⚛️┃physics

[quick rock]

still i would expect to have it work well with physics since its in fixed update.....

[timid dove]

Looks like you have some other script or code resetting velocity to zero

Most likely your movement script

[quick rock]

There is nothing setting it to 0. The only script manipulating velocity in any way is by addForce(direction*speed, forceMode.acceleration)

And the velocity is not 0 unless I'm standing still ever

And I'm printing this while moving around and I can see in the rigid body that velocity is rather high

[timid dove]

The logs seem to disagree.

[quick rock]

Well.

YOU WERE RIGHT :---)

Im sorry but i was hell bent on the thought of the physics handling itself messing up

well

great to know now

thank you so much for ur patience

[timid dove]

public Vector3 customGravity;

void FixedUpdate() {
  rb.AddForce(customGravity, ForceMode.Acceleration);
}```

You can either let this be additive or turn off normal gravity to replace normal gravity entirely

[timid dove]

Actually this looks like your movement code is causing an issue

You are likely setting your character's vertical velocity to 0 in FixedUpdate. The fix is to stop doing that.

[neon sedge]

I added a force, not set the velocitu

[timid dove]

Based on the way your character is moving it seems like you're doing something fishy

You don't have a parabolic motion curve going on

[neon sedge]

I did use Impulse, not force

[timid dove]

Your gravity appears linear, which indicates an issue with your code

[timid dove]

Why don't you just show the code?

The force mode wouldn't cause this

[neon sedge]

I am not on my computer, but I'll type if.

rb.AddForce(Vector3.up * jumpForce, forcemode);

//forcemode is set to Forcemode.Impulse in the editor.

Note that I modified rhe drag

[timid dove]

The drag is the problem

Set drag to 0 and the gravity will be fixed

[neon sedge]

How do I make the player's movement smoother?

Like, not slippery

[timid dove]

That's a really vague question

[neon sedge]

But also without addint too my friction

[neon sedge]

I'll just tinker around eith 0 drag and see what I want done

[timid dove]

Use higher acceleration forces and enforce a maximum speed limit

That will make it snappier

[elder patio]

Does anyone know how to fix this problem? When the ball hits the player, they get pushed back. I want it so that the player is static and the ball doesn't have influence over the player via force.

[timid dove]

use a kinematic character controller instead of a dynamic Rigidbody

[elder patio]

How do I do that?

Would that mean I have to reprogram all my movement?

[timid dove]

yes

[elder patio]

Ight

Thanks

[distant crater]

https://pastebin.com/RKGn31wR got an issue with ice physics in my project, especially on slopes. sometimes i slip and sometimes it's inconsistent.

https://streamable.com/gjexgy?src=player-page-share

[loud ridge]

How can I prevent this?I know I can make weapons trigger but I want them to collide.

[tender vapor]

for colliders to collide within the physics system, at least one has to have a dynamic rigidbody

https://unity.huh.how/physics-messages/collision-matrix-2d

[loud ridge]

I meant that stutter effect

that happens when those weapons get between the hinge joints

[steel pulsar]

Does anyone know how to fix this problem, I have given my chicken object a Pickupable layer and the chicken should directly transform to handtransform, why is the chicken not in the hand but in another position?

[timid dove]

Code?

[steel pulsar]

using UnityEngine;

public class ObjectPickupDrop : MonoBehaviour
{
public Transform handTransform;
private GameObject heldObject;

[SerializeField] private float pickupRange = 2f; 
[SerializeField] private LayerMask pickupLayer; 

void Update()
{
    if (Input.GetKeyDown(KeyCode.E))
    {
        if (heldObject == null)
        {
            TryPickupObject(); 
        }
        else
        {
            DropObject();
        }
    }
}

void TryPickupObject()
{
 
    Ray ray = new Ray(Camera.main.transform.position, Camera.main.transform.forward);
    if (Physics.Raycast(ray, out RaycastHit hit, pickupRange, pickupLayer))
    {
        if (hit.collider != null)
        {
            PickupObject(hit.collider.gameObject);
        }
    }
}

void PickupObject(GameObject obj)
{
    
    heldObject = obj;

    Rigidbody rb = heldObject.GetComponent<Rigidbody>();
    if (rb != null)
    {
        rb.isKinematic = true;
    }

  
    heldObject.transform.SetParent(handTransform);
    heldObject.transform.localPosition = Vector3.zero;
    heldObject.transform.localRotation = Quaternion.identity;
}

void DropObject()
{
    
    Rigidbody rb = heldObject.GetComponent<Rigidbody>();
    if (rb != null)
    {
        rb.isKinematic = false;
    }

    
    heldObject.transform.SetParent(null);
    heldObject = null;
}

void OnDrawGizmos()
{
    if (handTransform != null)
    {
        Gizmos.color = Color.green;
        Gizmos.DrawSphere(handTransform.position, 0.1f);
    }
}

}

[coral urchin]

Hey there, has anyone made a rigidbody character controller for very large entities and can share some best practices? I am wondering whether I should use rigidbodies at all, or use them with the kinematic option.

[honest river]

Looking into options to simulate 3D in 2D in a kinda-top-down ~45° camera angle.
The characters can move on the ground but then also jump:

Right now I have gravity set to 0, 0 in project settings for 2D.
I see a few ways to do it.
First:

1. Set the gravity vector to 0,-9.81 in project settings.
2. Normally, all the characters and moving objects have gravity scale of 0.
3. When the jump starts, set gravity scale to 1, launch character up, remember the position when the character jumps, then on each update in the jumping state check if character's position is the same or lower than the starting position. If it is lower, then switch back to the moving state and set the gravity scale to 0.

Second:

1. Set the gravity vector to 0,-9.81 in project settings.
2. Add colliders under each character that always follow them and only interact with that specific character. This probably means a lot of collision layers.

Third:
Switch to 3D physics.

I wonder if I should go 3D physics with 2D graphics here

https://cdn.discordapp.com/emojis/826227356477816864.png?size=48

[timid dove]

!code

[flint portal]

Posting code

📃 Large Code Blocks

Use links to services like:
https://paste.mod.gg/, https://hastebin.skyra.pw/ , https://paste.ofcode.org/ , https://paste.myst.rs/

📃 Inline Code

Surround code with three backquotes. Not quotation marks.
To format as C#, add cs to the first line:
```cs
// Your code here
```
Add a comment with a line number if there is an error message.

[timid dove]

It's likely a scaling issue

[dawn gazelle]

KITTY GAME

[dapper spruce]

I've seen how varying time steps for Physics.Simulate() can cause some weird break down in the simulation (e.g. objects at rest suddenly bouncing around), but it got me curious- why is that?

[novel light]

Having a fixed time step is just simply better for physics stability

Having continually varying delta time will give you inconsistent calculations.

It's easy to show that. 30 physics tickrate vs 60 physics tickrate will give you a different feel for the same vehicle controller, for instance. The calculations are simply not going to be the same, due to numerical errors, etc.

So, by having a non-fixed tickrate, you take this inconsistency into a whole different level.

[round solstice]

I'm experimenting this weird issue (each plank has a collider and a rigidbody and the planks on the middle has fixed joints to hold them meanwhile a force it's not applied

[timid dove]

Sorry it's not clear exactly what your issue is

I will say that the Rigidbody should probably be on the root object of that Pallet prefab, not a child.

[round solstice]

but each plank should be independent

[timid dove]

What is a "plank"? I only see "Pallet"s here

[round solstice]

the planks are vibrating

[round solstice]

parts of the pallet

[timid dove]

you need a root rigidbody \

and to have the children be rigidbodies connected to the root with joints

but it's not really clear why you need this complicated setup

What are you trying to achieve with this?

[round solstice]

this is a video that I recorded a few time ago, but now with the new (and bigger) terrain I get this issue

as you can see each part of the pallet is independent when it "breaks"

[timid dove]

I would just have a single Rigidbody and a single collider here, until it breaks. When it breaks, you swap in the separate pieces.

It will behave (and perform) much better

[round solstice]

I moved the pallets to 0,0,0 and now there is no vibration, I think that the issue is caused by the coordinates, does the physics system have issues with big numbers in coordinates?

[timid dove]

pretty much every system has problems when you get too far away from the origin

Best practice is to stay within ~1-2k units from the origin.

[round solstice]

well, then probably that was the cause, I was on 5k, ?, 3k...

[sharp tulip]

Is it ok if I use transform.rotate for a kinematic rigidbody? Said rb is a sphere collider and I'm wondering if there are any repurcussions to doing this

[bleak umbra]

@humble plaza you probably need to clamp the movement of your character to the hit distance such that your character will still move that last bit and not simply stop when there is any hit within max raycast distance

[unique cave]

Essentially the same as rb.rotation. I'd rather use the rb one but that doesn't do great job respecting physics either. For kinematic bodies though there exists rb.MoveRotation which modifies the .angularVelocity making interactions with other dynamic bodies much more accurate. The .Move... methods + kinematic body are usually preferred for moving and rotating obstacles that don't need to be dynamic

[gray sentinel]

How much of an issue is collision mesh complexity with performance?

Maybe that question is pretty worthless without context...

[timid dove]

depends on so, so many things

[gray sentinel]

Yeah... I hate to go down the route of prototyping it and it not work, but perhaps that's my only option.

[bitter haven]

So, I have a Rigidbody-based character controller that @thin dew is doing almost everything to develop.

But I'm running into a problem.

When I jump the player into a wall, the player sticks to the wall until the player is no longer holding in that direction.

[bitter haven]

Is anybody here?

[spare parcel]

can anyone help me, I'm trying to recreate ridge racer 4 in Unity, I have a ground with box collider and rigidbody, but for some reason the car is falling through the map. I don't know what should I do? The root CAR object has a rigidbody and a car controller script, the gameobject child BODY has a mesh collider with mesh renderer and then there are wheels with wheel collider.

am I missing a lore here on this simple car controller?

[timid dove]

Ridge racer definitely doesn't use a traditional physics engine so I would say you're probably on the wrong track entirely.

The way the cars hug the track it's almost definitely spline-based with custom physics.

[spare parcel]

No, I meant to say, how should I stop the car model falling through the mesh instead of standing on the mesh.

[timid dove]

Isn't "standing on the mesh" and "stop falling through the mesh" the same thing?

[spare parcel]

I'm so sorry, I meant, how can I make the car to stop falling through the mesh. Forget the first part "standing on the mesh"

[timid dove]

It will not fall through if there's an appropriate Collider

[spare parcel]

i Have a box collider on both the car and the ground!

[silver moss]

gameobject child BODY has a mesh collider
Is the mesh collider marked as convex or no?

[royal willow]

show the inspector for the ground and the car

[spare parcel]

I'll show in few hours because its snowing outside and I want to eat snow XDXDXD

[hollow basin]

How to set up the collider for a mesh that deforms?

[bleak umbra]

While technically possible to emulate such a thing at great cost (soft body physics). Unity only supports rigidbody physics.

[silver moss]

It's possible to fake soft body physics by using physics joints on the bones of a skinned mesh

[queen falcon]

I want to create Shield System like this, then how to impl this? (2d physics)

[spare parcel]

This is the inspector of root object Ground

This is the inspector of the root object SportsCar

and for some reason, when I hit BeginPlay, the car just warps through the ground, I don't know what's wrong?

nevermind, the tickbox in the ground's box collider "is trigger" was on. that's why the car was going through it.

[viral ginkgo]

you can do multiple raycasts

simplest way

[honest river]

How to make a rigid body on child object use parent game object position as a base?
So let's say we start with 2 game objects: parent and child. Child's local position is (0, 0, 0), then I start moving the parent game object, the child follows and keeps the local position of (0, 0, 0). Then let's say I do ChildRigidbody2D.linearVelocity = Vector2.Up * 10 and now the child rigid body moves up.
But then as I move the parent game object, I'd like child rigid body move with it, too, but also move in its' local space.

I tried joints: relative joint, but it hardcodes the relative position and I can't launch the child body with addForce or linearVelocity.

[honest river]

Basically this behavior:

https://youtu.be/PVtf3vg8BXw

But without the additional structure of objects. I guess, I'll have to manually translate the child's position based on parent velocity on this update in the late update

https://cdn.discordapp.com/emojis/581918526135074818.png?size=48

[potent valley]

Hello, I am making a simulation game, in 2D, top-down, where each tile in the tilemap is 1x1 meters.
Now I wonder, the velocity of Rigidbody2D, is it in m/s? Or other units?

[unique cave]

unity uses m/s for velocities, m for units, mostly s for time, m/s^2 for acceleration, N for forces etc.

SI in other words

[potent valley]

Thank you :)

[spare valve]

hi! not a unity specific problem, but i want to find out the acceleration produced by an engine of a given power.
I tried using both the power force and power kinetic energy and i am getting different results :( and i have no idea which one is correct :(

[timid dove]

F = ma is the basic equation

You'll need force in terms of newtons and mass in terms of kilograms

[spare valve]

I mean that i have power in terms watts, and want to get the force generated
I know P = F v is a thing but its kinda weird

[royal willow]

Do you want to calculate how much acceleration will come from how much RPM an engine produces?

[spare valve]

I guess... although how do i find the rpm? Lol

This whole area is really infamiliar to me

[royal willow]

Actually if you already know your speed, you can calculate acceleration off of that and calculate RPM off of that as well so basing it off of RPM is pointless.

If you go from 0 to 60 MPH in 3 seconds, you can do

(60 - 0) / 3 = 20 MPH per second or 1/180 mile every second (if you want a fraction)

sorry I wasn't responding quicker, I just did a ton of formulas just to now remember you could do this 😅

[spare valve]

Wait but im trying to calculate the acceleration to apply it to the object?
I only have access to the currect speed generally

[royal willow]

ah

[spare valve]

I feel like i am missing something crucial haha

[royal willow]

you could convert RPM to RCF (relative centrifugal force) which is a measurement of acceleration which you could use as acceleration

RCF = (RPM)^2 * 1.118 * 10^-5 * R(Radius of Wheel)

you could multiply or divide RCF to get a lower value of acceleration if needed

sample problem if you had an RPM of 2500 and a wheel radius of 11

RCF = (2500)^2 * 1.118 * 10^-5 * 11

RCF = 768.625

//Optional
Finalforce = 768.625 * 0.1

Finalforce = 76.8625

[spare valve]

Hmmm? Wait i thought acceleration would depend on torque not rpm?

Its all so confusing

[royal willow]

you can use RPM to calculate Relative centrifugal force which is a type of acceleration

That calculation above should be all you need if you know RPM

[timid dove]

The problem is that "power" doesn't directly convert to force (which is what you need to actually know to calculate acceleration), as there are lots of other factors involved, at least in the real world

We don't even know what kind of motor this is

Is it a rocket? Electric motor? Internal combustion engine?

What's the full context here?

[royal willow]

I'm going off of it is a Piston engine of some sort

but converting RPM to RCF should be okay

[timid dove]

Doesn't this depend on a lot of factors like the mass of the wheel, its radius, its moment of intertia, etc?

[royal willow]

From what I'm looking at all you need to know is the wheel radius and RPMs

[timid dove]

Doesn't the engine displacement matter too?

[royal willow]

RCF = (RPM)^2 × 1.118 × 10^-5 × r this is the formula I found 🤷‍♂️

[spare valve]

But thats for centrifugal force tho

Isnt that just the force which makes the body rotate around a center point?

[timid dove]

yeah that makes sense

[royal willow]

sure, but it is still the acceleration of the wheel, which you can use for newtons 2nd law to get a Force I would presume.

you could probably convert it to newtons if you want

[royal willow]

Also I would think you need this rotational force because you are trying to apply this acceleration to the object, so the object's velocity is 0 so you cannot do really any acceleration calculations with 0 velocity

due to how division and multiplication works

[wary vessel]

Okay, so in my game, the zombies im using are currently turned into a ragdoll on death, but does anyone know if there is a way to destroy the limbs, (example: if i shoot the body enough, the body is deleted, and the limbs are remaining, no longer restrained by the ragdoll skeleton)? - i cant really find anything online doing this exactly...

[royal willow]

Wait I think I just found the perfect formula which gives you a linear speed that could be converted to newtons

RPM = 2000 or 33.3 Hz (Need to know this)
Radius = 30cm wheel (convert to meters which is 30/100 which outputs 0.30)
F = Hz or Frequency
W = Angular Velocity
V = Linear Velocity

W = 2pi * F

W = 209.23 Rad/s

V = W * R
V = 209.23 * 0.30 = 62.769 m/s

Then 

F = M * A
F = 40kg * 62.769 = 2510.76 newtons

[spare valve]

Wut

Why is V become a, im confused...

[timid dove]

Visually you would need the limbs to be rendered by a separate MeshRenderer and destroy, disable, or detach them.

If you're using a SkinnedMeshRenderer, it's not straightforward.

[royal willow]

you turn Angular velocity into Linear Velocity, by multiplying W with the Radius in meters. then just do F = MA.

but that would be the entire calculation to get what you want

62.769 is the Acceleration and I just converted it into newtons for use in unity

[stuck bay]

I'm planning to make voxel game (block game that look like minecraft). How is unity engine performance at handling tens of millions of static (unchanging) box collider? Should i write my own physics or use unity physics? And if i write my own physics, then how do i make my custom physics work together with unity physics?

My first concern with that much box collider is memory consumption

[timid dove]

Using tens of millions of box colliders will not work at all

how do i make my custom physics work together with unity physics?
You wouldn't, generally.

You would pick one or the other

I believe most people who do minecraft-style games in Unity build dynamic meshes for chunks and use a single MeshCollider for a whole chunk

[stuck bay]

Ehh? Isn't mesh collider very laggy?

[timid dove]

That's a meaningless overgeneralized statement. No.

[stuck bay]

Hmm there is 1 million block per minecraft chunk (256 height limit), are you sure it is not laggy?

[timid dove]

Minecraft does not use individual colliders for each block

nor does it really use any off the shelf physics engine

But yeah pretty sure, I've seen lots of people make "minecraft in unity"

and they've all used MeshCollider in my experience

[timid dove]

note that the collider should be simplified to the greatest degree. For example a completely solid chunk would only have a collider with 8 vertices

(i.e. a single large cube)

[stuck bay]

Hmm interesting

Looks like i need to find tutorial to make minecraft physics on unity

Thanks

[timid dove]

There are many

just search YouTube or whatever

[stuck bay]

Yeah

[timid dove]

but yeah - procedural mesh generation is the name of the game

[stuck bay]

The worst case scenario would be the case where every block is not touching each other, which would easily cause hundreds of thousand of box collider

[timid dove]

well it's still one mesh collider, but yes probably, something like that.

[stuck bay]

And how do i handle that worst case scenario

[timid dove]

Don't worry about it until it's actually a problem

See if it actually causes issues.

I suspect it won't.

[stuck bay]

It is a problem tho, in minecraft multiplayer there is troller who modify chunk to make it laggy for other player, so i need to make sure in that worst case scenario the game won't be laggy

[timid dove]

Yes, you should make sure it's not

but don't jump to the conclusion that it's a problem before you try it.

[stuck bay]

Hmm alright

Thanks

[wary vessel]

As you can see, the zombie model has details inside, I'm guessing its made for this reason, and each of the limbs use skinned mesh renderer, which i can disable with a simple limb health script, but the thing is that if i remove this mid section as in the screenshot, the whole body still sticks together.

[wary vessel]

Is the only way to get the result im looking for, removing the mesh renderer, deleting the joint, and making each part their own root parent?…

[upbeat jackal]

Why?..

[timid dove]

Why what? What are you wanting or expecting to happen here?

[upbeat jackal]

I have marked isTrigger, but it still leads to collisions.

Here collision of my main object:

[timid dove]

My guess is your movement script uses raycasts or capsule casts or something to detect obstacles

setting the collider as a trigger has no effect on that.

My question is if you don't want to collide with these cubes why do they have Colliders and Rigidbodies at all?

[upbeat jackal]

isTrigger implies that the object will simply have a collider, which you enter and something happens, but there will be no collision. Isn't that right?

[timid dove]

Yes but that's talking about interactions with dynamic rigidbodies that are moving via normal means like velocity and forces

I don't think that's how your movement code works.

[upbeat jackal]

I can send my code

[timid dove]

Go ahead

I'm already 90% sure about what I will see

I would love an answer to this question though:

My question is if you don't want to collide with these cubes why do they have Colliders and Rigidbodies at all?

[timid dove]

an explanation of what components are on the player would also be helpful.

[upbeat jackal]

Objects or Hero?

[timid dove]

on the cubes

[upbeat jackal]

some objects are needed, others are not...

I'm get component to set the "isTrigger" and that interaction with the object takes place.

[timid dove]

Is there a language barrier here?

I don't understand what this sentence means

[upbeat jackal]

One second please...

[timid dove]

Anyway it seems pretty obvious that this is your problem:
#⚛️┃physics message

[upbeat jackal]

I'm setting up a collider in Unity and enabling the isTrigger option. As for the Rigidbody, I want it to interact with other objects but not with the player character. I've even written some code for that

Understand?

[upbeat jackal]

How to resolve this problem?

[timid dove]

Yes I understand.

The thing you should be doing is:

• Instead of disabling gravity, just remove the Rigidbodies or make them kinematic.
• Change your movement script to use a layer mask for its raycasts or capsulecasts that ignores the blocks

[upbeat jackal]

I have some suspicions about the layers because I have Post Processing set up

on capsule collider

[upbeat jackal]

Can I send my code of movement?

[timid dove]

I've been waiting for it

[upbeat jackal]

https://hastebin.com/share/icupekocoj.csharp

[timid dove]

yep like I said it's just manually handling collisions in this script with Physics.OverlapSphereNonAlloc

[upbeat jackal]

So, what should I do about it? 😄

[timid dove]

Anyway you already have a way to exclude objects from it:

    // Collision will not happen with these layers
    // One of them has to be this controller's own layer
    [SerializeField]
    private LayerMask _excludedLayers = default;```

why don't you just use that

[upbeat jackal]

One second please...

[upbeat jackal]

Do you mean assigning a specific layer to a certain object?

[timid dove]

I mean excluding the layer the blocks are on from your movement script with that field.

[upbeat jackal]

It worked! You are a very attentive person...

Now it remains to set up the interaction, but I'll probably do it tomorrow.

Thank you for your help!

One problem... When I am in the collider - the button does not appear

it checks by tag

If I set layer - button does not appear
If I set default layer - button apper

[open zealot]

is it possible for layer A to affect layer B but not the other way around?

in 2D physics

[timid dove]

Not really

they either affect each other or not

I actually made an asset for this for 3D but 2D doesn't have the appropriate APIs available to achieve something similar in the same way: https://assetstore.unity.com/packages/tools/physics/betterphysics-selective-kinematics-244370

You could make the objects in layer A kinematic, that would get you the effect you want, but it wouldn't allow e.g. layer C to affect layer A objects as though they are dynamic.

Could you describe the exact gameplay mechanic you are trying to achieve perhaps?

[open zealot]

https://gyazo.com/d5d392513f0ef0c58b36d1e84dd1de72 I'm picking up a bunch of water particles out of a bigger mass and I temporarily changer the layer of the grabbed particles so that the other particles off the mass don't block them from getting picked up smoothly. However I would love for the other water particles to still detect the collision of the grabbed ones so that the water doesn't drop instantly when you grab some or that you can smash the grabbed ones into the mass.

I guess making them kinematic would work but it's a lot of work for the desired effect so not sure if I'll go for it

[spare valve]

wheel colliders are very weird

[silver moss]

Pretty sure this is #🏃┃animation question, i don't see anything physics related

Unless it's an active ragdoll or something

[thorn geode]

ill post it in there might be smited for crosspost

[silver moss]

If you delete it from here it won't be considered crosspost

[upbeat jackal]

It can't see my object

the tag is set... I can send the code

[silver moss]

It can't collide*
Tags dont matter in collisions. Layers might matter.
Show the code you use to move the character and also the components of the character and the cube

[primal wren]

Hey, very beginner here
I'm trying to add very basic gravity to my game.
Problem is, when I add basic Rigidbody gravity to a sphere (1), it goes flying off in the air (2)
I thought it was due to clipping but I can confirm it is not

[timid dove]

The sphere is likely colliding with something, thus causing it to fly off

you say you can confirm it's not due to clipping

how are you confirming that?

You should add a script to the ball with OnCollisionEnter in it. It will tell you exactly which object(s) the ball is colliding with.

[primal wren]

I checked every side of. The ball floats over every surface.
Could it be bouncing to hard ?

[primal wren]

I'll try that thanks

Ok so it does collide when it falls. Show do I prevent it from bouncing so strongly?

Ah turns out I had messed the collision of the ground, so the ball was ejected to the top of it. Thanks !

[elder sedge]

Can somebody help with this issue? (I don't want my player to fly off of the slope and instead stick to it)

[hallow plover]

what scripts do you have for player movement

[elder sedge]

What do you mean? You want me to show you the movement code?

[hallow plover]

without code nobody can suggest you anything.

[elder sedge]

okay hold on

[river flare]

||||

[elder sedge]

here's the important bits (the left is what i use to move the player and the right is how i get the slope vector)

[hallow plover]

what are you getting if you put rigidbody. AddForce(Vector3.down * downWardForce); in place of targetDir.y something like this.

[elder sedge]

I'll give it a go and let you know

[elder sedge]

Tried it and you can't set a float value to a vector3 value

[hallow plover]

edited it. remove targetDir.y = and add force like rb.AddForce(Vector3.down * downWardForce);

i mean the entire line and replace it.

[elder sedge]

Yeah I know I'm currently tryin it

[hallow plover]

rb.AddForce(Vector3.down * downWardForce * rb.mass,ForceMode.Force); you can try even this also.

[elder sedge]

this doesn't work because: Too little of a downward force doesn't fix the issue on harsher slopes and too big of a force causes the player to slide down the slope automatically.

[surreal dawn]

Are there any good ways to learn about vectors in unity?? I’ve been trying to make an enemy AI and like when I see tutorials they’re doing normalized vectors or adding/subtracting vectors and I understand none of it. I really want to be able to understand it enough to work with it confidently

[hallow plover]

there are many other talented people here but they might require more details like how are you calculating slopeHit and where are you modifying exitingSlope etc.

[hollow echo]

There's an overview in the docs https://docs.unity3d.com/6000.0/Documentation/Manual/scripting-vectors.html

[surreal dawn]

Oooh awesum

Math the bane of my existence

[spare valve]

anyone know why the wheel collider wobbles like that?

the ground is pretty uneven, but i assume there is a way to make the wheels handle it?

[silver moss]

Pehaps the wheel mesh is not correctly aligned to its axis?

[unique cave]

I dont even think you can rotate a wheel collider even if you wanted (it will stay aligned with global y axis) so definitely seems more like issue on the mesh axis or the way you rotate it

[stuck bay]

i'm pretty sure vector is teached in high school? You can find high school text book that contain lesson about vector

[surreal dawn]

I don’t remember the lesson i had 5 years ago 😭

And i def didn’t learn what “normalized” is

[stuck bay]

cm'on it is just high school lesson it won't be hard. You can just find free online learning website that teaches vector

My autocorrect arr

[stuck bay]

if you dont know what subtracting vector means, then you lack the basics to learn what normalized vector mean.

[surreal dawn]

It’s a little different for programming, I know how to add or subtract a vector but I don’t know what use it has. I just read the document and learned it it wasn’t that hard

[stuck bay]

Vector in unity is just code implementation of vector in math

[surreal dawn]

Awesum

[spare valve]

Ah yeah it was an issue with the mesh in the end, model wasnt mine

Either way is there a way to make it more "stable"? As in driving better on uneven terrain

There are a lot of settings in the wheel collider and i dont know which does what

[elder sedge]

so does anyone have a fix for this....?

[silver moss]

If you are going to write a custom car controller, which is pretty complex, you should know how to look at documentation
https://docs.unity3d.com/6000.0/Documentation/ScriptReference/WheelCollider.html

You probably need to also add forces/torques to the rigidbody, or use a joint, to stabilize it

[spare valve]

well yeah but it is a bit difficult understanding how exactly each attribute affects the end result

i guess trial and error plays into it too

[mystic moat]

Have y 'all run into a bug where Unity sets the physics SDK to "none" randomly? It's happened to me twice now

[sterile ore]

somebody can tell me why when i turn off kinematic the characters just spins and fly like crazy?

[timid dove]

you mean when you turn it off?

Because when you turn off kinematic the object actually starts getting physically simulated

[sterile ore]

sorry

off

i correct it

[timid dove]

It looks like there's some internal interactions going on

[sterile ore]

the comment no?

[timid dove]

is this a ragdoll?

[sterile ore]

yes

[timid dove]

Probably all the pieces of the ragdoll are intersecting each other

and colliding

causing craziness

[sterile ore]

oh

that explains everything

actually i find a smart solution

unparent the player from the ragdoll when he dies

[uncut anvil]

How would you guys go about being able to use physics to open this (or any) hatch with a top down player?

[timid dove]

Do you really need it to use Physics?

[uncut anvil]

Not necessarily

[timid dove]

I would recommend writing a simple script to animate it opening and closing

and just calling that

[uncut anvil]

Well, ok, so yeah I guess it does need to be physics 😄

[timid dove]

Use a hinge joint

[uncut anvil]

I need the player to physically grab onto it and then be able to drag it open by moving their body

We use a configurable joint to hinge it open, so that's all good. What I need is a decent solution for being able to attach the player's hand to the handle and have his movement be able to open it. Similar to like Human fall flat grabbing things and moving them

I tried a fixed joint on the body, but I use the mouse to determine look rotation, so it ended up flinging the player around

[timid dove]

Using the mouse to determine look rotation seems unrelated to your problem

What you might have trouble with is how you are applying forces or moving objects with the mouse

[uncut anvil]

Well the mouse is a problem when it's rotating my player's body, and that is what the fixed joint is on. Then it rotates to the right and wants to drag the hatch door to the right :/

[uncut anvil]

Yeah maybe this, I'm kind of unsure..

[timid dove]

yeah that's a problem if you're just moving the player's Transform directly

The typical way to do something like that is:

• When you click, spawn an invisible kinematic Rigidbody at the clicked position
• Each fixedupdate, move the invisble rigidbody with e.g. MovePosition, according to where the mouse is

that way it's all within the physics engine and behaves better

[uncut anvil]

That's a good idea! I'm kind of doing the same thing now, except, there is no rigidbody, it's just a point between the hands (to put it simply). I will try that, thanks for the idea!

Wait, follow up question though. How would you attach the hatch's handle to that invisible point though?

[timid dove]

with a joint

fixed joint perhaps

[uncut anvil]

Ok, that's what I thought, thanks 🙂

[visual gale]

I am working on some golf ball physics. Since I have wind, loads of sphere spin, etc, unity physics doesn't feel like the right move. Does anyone disagree? ANyone got a link to a golf physsics paper or implementation I can read up on?

[pulsar canopy]

How can I rotate my rigidbody such that it respects collision? Torque doesn't do anything for me

[silver moss]

Torque should work. Show your rigidbody settings and the code you tried

[pulsar canopy]

I am using playmaker so I don't know if you can read that.

I really should just rewrite the char controller in c#

I know there is also rb rotation and moverotation

[pulsar canopy]

in case this helps you,

but it doesn't rotate

[silver moss]

Not really familiar with playmaker. Are XYZ supposed to be set to None?
I did look at this doc but it doesn't tell much https://hutonggames.fogbugz.com/default.asp?W317

And what does your rigidbody look like?

[pulsar canopy]

yes, you either map individual planes or all of them in the ''vector''

rb is kinematic if that's what you mean

you can see the vector live with the numbers below ''vector'' and ''mouse direction'' which means it works

[silver moss]

Kinematic bodies don't react to forces or torque

[pulsar canopy]

so what should I use instead?

rigibody rotate or move rotate

I assume one of those at least works with kinematic

[silver moss]

I don't know what those are but in C#, .rotation "teleports" the rotation and skips interpolation, while MoveRotation rotates and respects interpolation

Why is your rigidbody kinematic though? Are you doing custom collision checks?

[pulsar canopy]

excellent, so I need this?
https://docs.unity3d.com/6000.0/Documentation/ScriptReference/Rigidbody.MoveRotation.html#:~:text=Use Rigidbody.,in any intermediate frames rendered.

[pulsar canopy]

the reason I'm using a rigidboy at all is because I had to switch away from character controller and scrap it

it's a diving game where my character is elongated, i.e. its collider capsule is supposed to face in the Z direction

but stupid character controller capsule can't do that

wait, I am stupid

I should uncheck kinematic

and just uncheck use gravity

d'oh

stupido

now torque works ofc

[silver moss]

Most likely. You didn't give any good reason to keep it kinematic

Kinematic is for when you want to explicitly control the body without having forces and torques affect it

[pulsar canopy]

yeah yeah, I remembered

I just forgot

been a while since I worked with rigidbodies with my previous game

all good

thanks

[signal wyvern]

Anyone got any experience with velocity matching a player to an object, In a multiplayer setting where player is controlled by client & Object is controlled by server?

[timid surge]

so if I want to make a slippery stair that interacts with a collider and not the player should I have a setup like this?

[timid surge]

This is the player collider

[timid surge]

the stair colliders

[timid dove]

What does "interacts with the collider and not the player" mean?

[timid surge]

it makes a collider move which is set to the player

so if I run into it, it moves me around like on ice

[timid dove]

That sounds like an interaction with the player

So I don't understand the original statement

[signal wyvern]

Is it possible to have a one sided fixed joint? Where object As movement influences object Bs movement But object B has no influence over object Bs movement?

[supple steeple]

how do you guys get more accurate rb physics, my player controller slides and falls very slowly, just 2 examples but I've noticed a lot of other strange behavior

[timid dove]

This is likely due to your code

[supple steeple]

im just adding an addforce method on the rb based on the 2d wasd vector, i dont see how it can be my code unless im not writing enough code

[timid dove]

Then it should behave normally

[smoky wigeon]

Have I really hit the limits of the convex mesh colliders? When I turn on convex, for an already convex object (blender: Convex hull option) , it decimates it to a useless shape. Will I just have to make it not convex and take the performance hit?

[carmine basin]

having some issues relating to hinge joints - I'm giving them a target angle to move to via Joint Springs, but the joints snap the target position to 180 when they're above 180 or below 0

[bronze blade]

anyone know of a way to tell if a configurable joint has reached its angular limit?

[silver moss]

Why are you using blender's convex hull?
Unity/PhysX already creates a convex mesh if you enable "convex" in the mesh collider

Or is it that the blender's convex hull mesh is alright but the "convex" option in mesh collider messes it up?

Do you mind sharing this 3D file so I can take a look?

[pallid gale]

How can I get or set the top speed of a rigidbody? I know that once LinearVelocity.magnitude can show the top speed once the object is at its speed limit, but I dont know how I'd calculate that ahead of time

[silver moss]

The simplest way is to clamp the velocity with Vector3.ClampMagnitude

Idk what you mean with the "get" part though

You should be the one to determine the top speed

[pallid gale]

I just mean get as in "taking all the properties of the rigidbody, how can I calculate what the top speed would be"

[silver moss]

I don't think there are any properties that would determine a top speed

Nothing stops you from setting the velocity to (99999, 0, 0)

(Except of course being kinematic or otherwise constrained)

You could probably calculate a resulting velocity from some forces or acceleration though

[pallid gale]

well, with Mass = 1, Linear Damp = 0.2, Angular Damp = 0.2 the magnitude reaches 5
then Mass = 0.5, Linear Damp = 0.2, Angular Damp = 0.2 the magnitude reaches 10

It's clear there is a relationship between the mass and the top speed

What I'm trying to do is determine the % the current speed is, to what the highest speed could be, so that I can do something like apply some artificial boost but only when the speed is between some range

[timid dove]

It's clear there is a relationship between the mass and the top speed
No that's not how newtonian physics works

there is no limit to the max speed in newtonian physics other than those imposed by the limits of the software such as floating point representation precision.

The damping is definitely relevant but since there is no maximum force you can add, there's no maximum velocity.

If you imposed some constraint like "the maximum force is X per physics step", there's certainly a maximum speed you could reach with that limited maximum force per fixedupdate.

Basically:
Force adds to the velocity (how much it adds depends on the mass)
Linear Damping removes a certain percentage of the velocity each physics frame
And you reach your maximum speed when those two things become equal.

That requires you to know the formula behind how the damping is applied though. If it's the default Unity damping then you can look that up.

[velvet oracle]

hi everyone, I'm turning to people who know unity well because at the moment I'm trying to learn this software
I just don't really understand the 2d collision system (I'm talking about wall-floor-ceiling collisions, not coin collisions) if I understood correctly there is:
rigibody2D which basically allows unity to take care of everything in terms of physics (you just have to type 3 lines of code) it uses collider2D
raycasts: one line to make it really simple
OnCollisionEnter2D:
and finally there are overlaps

here's everything I know (that I understood)
but so which one should be used for wall-floor collisions (apart from rigibody I don't think we have much control)
or there are other techniques that I don't know
thanks

[timid dove]

but so which one should be used for wall-floor collisions
This is a really vague question

wdym by "used for collisions"?

what are you trying to do?

[velvet oracle]

actually I wonder if that's all there is to do this type of collision with gravity, rebounds...
because I have the impression that with the rigibody we don't have total control over the variables of physics
i don't know if it's very clear

[timid dove]

actually I wonder if that's all there is to do this type of collision with gravity, rebounds...
The physics engine gives you gravity, collisions, bouncing etc out of the box

you don't have to do anything

unless you want to customize it somehow

That's why I ask - "what are you trying to achieve"?

[velvet oracle]

just practice seeing all the ways to do collisions in 2D to get an overview

[visual gale]

Just got "Triggers on concave MeshColliders are not supported" ... is there a general way to fix my colliders? This is terrain for a golf course

[timid dove]

Why would the terrain be a trigger?

make it not a trigger

[visual gale]

Idea was to do manual physics during ball flight until it interacts with something, then switch to unity physics.

[unique cave]

That doesn't do very good job explaining why the terrain would be a trigger though does it? I also don't see a reason not to just use unity's physics during flight and adding your own forces on top if you need features like curving path based on spin or more sophisticated air drag model (or do you do that already, I'm confused). It sounds like you are making the environment triggers because you don't want the ball to interact with it until the ball hits something, but that makes little sense since the ball won't interact with things it doesn't hit anyways. Triggers also sound terrible solution for capturing hits for a high speed ball since triggers don't respect the collision detection mode (continuous collisions) in any way so there's a chance the ball tunnels through geometry without any notice and even when you get a trigger hit, it won't give an accurate representation of where the ball should have naturally hit something (because the ball might already be way below the surface of the object)

[timid dove]

I don't see what that has to do with the terrain being a trigger

[carmine basin]

why are there only like
three threads on the internet about wheels?
Im finding it hard tuning a vehicle's friction. If I move it any slower than crawling pace, it spins 720 degrees and then regains grip

unless the trick is to only drive on perfectly flat ground

yeah i can't seem to get the car to go faster than a few mph without spinning out with the stock wheel settings

[smoky wigeon]

Yes, the convex mesh from blender is just what I want but when I turn it to convex in unity we get the lower poly mesh that is too basic for collision. Unitys "convexing" is really poor and I can't find any parameters to adjust it

[silver moss]

I tested with a similiar shape and the convex mesh looks fine

Your mesh might have some degenerate triangles

[silver moss]

Okay I replicated your problem:

This happens when the mesh is very small. I did this by scaling it to 0.001 in blender and then to 1000 in unity

At 0.01 -> 100 it seems to be somewhere in between

This is probably due to some floating point precision limits. Could also be an optimization that PhysX/Unity does to small meshes

[short bolt]

Is interpolating physics broken for you too? Every rigid object that I set to interpolation, interpolate, is acting weird. If the object does not interpolate it acts correct, but the FPS of the animation is bellow 50 I think

[timid dove]

Works for me

[short bolt]

Which version are you using?

[timid dove]

6000.0.32f1

[short bolt]

It "works", just a lot worse

It acts more chaotic

[timid dove]

My guess is it's something in your scene or code

[short bolt]

Like what?

If the code is exactlyt he same, but the only difference is the setting for interpolation?

For instance, when not interpolating, if you drop an object in a certain way, it will almost always drop the same way. While with interpolating, it will randomly bounce around more

It's less consistent and stable

I do not seek consistensy, but it's an example of how more chaotic it is

I will try 32f1 though

Oh wait, I am actually using it now heh

[timid dove]

If that's true it's a bug you should report

interpolation is only supposed to be visual

it's not supposed to affect the physics

[short bolt]

That's what I thought. I really don't think that's how it works.
The physics have had a lot of issues in version 6, thought hey fixed some of the previous issues.
I will wait a little to see if anyone else has this issue.

I mean, early versions of 6 had a lot more issues in physics

[smoky wigeon]

Ah, outstanding. I think I was playing with obj files that add in a 100x scale, if I recall, which would explain why it is failing for me. Not a great system to fail like that but at least I know some work with scale can solve it. Much appreciated!

[high slate]

Hey guys, hoping to get some help regarding ragdoll and fixed joints.

I've currently got a simple character controller and animations and I've created another character with ragdoll physics attached by a fixed joint to the palm of my player's hand. The Physics for this is incredibly janky and I'd love to be able to reduce that somehow.

Things I've done:

• Set all rigidbodies to Interpolate
• Set all rigidbodies collision detection to Continuous
• Set all character joints to Enable Preprocessing
• Given every rigidbody a physics material with dynamic friction and static friction reduced to 0.2
• Increased the Default Solver Iterations in the project settings. (I have tried a range of values, including the max of 255. The video shows a value of 16)

What can I possibly do here to make this less jank?

[warm oar]

Hey, i'm new about this thing but i've created a building for VRC world. Everything was fine until i was created glass wall. Then beign troubles. I used for it Mesh Collider but that makes me stuck into it. I was searching about any solve but not found. Know someone, how to fix, please? Have some thick walls which are good but the thin stuckin me.

[short bolt]

Is it possible that your colliders are intersecting? That could cause issues. If they are inside each other

[high slate]

Well, I suppose that's possible. The ragdoll will malfunction like that if the colliders overlap I assume. I understand I could use the ignore physics collisions on layers, however I do want this layer to collide with itself unfortunately, so would the solution be to simply reduce the size of all my rigid body colliders? (Even though by default they aren't overlapping, but still decrease the size to create a gap?)

[short bolt]

Yea, it does not need to ignore it's own collider. I am saying if the colliders are too "fat", it will explode like that, because they cannot reach resting position, ebcause they will always pressue each other.

Give it at least a little space between each collider

[high slate]

Got it, thanks for the advice

[unique cave]

Enable Collision might be of importance. Disabling preprocessing for the joints may help the stability of the simulation too

[timid moth]

what is your opinion to the best Method to create a dash ability

[forest beacon]

Hi there

I am a beginner. I am trying to implement a plate balancing game

I have encountered something weird, which is although the ball do falls on my plate, but when I tilt the plate with my scripts written, the ball didn't roll, but remain in the air

using UnityEngine;

public class PlatformController : MonoBehaviour
{
    public float tiltSpeed = 10f; 

    void Update()
    {
        float horizontalInput = Input.GetAxis("Horizontal"); 
        float verticalInput = Input.GetAxis("Vertical"); 

        Vector3 tilt = new Vector3(verticalInput, 0, -horizontalInput); 
        transform.Rotate(tilt * tiltSpeed * Time.deltaTime);
    }
}

my scripts

what happened

or probably something goes with the collision setting?

my settings for the plate

my settings for the ball

[silver moss]

Likely because you are using transform to rotate the plate, it will bypass physics pretty much

You should either:
A. Add a non-kinematic rigidbody to the plate and use torque or a joint to rotate it. The mesh collider should be marked convex. You can lock the rigidbody position in the constraints tab
B. Add a kinematic rigidbody to the plate and control its rotation with MoveRotation (collisions might be unreliable)

[forest beacon]

got it, let me update my codes

[old haven]

Hi, trying unity after godot.
Having pretty simple question. Why CharacterController.Move should be called in Update instead of FixedUpdate (Is it correct to call it PhysicsUpdate?)? My concern is wouldn't it cause to many physics engine calls if call it every frame?

[timid dove]

No

It's designed to run in Update so that the movement is not jittery

[unique cave]

Yeah, I don't think CC supports interpolation so it would not look smooth on high framerates (if it was moved on fixed update)

[timid dove]

Exactly

[novel light]

you can implement interpolation yourself, it's extremally simple

[unique cave]

You can but I'm not sure that's very beneficial for character controllers (over just moving in update)

[novel light]

for a single player game yea, not very important, but i still would do it so that the logic runs perfectly the same regardless of fps

[unique cave]

The difference is likely not noticeable but I get the point, that would be totally reasonable although I likely wouldn't care enough to do that myself in any of my casual projects

[slim sierra]

can someone please help me? i wanna make a rope physic with spring joint but this is keep happening

[round forum]

Im not sure if this is the correct thread but it is related to physics so Im giving it a shot.
The bedside table has a mesh collider that is equal to its size, so when I drag and drop prefabs ontop of it gets in the right position, however I have a second child box collider that extrudes above the table, this is to prevent the player from jumping ontop of objects, this boxcollider only affects the player and nothing else in the physics tab, however it also block the placements of prefabs ontop of the table, so I cant drop and drag prefabs manually over the table anymore, instead they slide along the invisible box collider, is there any solution for this?

https://i.gyazo.com/8743798f07cb897f2353427e04cc6b18.png

[silver moss]

If the box is on a different layer, just hide that layer here in the top right of unity:

[torpid hollow]

Hi, so i have this configurable joint sphere that is supposed to track the cameras local position using target position but there is a weird offset that i do not remember setting? I also send my code and more of the config joint

[round forum]

Oh yes perfect!! thanks 😄

[pallid gale]

In my train vehicle, I have it so you can freely drive it around via the users input, and I've added a new feature that allows the input direction to be locked (shown as the red line). As if the wheels have been attached to some rails, only moving in the rail direction and not to the side.

What I cant figure out is how I can constrain the position of the locked section, so it is unable to move to the side even when it has external forces pulling it in that direction. Right now I can enable the lock, but the train is still able to be pulled perpendicular to its locked heading

[timid dove]

Better to use a spline library and handle physics manually than trying to use Rigidbody

[pallid gale]

eh, having the position effectively just be interpolated along a spline doesnt really sound that great

[timid dove]

Why not?

You can easily simulate basic physics with that

F = ma

And have a velocity

[pallid gale]

one problem is the entire vehicle consists of multiple different bodies that are connected together

if I was to switch the wheel section from a rigidbody, over to the non-rb locked version, I'm not sure how I'd handle the other parts connecting to it and allow them to still apply forces to the locked portion

only other thing is if I worked with splines, I could implement a custom prismatic joint, because a joint like that would achieve the desired behaviour

[timid dove]

Is that needed? What kind of game is this?

[pallid gale]

it would be, yeah. Its basically an action roguelike where you drive a train around to defeat enemies

I plan on having tracks that you can drive either end of the train onto, locking it on the track as there will be different mechanics that become available in that mode

or with both ends on 2 tracks, youd get multi-track drifting

[silver moss]

But you will be also allowed to move freely, not on tracks?

Or no?

[pallid gale]

off the tracks its like that video, full freedom of movement

but the side thats on the track gets fixed and effectively the other side would orbit around it

[pallid gale]

if it was possible to do something like the constraints here, but Freeze Relative Position X I could see that being ideal

I assume Unity does more stuff underneath with the constraint, rather than it essentially just being the same as body.linearVelocityX = 0;

[silver moss]

Relative to what, though?

You'd effectively do that with a joint

[pallid gale]

relative to transform.right or maybe .up

freezing the X axis would use the worlds X direction like the black arrow. But the relative X direction should be the red arrow

heres a more accurate video showing what I'm after

then just imagine the same idea but if the rail was to curve rather than only be vertical

[timid dove]

You could try Articulation bodies (which would mean 3D physics) but I really think you're looking at something spline based and custom

[pallid gale]

I'm not familiar with articulation bodies, but if they'd be able to achieve the sort of thing youd expect with rails like these objects, then they could be worth it

but if not, I suppose I'll have to implement a completely custom type of joint

conceptually, its not that complicated of a feature. But I'm sure implementing it is another story

[timid dove]

They could do that but only linearly I think

There's a prismatic joint

[pallid gale]

for any sort of curve, its easy enough to find the location A on the curve thats closest to B

I wonder if it could be done by just giving an object at A that uses the typical spline/path following like most people do with train tracks, then attach B to it via something like a distance joint

[fading thistle]

this is my code for my player why the jump buffer dont work some help ??
https://paste.ofcode.org/XgwgmgkZeE9GG6RGWg94CH

[timid dove]

What does "the jump buffer don't work" mean?

What are you expecting the code to do?
What is it doing instead?
What debugging steps have you taken to investigate?

[proper dome]

Hey!

Maybe the wrong channel, but;

I'm trying to find an alternative to detecting environment interactions (Unity 2D). The use case is when players or enemies touch grass sprites they will trigger a tween animation. Right now the POC simply uses circle colliders and it is obviously not very performative, especially since our game is a horde-shooter with lots of enemies.

We really don't want to start implementing DOTS now.

Anyone got any idea how to work around this issue (outside not having any environment interactions)? Was thinking this might be common. Maybe some sort of shader, masking or camera layering.

[timid dove]

Are you using tilemaps?

[proper dome]

No, we use some wacky self-made solution

I'm actually not even the artist or the programmer on that, but i decided i wanted to start figuring out a solution

[timid dove]

Tilemaps are almost certainly more performant than a solution that involves many individual SpriteRenderers, if that's what your solution does. And if you use a TIlemap you can just check nearby tiles to see if they are the interactable ones

[proper dome]

We might make a sprite based collision technique to what you're alluding to, as in just subdividing our playable area into a grid.

As you can tell, we are really into making things more complicated for ourselves!

[supple steeple]

So I have a weird issue. I'm applying a impulse force up to make the player jump, but if the player is in motion horizontally the jump height is reduced.
I'm just doing(pseudo code)

Rb.addForce(vec3, impulse);
}```

[timid dove]

Could be anything from:

• Other code modifying the vertical velocity
• vertical momentum gained in other ways such as running horizontally up a hill

SHowing us a video and the full movement script would help

As well as the inspector configuration of the player and which components it has

[supple steeple]

Alright I'll do it later sense I'm at work, ty tho

[timid dove]

one other thing might be if you have linear drag

[supple steeple]

No drag, really it only wasd apples using addForce and rot with mouse vector, and a stair climbing that doesn't add a force but teleports based off of the stair height, but I'll prov8de the script later

[silver moss]

My guess is you are clamping the whole velocity vector

[timid dove]

Yeah if you're doing some kind of clamping/speed limit thing that'll do it. That falls under "some other code modifying the velocity"

[supple steeple]

Could it be because I'm setting a max velocity

[timid dove]

yes absolutely

Sounds like your code is working as expected

[supple steeple]

I guess so, but how to I limit the players speed

[timid dove]

Change your code that limits the velocity to only care about horizontal (x axis) velocity

[supple steeple]

I'm guessing I do that with if vel.x > Val, vel.x = val

[timid dove]

Probably just

myRb2d.linearVelocityX = Mathf.Clamp(myRb2d.linearVelocityX, -maxSpeed, maxSpeed);```

[torpid hollow]

this was fixed coz i forgot to add a a rigidbody to the component that time lol

i've done it before so many times that i forgot this time

[pallid gale]

circle based collision is just a matter of checking if the distance from A to B is less than the radius

once the distance from the enemy to the player, is less than the radius, you know the enemy has "collided" with the player

[autumn ridge]

guys why is it when I add a Mesh Collider to my custom built Mesh polygon, it generates a square collider?

[autumn ridge]

figured it out after 5 hours

Look at this sexy Mesh Collider. Basically, if you feed the Mesh Collider Vertices all on the same Z Pos. I think it returns a quad. BUT if you feed it a 3D array of Vertices, you get this beautiful artwork.

[pallid gale]

I've got my kinematic body attached via a hingejoint to a counterweight body. What I need is to be able to control the direction my kinematic body moves in, so it feels like its actually being influenced by the movement of the weight. The video shows what it's like without any influence

void SolveUnbalance()
{
  Vector2 armDirection = (_counterweight.transform.position - _rb.transform.position).normalized;
  // get the direction from the center to the counterweight
    
  Vector2 perpendicular = Vector2.Perpendicular(transform.up);
  // find the direction that is perpendicular to the direction the triangle faces
    
  _unbalanceDirection = Vector2.Dot(armDirection, perpendicular) * perpendicular;
  // aim from the triangle towards the side that the counterweight is

  _unbalanceDirection = Vector2.Lerp(transform.up, _unbalanceDirection, Mathf.Abs(_unbalanceDirection.magnitude));
  // not a great approach, but this will steer the direction towards the counterweight
  // that way the more to the left or right the weight is, the more influence pulls it that way
}```This has been my attempt for it, and it only sort of works

[timid dove]

Kinematic bodies aren't affected by forces

If you want that kinda thing make it not kinematic

[pallid gale]

I know, which is why I'm trying to solve it

https://en.wikipedia.org/wiki/Pendulum_(mechanics)#/media/File:Oscillating_pendulum.gif I figure its sort of similar to how pendulums behave

huh, it embedded the wiki page not the gif itself

[timid dove]

The clever way to do this I would say is to check the velocity of the counterweight each FixedUpdate and work backwards to figure out what force the joint must have applied to it to change its velocity from the previous frames velocity to the current.

Then apply Newton's second law:
Every force has an equal and opposite reaction force.

That opposite force is what needs to be applied to the main body

You could also just make it not kinematic and get that for free

[pallid gale]

I've been using a dynamic rigidbody, it's nice that this effect does come for free, but it's pretty tough to fine tune the bulk of my movement code

it's greatly improved now im using a kinematic, its just this pendulum type effect thats been the first hurdle. End of the day it is just some additional math to be solved to get what I need

What I'm having trouble visualising is what direction the counterweight would push the triangle. Would it be the angle perpendicular to the joint (green line), ie the same way the box would be rotating towards (red arrow)

Or would it actually be the acceleration direction like what the gif shows?

[pallid gale]

One a problem with finding the velocity of the counterweight is I have to subtract the triangles velocity, otherwise the counterweight might not be rotating, but it inherits the velocity of the triangle.
Vector2 currentCounterVelocity = _counterweight.linearVelocity - _rb.linearVelocity;

That becomes a problem the moment I apply the reaction force, because the next time it updates some value begins to accumulate and the forces "explode"

[timid dove]

No no

this isn't what I meant.

Something like this:

Rigidbody2D counterweight;
Vector2 previousCounterweightVel;

void Start() {
  previousCounterweightVel = counterweight.linearVelocity;
}

void FixedUpdate() {
  Vector2 currentVel = counterweight.linearVelocity;
  Vector2 differenceFromLastFrame = currentVel - previousCounterweightVel;

  // Work backwards to calculate what force must have been applied to the counterweight
  // How much did it accelerate?
  Vector2 acceleration = differenceFromLastFrame / Time.fixedDeltaTime;

  // To accelerate the counterweight that much we need to have applied a force proportional to its mass
  Vector2 forceApplied = acceleration * counterweight.mass;
  Vector2 equalAndOppositeForce = -forceApplied;

  // Now apply equalAndOpppositeForce to your "main" body.
  // The details of this are unclear to me since it's kinematic.
  // But if it was a dynamic body it would be mainBody.AddForce(equalAndOppositeForce);

  // store the velocity for next frame.
  previousCounterweightVel = currentVel;
}```

[pallid gale]

I havent applied the force yet, but the yellow line is what equalAndOppositeForce gets calculated as. Does that seem right?

[timid dove]

more or less, yes

[pallid gale]

_desiredForce = _facingDirection.normalized;
_desiredForce *= speed;
_rb.linearVelocity = _desiredForce;```The actual velocity for the triangle is pretty simple, though the [documentation](<https://docs.unity3d.com/6000.0/Documentation/ScriptReference/RigidbodyType2D.Kinematic.html>) has me wondering why you'd choose to deal with manually setting a position rather than setting the velocity to alter it

`This type of Rigidbody2D can be moved by setting its Rigidbody2D.velocity or Rigidbody2D.angularVelocity or by being repositioned explicitly.`

[timid dove]

Sometimes you may wish to teleport it

[pallid gale]

ah, true

[timid dove]

But yeah the ability to set velocity on kinematics in 2D is nice. 3D doesn't have that

[pallid gale]

3D uses PhysX doesnt it?

[timid dove]

yes

[pallid gale]

I cant seem to apply the counterweights force, the moment I try and use it it quickly "explodes"

[timid dove]

Well how are you doing that

given it's kinematic

[pallid gale]

_desiredForce = _facingDirection.normalized;
_desiredForce -= _unbalanceDirection;
_desiredForce *= speed;


_rb.linearVelocity = _desiredForce;```
more or less like that

[timid dove]

What happened to the force we calculated before?

[pallid gale]

which would point the blue vector towards the yellow to give it the white direction

[timid dove]

and why is velocity being set directly to a force?

[pallid gale]

its named _unbalanceDirection

[timid dove]

Force and velocity are not the same

[pallid gale]

poor naming choice on my part

[timid dove]

it would be like:

_rb.linearVelocity += (_unbalanceDirection * Time.fixedDeltaTime / _rb.mass);```

not sure why the facing direction or speed should come into play

unless we're mixing up your existing movement code with this new stuff here

(also assuming this is in FixedUpdate)

[pallid gale]

private void FixedUpdate()
{
  SolveDesiredRotation(MaxRotateRadian); // all this does is rotate the triangle to face the joystick/WASD input
  
  _desiredDirection = _facingDirection.normalized; // forces can only be applied in the direction the body is facing
   
   CalculateMomentum(); // handles the accel/deccel to build speed over time
   _desiredDirection *= Momentum; //Momentum is multiplied against fixedDeltaTime
   
  _desiredDirection *= 1f; // scale the speed 
  
  SolveUnbalance(); // the counterweight calculation to find _unbalanceDirection  

  _rb.linearVelocity = _desiredDirection;
}```Thats a more broad overview of the whole process

[pallid gale]

The way I'm thinking of it, there's the direction the triangle is moving towards that the player controls, but as the counterweight has it's own unbalanced force, I need to adjust the triangles velocity, so it combines both the players input and the velocity from the counterweight

[whole tusk]

is it more efficient to use overlapSphere or a bunch of raycasts

[novel light]

How many raycasts?

If it's like 10 raycasts, then I would expect OverlapSphere to be faster

[silver moss]

Not really something we can answer

It would also depend on the size of the overlap/length of the raycasts - which affects how many objects the physics engine needs to check

[silver moss]

What's the use case?

[slow sluice]

I have one big rigidbody cube, and above it, one small rigidbody cube, each with 0 rotation. I start the game, the small cube drops onto the big cube, but as it lands it rotates a little before settling. In theory, surely there should be no rotation, since they're both box colliders with no rotation, right?

[slow sluice]

If I set angular damping to 0, the box fully flips over on collision. How can this happen when a perfectly flat cube falls on another perfectly flat cube?

[silver moss]

Is one of the boxes a child or another box? Or are they children of anything else?

That definitely sounds odd

Does it happen in an empty scene where you just add two cubes?

[slow sluice]

I just tested it in an empty scene and now I'm getting the behaviour I expected

So clearly something else in the scene in messing with it, even though the cubes are far away from everything else. Reenabling objects one by one to figure out the problem

[whole tusk]

well, not my project. i just saw this and thought it looked like a performance hazard
#💻┃code-beginner message

[warm patrol]

Hey, do anyone know why do my objects fall through a mesh collider? If an object touches the floor with a single face it just falls through. This doesn't work with other colliders. What should I do?

[silver moss]

Mesh colliders are not supported for rigidbodies unless you mark it convex. You should have warnings/errors saying that in the console

[fair iron]

i need a formula/calculation for calculating force of each thruster depending on center of mass

[timid dove]

To achieve what?

[fair iron]

non fliping

[timid dove]

So you mean you want net zero torque

[fair iron]

mostly, yes

[warm patrol]

oh, thanks

[pallid gale]

for a kinematic rigidbody2d, Im not quite sure the right way I should move it around. I could either calculate the velocity needed to be set as linearVelocity or I could calculate a position with the velocity and move the transform.position according to that

I also don't know if I need to calculate acceleration or not in order to find the correct velocity

[timid dove]

If you want it to interact properly with other objects you will not use the Transform

You would use the Rigidbody2D

[pallid gale]

Once I set the linearVelocity and the angularVelocity of a body, like this:

rb.linearVelocity = rb.transform.up.normalized; // velocity always aims in the direction the body faces
rb.angularVelocity = GetInputDirection(); // the body will turn towards the joysticks direction```How would I find the "real" velocity (I'm not sure what to call it), as in what direction will the rigidbody move towards when you consider both the linear and angular velocity

Say in this, the point would travel along the dotted path over time

[silver moss]

Hmm, it could be as simple ascs var rotatedVelocity = Quaternion.Euler(rb.angularVelocity * Mathf.Rad2Deg) * rb.linearVelocity;
Untested

Might need to account for angularDamping too

[pallid gale]

float newAngle = _inner.angularVelocity;
Quaternion rotatedVelocity = Quaternion.Euler(0f, 0f, newAngle);
Vector3 dir = rotatedVelocity * _inner.linearVelocity;
Debug.DrawLine(_inner.transform.position, _inner.transform.position + dir, Color.yellow);```The yellow line is aiming in the same direction that the joystick is, but I dont think it's actually showing the right direction. 
In the image, even though the yellow line aims to the right, the next step of the simulation would actually have the red line match the white line as it only ever turns a small amount each step

I omitted * Mathf.Rad2Deg because I already use it when I set the value ofangularVelocity

[silver moss]

I forgot to account for time in my example. How far ahead are you trying to predict? One fixedupdate?

[pallid gale]

float newAngle = _inner.angularVelocity * Time.fixedDeltaTime; seems like that was the missing part!

[silver moss]

🤔 If you read from angularVelocity (radians) and use it in Quaternion.Euler (degrees), you need to convert

[silver moss]

Yeah

Is _inner not a rigidbody?

[pallid gale]

it is

[quartz wyvern]

is there a way to create physics colliders in scene using unity's physics stuff without fully using dots? i want to create a static scene of colliders for boids to raycast in, don't really like using the gameobjects approach

[timid dove]

It's not clear what your definition of "fully using dots" is, but you have to use DOTS to use Unity Physics

you can freely use GameObjects and Components for the rest of the game even when using DOTS for the Physics part

you'll just have to write a SystemBase system or two to glue things together

[quartz wyvern]

basically i have an array of meshes and im rendering them like this https://pastebin.com/Uai7UbPQ

because they are proc-generated i there for dont use gameobjects for it. but i still want to do raycasts so i wanted to also add an array of colliders but you have to instantiate gameobjects with regular colliders so i wondered if i can use unity's physics api and query an array of those colliders without using Systems and all that jazz

in simpler terms i just wanted to use a more pure math representation that i can query with a ray cast without all the bloat of using dots or gameobjects

[timid dove]

I think you would need to pretty much import a separate physics library and directly use its API (or write your own) to avoid that overhead completely.

It's probably not worth the hassle

The DOTS stuff is pretty lightweight.

[wind meadow]

is it normal for layermasks to allow physics cast (line or circle for example) to still get other colliders whose layers aren't in the mask?

nevermind

has to do with argument ordering for the optional parameters causing my layer mask to be fed into the distance optional parameter

[quartz wyvern]

oh my bad wrong channel

[silver salmon]

Hi, guys. What is the best way do set colliders for the ground? I made ground in blender and it is just a curved plane. I add mesh collider and have problems with dropping objects on it. They just go through the ground sometimes as collision detection for planes is not good. What can i do? Do not want to duplicate box colliders under it as it would be a lot of work(

[unique cave]

"collision detection for planes is not good" is something that I'm not aware of at least. Usually tunneling happens when fast moving objects without Continuous Collision Detection hit thin objects. Make sure that is enabled for you rigidbody, otherwise you will get tunneling effects no matter which collider type you use

So Collision Detection -> Continuous

The only problem with planes that I know of is just that they are infinitely thin so if the object gets over half way through within one physics frame, the physics system will force the object to the other side and not back to where it came from. When continuous collision detection is used, the physics system calculates continuous path for the object and stops it before it gets to tunnel through obstacles

If I didn't make myself clear on that, the Collision Detection mode should be set for the moving object, not for the ground (which doesn't likely need rigidbody at all)

[silver salmon]

yes, i tried continuous detection, still bad

[unique cave]

Could you record a short video clip of the effect happening as well as the setup you have?

It's really hard to get the full picture of what's happening and what might be causing it from textual description alone

[silver salmon]

i dropped it 10 times before, was completley fine

[unique cave]

What type of collider does the sword have?

[silver salmon]

mesh, convex. Bedore it was box but same effect

[unique cave]

and it has continuous rigidbody right?

[silver salmon]

as you can see it hits the ground with the handle, but then goes below

[silver salmon]

i played a little bit with settings, now it discretre, but with other types its the same

i will test one more time

with continious

yep, same

its fine with box colliders or mesh convex, but i cant make the whole ground convex as it adds wrong collision surfaces

[silver moss]

You shouldn't have to make it (the ground) convex anyway

[silver salmon]

and this is multiplayer, using photon. But it happens localy..

[unique cave]

I have seen similar effect happening when small/thin colliders collide with large mesh colliders, I'm currently trying if I can replicate that

[silver salmon]

i thought maybe i was missing something obvious but looks like there is no common solution

[unique cave]

The only difference is it only happened for me with very extreme values. I have not seen it happen in scene like this where the speeds are low and objects of reasonable size

Have you tried using the Physics Debug window to see if it reveals something of interest?

[silver salmon]

never tried..

when i make collider convex its fine, as it becomes to have thickness

[unique cave]

It would also be good if you could place the objects on the scene in edit mode in a pose where the tunneling happens every time so it would be easier to see what is happening there frame by frame (in the physics debugger for example)

[silver salmon]

in the scene view i move the sword under ground and it falls. But if there is some thikness, it pushes the sword back on surface as i release mouse button

looks like thats how the physics worls, nut you said you have this isue only with big values

[silver moss]

Which collider are you talking about?

[unique cave]

That does not sound normal though. I'm unable to recreate the effect in unity 6 with reasonably sized objects

[silver moss]

Yeah, something's off

@silver salmon Are you sure that the sword has no scripts or other components (like animator) that could be modifying its transform?

And it's not parented to another rigidbody or anything silly like that?

+What is the scale of the sword object in unity?

[silver salmon]

ground collider

[silver salmon]

after instantiating it is being attached to an empty parent SpawnedItems with coordinates 0,0,0. Scale is set to 0.8 right after instantiating

[silver salmon]

and there is no animator

[unique cave]

@silver salmon Did you use the regular Continuous collision detection mode for the sword or something else. The only mode (other than discrete) where anything remotely similar happens is Continuous Speculative but even then it required really thin objects to occur

[silver salmon]

i tried all of it, and yes, the ground is a plane, so it is thin

[unique cave]

no, not the ground, the falling object must be thin for it to happen for me

With Continuous I need to set the scale of the object on one axis to around 0.00002 in one axis for it to go through but at that scale range even floating point issues start to arise

[silver salmon]

not thin

maybe there is something else..

[unique cave]

If some of the faces on the ground are inverted, this could also happen I assume, but you would have likely noticed that by now due to other issues wouldn't you?

The handle still colliding doesn't seem to suggest that either (would need to be very specific triangle that's flipped but in theory possible)

[silver salmon]

dont know, i think it will fall under ground every time but it happens only 1 of 10..

[unique cave]

You could try to set up a test scene where this would happen automatically every time you press play so it would be easier to debug (wouldn't be 1 of 10)

[silver salmon]

yes, ok, will continue to investigate, thank you!

[snow surge]

Just making sure -- The spring in this case will push the cart along the blue vector, not the red one, right? (ignoring the torque applied from the offset force)

[timid dove]

The red vector force comes from the inclined plane not the spring.

[steady thunder]

Does anyone know why my player jerks so much when I move around?

private void Update()
{
    moveDirection = transform.right * horizontalInput;
}

private void FixedUpdate()
{
  // Apply horizontal movement
  _rigidbody.AddForce(moveDirection * horizontalMovementSpeed, ForceMode.Force);
}

[silver moss]

Try enabling interpolation on the rigidbody

[steady thunder]

it works

Thank you

[snow surge]

So the cart would be pushed along the red vector? I thought the spring pushing would coerce an equal and opposite force from the ground regardless of the normal of the contact point?

[unique cave]

the wheel would

[unique cave]

It will, but the wheel is pushing against the slope and therefore the slope pushes the wheel along the normal with equal force. They are two different forces. If the other force didn't exist, even if you made the slope 89 degrees steep, the cart would not slow down at all and only be pushed upwards which if you think about it, wouldn't make much sense

[snow surge]

Ah I suppose if I think of it like a bouncy ball it makes more sense that way. good point. Thank you!

[unique cave]

In reality you would get something like this happening of course but I assume you are not simulating the spring in that dimension

[snow surge]

No not quite. It's a kind of simple suspension system

Assuing the wheel is on some kind of piston that limits its movement to vertical

[unique cave]

That's what I thought. Still you would get the red component pushing the wheel towards the normal, some of which will result to the spring contracting and some will change the velocity of the cart itself

If you ran into 89 degree wall, the spring would only minimally contract and the cart would most certainly stop or do a front flip over the obstacle

[snow surge]

Oh, to be clear this example has all objects still at the initial stage

just a compressed spring

[unique cave]

With any reasonable scenario with a car having a bumper in front, I'm not even sure you would need to account for that outside the spring contraction (the maximum angle between the car body and the slope would be minimal) but in theory it should happen that way

@snow surge With a monster truck for example, I could see a scenario where the spring force wouldn't be sufficient to resolve the collision (with some trucks even a straight wall could be hit wheels first). By the spring force logic alone the truck would be sent flying upwards while in reality that wouldn't be at all what would happen (it would just face plant the wall)

[stuck bay]

Hi, I am trying to create these awesome animation thingies using a Ragdoll (trust me this is more of a physics/transforms based question)

The main concept was for the ragdoll to have configurable joints with spring force (I think it is called Angular YZ drive etc), that would make it so that the character does not fall. Another mixamo:hip bone hierarchy at the same time would play it's own animation, and the configurable joints target rotation would be changed to that another mixamo:hip. This would enable for a more "physics based" motion.

I tried achieving this, using this tutorial: https://www.youtube.com/watch?v=iNLQCwCHEBM

However, Unity freaks out and does not accurately rotate the limb in the desired direction. I have no idea if I am missing something from the tutorial or just doing something really dumb. If you need more details, screenshots, etc. head over to my dms id really be grateful for the help!

Not sure if it really matters, but both of these do not face the same direction when all rotation axes are 0

[uncut dawn]

So I'm trying to make a Sonic style game, but I'm having trouble getting loops to work.

Movement Script: https://pastebin.com/pHEWCfp0

Player Stat Script (Included because it has the gravity calculations): https://pastebin.com/BTsKfrGK

[timid dove]

There are a lot of things wrong with the code. I don't have a specific thing to call out as "the reason" it's not working, but there's much to fix:

1. both of your raycasts are hardcoding Vector3.down as the direction to cast, rather than casting in whatever direction the player's feet are.
2. transform.rotation = Quaternion.Euler(Vector2.up); makes very little sense. You are confusing direction vectors with euler angles here seemingly. This will set the player's rotation expressed in euler angles to (0, 1, 0). That's 1 degree of rotation on the y axis, and 0 on the other axes. If you just want to make the player upright that would be either transform.rotation = Quaternion.identity or Quaternion.Euler(0, 0, 0)
3. transform.rotation = Quaternion.FromToRotation(Vector3.up, LoopHit.normal); also doesn't make sense, as thius is setting hte player's rotation to the _rotation that would go from the suyrface normal to the player's current rotation... it's kind of a misfire. What you want is actually the rotation where the surface normal is "up". That would be something like this:

Vector3 forward = transform.forward;
Vector3 projectedForward = Vector3.ProjectOnPlane(forward, LoopHit.normal);
Quaternion newRotation = Quaternion.LookRotation(projectedForward, LoopHit.normal);
transform.rotation = newRotation;```
4. your basic movement code: `RB.velocity = new Vector2(MoveVelocity.x, RB.velocity.y);` makes an assumption that your player is always standing upright, which obviously is not true when going through a loop.

[uncut dawn]

Thanks for the response! I'll try to fix the stuff you pointed out, but I have a question about the code you gave me.

Isn't Vector3.ProjectOnPlane supposed to take 2 arguments and not just one?

[timid dove]

yes that was an omission on my part, I fixed it.

[uncut dawn]

Thank you

[stuck bay]

Anyone?

[silver moss]

Do you still have your ragdoll bones parented to each other? They should not be

Make them "siblings"

Either with no parent or a parent that never moves

Also, make sure that no animation is affecting the ragdoll bones directly

Btw that tutorial is questionable. In the final result, the legs rotations are inverted. 🤦‍♂️

Active ragdolls are more complex than "joint.rotation = targetBone.rotation"

For me, this extension has been helpful for setting the target rotation correctly:
https://gist.github.com/mstevenson/7b85893e8caf5ca034e6

Make sure to read the comments

[stuck bay]

Okay let me look into it and lyk asap

Thanks!

[pallid gale]

I've been setting angularVelocity in a rigidbody2d to allow it to rotate how I like, but I've figured its probably better to start using AddTorque() and now that I am doing that, I cant seem to get any of my existing rotation code to produce the correct torque value.

The documentation says that in order to get a specific change in degrees that I should do (angularChangeInDegrees * Mathf.Deg2Rad) * body.inertia; . Which makes me think that is all it should take to use the code I have which already is producing the angular change I want in degrees

[timid dove]

torque is analagous to force
angularVelocity is analagous to linear velocity

there is no such thing as an equivalence between the two, because they are apples and oranges

there is no such thing as a torque that results in a particluar angular change in degrees

torques produce velocity

and the velocity determines how quickly the orientation of the object changes

So all torques will eventually result in some number of degrees of change, but there's an additional variable in there - time

The sentence:

However, if you require a specific change in degrees, then you must first convert the torque value into radians by multiplying with Mathf.Deg2Rad then multiplying by the inertia.
Is poorly worded. What they meant is "degrees-per-second"

not degrees

And there's little reason to use AddTorque there instead of just setting the angular Velocity directly if your goal is to just reach a desired angular velocity

[pallid gale]

For my use all I care about is for it to do something akin to "rotate from the current angle to a target angle over a period of time"

I'd assumed AddTorque was a better way to do that as it might do a few extra things underneath that would make the angular velocity be set correctly rather than the value I'd find if I was to set it directly

it says it scales by inertia which I've never been doing, so I might have been accidentally setting the final value wrong

it seems like I've messed up the turning circle somehow 😅

[timid dove]

"rotate from the current angle to a target angle over a period of time"
Definitely just set the angular velocity manually

You will go throlugh that whole torque calculation thing and your end result will just be the same exact thing as if you just set the angular velocity manually

The only effect that AddTorque has on the body is to change the angular velocity

[pallid gale]

What would the intended use be for the method?

[timid dove]

if you don't have an intended target angular velocity and you just want to apply a certain amount of torque.

For example a game where you fly a spaceship and you can hold Q to fire your attitude control rockets

you would AddTorque in FixedUpdate while Q is held

[pallid gale]

ah, that makes sense

[lone atlas]

Hey, folks. I'm quite new to unity, so appopolgies if I come across a bit green. I've modelled a banana peel , and would like the peeled pieces to have almost ragdoll physics, but I can't seem to find any tools for ragdoll'ing that aren't specifically for humanoid characters. Any ideas? :)

I've had a search online, but everything seems to be for characters, rather than props.

[copper sentinel]

The tool you are referring to is only to simplify the process of adding rigidbodies, colliders and joints to the bones of humanoid character. You can add and configure these components yourself to pretty much anything. Check documentation to find the best components for your use case. If you want, you can also create some kind of humanoid character ragdoll in order to check how all the joints are configured there.

[lone atlas]

Thank you! Figured it out in the end, is just a case of playing around with the values now :)

[stuck bay]

When I do set a target rotation, do I do it only for the bones that the ragdoll wizard setup with rigidbodies or do I start adding the script to every single bone

[silver moss]

You set target rotation for each joint that you want to move

[stuck bay]

Got it.

[graceful harbor]

I got this 2D homing fireball but when I apply the follow script it follows the player but the rotation is all wrong it's like the trail has a weight.

The trail is being dragged downwards

[timid dove]

transform.right = direction; is all you need, not anything with the Cross products

[graceful harbor]

So I only change linearVelocity and transform.Right and not angularVelocity?

[timid dove]

Is there a good reason you would need the rotation to be physically simulated? Isn't it just a visual thing?

[graceful harbor]

Not really, it is just a visual thing but that was the solution I found online since I'm not too good with physics calculations

Is this updated code the proposed solution?

[timid dove]

Yes

Though it might break interpolation now that I think about it. Two solutions to that would be:

• use Atan2 on the direction vector to convert it to an angle and apply that to rb.rotation
• put the sprite on a child object and rotate the child only as here

[graceful harbor]

I solved it by making the main sprite a circle and by seperating the trail to its own animation so it doesnt get affected. Thanks anyway :)

[lilac patrol]

anyone have any idea how to implement tank movement (without actually simulating each part of the track with hinges)? I've tried using wheel colliders but I had some issues with that, now I'm just calculating the angular velocity of each track (left/right) but I'm having trouble applying that as a force to the tank. either there's too little friction and it slides once it starts moving, or it can't move at all.

[silver moss]

Show your current code too

[lilac patrol]

this is what I'm currently using to apply forces to the tank (had to use really big numbers to even get it to move)

        float velocity = (treadLeft.treadSpeed + treadRight.treadSpeed) / 2f * 100000f;
        float angularVelocity = (treadRight.treadSpeed - treadLeft.treadSpeed) / 0.5f * 100000f;

        rigidbody.AddForce(transform.forward * velocity, ForceMode.Force);

        rigidbody.AddTorque(Vector3.up * angularVelocity, ForceMode.Force);

should be transform.up *

[silver moss]

Probably not a fix to your problem, but instead of torque you might want to try AddForceAtPosition for each side of the track

So turning right would be achieved by adding forward force on the left track and backwards force on the right track

That's sort of how tanks work right?

[lilac patrol]

that's another one of the things I tried haha

didn't work too well either

I think it's mainly a problem with friction

since the tracks are just static colliders

[silver moss]

Btw the reason you have to use large numbers like 100000f is probably because your tank has a large mass

And ForceMode.Force divides the force by mass

https://www.reddit.com/media?url=https%3A%2F%2Fpreview.redd.it%2F6y8myu3ruxo71.png%3Fwidth%3D1080%26crop%3Dsmart%26auto%3Dwebp%26s%3D7aa3db9af9a51d692d2f79e5bce2474a07699418

[lilac patrol]

true

[lilac patrol]

I had that open haha

[silver moss]

Acceleration might be easier to work with here

Also, what do your colliders look like, specifically on the bottom that is in contact with the ground?

And the physics materials?

[lilac patrol]

like 5 sphere colliders

on each side*

[silver moss]

Might want to use low friction on the physics mat so you get better control over the velocity with forces

[lilac patrol]

I've tried a bunch of combinations of static/dynamic but can't seem to get it right

[silver moss]

Making your vehicle a "hover craft" is pretty common too: Don't have colliders for the wheels, just use physics queries (RayCast, SphereCast) to check the distance to ground and add force to hover above it

[lilac patrol]

if the friction is too low, it'll slide when the tracks stop turning

[silver moss]

This way the collisions don't interfere with your velocities

[lilac patrol]

hmm, I could try this

thanks

[silver moss]

Oh one more thing, try messing with rigidbody.centerOfMass

Maybe expose it as a variable for debug/dev purposes

[lilac patrol]

hmm ok

what would be a good way to stop it from bouncing too much up and down with the "hovercraft" approach?

I guess I just have to write a full suspension system at this point

guess that would be fun too

[silver moss]

Damping/not adding too much force when not necessary

[lilac patrol]

yeah I'm trying to implement damping now

[vale dragon]

Seemingly simple question - how do you detect trigger-type collisions on fast projectiles?
I looked this up for a while and the solution often cited is to use Rigidbody2D.Cast. But when I tried that, I eventually discovered that only works on NON-TRIGGER types of colliders. That seems to defeat the whole point of that solution, as for non-trigger/solid colliders you can just use Continuous collision detection.

I know I can just set the projectile's collider type to non-trigger and set its layer overrides to exclude "Everything", but this feels like a weird solution and my gut is I'm doing something wrong here.

any advice?

[lilac patrol]

how do you detect trigger-type collisions on fast projectiles?
can you clarify what you mean by this? what is the situation you're describing?

[vale dragon]

Suppose you have a bullet and you want it to deflect off another collider at an exact angle. Solid (non trigger) collider won't work for that, so better would be using the bullet's collider as a trigger, and when the bullet's OnTriggerEnter code fires, activate a change in the bullet's velocity to the angle you want.

That way it doesn't bounce off the other collider with any weirdness, because you can precisely control what it does.

[lilac patrol]

I'd use a raycast and calculate the new angle from the hit normal

[vale dragon]

Huh, good to know. I'm surprised I kept seeing suggestions to use cast

I'll keep that in mind. Thanks!

Actually, what if it's not something small enough for a single hit point? Like what if it's a big sprite moving fast, like a ball or a car or something?

Or even a really cartoony, wide bullet?

[silver moss]

SphereCast etc.

[vale dragon]

What if it's a complex shape like a vehicle?

[silver moss]

Rigidbody.SweepTest

[vale dragon]

For 2D?

[silver moss]

Collider2D.Cast for each collider

There's a physics query for almost everything

[vale dragon]

Exactly, Rigidbody2D.Cast is what I was trying to use before. But I found it only works with solid/non-trigger colliders. I'm trying to figure out how to use that for trigger-type colliders

Assuming it's a situation where you don't want it to be solid/non-trigger. Suppose a fast moving lightning beam or something

[silver moss]

What is the purpose of the colliders at this point?

[vale dragon]

Calculating if the object hit something, taking into account the object's exact shape

[silver moss]

You might be able to use Rigidbody2D.Cast/Collider2D.Cast and set the colliders to non-trigger temporarily, but that sounds like a hack

[vale dragon]

Yes, exactly. I found I can slightly less hack it by setting exlusions to Everything in the collider.

But still, a hack. Anyway around hacking this?

It seems like it should be straightforward

[silver moss]

I was going to suggest using solid colliders but with collisions disabled via their layer, but seems like Rigidbody2D.Cast/Collider2D.Cast can't take a custom layermask parameter so i guess it wouldn't hit anything then

Can't think of a non-hacky solution right now. I'll let you know if something comes to mind

For now, I would probably just cast with solid colliders that are disabled normally and enabled only for the cast

[vale dragon]

I feel a lot better knowing I haven't overlooked something obvious then 🙂

[vale dragon]

Ok, I'll give that a try

Thanks

[vale dragon]

@silver moss So I did a full test and it mostly works? The cast works terrific for detecting the other sprite, and the red sprite never even vaguely clips through.

However, it looks a bit odd visually for the opposite reason. In a simple test here, I'm trying to make high-speed bouncing work, but WITHOUT actual bouncing, instead using code to reverse Y velocity when a hit from the cast is detected. I have the red sprite's collider set to solid (non trigger) and am keeping it that way for simplicity's sake now. There are no physics materials attached to either red nor green sprite.

But it doesn't look like it goes all the way down. How far down it goes even varies from bounce to bounce.
It looks like the framerate's been capped to 30, but the inconsistency looks even more pronounced on my 60ish hz display:

I coded the moving sprite's casting according to how I've seen recommended online mostly. The direction of the casting is determined by the normalized velocity, and the distance it casts is determined by the speed of the sprite * Time.fixedDeltaTime:

private void FixedUpdate()
{
    float distance = rb.linearVelocity.magnitude * Time.fixedDeltaTime;
    int numberOfCollisions;
    
    numberOfCollisions = rb.Cast(rb.linearVelocity.normalized, hits, distance);
    Debug.Log("number of collisions is " + numberOfCollisions);

    if(numberOfCollisions > 0)
    {
        FastTriggerEnter();
    }
}

private void FastTriggerEnter()
{
    rb.linearVelocity = new Vector2(rb.linearVelocity.x, -rb.linearVelocity.y);
}

[silver moss]

But it doesn't look like it goes all the way down.
This makes sense because you are casting from current pos to next pos, instead of previous pos to current pos

It never gets a chance to reach the collision point

Because you invert its velocity when it's "about to collide"

To me it would make more sense to cast from previous to current position. If hit, set position to the closest hit position and revert the velocity or whatever you want to do

What's the usecase here though?

[vale dragon]

Use case was I was trying to make a bubble that a circular player object could jump on, and when you jump on it it would perfectly invert your Y velocity so you could chain jumps together. At the same time, when the player bounced on the ground, it needed to set the player's exact bounce up speed, IF the Y velocity was too much. This is because you could let gravity drop the player, or press a key to slam down fast, and I wanted a controlled bounce if you were going down super fast.

I also remember having issues with bouncing on bubbles not being consistent 100% with the velocity of the bounce. Usually it was perfect, sometimes it was like a dead stop. I think I was trying to find a more consistent solution

[vale field]

Hi, might anyone know why my script isn't working as expected? I've made a trajectory predictor, by following the implementation in this forum:
https://gamedev.stackexchange.com/questions/71392/how-do-i-determine-a-good-path-for-2d-artillery-projectiles
As of now, the path that my object travels is slightly lower than the one indicated by the line renderer, so I believe that I've either miswrote the equations or there is some aspect of the physics system that I haven't taken into account. Here is my script:
https://scriptbin.xyz/norokoniba.cs

I know very little about unity's physics system, so I'd love an explanation as to what went wrong here

[vale dragon]

That's a great idea. Thanks!

[silver moss]

Make sure to move it according to hit.distance, not hit.point

[vale dragon]

What would go wrong with hit.point?

[silver moss]

If you move the object to hit.point, it will now partially clip into the collided object

You should just add cast direction * hit.distance to the position instead

This way you move it "until it hits"

So that the collider's edge ends up on the hit point

[worn mantle]

hello! Have a quick question here, i made my player characters collider bigger then the player, so it actually floats ABOVE the ground

i have a raycast checking if theres ground within 10m below my character

[timid dove]

ray.GetPoint(distance) is ideal

[worn mantle]

but here, my character is UNDER the ground

ground is set at 0, and the pivot point of my transform is where the move component is

[timid dove]

Make sure your tool handle position is set to Pivot not Center

[worn mantle]

it is,

[timid dove]

The player should never be below the ground in the first place.

[worn mantle]

yeah, and its only underground in this specific area

[timid dove]

If this y is negative then your ground is not at 0

Or it's not a flat plane or something

[timid dove]

That's a map magic terrain

[worn mantle]

it is just a couple of Unity terrains behind the scene

[timid dove]

Which is a heightmap basically

[worn mantle]

i can paste the terrains themelves, one sec

[worn mantle]

so just for clarification, its not to do with collider of player , but something to do with MagicMap, a 3rd party asset? If so i will figure things out myself

[silver moss]

True that, I always forget it exists

[waxen kindle]

Why doesnt the scale effect the raycast this is very weird

This is really realy weird

Btw i alsi tried collisions but only raycast is broken

[timid dove]

what do you mean by that

can you explain what we're looking at?

What raycast is involved here?

[timid dove]

not mapmagic necessarily just the height of the terrain

[vale field]

Would anyone mind helping me out😅?

[light cradle]

i want to create rotation for my spaceships with add torque (physics are important, yes) and it's in space so there is no drag.
problem is idk how to make it not overshoot the rotation -> so it always adds the exact torque/deceleration counterforce to smoothly stop exactly at the target rotation (i am trying to do this for so long now i am losing my sanity again)

[waxen kindle]

İ do a raycast to the layer of those 3 toruses however even if i change the scale of these objects it raycasts like the scale is 1

[timid dove]

I'm not sure what you mean by it Raycasting like the scale is 1.

What are the details of the Raycast?
What results are you expecting from the Raycast?
What results are you seeing instead?

[waxen kindle]

Nvrmind i fixed the issue

[sick marsh]

i'm doing some collision checks for a custom character controller using com.unity.physics, and i'm trying to use cast my character's capsule into the collision world to determine a maximum position that i can move to without penetrating other objects

the problem i'm running into is that ColliderCastHit.Fraction returns a fraction such that if i move by that fraction, the next CapsuleCast will consider the cast to hit at Fraction=0, leaving me stuck

i've checked the source and this seems to be a trivial consequence of how Unity.Physics.ColliderCastQueries.ConvexConvex is implemented, as it advances in discrete steps and outputs the first fraction where a collision is already happening

i can work around this by modifying the source to add a ColliderCastHit.LastFraction property which simply stores the last fraction at which a hit wasn't yet detected (a hit being detected by distance < 1e-3)

is there a better way to get a pre-penetrating fraction for collider casts? i'd prefer not to have to modify the package source if i can avoid it

[errant meadow]

Hey all, I am trying my hand at a 2d platformer and I'm trying to stop the player from being spiderman...

The player is a rigidbody2d set to dynamic, I figured out how to stop it from sticking to ceilings with a 'grounded' check, but what about walls? The solutions I see online are to use a frictionless material, but that results in sliding!

Just wondering if there are any common solutions to this problem, thanks!

[errant meadow]

ok I fixed it with a second box collider that goes around the main one but has no friction, still wondering if theres a neater method though

[zenith parcel]

when I walk against a wall or anything that collides with the player, it starts to spin. How do I make it tight and not spin?

[timid dove]

constrain the Rigidbody's rotation

[dull shuttle]

how to make an object with HingeJoint to swing infinitely?
if this component can't do it then tell me please which speed rotates the object when the component is added and how to get/set it

[turbid ember]

anyone experienced with articulation bodies I have a question

do child articulation bodies of the parent articulation body ignore collision within itself?

[turbid ember]

in addition also, is it possible when a collider (with a rigidbody) imparts a force on another rigidbody, can one check for this collison and nullify the imparting force?

[timid dove]

Yes with this free asset: https://assetstore.unity.com/packages/tools/physics/betterphysics-selective-kinematics-244370
Or with this API: https://docs.unity3d.com/6000.0/Documentation/ScriptReference/Physics.ContactModifyEvent.html

[rotund sentinel]

Hey everyone,

I'm trying to get my first-person character controller to move smoothly on a moving and rotating platform (like a ship) without making the player a child of the platform.

Right now, I’m using a raycast to detect if the player is standing on the platform (by checking a specific layer), and then I apply the platform’s movement and rotation delta to the player in LateUpdate().

What I’m Doing

1. Raycast Down from the player's feet to see if they are on a platform (using a tag or layer).

2. If they’re on a platform:

Store the platform’s position and rotation from the last frame.

In LateUpdate(), calculate how much the platform has moved and rotated.

Move the player by the same position delta and rotate their Y axis to match the platform.

3. If the player jumps or steps off, the raycast no longer detects the platform, so they stop moving with it.

The Problem

It kinda works, but the player slides when the platform moves, especially when it turns only. Since I’m using a CharacterController, I’m applying movement using Move() instead of setting transform.position, but it still doesn’t feel right.

[frigid pier]

@rotund sentinel Don't do weird spacing pushing away other messages. It's also highly unreadable as a whole.

[turbid ember]

nice I'll look into this, are you able to answer my other question? #⚛️┃physics message

if not thats fine

[timid dove]

I don't know anything about articulation bodies

[turbid ember]

noted, thanks for the assist anyhow

[rotund sentinel]

Spacing?? Can you explain what you mean by that?

[turbid ember]

lots of spaces inbetween your text

[random moat]

Does anybody know if it's possible to fully set the state of a rigidbody to a recorded state, including it's contacts? Obviously you can set it's position/rotation/velocity/angular velocity, but this leaves out any contacts it may have had which means it's state is not identical to the recorded state.

The only idea I've had so far that might accomplish this would be to actually record it's state 1 physics update prior to the desired state, set it's position/rotation/velocity/angular velocity to this state and then run PhysicsScene.Simulate() once. I believe this would cause it to then generate contacts, but I don't know if they'd be identical to what would've been generated originally. If anyone has any thoughts on solving this kind of problem I'd love to hear them.

[timid dove]

I'm not sure I would agree about contacts being part of the state? What's the end goal here?

[random moat]

I want to be able to modify the impulses of the contacts generated for a particular rigidbody via Physics.ContactModifyEvent. However, for the particular modifications I want to make, I need to know what the impulse is going to be for each contact. From what I can tell, ContactModifyEvent occurs before the impulses are calculated as there isn't any way to get the impulse of each contact in ModifiableContactPair.

So, what I'm hoping is that if I were have a duplicate rigidbody in a separate physics scene that I apply the state of the original rigidbody to and then simulate one frame for, I will be able to record all the contacts generated for the duplicate rigidbody including the impulses for them. Then, when the original physics scene is simulated and I receive the ContactModifyEvent, I would be able to match up these modifiable contacts to the ones I recorded for the duplicate rigidbody and know what their impulses will be. I am hoping that if the position/rotation/velocity/angular velocity of the duplicate rigidbody are identical to the original, then the same contacts would be generated.

However it seems that if you set a rigidbody's position/rotation, even via MovePosition or MoveRotation, this affects how the contacts are generated, even if you set them to the position/rotation the rigidbody is already at.

[silver moss]

You want angularVelocity if you want to set a rotational speed

[elder citrus]

How fast can a physical bullet object go in Unity before it starts to phase through other objects? I am somewhat reluctant to use raycasting for bullets

[silver moss]

If you don't use continuous collision detection, the bullet can phase through a wall if it moves a greater distance than the thickness of the wall (+the radius of the bullet) in one physics step

So you want CCD at least if you go with rigidbodies

[elder citrus]

Alright, thanks

[silver moss]

The effect is called "tunneling" FYI

[elder citrus]

CCD works at any speed or will it still break if speed still goes high enough? I copied the bullet speeds from muzzle velocities of real life guns and some of them fire their bullets at speeds just over a kilometer a second

[elder citrus]

Ah, thanks for letting me know

[timid dove]

For such velocities a SphereCast or CapsuleCast may give you better results.

[silver moss]

I do prefer casts for bullets that travel at realistic speeds

Raycast works for most cases, you only need spherecast/capsulecast if the bullet needs to have some volume

If you potentially have a lot of bullets active you can look into RaycastCommand. Way cheaper than having hundreds of CCD rigidbodies flying around

[elder citrus]

Thanks for the advice!

[rotund sentinel]

What is the best solution for making a character move on a moving & rotating platform without making it a child?

I've experienced issues with CharacterController not inheriting platform rotation properly, causing drift. Is there a way to fix this, or is Rigidbody the better option? Which one is the best approach for smooth movement and performance?

How would the logic work for handling this correctly?

[silver moss]

With a rigidbody, when calculating the character's velocity each frame, you can add the platform's velocity to the character's final velocity before applying it

[timid dove]

Have you looked into off the shelf solutions like https://assetstore.unity.com/packages/tools/physics/kinematic-character-controller-99131 ?

KCC has this problem solved pretty well

[dull shuttle]

thanks (bruh i forgot to send)

[pallid gale]

I'm not sure how to describe the problem, but I'm having trouble figuring out how to get the movement to feel "tight". See how in the video after it completes a 90 degree turn, instead of moving in the new direction all the existing velocity pushes it sideways

[timid dove]

Your object has momentum

that seems to be the entirety of your problem?

If you don't want momentum, you should be manually overwriting the velocity instead of adding forces.

alternatively - you need to apply much higher forces if you want to get faster acceleration.

[pallid gale]

I'm wary about changing the code to go from adding force to overwriting velocity, mostly as I'm not sure if that would be a significant change to the code I've already got

I think if I could increase the "friction" of it, that could help straighten it out once it makes the turn

[timid dove]

yes another option is:

• applying higher forces overall
• applying forces that counteract your current sideways momentum
• limiting top speed either with another drag calculation or rb drag or straight up clamping

Basically some combination fo all this

but you just need to think of it in basic physics terms

to counteract any kind of velocity you don't want, you need a force

[dull shuttle]

i made some swinging object and i don't understand how to get it's angle
i used transform.rotation.eulerAngles and for some reason it swings between 290 and ANOTHER 290 angle, why?
i mean it goes to from 290 to 270 than again increases to 290

[timid dove]

what do you plan to do with the angle

basically using euler angles to do logic is always a bad idea

especially reading them back from a Transform

euler angles are pretty much cancer

[dull shuttle]

how to get an angle that i can see in Transform component?

[timid dove]

You can't get that one

it's visual only

and it's a bad idea

what is the END GOAL here?

There are better ways to do whatever it is

[pallid gale]

would it be the same principle as a helicopter rudder?

[timid dove]

All newtonian physics works the same

f = ma

what you have here is torque which is for rotation which is really not that relevant here

because we're not concerned with angular velocity

[dull shuttle]

to make the object swing infinitely by applying force when it reaches equilibrium point (i mean the point where it ends up when stops)

[dull shuttle]

that was the answer to this question

[timid dove]

You mean down

so it depends on the orientation of the thing but

presumably the rest point is at "0" rotation?

[dull shuttle]

no

[timid dove]

Can you show the object in scene view with its gizmos so we can see its orientation?

(with gizmo rotation in local mode)

[timid dove]

ok so it's weirdly rotrated but if i'm reading this correctly you want to know when its forward direction (blue arrow with the move tool enabled) is facing down it seems?

So you can do something like:

float differenceFromDown = Vector3.Angle(transform.forward, Vector3.down);```

[dull shuttle]

*it will face up in that position

[timid dove]

it's hard for me to tell with the rotation tool selected

[dull shuttle]

that's why i told you it

[timid dove]

Then it'd be:

float differenceFromDown = Vector3.Angle(transform.forward, Vector3.up);```

And if you need a signed angle you can use Vector3.SignedAngle

[rotund sentinel]

Hmm, never consideret it, but it seems im forced to use it now. I may try to make my own one firstly

Thanks for the tip

[dull shuttle]

i wrote this before

if (transform.rotation.eulerAngles.x < -90 && !speedsaved)
{
   angVel = rb.angularVelocity;
   speedsaved = true;
}```
how to use what you wrote here?