Creating a water buoyancy system

Leakbang

I want to create a simple but robust water buoyancy system that makes some objects float in a body of water. Initially, I thought it is possible to do all of it with a simple Area node without any code but I found out that it will be more complicated than that. The idea is that heavier rigid bodies fall down easily while lighter objects float for longer. Also I have in my mind to differentiate between objects that can or can't float by using groups. For instance if the object is in group "wood" then halve the rigidbody weight temporarily so that it floats. But I don't know how to tackle this. Initially I used the on_body_entered trigger to add a force to the entered object but that didn't work, because the force is only applied for a single frame and that's it. I also thought of storing the objects in the area in an array, but I don't think that it is performance friendly and robust. I'm at a loss. I would greatly appreciate your ideas and help.

xyz

Rigid bodies have Gravity Scale property. Why not use that to fake buoyancy.

Leakbang

I just tried that, but it gives more the feeling of space rather than a body of water. Still an upwards force needs to be applied to the rigidbody.

cybereality

Can gravity scale be negative?

Leakbang

So I came up with this code: When a rigidbody enters this area, it applies an upwards vertical force to it equivalent to it's weight.

extends Area

var Vmass
var Rbody

func _process(delta):
	if Rbody:
		Rbody.add_central_force(Vector3(0, Vmass, 0))


func _on_Area_body_entered(body):
	Rbody = body
	Vmass = body.weight

But the problem with this is that if there is more than one rigidbody in the area, it won't work. It essentially only can apply the force to one rigidbody. How can I make it apply the force to multiple objects at the same time?

cybereality

You are saving body into the Rbody variable, so it will only receive the last body that entered (only one). You probably want to add body to an array, and then in process loop through the array. Also, in that case you will need to remove the body from the array on area exited.

Leakbang

@cybereality said: You are saving body into the Rbody variable, so it will only receive the last body that entered (only one). You probably want to add body to an array, and then in process loop through the array. Also, in that case you will need to remove the body from the array on area exited.

I originally thought of it as well. But my assumption was that putting the objects into an array is not optimized. Wouldn't it become a problem if there are a lot of objects in the array?

cybereality

Well it depends how many objects, but up to a few dozen should be fine. The array itself and looping is super fast, but the physics can slow down with too many bodies. But the code needs to live somewhere, so if you are applying forces, it might actually be more optimized to loop in a central location (rather than having each rigid body have it's own buoyancy code).

Leakbang

@cybereality said: It might actually be more optimized to loop in a central location (rather than having each rigid body have it's own buoyancy code).

Yes I agree with that. That's what I'm currently doing as well. Thank you a lot!

Here is the updated version of the code above:

extends Area

var Vmass = []
var objects = []

func _process(delta):
	if objects:
		for i in range(0, objects.size()):
			objects[i].add_central_force(Vector3(0, Vmass[i], 0))

func _on_Area_body_entered(body):
	objects.append(body)
	Vmass.append(body.weight)


func _on_Area_body_exited(body):
	objects.erase(body)
	Vmass.erase(body.weight)

Hope this can help others out.

xyz

Since you apply permanent constant upward force, that's exactly the same thing as scaling the gravity, but without a need to run any script loops per frame. Gravity scale can also be negative. body.gravity_scale = (gravity-buoyancy) / gravity

Leakbang

Yes, but this allows for a more flexible and realistic system. As this is how it behaves in real-life. To make this system behave to real-life water bodies, I somehow need to figure out, how can I apply more force the deeper the object goes.

xyz

Same way you can apply less gravity ;) It's a bit more efficient if you need to run many objects as no new forces are introduced into the simulation. Buoyancy always directly counters gravity, so no real need to introduce another force vector when you can just adjust magnitude of an existing force vector.

cybereality

You'd just need to take the difference from the area position (offset to the top of the water) and the body position on the y axis and apply the force proportional to the distance.

cybereality

I haven't benchmarked it but I would assume using gravity scale might be a little faster than apply forces. But for accurate water you need to update the gravity for each object as it falls into the water, so I don't see how you can avoid the loop.

xyz

Buoyancy is equal to the weight of displaced fluid. So if you want a realistic looking effect you need to take apparent volume of the submerged object into account as it directly affects the amount of displaced fluid. Also, once the object is fully submerged the buoyancy force stays the same. It does not increase with depth.

Leakbang

@cybereality said: I haven't benchmarked it but I would assume using gravity scale might be a little faster than apply forces. But for accurate water you need to update the gravity for each object as it falls into the water, so I don't see how you can avoid the loop.

Yes you are probably right. I'd reckon that gravity scale is also faster. Also there is also the case of objects being stacked on top of each other. It means that the more objects are stacked ontop of each other, the more force the object on the bottom is applying. I did some searching and I can't find a way to calculate the amount of force. However, setting the gravity scale of the objects to -1 seems to deal with this issue nicely too. The problem with gravity scale is that it pushes the objects up and when they reach the top, they keep wobbling back and forth into the area.

Leakbang

@xyz said: Also, once the object is fully submerged the buoyancy force stays the same. It does not increase with depth.

Yes you are right. I meant that the water density also can effect it. I believe that the water density increases with the depth and the stationary floating object has the same density as the water (at that level) so the floating object is heading towards the water level with the same density.

xyz

No, the water density is considered constant. Each object that floats is actually partially submerged, to a degree that the weight of fluid it displaced becomes equal to object's weight. Thus equilibrium is established, gravity is completely nullified, and object floats in a zero-weight state. If object is more dense than water this can never happen so it sinks under influence of diminished gravity. For a fully submerged sinking object, gravity magnitude is lessened exactly by the amount of maximal possible buoyant force, and it stays the same until object reaches the bottom.

So submersion depth is relevant only until object is fully submerged. After that, the buoyant force becomes constant and you don't need to update it.

Megalomaniak

IIRC water, for simplicity sake at least is considered incompressible?

Here googled it: https://www.usgs.gov/special-topic/water-science-school/science/water-compressibility

Water is essentially incompressible, especially under normal conditions. If you fill a sandwich bag with water and put a straw into it, when you squeeze the baggie the water won't compress, but rather will shoot out the straw. If the water compressed, it wouldn't "push back" out of the straw. Incompressibility is a common property of liquids, but water is especially incompressible.

cybereality

So I tried messing with it. The buoyancy works pretty good, with a fully submerged object. But I'm not exactly sure how to adjust it when reaching near the edge of the water. I can get the depth just fine, but simply adjusting the gravity scale doesn't seem to work right. It just ends up oscillating around the edge. And damping linear velocity sort of hides this, but it's a hack and also doesn't work great. I can look into it more later, but maybe applying forces when near the edge of the water would work better.