've been working on a lowpoly proxy to calculate the movement of the particles, but thats the low rez one that has 20k+ polys.
I am thinking that I may have to animate the helicopter and then make the proxy only for the areas that I have fuel spill.
Actually I have to show the particles going through the pipes below. I am going to have the helipad and pipes turn transparent, to show specifically the path of the fluid, since that is what this animation is supposed to show.
I have a backup plan in mind with animated clipmaps if this doesn't work out, but it will be a real pain.
07 July 2003, 12:08 AM
but i give you credit for trying!
hmmm, yes your best bet is to first blow up the copter, figure out the exact areas that need spillage...then make only those areas collisions objects, and DEFINITELY dont use crazy hires meshes like that cutaway!!!...simplify.
depending on your scene you can fake the flow once the pipes are filled with maps like you said...
does it need to be realisitc?
nice model by the way!
so whats this for? courtroom litigation?
07 July 2003, 12:31 AM
Yes I confirm, you're nuts :p
In that type of work I usualy think the movie as a number of sequences instead of a unique long sequence.
If (but perhaps that's not possible here) you just need some close up of fluids running through the pipes on a certain distance, then simulate just this part instead of the whole installation.
My 2 euros...
07 July 2003, 03:38 AM
Well if you think that's nuts, you should see my deadline.
The shot I am going for is a wide shot showing a sort of x-ray of the platform with the flow of the fuel highlighted inside.
The purpose of the animation is how this deck moves the fuel away from the crash, via the gutters and pipes and leaves less fuel sitting around to burn.
It's a product demonstration, the client makes these helipads.
I just wanted to do a few tests to see how hard this was going to be with PFX. I did a test on the 20k poly section that worked pretty ok. Hopefully I will be able to restrict the area for collision detection to more manageable section.
So you're saying 147k polys is a little heavy for an object? :rolleyes:
On to the helicopter now.
01 January 2006, 04:00 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.