Mysteriously repellent physics objects

I was messing around with some rather oversized *kofkof* bricks last night and noticed something a little odd. I don't know if these count as bugs or what...

Make two cubes. Any size will do (however, at least half as large as you would be useful).
Activate physics for them both.
Stack one atop the other.
Peer in-between and note that they seem to refuse to actually touch each other--or even come closer than an inch or so.

Oddness.

The other thing is that hollowed objects appear to count as solid all the way through with respect to physics.

The "bounding box" for a cube is or can be slightly larger than the cube itself. This is not expected to change until the havok1 physics engine is replaced which was suppposed to happen in 2004, so I wouldn't hold my breath until then.

The physics engine used in SL is an old one. As such, it follows older principles of physics. The gap you see is actually filled with aether.

http://en.wikipedia.org/wiki/Aether_(classical_element)

OK, here's another interesting 'feature'

Make a 10m open topped transparent physical box (i.e. 5 sides and an open top)

Now rez enough 0.5 m physical spheres in the box to nearly cover the floor.

Go away for a bit.

Return.

Brownian motion anyone?

I have no idea why - and don't ask what I was up to. The spheres were being returned two days later from up to 5 sims away...

PS, the returning balls were after I did something really silly with a physics disrupter, the box demo is 'safe', you can even fit a lid if you are worried. For my next trick, the same but with a hollow cylindrical box...

It is actually sorta like brownian motion or perhaps the uncertainty principle and quantum tunneling. The physics system wants to settle objects to zero-energy but isn't perfect about detecting collisions. Therefore your spheres tunneled through the box much like an electron passing through a potential barrier.

Well actually it has more to do with time-stepping physics (which all physical simulations have to do) and imprecission in collision detection. You would likely find larger spheres and thicker walls would have prevented or mitigated that effect.

Okay, but the problem still remains of how to make things actually look as if they're touching.

Do I need to just focus on making everything look cel-shaded and set the outline object to be phantom or what? Any ideas other than that one? (I think I'm on the right track)

When the grid opens back up, I'm gunna start selling hollow spheres to new-age avatars, and call them "Spirit Balls"


edit: wait a sec -- it's open? ....nope

Well I didn't 'loose' any until I was silly with a physics disrupter script. What I was referring to was the seemingly endless gentle movement of the balls.

I would suggest that you forget about physics entirely and use the kinematic functions instead. Not knowing what you are shooting for, it is hard to say more.

Even making wild ass guesses from your allusions above, if you were making shooting targets, you can still have the target detect a collision with the bullet, exchange momementum information with the bullet and move itself appropriately. Much more deterministic than SL physics, but then again, so are bullets to a first approximation.

Second time I've laughed like an idiot this morning! Thank you.

Actually what I had in mind were things like giant drooling spiders to see if I can actually (finally) write workable scripts that communicate between parts of an object (so it can crawl over things without having legs hanging in the air), and maybe some pocket monsters. The gaps between objects just seemed strange in the extreme.

Based on the scripted quadrupeds I've seen to date, I think you are looking at an ambitious first project. D&D Dogs might be a good place to observe how some of the better attempts at this have gone. Good luck, the world needs more giant drooling spiders.

<tangent>

"Therefore your spheres tunneled through the box much like an photon passing through a potential barrier."

</tangent>

<normal>

"Therefore your spheres tunneled through the box much like a photon passing through a potential barrier."

</normal>

Hey buddy, I didn't take a load of scattered photons in the gonads from a dropped packet of near light-speed electrons at Brookhaven National Synchrotron Light Source so that you would be safe in your little world to lecture me about the properties of biased PIN diodes in low-pressure liquid helium at 2.7K and the quantum tunneling effects of electrons while you sit safely in front of your CRT bathed in a gentle shower of photons and mostly attenuated X-rays. Hmpf!

Well okay, it was a small packet, and my kid hasn't grown a third arm yet, and I'm not Dr. Coon, but all the rest is true, except the macho attitude of high-energy physics geek street cred.

( )

Pssht. I can do one better. Make a column about as high as you are, and then make a sphere of the same x,y dimensions. Put the sphere half in and half out of the top of the cylinder, hinge the two objects together, and then enable physics.

Prims apparently can and will *explode* outright. No need for the subtleties of brownian motion.

*Scuttles of to the nearest sandbox to try it out!*

The first one fell over, 'exploded' and went off world. After that I had to 'edit linked parts' the sphere to make them explode. One just spun round very fast.

What's going on here?

Well, apparently if you slam prims together hard enough, they break apart into something called "triangles" which only exist for a femtosecond before releasing a burst of energy.

I suspect this is why the thing won't normally let you enable physics on interlocking objects. For some reason the combination of cylinder and sphere I happened upon seems to fool it into allowing such things.



Now, if I found a way to make normally 100% stable prims blast themselves to smithereens in an hour or two of poking and prodding, imagine what kind of apprehension I have been approaching the subject of self-replicating objects with.

Ah hah! All is as clear as mud.

BTW, please don't encourage me to do stuff like that during a grid crash again... For a horrible few minutes I thiught I might have done it

Interesting that 0.5m column/sphere pairs are fairly unstable, .025 and 1.0m ones much less so. I feel the need to experiment further...

"Triangles" are purely theoretical - they're more a mathematical conveniance than anything. I don't think they really exist - I mean, really, what kind of particle (sub-primecular or not) comes in red, blue and pie flavours?

Think I may have to follow some of you folks to the sandbox myself, haven't seen that before. Could be interesting . . .

Hey man... I didn't make any promises about it being *safe* to do this. I felt like I should make myself a blast shield to stand behind after the first one blew up.

Ak! Not safe? You might have SAID...

Hey, life as a software tester becomes pretty boring if you tiptoe around "extreme" uses of normal functions in software. LL and the Wiki document clearly the things they think are unsafe to play with, so for as far as I'm concerned if I'm not trying to to deliberately crash things, and no one's thought of it before, I'm just exploring new spaces.

(So...who wants to help make a prim supercollider?)