Old Density-Collision Bug (why?)

[Paradigm 29]

An old bug in Phun and Algodoo where low-density objects are compressed by high density objects even though all objects in phun/algodoo are rigid bodies.

I can understand why at high speeds there is object penetration (because of the rate that the game processes collisions), but why does a slow moving (nearly stationary) object penetrate? I just would like to hear the official reason for the problem.

[scene]34171[/scene]

[Mystery]

Also adding to this same thing but with hinges. High density object attached to hinge fine, low density object attached to a hinge is all bouncy and springy.

[gradyfitz]

An old bug in Phun and Algodoo where low-density objects are compressed by high density objects even though all objects in phun/algodoo are rigid bodies.

I can understand why at high speeds there is object penetration (because of the rate that the game processes collisions), but why does a slow moving (nearly stationary) object penetrate? I just would like to hear the official reason for the problem.

[scene]34171[/scene]

I don't think this is a bug really, I think of it like if you put a block in water, the block is going to try to push through the water if it's quite dense, the difference here is that they are rigid bodies, so they can't split around the dense object, I think this may also be a side effect of light objects trying to lift into the air , if a splittable area were used (like a few hundred boxes or circles of smaller size), the heavier object is able to get to the bottom easier (as the area is splittable).

I don't think this is a problem, but I'd be interested to see what the developers think of this .

[davidz40]

I think that bug is caused by the fact that very high force (dense object) applied to very light object creates extremely high accelerations, and algodoo has not enough fps to handle them properly, just as with speed over-penetration.

[Paradigm 29]

I think that bug is caused by the fact that very high force (dense object) applied to very light object creates extremely high accelerations, and algodoo has not enough fps to handle them properly, just as with speed over-penetration.

The acceleration of both blocks is the same. Gravity is pulling them both down at 9.8m/s/s, so regardless of their mass the game should have enough time to calculate the collisions of the heavier one if it can calculate the collisions of the lighter one.

...

I don't think this is a bug really, I think of it like if you put a block in water, the block is going to try to push through the water if it's quite dense, the difference here is that they are rigid bodies, so they can't split around the dense object, I think this may also be a side effect of light objects trying to lift into the air , if a splittable area were used (like a few hundred boxes or circles of smaller size), the heavier object is able to get to the bottom easier (as the area is splittable).

I don't think this is a problem, but I'd be interested to see what the developers think of this .

Of course if there were a bunch of small circles you'd expect a dense object to push them out of the way, but that's not the point. These objects are rigid bodies...

If you made a table out of wood and then you put something very heavy on it like a refrigerator or a car then the table would not shake violently. If it were able to support the weight then it would stay still. If it couldn't support the weight then it would just break.
Basically, we're talking about solid objects here - not fluids.

It'd be nice to have an official response to put it all to rest though

[Mystery]

Well if you think about it if they did fix this how many scenes would stop working?

[Sonic]

I don't know. Probably not many though. I've never seen it used for anything.

[standardtoaster]

I used this "glitch" in my polygon cloning scene. I might be able to update it if this was fixed but I think that it's fine the way it is for now.

[ShnitzelKiller]

Well I don't like that light rigid bodies are basically treated as fluids. There are very light but very sturdy materials, and this is then something not doable in phun.

[Laraso]

They aren't treated as fluids. Where did you read that from?

[Chronos]

They aren't treated as fluids. Where did you read that from?

Lrn2read. He said basically. He never said they were, just that it seems like it. [i]Italics[/i ftw.

[Tank2333]

i like that phun handles the collisions like this (for extreme low density its very weird though)
because its a little bit like softbodies and u cant build a mashine that weights 0.1g and pulls a 1000kg heavy thing if u know what i mean
it makes it a bit more difficult and realistic

but a little bit less penetration would be nice though

[Versieon]

Thats totaly not what she said....

[Laraso]

Lrn2read. He said basically. He never said they were, just that it seems like it. [i]Italics[/i ftw.

Learn to spell... And while you're at it, learn to contemplate.

In my vocabulary, the definition of the word basically (the way he used it) is "same as, alike". The way you made it seem like he said it is that he said that they are treated like fluids, but they really aren't.

They aren't treated like fluids AT ALL.

Why don't you look at this scene and tell me which one is more fluid like, the water or the small circles.

Is It A Fluid?

Laraso

Rating:
Filesize: 109.41 kB
Comments: 5
Ratings: 4

EDIT:

but a little bit less penetration would be nice though

But then my needle gun will be obsolete...

[Paradigm 29]

Is it safe to say that Phun and Algodoo do not simulate physics on the atomic level? If that's the case, then then there's no point in arguing whether or not this is a bug.

And if you want to somehow make the case that Algodoo and Phun are simulating physics on the atomic level by allowing penetration of rigid bodies, then you would have to submit to the fact that the simulation is not accurately depicting the real-world results of such a physical reaction. Meaning that this is a bug either way.

[ShnitzelKiller]

i like that phun handles the collisions like this (for extreme low density its very weird though)
because its a little bit like softbodies and u cant build a mashine that weights 0.1g and pulls a 1000kg heavy thing if u know what i mean
it makes it a bit more difficult and realistic

but a little bit less penetration would be nice though

That's what I was saying! I never said that light objects should be able to "pull heavy objects", I meant placing a light bridge over a gap and dropping a heavy object on it shouldn't make the heavy object fall through it. These are RIGID bodies. They need more rigidity.

[Laraso]

The heavy object should never fall through the light object, UNLESS:

1. The objects do not share any collision groups.
2. The heavy object is a killer.
3. The light object is as dense as air, and the heavy object has a mass the equivalent of Earth.
4. The heavy object is a needle shot out of my needle cannon.
5. The light object is built of many smaller objects, put together by hinges with an impulse break limit of 1.
6. The heavy object is falling crazy fast!
7. The light object is as thin as paper.
8. I suppose if you turn the gravity on really high, it could have some sort of weird effect and make the heavy object fall through the light object. (Yes, I am looking for ways to do this now.)
9. You used the drag/move tool, forcing the heavy object through the light object.
10. Thyme.

[davidz40]

I did some experiments with forces visuaisation. I found out that static situations (like two boxes lying on each other) doesn't have completely static forces. The forces vary slightly, and keep some sort of dynamic balance. (in case of boxes, normal forces on left/right side of the box are slightly different). But when light object is pushed with considerable force (like weight of heavy object) the resulting acceleration that's inversely proportional to mass is so high that the object reaches over-penetration speeds. In case of said boxes, unbalanced normal forces try to rotate the light box, and normal forces from ground that should normally counter this are a bit unbalanced too, so the box rotates anyway.

[KarateBrot]

the wiggle of light geometries seem to happen because they can be accelerated very easily by great forces. because of this great acceleration they gain much speed within on calculation step which causes such deep sinking into planes and other solids.
i guess this "bug" can't be solved unless a lot more derivatives of position are included like jerk, jounce, crackle... (advanced Taylor Series)

[Leborothes]

Yeah I have a problem with that problem, Because of it I can't get my gun to work properly, the casings go like half the way through the thingy that it's supposed to go through and then back, reducing the force with wich it pushes the thingy back, making my gun jam every now and then
Oh and paradigm, you're awesome

[KarateBrot]

you can fix it by increasing the frequency. Or the next time build with a bigger scale that should work too.

[Leborothes]

It's already in huge scale, it's like 50 meters long w/o barrel, so size isn't the problem. Also done pretty high ensity on the objects, but don't want to change them too high or something else might go wrong.
And what frequency are you talking about here?

[KarateBrot]

I'm talking about the simulation frequency. Normally it's 100Hz so algodoo calculates 100 steps per second. If you increase it the simulation gets more precise.

Try it:
Open the console (F11) and type in

Code: Select all

sim.frequency = 600


If you're using phun type in

Code: Select all

sim.timedelta = 1.0 / 600


But don't forget to change it back to 100Hz after downloading other scenes because they usually were built for use with 100Hz so they maybe behave different at other frequencies.

[Leborothes]

Thanks alot for that code, it worked and solved that problem