⚛️┃physics

[timid dove]

What's it supposed to be doing and what's going wrong?

[fiery chasm]

its supposed to detect the ground in front of AI and stop him if he's going to fall. just waiting for my project to compile to show relevant scs

debug ray properly working, collider has right tag, and i checked, layers are set to collide w each other

[timid dove]

But you haven't actually said what's going wrong

Also, the layers collision setting is irrelevant for raycast

[fiery chasm]

oh right sorry, just never collides with the ground

[timid dove]

What does it collide with? Have you done any logging? How do you know it isn't hitting the ground?

[fiery chasm]

it never sets off the if statement, i am just double checking now that it isnt colliding with itself, i tested earlier and it wasnt

[timid dove]

So why don't you add a log that prints the object that it actually is colliding with?

I will note that your code doesn't seem to be taking into account the possibility that nothing is being hit and will throw an error in that case

Are you seeing errors in the console?

[fiery chasm]

just logged it, it was colliding with itself🤦‍♂️

what is the best solution for this? simply offset it or make the collider another layer?

or something else?

[timid dove]

Use a layer mask on the Raycast

[fiery chasm]

ok cool thanks so much 🙂

[mellow idol]

so how do i make things not fall through the ground

No tutorials are working and i don't know why

[timid dove]

Give them and the ground the appropriate physics components and make sure everything is moving via the physics engine

[mellow idol]

I have literally no idea what any of that means

no matter what i do it just wont work

[timid dove]

Probably related to you not knowing what anything is.

Anyway your description of the problem isn't really giving anyone any clues to help you

[mellow idol]

ill write down everything i've tried

[timid dove]

You would need to share with us exactly what you tried, how you set everything up, then we can point out what's wrong

[mellow idol]

ok

well i've tried turning off is trigger on both colliders for the terrain and capsule colliders, i've tried giving the terrain rigid body, the capsule rigid body, setting the colision to continuous on both, giving the terrain a box collider, giving the capsule a box collider, coding my own collision (it wouldn't even start after that) and thats about it

[timid dove]

1. Yes the colliders need to all NOT be triggers
2. Please share screenshots of the inspectors for your "capsule" and the terrain

[mellow idol]

I just have a capsule with this code for movement, but the jumping doesn't work anyways because it can't touch the ground to begin with

using Microsoft.Win32.SafeHandles;
using System;
using UnityEngine;
using UnityEngine.InputSystem;

public class MovementScriptBaseCharacter : MonoBehaviour {
    public float movementspeed = 5f;
    CharacterController controller;
    float gravity = -3f;
    float velocity;
    [SerializeField] float gravityMultiplier;
    public float velocityMultiplier;
    public float jumpPower = 5f;
    PlayerInput playerInput;
    InputAction moveInput;
    InputAction jumpInput;

    static bool IsDown(InputAction action) => action.phase == InputActionPhase.Performed;

    void Start() {
        controller = GetComponent<CharacterController>();
        playerInput = GetComponent<PlayerInput>();
        moveInput = playerInput.actions.FindAction("Move");
        jumpInput = playerInput.actions.FindAction("Jump");
    }

    void Update() {
        CharacterMovement();
    }

    void CharacterMovement() {
        Vector2 direction = moveInput.ReadValue<Vector2>();
        transform.position += new Vector3(direction.x, velocity, direction.y) * movementspeed * Time.deltaTime;
        if (controller.isGrounded) {
            velocity = -1f;
        } else {
            velocity += gravity * gravityMultiplier * Time.deltaTime;
        }
        if (velocity == -1f && IsDown(jumpInput)) {
            velocity = jumpPower;
        }
    }
}

oh

thats much longer than i thought

[timid dove]

Your code is moving the object directly via the Transform

[mellow idol]

Yeah

[timid dove]

Well that's why it doesn't work

That bypasses the physics engine entirely

[mellow idol]

wait then what do i move it with then

[timid dove]

Use the Rigidbody

[mellow idol]

But i can't change the rigid body's gravity

[timid dove]

Or a CharacterController

[mellow idol]

Hold up but i thought the transform position WAS the character controller?

[timid dove]

Physics gravity can be controlled from the physics settings or in your code

[timid dove]

CharacterController is a specific component by that name

[mellow idol]

Well i do have a character controller set up so ill try to do it

[timid dove]

Maybe you do but your code isn't using it

Your code is directly changing transform.position

[mellow idol]

can i just replace transform.position with controller.Move

[timid dove]

That depends on if everything else is set up properly

[mellow idol]

Well i have the character controller set up so ill see if it works

yeah it cant assign move cause its a method group or something, idk what that means so ill just try to put the gravity in a whole different code block i guess

[timid dove]

That means you tried to call a function without the ()

Move is a function you can't do Move = something

You have to call it

Like myCC.Move(location);

[mellow idol]

i guess but now an issue is that the direction takes too many arguments so it doesn't work

[timid dove]

Well naturally you have to give it the correct arguments

[mellow idol]

I can't specifically use my velocity because the direction is from a 2d vector and whenever i try to use a 3d vector it gets mad at me for not assigning up and down

[timid dove]

It should just be the whole expression you were "+=" ing before

[mellow idol]

and i don't want people to be able to just float up or down

[timid dove]

You seem to just be struggling with the basic C# syntax of this

[mellow idol]

well im not exactly focusing since i'm still trying to code in the middle of this conversation to fix it

which luckily enough i finally found a fix for it

[mellow idol]

i didn't know you could just delete up and down on the 3d vector action so now i dont get a barrage of error messages probably and my movement is actually working now

[mellow idol]

Yeah I didn't know i had to use the controller for physics 😭

but yeah thanks for telling me about the thing with transform.position, i'll probably only use that if i have to add teleporting or something.

[tardy rune]

Does anyone know why it keeps making the reload sound when I'm not even touching the gun?? I'll guess I need to change it back to a physics joint?

[neon island]

There is nothing in that code that refer to a reload sound and when freeFloating is false, localDifference will always be equal to 0. I don't know what this code actually do

How to create damping between two object using code?

[tardy rune]

The reload sound is trigger by another triggercollider/script when the gun slide is pulled back. Visually it stays in place but the gun keeps reloading/triggering the sound.

[neon island]

You should take a look at this (it explains the update order) : https://docs.unity3d.com/6000.1/Documentation/Manual/execution-order.html . What you are doing with the code is setting the position of the transform during the Update, but the trigger actually triggers just after the physic simulation and before the Update, so the Rigidbody might move during the physic update and trigger the reload but you wouldn't see this since when the frame update, you set the position. What you should actually do is lock its local position with physics so it never touches the trigger when you don't want it to

[neon island]

You can probably lock the rigidbody by disabling it when needed and setting its "local" velocity to 0. But I don't know, I never tested it before

[tardy rune]

Thanks for the tip, I used IsKinematic and it worked.

[tribal dragon]

I'm trying to use Spring Joints to create a some sort of lining for my fishing rod - but they are too springy/bouncy. What values I should set for it to behave properly? Or using Spring Joints in such way is incorrect?

[timid dove]

What does "lining" mean for a fishing rod?

[tribal dragon]

It's like a very thin rope, in a way

[timid dove]

Oh a line

It really depends on exactly how this fishing mechanic is supposed to work

I'd say the vast majority of the time trying to make it physically realistic is a bad idea

[tribal dragon]

Honestly, it's mostly for visual sake and all

[tribal dragon]

It's not about it being realistic: I just don't want the points to bounce up & down, but still react with physics
(i.e. when the player cast his fishing rod and the lining/rope flies forward in the sky, until the end hits the water)

[timid dove]

I would definitely not do this with a bunch of rigidbodies and joints

[tribal dragon]

Any suggestions for an alternative?

[timid dove]

verlet integration, trailrenderer, bezier curves

[worn mantle]

Ello, just wanted to get confirmation on a problem ive been having. My game has the ability to place objects on the ground, and checks to see if anything is colliding with it (ground included, because i will have slopes in the game). I have the buildings placed such so that they are "touching" / edge colliding with ground. Is the solution here to multiply the y size by like .99 when doing boxcasts? Or is there a better solution to handling edge touching of different items. (PS: I want the boxcast to see if theres any part of the ground above where its placed) so just want to ignore "edge" cases for now, can implement better logic if i want to be able to place on a slope

[timid dove]

If you want to have the buildings aligned perfectly, you could use a grid-based system with snapping. Otherwise, the .99 scale trick on the OverlapBox call is good IMO.

[worn mantle]

Ahh gotcha, i do have a grid system, but the grid is more of a helper for the player, rather then dictating where items can be placed. thank you for the reponse

[lusty shard]

How much of an issue is it if my player is at a y position of -5.960464e-08 when it loads in and falls using a player capsule and rigidbody?

[timid dove]

Why would that be an issue?

[lusty shard]

Im unsure if it will affect certain calculations 🤷‍♂️ . This is my first unity project

[timid dove]

What do yoou mean

what kind of calculations and how would they be "affected"?

[lusty shard]

Idk ground checks, uneven surfaces

thats the thing I dont know if it would affect any

[timid dove]

what about them?

[lusty shard]

its jsut a general question

[timid dove]

Are you aware the number you wrote, written out as a full decimal is:
-0.00000005960464

[lusty shard]

yes

[timid dove]

it's effectively just 0

so I'm really not sure what your question is

[lusty shard]

So its not an issue

you answered it

[timid dove]

I mean I don't know for sure, because I'm having trouble understanding what kind of issue you think it would be causing

if you are trying to compare that number to be PRECISELY 0, for example, then yes that would be an issue

[worn mantle]

i mean if you wanna use it for a collider size or a transform scale, then it would also be a issue

you got to give what context you think it might be an issue in, because things depend on other things. If it was ALWAYS an issue, unity wouldnt have the ability to store / use that float

[final parcel]

Anyone know why its giving an error?

[timid dove]

You can mouse over errors to see what they are

Or look in the console window

[pulsar canopy]

Tessellation doesn't affect raycasts, right?

Meaning, a ray will hit the original terrain mesh before tessellation

[unique cave]

shader displacements in general never get send back to CPU and therefore have no effect on collisions

[cosmic bear]

My player ship seems to be rotating to the Z axis when colliding with an object despite me having already froze all points of rotation

[timid dove]

Show your code

[cosmic bear]

https://pastebin.com/MuEM7Xbu Certainly. This is the code that controls movement

I purposefully set the object the player is colliding with low mass so it doesn’t affect its rotation that much

[inner thistle]

rb.freezeRotation = false; at the end of the ApplyRotation function unsets all those constraints

[timid dove]

Yep - it's like Nitku says. You're turning those constraints off in your code

you're also rotating the Rigidbody directly via its Transform, which is a bad idea

[cosmic bear]

Ah silly me! My reasoning for making it false was that I suspected it would eliminated the rotation of the X and Y which are needed to even move certain directions in game

Thank you!

[cosmic bear]

I see. Any more efficient alternatives?

[timid dove]

it's not a matter of efficiency

[inner thistle]

Rigidbody constraints don't apply when manipulating the transform directly

[primal dagger]

what is the best way to add collision to large environment models? like a winding ravine path or cave system

[unique cave]

I would generally consider mesh colliders for their ease of use, for detailed meshes it can have an impact on performance so using separate more simplified mesh for the collider may be appropriate. If you have different LOD levels for the mesh, you can use less detailed mesh from there or make a new one (modelling tools often have tools to simplify existing meshes so you don't have to do that by hand)

[primal dagger]

my meshes arent very complex(psx style game) so i thought to use mesh colliders too but i seem to have misjudged how they worked
it kind of just tried to encapsulate the entire path in a collider which didnt work too well

[primal dagger]

do i have to split up the mesh into pieces that can work with the mesh collider or did i do something terribly wrong?

[unique cave]

I'm not sure what you mean though make sure the collider is set to non-convex

[primal dagger]

alright i got it working

thanks

[fiery chasm]

is it common to have problems with tilemap colliders? i have a basic 2d platformer set up and the ennemies or players sometimes get hitched on what i can only assume is a tiny bit of collider sticking out

[timid dove]

It's common when you neglect to use a CompositeCollider2D

[fiery chasm]

thanks ill look into it

[dense basin]

I am porting a game currently to consoles and we failed lotcheck on Switch due to some random crashes. Now after some testing I noticed the memory is increasing every couple of seconds although nothing seems to happen in the scene. And I noticed that when the game was paused it stopped allocating memory. I spent the whole day to figure out what the issue might be looking at logs, profiler/memory profiler to no avail until I started to conquer and divide in the scene and started to delete gameobjects one by one to see if it makes a difference. and it came down to Physics2D.
Some gameobjects that just had physics components attached were the reason for it. The Physics2d section in the profiler confirmed it, the “Physics2D used memory” count was increasing every frame as long as the game was not paused. I looked at the gameobjects with the components and nothing really was fishy about them except one thing: some of those had multiple colliders, meaning 1 gameobject had like 4 boxcolliders or/and two hinge joints etc

Is this the reason? I myself wouldnt setup GOs like this butgot this project from a publisher and thinking that Unity should be able to handle multiple comps like this, no?

[timid dove]

Having multiple colliders on one object is not unusual or unsupported or anything

[dense basin]

I know, that’s why I am wondering what actually the reason is that Unity allocates memory every frame without releasing it at some point

[glossy prism]

anyone have good experience with active ragdolls?

[dense basin]

Calling GC.Collect manually also does not help.
I wonder if it’s some setting on any of the physics components leading to this, it feels like an engine bug

[glossy prism]

need help bros

i am trying to make active ragdolls playable and controllable from player input, i bought an asset pack with some prebuilt scripts because my scripts were too rigid and too confusing for me i hate quaternions and shit. Anyone who can help, it would be much appreciated.

my dms are open or ping if anyone has advice

[vagrant root]

my player just keep falling through the game i made a tile area and added tile collider and all and put rigdbody and capsule collider its a 2d game but my character keeps falling
can anyone guide how can i fix this

[peak portal]

Rate this game on a scale of 1 to 10

[steady elm]

Hi, can someone please help me with my tank controls/collision? :') I don’t know how to stop my tank from behaving like this when it hits a wall, and I’m not sure what to look into. Does anyone have an idea of what I should research to fix this problem?

[inner thistle]

That kind of thing usually happens when modifying the transform directly but you'd have to show the movement code

[random palm]

Do you use a Rigidbody on the tank?

[steady elm]

Here is it. I used MovePosition since I read somewhere that it was probably better than just translate

// Handles tank movement and rotation
protected void HandleMovement()
{
    float delta = Time.fixedDeltaTime;

    // Move the tank forward/backward based on input
    Vector3 movement = m_rigidbody.position + transform.forward * m_inputs.GetForwardInput() * m_tankMovementSpeed * delta;
    m_rigidbody.MovePosition(movement);

    // Rotate the tank based on input
    float rotationAmount = m_tankRotationSpeed * m_inputs.GetRotationInput() * delta;
    Quaternion rotation = Quaternion.Euler(0f, rotationAmount, 0f);
    m_rigidbody.MoveRotation(m_rigidbody.rotation * rotation);
}

[steady elm]

yep the second screen are the tank components

[random palm]

Have you tried using AddForce instead of MovePosition? I use this for Rigidbody2D, but use MoveRotation as you do.

I also had to adjust the 2D physics parameters to never get stuck in walls, but there are less parameters in 3D to play with... Just to give you some input of what maybe to look at.

[steady elm]

I'll try with add force then !

[steady elm]

AddForce() solved the problem ! but AddTorque() causes a new one :')

[granite quiver]

is it possible to pass an array of colliders into the Physics.Raycast function in order to ignore them?

can't use layers to do this btw

[timid dove]

no, only a layermask

[granite quiver]

darn, well I appreciate it anyhow

[timid dove]

You can always look at the documentation if you're unsure about the parameters of a method:
https://docs.unity3d.com/6000.0/Documentation/ScriptReference/Physics.Raycast.html

[lean wagon]

Hi ! I followed a tutorial on YT to create a blender character then export and create a ragdoll. But every time some of the spine parts looks like they stay at the original place they "spawned". Also the head "hitbox" is like 20 times too big.. On blender nothing looks glitchy in pose mode. Do someone has issued a similar thing ? thank you !

oh and I don't have a "head" or "def head" ? .. 🧐

[lean wagon]

Ok so i did not generate rig and it worked !

I may have missed something, but atm my problem is solved

[timid dove]

This is expected

If you check the docs it mentions it only works on convex mesh colliders

https://docs.unity3d.com/ScriptReference/Physics.ClosestPoint.html

[iron basalt]

Over the past week, I’ve been working on mapping Unity’s terrain heightfield onto a PhysX scene. I initially pulled my data directly from the TerrainData API:

var data = terrain.terrainData;
int tileRes = data.heightmapResolution;
var heights = data.GetHeights(0, 0, tileRes, tileRes);

Despite using GetHeights (and also trying the underlying heightmapTexture), I observed discrepancies of up to 0.5 units on sloped areas. Only after implementing a custom exporter—one that raycasts every point in world space and reconstructs the heightmap from those samples—did the PhysX collider match Unity’s terrain precisely (sub 0.01). Unfortunately, that raycast-based approach is quite slow.

Is there something I’m missing in how Unity processes its heightmaps for collision? Beyond the known bilinear interpolation applied by SampleHeight, does Unity perform any special treatment when generating the heightfield collider? The terrain itself was created using Unity’s built-in Terrain Tools rather than imported height data. I’m also curious how others have handled server-authoritative movement: did you rely on GetHeights/heightmapTexture, or did you employ an alternative method to generate a PhysX-compatible heightfield?

[timid dove]

I observed discrepancies of up to 0.5 units on sloped areas.

Well yeah GetHeights only gives you the heights at integer "pixel" coordinates. Everything in between those coordinates is interpolated

You can use https://docs.unity3d.com/6000.1/Documentation/ScriptReference/TerrainData.GetInterpolatedHeight.html or https://docs.unity3d.com/6000.1/Documentation/ScriptReference/TerrainData.GetInterpolatedHeights.html for the interpolated heights of course.

[iron basalt]

Unity can't feed physx interpolated heights though, it takes a heightmap which is going to be per pixel heights.
I'm not comparing height samples, I'm comparing collision positions.

If I take the heightmap unity is supposedly using, and import it into the same physx version, I would expect a raycast to collide at the same heights

Unless unity is like overwriting the raycast with its interpolated height when it detects a terrain collision?

[timid dove]

it absolutely uses interpolated height for raycasts

it uses interpolated height for all physics interactions

otherwise the terrain would be minecraft shaped

[iron basalt]

Well yeah because physx does that

This would be unity applying something after

We are not talking visuals here

Unity may apply something after I'm not sure, but physx itself interpolates the height

Or else physx would be minecraft shaped

GetHeight(
GetInterpolatedHeight(
Appear to use direct sampling from the heightmap

But physics interactions are not the same, they do not rely on this,

There are two seperate things going on with terrain, there is direct sampling, and then there is actual physics collisions handled by physx. Physx is not mapping a blocky terrain even though you feed it in direct points. The whole reason unity needs the interpolated height method is because its sampling from the heightmap

[timid dove]

I don't feel like I fully understand your issue

afaik Unity directly uses the PhysX terrain for everything

and doesn't do any additional processing or interpolation

I observed discrepancies of up to 0.5 units on sloped areas.
Can you elaborate more on this? How and where did you observe this?

[iron basalt]

Yeah and that's what I also thought, physx doesn't actually let you do anything besides change mesh cooking params and setting tesselation on certain points

[iron basalt]

If you dump the heightmap, in any of the various ways unity provides. Then create a physx scene, and create a height field at the same origin, size ect. Slopes will be off by 0.5+

And when I say slopes are off, I mean I'm picking a point, say 0, 0 and I'm doing a raycast from above. Unity might report collision at 427.3145, but on a physx scene with the same (or atleast I assume as the same) parameters, I might get 427.125

Doesn't seem to be a contact offset as its not consistant at all

[timid dove]

My guess is you're doing something wrong somewhere in the translation process

perhaps Unity and Physx use a different coordinate/ordering system for example

like Unity starts top left and Physx starts bottom left, something like that

[iron basalt]

Yeah that was one of my inital theories but if that were the case the whole terrain would be off not just slopes

and dumping my own heightmap at a 1:2 scale via raycasts actually got me sub 0.01 precision to unities reported raycast collisions

But it just doesn't make sense why I would need to do that

Also I'd like it to be within floating point errors 😂 0.01 is kind of high

[tight tusk]

Help.... i excluded the layers in the collider and rigibody and it is still affacting the player capsule collider.... making the movement slow and weird
but simply removing them causing not being able to get the tail physics i want
what should i do?
(ignore the tail go through floor bug, i kinda went to project settings and in physics settings i disabled tail layer collision to everything to tail layer, will fix after)

[wispy hemlock]

Not sure if this belongs here or in Animation tbh, but I'm using a IK Chain constraint to drive a bone chain, but I would like bones in the chain to interact with some world objects (like wall/objects that move etc), is it possible to add colliders to the bones and have them 'pushed' out of the way by said objects?

[iron basalt]

i know its possible as i saw someone on reddit do it to push the gun in their hand back when up against a wall to prevent clipping

no idea how they did it though

i tried finding the thread but wasn't able to

[iron basalt]

Where are my terrain and physx experts?

[bleak umbra]

Are you by chance making incorrect assumptions in how to map between a world coordinate to a terrain heightmap pixel to your other scene position?

[wintry pewter]

Got one question about constant force vs gravity
My player has a rigidbody, and I use a script that makes them float above ground using force, with the legs using IK to touch the ground
Leftmost pic is when gravity is enabled
Second pic is with gravity disabled, but with constant force of -9.81 applied, which should be equal to gravity amount
Am I not understanding something about how constant force works?

[quaint birch]

I'm uh... sucky at physics... but could it be mass-related? Like, maybe you need to multiply the force by the mass of the object for it to be the same?

[timid dove]

The mass matters as pulni said

It would only be equivalent to gravity if the object's mass was 1

[wintry pewter]

Ahh, that explains it
Thank you

[wispy hemlock]

Aah okay. Thanks anyway dude 🙂

[iron basalt]

Hmm not sure I'm exactly following what you are asking here.
I'm placing them at the same origin, with the same sizing.

I have a unit test of about 70 points dumbed via raycast from unity, all flat terrain match, slopes are off by a hair (and more if using unities provided textures/heights). I have tried shifting the terrain in the x, and z origin by increments (0.1, -0.1) to see if it was off by a set amount, this yielded results indicating my placement was correct, as then flat spots that did match unity, no longer do as the terrain has shifted.

it is a complex terrain, and with 70 points tested, any sort of positioning/coordinate errors would show in a different way than slight slop height differences as flat areas return with no issue. The height differences actually go up depending on the magnititude of the slope. A cliff will return differences that are way larger than an area with a slight incline. The slight incline might be 0.001 off, but the cliff (if using unities provided heights) might be off up to 0.6

[pulsar canopy]

I have a stupid question. Can you slap an ortographic camera in a 3d scene and use rigidbody2d, vector2, etc to get the peformance benefits of 2d?

[random palm]

Ofcource you can. It is perfectly possible to have movement 2D and visual 3D.

@pulsar canopy

[inner thistle]

There's only one type of scene. The only difference between 2d and 3d is the camera and that you typically use sprites in 2d and models in 3d

[pulsar canopy]

I mean, will you get performance benefits out of the ortographic camera? I'm sure calculations in 2d and rigidbody 2 are faster, though

[inner thistle]

You can try toggling the camera between ortographic and perspective and seeing if it has any effect on the fps. I doubt it does

[random palm]

I don't think so, what you chose to show is what is shown.

[pulsar canopy]

What I'm getting at is, a true 2d is always more performant, other things being equal?

[timid dove]

the fact that the camera is orthographic really isn't going to change the performance of rendering.

it will still be rendering the same objects with the same shaders etc

orthographic is not "true 2d"

it's just a camera projection that doesn't have perspective effects

you're still rendering a 3D object

[bleak umbra]

for example, pixels on a heightmap define vertex positions. lets assume a 100m terrain and 1025px heightmap -> 1025 vertex positions but 1024 intervals. Pixel 0 of a 100m terrain is at 0 and pixel 1025 at position 100, if you however, for some reason, forget that last pixel, or map 1024 to 100, in your other scene, you would get slight offsets on sloped terrains, but not on flat ones for the same world coordinate.

[pulsar canopy]

Right. So true 2d is always more performant, other things being equal?

[timid dove]

What does "true 2d" mean?

[pulsar canopy]

I mean just using sprites

[timid dove]

it's really comparing apples and oranges

It's like... "what's more performant, an airplane or a car?"

[pulsar canopy]

Why

[timid dove]

or maybe "a car or a boat?"

Because it's lacking context

you can render a plain old Quad mesh with a MeshRenderer vs a sprite on a SpriteRenderer in a scene alone

pretty sure the Quad would be faster

And that's "3D"

They are different tools for doing different things.

[pulsar canopy]

Yes, but I'm talking about a real 3d scene with 3d models vs a 2d scene...it's always more perforrmant, regardless of the canera perspective if the objects are 2d rather than 3d?

[inner thistle]

That's still kind of a nonsensical question because it depends on the sprites and models, and if you're planning to switch from 3D models to sprites only because it's "more performant" then that's just silly

but yes in general rendering one sprite is probably usually more performant than rendering one 3d model

[pulsar canopy]

It's a curiosity thing, more than anything

I'm considering doing a 2d game and I don't know how to draw and am much more familiar with everything in 3d

Therefore the answer is probably irrelevant as I'd choose to do it 3d cause it's the better choice for me

But I like knowing things

[timid dove]

no

[iron basalt]

Ah yeah I see what you are saying.
I’ve double checked my spacing logic and it already accounts for the (resolution - 1) intervals

    w.sampleDX = meta.Size.X / float64(resX-1)
    w.sampleDZ = meta.Size.Z / float64(resZ-1)

[pulsar canopy]

And the caveat there I suppose is very high res sprites vs low poly models with colors instead of textures

Yea I get that part

[frigid echo]

i thought that i have disabled gravity, but for some reason there is a resistance when i move my character down?
what should i do to fix this?

code, based on a tutorial video

[maiden snow]

show inspector

and move the code inside FixedUpdate into Update(at very end)

[frigid echo]

ok

[frigid echo]

but why?

[maiden snow]

because it does not belong into Physics, you are not using any physics there

check the unity docs Charachtercontroller.Move you will see that too

oh wait you also have arigidbody there

[frigid echo]

oh ok

[maiden snow]

this wont work well together., you either use the cc or rb but not both

[frigid echo]

cc is character controller?

[maiden snow]

ah never mind i see it now..

so leave the code inside FixedUpdate, the freaking script name confused me totally.

[frigid echo]

so revert this change?

[maiden snow]

yeah please. what you had prevoiusly was totally fine i thought we are using a cc but totally forgot that 2d doesnt have one and the script name confused me.

[frigid echo]

wait

i might have done a mistake

hold on

yep

i forgot that crouch is down

so when i press down, not only it goes down, but it also slows the player down

im sorry

[maiden snow]

let me check brackeys video.

keep in mind the tutorial is from 2019

[frigid echo]

okay

btw now it works as i wanted it to be now

[maiden snow]

nice

if you have any other questions feel free to ask them here

[frigid echo]

will do, thank you

[stuck bay]

Why is my model falling so slowly?

[maiden snow]

do you have any code?

[stuck bay]

Nope, this is only physics

The gravity settings are all default at -9.81

[maiden snow]

try a rigidbody reset

[stuck bay]

it goes upward now and at the same speed

like its ascending instead of descending

Oh I had an animator object and it was affecting it mb

[maiden snow]

glad you fixed it

[frigid echo]

i have an issue:
sometimes the overlap detection does not fire when player and enemy collides
im not sure whether i am doing this wrong, or that there is something that i miss

[inner thistle]

OverlapBoxAll will only detect things inside the overlapbox but the enemy isn't going to be inside it because the overlapbox is exactly the same size as the player's collider and the enemy bounces off the player

If you want to detect the collision then just use OnCollisionEnter2D

[frigid echo]

ohh ok, ill try it

works great, i'll look more into the params of it
thank you

[frigid echo]

collision/collider is for when the object bounce / touch, while overlap is when objects pass through each other
am i correct?

[silver moss]

Maybe the object is just really large so it appears to fall slowly relative to the default gravity setting

[stuck bay]

Ive resolved the issue but I appreciate your help

[stark horizon]

hey yall, what is the best way to implement distance based knockback where no objects can cause the knock object to slow down and instead give way to the knockobject ? Im using rb.moveposition() to get my knock back my game object, however when it collides with another object, it will slow down, stop and not reach the distance its supposed to be knocked to. Turning on Kinematic body types will cause phasing issues so its a no go.

in this clip, the blue ray represents where the enemy actually ends up and the yellow ray represents where its supposed to end up. In the beginning when the enemy doesnt have another enemy standing behind it the blue & yellow ray aligns however when it starts to collide with another guy then he falls short of the knock distance

[paper dirge]

How can i set the zero of articulated revolute joint?
If the joint is in x axis and if i rotate the joint in that axis, the zero position doesnt change.

[stray belfry]

i use a tilemap collider on my grid map, i get stuck in nothing, same for enemies, cuz its "uneven" for some stupid reason.

i use a composite collider, some raycast fail because they are cast inside the ground, bu dont detect it cuz the composite collider is not filled. additionally, it doesnt work well with A* pathfinding graphs, need much more tweaking and such.

what am i supposed to do?

[timid dove]

Use a CompositeCollider2D and use polygon mode

[honest folio]

Is there any good videos online anyone would recommend for physics? (Vector3, Quaternions)

[timid dove]

Vector3 and Quaternions are not directly physics concepts

more like, just spatial reasoning

[honest folio]

Oh yeah I guess so

[rotund pelican]

watch the end of the video to understand. Sudden jolts of force send fixed joints into panic which causes contraptions to "blow up". The playerbase HATES this instability and i wanna fix it. How?

Would a config joint with all params locked help?

[rotund pelican]

tough crowd

[vapid lava]

Moved to correct channel

I didn't mean to ask this here

[rotund pelican]

implemented a new "soft weld" system that just makes a config joint with really high drives

works pretty well

are there any better alternatives? i want to keep using fixed joints but i dont want these explosions to keep happening when people are trying to make stuff

[slate orchid]

I have a problem with the car. When I open the game, it flies. I also added a Collider Wheel and things are fine, but after 3 or 4 seconds, the car flies.

[rotund pelican]

what

what car are you talking about

[slate orchid]

I fixed thz

Thx

[tawdry wave]

why is there no plane collider? it's like the box collider but more efficient because there's only one face

[unique cave]

For one, PhysX which unity uses by default doesn't really have support for them. The only plane collider PhysX knows is a plane that extends infinitely (like mathematical planes do) and can only be used for static objects. It's an interesting question why unity haven't decided to make support for that but I suppose that wouldn't be super useful for most anyways.

Box colliders are super efficient, you don't usually have to worry about that. The performance of a collider isn't really defined by it's faces (but rather many complicated factors), sphere colliders are often the most efficient collider shape even though they have 0 or infinite faces depending how you define it (they are mathematically super simple to model by using their center and radius and hence the performance). The AABB of a proper plane and a thin box would be very similar so there wouldn't be huge difference from there either.

Why there are no plane colliders that are bounded to an rectangle and can be dynamic then (e.g. useful for playing cards)? Apart from the previously stated lack of support by PhysX, it would be very unoptimal shape to simulate in a discrete simulation potentially causing a ton of tunneling issues especially when they interact with one another. You can see these issues in action if you create very thin box colliders and stack them on top of one another. The little I know about physics simulations, I don't think these issues are easily fixable for tailored plane colliders either.

Is there some specific concern that you have with the box collider and what would a plane collider fix in your case? You can always leave unity a feature request if you feel like it, but given that one hasn't been implemented to this date, it would be unlikely to happen in the near future unless they either update or change physics engines altogether (don't know how this is handled in other engines or never PhysX versions)

[native grotto]

I have a car character locked on the Y axis for positions, and Z and X axes for rotation. It works great on its own.

But when I add a snow plow to it, it’s like it resists turning all of a sudden. The player and plow have no collision between them enabled. Can’t do trigger on the plow as it must push snow around.

The rigidbody is on the character prefab. Both the character and plow have their own collider on separate child objects. The plow is part of the character prefab.

I apply force by doing addforce. I already tried setting the center of mass in code, but that didn’t work.

Any ideas? For some reaso, when I changed some parameters, now the character rotates on the X and Z axes even though it is frozen. Still with the slow turning that is weird as if the plow interferes…

[nova ore]

I need help, I can’t grab any objects that were supposed to be grabbable, what is going on?

[austere salmon]

Hi, I have a rigidbody and capsule collider with free y axis rotation. On slopes there is no issue going up or down but when moving along the slope at a level height the character controller slides without friction and the experiences a frequent jitter in y axis rotation. Anyone know why this may be?

[timid dove]

You'd have to show your code

And the Rigidbody inspector

[timid dove]

Where is "targeteuler" coming from?

!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.

[austere salmon]

public void OnLook(InputAction.CallbackContext ctx) {
    Vector2 delta = lookSensitivity * ctx.ReadValue<Vector2>();
    targetEuler += new Vector3(-delta.y, delta.x, 0f);
    targetEuler.x = Mathf.Clamp(targetEuler.x, xEulerMin, xEulerMax);
}

[tawdry wave]

i've heard that mesh colliders are terrible so i was hoping there would be some alternative for level geometry that is static, and rectangle-shaped surfaces could use a plane collider instead i suppose. everybody keeps saying to put box colliders on everything but that is just insane, what is one supposed to do then?

[unique cave]

Couldn't you use flat-ish box collider for a rectangular mesh? Who told you mesh colliders are terrible? They are generally bit worse sure, but often times the collision shape doesn't even matter due to how the broad phase/AABB check eliminates most of the potential collision pairs. Generally you should use the primitive colliders (box, sphere, capsule) whenever they give reasonable approximation of the shape. Using mesh colliders is totally fine in most cases though. For a super detailed mesh, I would consider a lower LOD level for the mesh collider, you often don't need nearly all the triangles to get good enough physics precision

If you could specify what kind of objects/meshes you a trying to get colliders in (what shapes? how many? dynamic or static? used for what? etc.), it would make it much easier to give any specific suggestions.

[tawdry wave]

well i saw some reddit posts like https://www.reddit.com/r/Unity3D/comments/11cd0tf/mesh_collider_uses_up_too_much_performance/ and https://www.reddit.com/r/Unity3D/comments/ctw21o/a_question_regarding_when_to_use_mesh_colliders/ and it seems most people are saying that you should manually place box colliders for everything but it would take ages wouldnt it?? for example an enviroment like this image: (its not mine but its an example of stuff i want)
i cant imagine how people expect me to make box colliders for this scene

but also even a house like this would take ages to manually place box colliders

[unique cave]

I wouldn't read too much into those posts, seems to be mostly people that "have heard" from somewhere. Both the original posters didn't find any issues with mesh colliders themselves, just that they were worried. This kind of scene I would make out of combination of collider types. If the terrain system was used, a terrain collider for the terrain would be quite obvious choice. The boulders I would either make out of primitives or use convex mesh collider (potentially for lower LOD mesh). With the poles and the trees even simple capsules could be sufficient. For the railing for example I would use a simple mesh collider (again with simplified mesh if needed)

[unique cave]

More like couple minutes if you only need collisions with character controllers. If you needed the collider to be exactly the shape of the mesh for something like precise bullet decal positioning, I would just use mesh collider, that seems to be very low poly mesh anyways so I wouldn't be too worried.

[tawdry wave]

but if i want bullets to collide properly and with precise decal positions then isnt the only way to make everything into mesh colliders in this entire scene? including the trees and poles and boulders. and would that still be fine for performance?

[south granite]

In fully developed scenes they can have a pretty notable impact on scene start times though

not something to worry about too much in advance however

[unique cave]

I was just typing a message about that. The performance isn't generally the issue for low poly mesh colliders, but they can introduce other issues regarding loading times and memory consumption from what I have heard

[south granite]

though that can be mitigated by spreading them out across frames i think

its their initialization off of Awake() that can creep up at scale

[unique cave]

Other than placing a ton of primitives, yes. I would start from the easier solution of mesh colliders, profiling in your own scene is the only way to find out what the cost is

The reason for those loading time problems I believe is that PhysX builds some clever spatial partitioning structures for each static mesh collider which makes them very fast to look up (good runtime performance). I don't know if it is possible to force those preparations to happen beforehand (during building for example)

[tawdry wave]

hmm that is interesting, i wonder why it doesnt do that already. i also know that unity also has havok physics as an option, i wonder if that would be a lot better with mesh colliders?

[native grotto]

Anyone?

[timid dove]

Adding colliders changes the moment of inertia of the object

because it changes the shape of the object

[native grotto]

Is there an easy way to account for this? I really wanna make my snow plow...

[timid dove]

you can manually set the inertia tensor to whatever you want

You can also disable the automatic intertia tensor checkbox

or through script https://docs.unity3d.com/6000.1/Documentation/ScriptReference/Rigidbody-automaticInertiaTensor.html

[native grotto]

Amazing, that was it!! Thanks

[orchid sky]

Dynamic Rope

[late smelt]

jiggly asf lol nice work

[solemn plover]

Hi everyone! I have a very basic humanoid rig that I'm working with that uses configurable joints and some custom torque code that rotates the joints to the desired angles using a simplified idea of muscles.

The issue is that when the joints have reached their angular limits, any additional torque onto the limb will cause the rig to have "god forces". An example is that the rig could spin on the ground when at maximum angular limit and continues to extend the legs. The desired behavior is similar to what we have in real life - if you extend your leg while it's at its limit, it won't extend any more even though you're trying to. Again what happens in my simulation is that it produces unrealistic forces that will spin the rig around.

So far I have tried limiting the torque that can be added when the joint is at its angular limit, which sort of works but something is telling me this is a workaround that will give me more headaches in the future.

[young gate]

Does anyone know of anyone or game that has implemented rotating Articulationbody wheels?
Actual rotating collider, I've reached out to a few devs that may have done it but they had all given up on actual rotation and just had visual rotation

I've done all sorts of experiments to see if it was possible but rolling friction just seems to get in the way

[wise helm]

anyone know how to fix a rigidbody bouncing when the player rigidbody moves into it

[wise helm]

figured it out, was a simple code toggle

[worthy heath]

Does anyone know how to transfer rotation between two bodies with configurable joint in unity? I am making a simple ball joint that will also transfer rotation

[tardy rune]

I have a physics based game in zero-gravity and currently when I shoot a bullet at a floating box (both have rigidbodies) the box moves a bit from the impact, this is normal behaviour ofc. But why doesn't the box start rotating when shot away from the centre of mass? Like when I shoot near the top of the box it only moves, but doesn't start rotating around it's centre of mass.

[analog scroll]

very begginer stuff(just came back to unity and I don't remember a lot):
Why is the object floating a little bit above the collider(colliders fall until they touch)
the floor is made out of tiles with colliders btw

it's very inconsistent too

the floor collider just randomly got bigger when I wasn't looking, but it still floats above that collider

[timid dove]

It's the Default Contact Offset in your physics 2D settings

[analog scroll]

thanks, it's fixed

why does it exist though?

[foggy falcon]

I'm trying to use animations from mixamo, but the two character I applied the animations to are not the animations I saw on the site.

[timid dove]

Sounds like a topic for #🏃┃animation.

[tardy rune]

Does anyone know??

[inner thistle]

You'd have to show the inspectors of both objects. But if the box moves only a bit instead of gaining permament velocity there's some dampening mechanism that might dampen the rotation as well

[tardy rune]

Yes it's only a small force, is there a way to turn off that dampening?

[dull jungle]

the rigidbody has linear and angular damping properties

[inner thistle]

There's no way to know what it is without knowing your setup, you'll have to show the inspector and any relevant code

[tardy rune]

Yea sorry I thought I already tried turning off the angular damping, somehow that didn't work last time. But set it to 0 and now it works. Weird. Thanks for the help tho!

Bruh what now it doesn't work again?!

I'll make some screenshots 1sec

[dull jungle]

did you change it in playmode? those changes dont save

[tardy rune]

I'm in playmode rn with 0 and it doesn't rotate, I'll send some pics

Yeah now it doesn't rotate anymore, so weird

Every time I try to record it on video it rotates wtf...

Ok got it on video now

[timid dove]

Looks like the object you're shooting at has a mass 1000 times that of the bullet. Unless the bullet is moving extremely fast, it's going to barely affect the thing

[tardy rune]

1 sec ill send the video

This time it did work

This time not working

It's so weird, one time it works, other time it doesn't

[tardy rune]

I sent some videos, maybe you know what it happening?

[timid dove]

that's a bit weird, assuming nothing changed between

[tardy rune]

Yeah I changed nothing

Bruh I caught it working and not working in a single take... wtf...

[timid dove]

I mean - you manually edited the damping and the position of the object between.

That surely messes with some things

it probably doesn't happen if you don't do that?

[tardy rune]

Ill try again later. I only set it to 1 and back to 0 to stop it from rotating.

[tardy rune]

Ok this is really weird. @timid dove @inner thistle @dull jungle
Just to be sure I made a test build. I put the boxes in a row left to each other and set the Angular Damping different for each of them. From left to right it goes from 0 to 0.6.
I shot all of the boxes and reloaded the scene a couple times, each time the results were different. Also some strange results, sometimes the box with 0 damping didn't rotate but a box with more damping did.
The last video shows it the best with very inconsistent results.

[dull jungle]

how fast is your bullet, what collision detection method are you using?

[tardy rune]

See this image for settings. Boxes are Discrete, bullet continuous dynamic.

(bulletForce = 1)

[dull jungle]

I would assume that the only rotational force you get is if the box has friction with the ground it is placed on based on what we see here, can you create a physics material with 0 friction? (and minimum as calc method)

[tardy rune]

The boxes are floating midair in zero gravity, they are not touching anything until they reach the wall.

[dull jungle]

oh looked like they touch the ground

how detailed is the mesh of the boxes?

[tardy rune]

This last video shows it the best with very inconsistent results.

[tardy rune]

Box collider

Bullet has sphere collider

[dull jungle]

are you deleting the bullet immediately on impact?

[tardy rune]

It gets disabled immediately yes (pooled)

[dull jungle]

can you try leaving it active, just to fool around?

[tardy rune]

Ok 1 sec

[tardy rune]

It's still weird

[dull jungle]

What is the mass of your bullet?

[tardy rune]

0.01

[dull jungle]

try raising it

[tardy rune]

ok

Yeah it seems more consistent now

I'll try playing with the values a bit

Ok I think the problem is the speed of the bullet

I increased the mass and it looked normal. But because of the mass increase the bullet went slower, so I increased the force to bring it to the old speed again and now it doesn't work again, even with the higher mass.

[foggy falcon]

Thank you I didn't see that tab I was looking for it.

[timid dove]

do you have some kind of logic on the bullets when they collide?

or in general

[tardy rune]

Yes, but I put return on a tag so it wouldn't destroy them on the boxes. But I think it's the speed

So I guess no spinny boxes in my game for now :(
I think this is just a weird unity thingy. But weird how it does give linear velocity, you would expect it to get no velocity at all right?

[dull jungle]

you can also manually get the impact point and apply a force by code

make the bullet a trigger instead and handle the collision force yourself

[spiral nexus]

Not sure if it's correct chanel to ask but how can i make unity not cutting edges on textures with missing/fully transparent pixels? It's causing random player jumps while moving on flat platform. I've tried to use same tilemap with default square sprite and everything was fine so i'm sure the problem here

[timid dove]

You can set a custom physics shape for any sprite in the sprite editor

[spiral nexus]

Found it, thank you

[wheat pike]

hey guys, ive been having this issue, im trying to make a ragdoll after a hitscan attack on unity, the animations where fine but the character cant get up, i tried using the animations to get him up, i tried to change from rigidbody and collider to character controller nothing, the ragdoll itself works fine but it cant get up

also i made a code to disable the collider for a second after it stays on, which is my theory of why it doesnt get up, but it still didnt disable it, am i missing something?

btw i need that collider to trigger the ragdoll, but its also not allowing me to get my character up

[proud condor]

I am making characters on ship

my character and ship are both rigidbody component

characters should be able to move freely on ship

how exactly i can resolve movement on top of ship?

[timid dove]

You keep saying "get the character up", what's exactly are you doing to try to do that?

[covert crypt]

Anyone see the AVBD physics simulation paper? It looks like a more stable physics system that could be used for general physics sims or smaller particle sims like cloth physics

[whole tusk]

oh yeah i saw that, it looks interesting. from the example i saw, it still has issues with clipping at high speeds. but it's much better at handling ropes and chains.

[small cliff]

why is the pillar not falling down? gravity isnt affecting it, any ideas?

[timid dove]

is your game paused or Time Scale set to 0?
Is your physics simulation mode set to auto/FixedUpdate?
If your physics engine set to use PhysX?
Is there an invisible collider underneath it?

[small cliff]

this was the cause

[timid dove]

There ya go - simulation mode

[pseudo mantle]

Anyone done moving platforms for an FPS game? Bonus points if it uses or works with the Unity Starter Assets FPS asset

[timid dove]

Kinematic character controller is the way to go

(free on asset store)

[pseudo mantle]

What if I want to use the Starter Asset one that is not kinematic?

I appreciate that that might be easier but Id like to solve it for both/all use cases too

[wheat pike]

sorry for the late answer, there is an animation to get the character up, and there is a script to try to make him upright however it doesnt work, you see when he gets hit he goes into ragdoll mode, then fallen animation and then get up animation, but he stays sideways

[timid dove]

Starter asset uses a character controller which behaves kinematic as well

[young gate]

Is there any generally accepted way to calculate forces required for a desired velocity on an articulationbody that's part of a hierarchy? I can go off combined mass but that does not factor in anything related to joints

[uncut flower]

Why Rigidbody.angularVelocity won't exceeds 5.041452?

private void FixedUpdate()
{
    Rigidbody rb = cube.GetComponent<Rigidbody>();
    rb.angularVelocity += new Vector3(1f, 1f, 1f);
    Debug.Log(rb.angularVelocity.x + "," + rb.angularVelocity.y + "," + rb.angularVelocity.z);
}

[timid dove]

angular velocity is limited by physics settings

the default is something like 7 rotations per second

[uncut flower]

in project settings?

[timid dove]

yes

[uncut flower]

alright thanks

[simple token]

what happens when a bouncy object hits another bouncy object?

double the bounciness?

[wooden heart]

Does anyone know why this is happening idk how to fix it

[timid dove]

your 3D model is oriented incorrectly

the direction your object is moving is the forward direction of your object

If you switch to the Move tool and switch the tool handle rotation to "Local" (not Global) you will see:
Blue == forward
Green == up
Red == right

To fix this you need to reorient that "root" child object so that it is facing the forward direction of the parent object.

[wooden heart]

I fixed it, but new problem now. it flys but not well and its super wonky when it comes to its physics, if i pitch up it turns for a bit but then starts spinning

[timid dove]

convincing realistic airplane physics are not going to be simple

[wooden heart]

yeah prob the worst problem rn is that i got my jet flying but when i turn and my plane is facing the new direction it is still moving in the old direction so like sideways basically instead of moving forward... and idk how to fix it cause ive tried quite a bit of stuff

[pulsar summit]

Real quick question. Is there a method where you can make all 3D objects parented under a empty, follow the same rigidbody as if it was 1 object

[unique cave]

That sideways movement is called sideslip which is totally normal in all aircrafts when rotated via the rudder (control surface at the tail that directly changes the yaw). The sideslip angle (angle between the plane axis and the relative wind) then causes drag on the tail and the fuselage that causes lateral acceleration that then minimizes the sideslip. You can search for yourself how the lateral force caused by sideslip could be simulated.

Worth noting is that planes rarely turn (not sharply at least) purely by using the rudder. Much easier way to turn is usually to roll the plane and pull up to turn. It depends a lot on the type of game and level of realism you are going for but many arcade type flying games don't simulate rudder at all because the roll and the pitch are enough control to fly the plane anywhere. Real planes don't usually use rudder to turn, it is mostly useful for correcting for yaw in case of engine failure or turn for example.

[unique cave]

Wouldn't you just add rigidbody to that empty object?

[forest thunder]

hey is there a way to make a reliable collision detection using Physics.ComputePenatration ?

[light sinew]

technically not a physics question, so im sorry if this is the wrong place for this

been using KCC for my character controller, which is great, but I've been having this problem where my character 'sticks' to things when they jump over/next to them, which sort of 'cancels' the jump and doesn't feel great (player expectation if you press jump is to get the full height, not to randomly stop halfway through the jump).

I tested with the KCC example and it doesn't have this issue. If I make the capsule collider on my character much wider it does fix the issue, but making the example collider smaller doesn't cause it in the KCC example scene, so I don't think it's that. (my thinking was that the character's long and spindly collider might be causing this but that doesn't seem to be the case.)

I don't really know what I could have done that changes this, since I haven't changed the kcc motor code at all, and my character's motor settings are largely unchanged from the default...

this post is kind of a hail mary since I'm not really sure what to do, any help would be appreciated

[maiden snow]

sorry to ask, but do you really need the kinematics

[covert jackal]

Hey all! I need to somehow make a realistic aerodynamic physics system working… I would buy an asset but most of them are tailored to airplanes and I’m working with rockets and I’m worried that they won’t work for what I need specifically

Edit: holy moly what a run on sentence. Hopefully it’s still readable

[light sinew]

yes? haha

I'm not about to remake my entire character controller to fix 1 issue 😭

[stoic cloud]

I'm having my own grid in Unity, yet I'd like to combine it with 2D physics. Is my best bet using a composite collider? Some physics libraries seem to have custom colliders where you could implement collisions between other shapes and the grid, but I can't make one in Unity right?

Well, or polygon colliders I guess. Not sure which one is faster.

The thing is checking against the grid would be waaay cheaper.

[timid dove]

What do you mean by combining a grid with 2d physics exactly?

[stoic cloud]

Using normal 2D colliders and rigidbodies and the grid is basically the terrain that you collide with. It's also edited at runtime cause it's a sandbox game. So I have chunk and those chunks need collisions for each solid tile. Kinda like a tilemap. The thing is having a composite collider with 1000s of box colliders, one for each tile, sounds super expensive.

[timid dove]

So why not use a tilemap?

[stoic cloud]

Cause I use texture arrays for the type of cell. I also want greedy meshing to improve performance. I think a tilemap is also just a composite collider per chunk using each tiles collider.

Also a tilemap renders a cell no matter if there's something in it or not. I have layers that only add a quad to the mesh if there's e.g. a plant in it or so. Overall I get drastically better performance with my own mesher.

[timid dove]

Cause I use texture arrays for the type of cell
So does the tilemap? Or something similar at least

I also want greedy meshing to improve performance
Tilemap performance is very good, I'm sure they're doing something like this under the hood.

I think a tilemap is also just a composite collider per chunk using each tiles collider.
Not sure what you mean by this

kind of sounds like you're reimplementing tilemap

[stoic cloud]

Please don't derail my actual question or my question gets burried. 😦

[timid dove]

Anyway you can always make your own PolygonCollider2D

I don't feel I'm derailing anything here

[timid dove]

depending on how your game actually works you can certainly do your own grid checks instead of using the physics engine

[stoic cloud]

It's ok if you feel like that. You are free to do what you want. It's just not what I asked.

[timid dove]

Alright I'll bow out then 👋

[safe flare]

What am I supposed to see when I enable Show Contacts and Show All Contacts in the physics debugger? nothing seems to change.

[timid dove]

Make sure gizmos are enabled too

[brazen dock]

I assume I might be the millionth person asking this, but is there a way to prevent dynamic rigidbodies from 'pushing' each other, as if one had high mass?

[timid dove]

There's an asset that lets you do that https://assetstore.unity.com/packages/tools/physics/betterphysics-selective-kinematics-244370?srsltid=AfmBOorRmWsuF_72PeThzuSkXD_n12cLck3vKmLy1nfIO4cjeXGuOxHQ

[lime void]

ok so i've got this basic ship set up where it'll rotate based on the mouse position, but whenever i do so the boxcollider2d attached to it has this really weird behaviour

the actual script and collider are attached to a child object of the sprite

[timid dove]

Show how the hierarchy is set up and which objects have which components

And how everything is positioned/scaled etc

[lime void]

the SmallShip object has a rigidbody

and the brain object controls its movement

[timid dove]

Which object has the collider? Which object is rotating? Which object has the Renderer?

Showing the inspectors of all these objects would be nice

[lime void]

the third one is the collider before the ship starts moving

which works fine

[timid dove]

These second two screenshots are useless

Again I want to know which objects have which components

Screenshots of the full inspectors for the several objects would be helpful

But the first screenshot at least has a pretty telling issue

It's scaled non uniformly

Ideally the root object is not scaled at all, or is only scaled uniformly on all axes.

The renderers and colliders should be on leaf objects (objects with no children's), and then they can be freely scaled.

[steel sky]

Hello - I am making a first person character controller with a dash. My issue is that if you dash towards the ground, you get launched into the air, which I don't want to happen. Instead, dashing towards the floor should just stop your movement once you hit the floor.

I have tried adding a physics material with bounciness = 0, and that did not work. Anyone got any other solutions I could use? I was thinking perhaps I could lock the players motion to the plane of the dash, but rigidbody constraints only work on the coordinate axis.

[brazen dock]

Thanks, but is it also applicable to 2D too? I might be able to just modify its source code to make it work 2D, but just in case if there's 2D version for it, or an easy way to switch it to 2D mode

[craggy holly]

if someone can explain this please
i have projectiles that will be shot from enemies
player also shoots projectils
both projectile colliders are set as trigger, because some of them, i want to go through enemies
others will destruct on first trigger

my player, has 2 colliders, 1 as trigger, and one normal collider with rigidBody to not allow player to move through obstacles

i have also pickup items, that have only 1 trigger colliders
when player goes to pick them up, it detects trigger twice, because player has 2 colliders even though only 1 set as trigger

this is starting to confuse me a bit and idk what's a good approach here

or if what im doing is correct

[sharp socket]

Is it possible for the ocean in the hdrp water surface system to make waves colliders? so when player jumps into them it triggers he jumped into water and starts swimming .? or any other idea how to do trigger that he jumped under that water surface (im new to unity so if i speak nonsense i apologize)

i tried many things like box colliders but the game has waves that are above the box collider and it looks weird

(after 6h of thinking i got it)

[timid dove]

Not for 2d unfortunately

[brazen dock]

Aww 😢
Still thanks for answering

[stuck bay]

are soft body proxies plausable?
like instead of simulating my characters mesh, can I sim a low poly version of specific body parts and bind the verticies of my character to follow it?
asking for a friend totally not about to do anything weird

[timid dove]

Sounds like you're just describing a ragdoll, which of course are possible and common

[smoky moss]

Hey all, I'm trying to make a orbital trajectory predictor (think a very simple KSP), and I figured the best way to do it was a simplified physics scene in the background, plotting the positions, and reading it back to the main scene. It's generally accurate, but quickly diverges with some really odd behavior on certain orbits, especially elliptical ones. Any advice on how to debug this, or has anyone seen something like this?

I'm running PhysicsScene2D.Simulate(step) with step being the same as the physics engine timestep (0.02 default). I'm manually calling my gravity functions for each object before moving on to the next timestep. I've also tried predicting it with a RK4 integrator that doesn't use the physics engine, and that gets similar results that are generally accurate but then freak out on certain frames.

[timid dove]

showing code would be a good start

[smoky moss]

Ah, thought I attached it, my bad.

[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.

[smoky moss]

https://paste.mod.gg/ywvqtlpdjhzn/0

[whole dagger]

Hello everyone!

i have these two RBs connected by the configurable constraint

they are siblings, not in a parent<>child relationship

I'm trying to manipulate the driver on the constraint so that the cube will move to the center point

this works but i also want it to apply enough force that it can lift the heavy car to do so!

the car mass is 1500 and the cube is 150

right now it doesn't seem able to do this, even at very very high spring rates

so it'll wiggle, but it's unable to exert force on the other body through this spring i suppose

how could I achieve this?

[rotund pelican]

i downgraded from unity 6 to 2022.3 and uh

how does this even happen

ill show the physics settings in a bit

[unique cave]

Show also the rigidbody settings maybe. I don't remember what it was exactly but I remember somebody had a problem earlier where the serialized per rigidbody values didn't translate between versions (especially starting from unity 6)

[rotund pelican]

wdym?

[unique cave]

Inspector of the object in question

[rotund pelican]

this happens with most objects, though i might have fixed it idk ill check later

[lapis geyser]

Can anybody tell me a way to avoid a moving object to penetrate a wall?

[cerulean lava]

I THINK this comes under #⚛️┃physics but if I'm wrong, LMK.
I've got a basic auto runner script which lets you jump right now. It handles custom gravity as well. Right now when you jump, there is a very small window where you are still colliding with the ground so you can input another jump and it functionally double jumps. This is not intentional. My fix for my other projects is to check if the velocity is upwards before adding the grounded state to the character. However, this doesn't work when the player is moving up slopes because there is an inherent upward velocity of the normal force. I tried to isolate it using Vector3.Project but it didn't work and left me with non zero values when I wasn't inputting a jump. I'm not sure what's wrong, can any of you wizards help me out? I've got a PhysicsMaterial with 0 Friction on the player

This is what I tried:
Vector2 SlopeTangent = new Vector2(GroundNormal.y, -GroundNormal.x).normalized;
Vector2 SlopeVelocity = (Vector2)Vector3.Project(RB.linearVelocity, SlopeTangent);

Vector2 NonSlopeVelocity = RB.linearVelocity - SlopeVelocity;

[timid dove]

Make sure you're only moving your Rigidbody via the Rigidbody

[timid dove]

just tune your grounded check to be tighter

how are you checking if you're grounded currently?

[cerulean lava]

I can attach the script it is fairly small and self explanatory

[cerulean lava]

I'm using a Collider Cast

[cerulean lava]

I've tuned the Grounding Check so it is literally as tight as can be with the method I'm using at least

[simple locust]

In my scene, I have a game object that consists of an outer circle, an inner circle, and a dot (neutron) at the center. I originally wanted 2 main things to happen: For the dot to get movement input to move around the inner circle, as well as follow the main game object of this whole construction (the outer circle is the parent). I made the inner circle a child of the outer circle since it just needs to follow it, but the neutron is where my problem started. At first it was dynamic and I've tried combining addForce and linearVelocity to try and make something work, but the physics ended up fighting each other and it looked jittery and unstable. I think my best approach is to simply make the neutron kinematic since it doesn't really need any sort of dynamic physics (apart from collision which can happen with kinematic objects anyway, and it only collides with the also kinematic inner circle), but when I make the neutron a child of the outer circle - It changes the position in a circular manner because of the rotation of the outerCircle. I've tried handling it in LateUpdate to maybe get it to stop, but it just ended up causing more problems. I might be looking at it from a wrong angle but in general - I can't seem to find a way for the neutron to both have stable movement input while also constantly following the position of the overall game object (I can't use transform.position, since it didn't go well with physics).

Does anyone have any ideas or directions I can go in?

[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.

[cerulean lava]

https://paste.myst.rs/bbyzmt8x

[solid blade]

https://streamable.com/eszn1z

fellas, if any of yall understand this, i would be eternally grateful. as shown in video, my bullet is not bouncing how i'd expect it to on these colliders--it should just bounce off once and follow the tracer--or so I thought. Does anyone know why this is occuring?

it feels like the force almostt isn't being redirected properly

[upbeat jackal]

Um... How to fix this?

[timid dove]

Either:

• fix the hierarchy and position the wheels properly
• fix whatever code you're using to position/pose the wheel objects when the car is in action

[upbeat jackal]

Can you tell me what I should do, please?

[timid dove]

I just did?

[solid blade]

https://streamable.com/eszn1z

bump on this if anyone has any ideas

have a frictionless physics object but it's moving pretty weirdly

not how you'd expect

heres another video where you can see it's only on less sharp bounces

[timid dove]

For this kind of game I think I would probably precompute the path with Raycasts and then animate the object along the path rather than hoping Rigidbody will behave predictably

It looks to me like you are already doing step 1

[solid blade]

thats a good idea i hadn't thought of that. agreed too, because the bouncing is a major part of the game

the tracer does that, yeah

just out of my own curiosity do you know why that is even happening?

never seen it before

[elder sedge]

Unity 6. I need help fixing this, or figuring out why it's happening. This capsule, bounces up and down on a mesh collider floor, it's got a frictionless physics material. This ALSO happens on my player which also has a capsule collider, it's driving me mad lol!

[elder sedge]

It only happens on edges btw.

[sand chasm]

Try marking it have friction

[tepid sluice]

Does anyone know why my bullets suddenly speed up and speed down?

I’m currently using an impulse add force method for them

[timid dove]

Not without seeing your code

Presumably a bug in your code

[tepid sluice]

This is the main code for setting up the instance

[timid dove]

The code that calls this passes in a velocity

The issue is probably with that

[tepid sluice]

I’ve tried to put the rigidbody to sleep and waking it up again but the same result still occurs

[timid dove]

Not sure how that's relevant to what I mentioned

[tepid sluice]

Wait so what does it mean

[timid dove]

It means you're passing in a faster velocity sometimes

We should look at the code that calls this function

[weary anvil]

Try modifying velocity, not linearVelocity

[tepid sluice]

Isn’t velocity deprecated

[tepid sluice]

The function below is the one that applies the stats

[weary anvil]

In Unity 6, yes. If you are using Unity 6, I think linear is the correct usage

[timid dove]

So again this velocity is coming from some higher up function

We need to keep going until we see how it's being set in the first place

[tepid sluice]

I’m not on my pc rn so I don’t have the specific function but I’ll get on in a bit

[timid dove]

Alright I'll probably be on a boat by then. Just make sure you're not doing something like multiplying the velocity by deltaTime

[weary anvil]

I suspect you are compounding velocty. Does the speed increase every respawn?

[tepid sluice]

Damn

[tepid sluice]

No

It’s a for loop which uses a referenced array of scriptable objects holding the stats of the bullets

Also using coroutines for cooldowns

[timid dove]

The way you're using a coroutine there is also a little problematic because WaitForSeconds is not a precise thing. You're not going to have even bullet patterns especially when the framerate is fluctuating. That's a separate problem to solve though

[tepid sluice]

heres the code for the attack pattern

[next salmon]

pivot must be at the center of the 3d wheel.

[heavy jewel]

I made a bullet penetrating raycast algorithm that works basically like this

while(traveled < maxdist and penetrateddistance < maxpenetrationdistance)
  if not inside something
    raycast in direction by maxdist - traveled
    move current position in direction by stepsize
    set current collider to hit collider
    if didnt hit, break
  else
    raycast backward from current position 
    thickness is the distance from entry to exit hit
    if we run out of penetration distnace, we mark this penetration as not going all the way through
    if we hit something, add a penetration to the list with appropriate values
    otherwise, move current position in direction by stepsize and continue

and when i shoot a raycast to test penetration, it works just fine. it detects entry and exit points, however sometimes it erroneously creates thousands of points when hitting a certain thing, any advice on soemthing better than this would be good thanks, this algorithm kinda feels janky. anyway when doing this it never generally MISSES a wall entirely. i added this script to my bullet though, which kinda moves its position by its velocity and then casts one of these rays from last to current, but somehow it seems to randomly miss walls, going straight through. if i do normal raycast, these walls are not ignored and the bullet hits them just fine, but JUST for this it breaks, seemingly randomly, although i noticed it more when it hits at a shallow angle. any ideas why???

heres a demonstration, the gizmo with lines and circles is my test ray that clearly detects everything just fine, the particle system is wehre the bullet hit

[rare carbon]

Hello! I am making a 2.5D platformer (2D sprites, 3D environments) in Unity. I am using a state machine for the movement. I am having some problems with physics, my jumps are too small, but when I try to enter a somersault state (like in Metroid), in which is activated when the player is jumping while moving, the player is launched across the test stage as if it were teleporting, sometimes ignoring the physical barriers I put up. I have used forces and I have not directly manipulated the velocity. There are times when my character trips over like one slight pixel on the ground that I put together with cubes. I have noticed after debugging, when I try to move the character, it doesn’t move after that fact. I checked the debug logs and every time my character trips, there is no buildup in the velocity. My UpdateState method is tied to fixed update for physics. Before I share my scripts, Are there any physics based functions I should avoid?

[timid dove]

For tripping over a slight pixel that should be fixed with composite colliders

[forest thunder]

is there a way to get collision points without a rigidbody ?

[timid dove]

Raycasts?

Otherwise - custom physics engine

What's the use case?

[forest thunder]

i am creating a custom car physics .its almost done just the collision is remaining

[forest thunder]

how exactly

[timid dove]

How exactly what

How do you use Raycast?

[forest thunder]

yeah

[timid dove]

Have you checked the docs?

[forest thunder]

nope

rn i am trying with ComputePenetration

[timid dove]

That would be a good place to start

Check out other physics queries too

Such as the other shape casts, and overlapXxx

Though only the casts will give actual points

[forest thunder]

i want the points for collison response

for raycast i will need to check i all directions and i cant find anything related

[timid dove]

Wdym in all directions

[forest thunder]

for collision detection i would have to cast rays in all direction to check any collision .right?

[timid dove]

Why not just in the direction the object is moving?

Also don't forget about BoxCast or SphereCast

[forest thunder]

What if the obj collides sideways

Like another car colliding

[timid dove]

What does "collides sideways" mean?

Anyway why are you trying to reinvent the wheel here?

[timid dove]

That other car would detect the collision since it's moving that direction

[forest thunder]

If u mean the custom physics, then for optimization and full control

[tacit laurel]

if the trigger is deactivated then reactivated with the collider still inside, does ontriggerenter be called upon reactivation?

answer: yes but only if the other collider has a rigidbody 😡

[orchid sky]

🙁

[sly crystal]

what ?

[timid dove]

It seems unlikely to me you will improve on the performance of Physx.

[timid dove]

What should we be looking at here?

[sly crystal]

umm no idea. Had to fix it myself. Lil boxes can now fit on the back of the truck, but no collision with the back of the truck (they fall off)

edit: fixed 7 hours ago (aroung 7:3X PM. It's 2:12 AM now)

[deep kettle]

Hi all, I just posted in #🌐┃web but it turns out to be an issue in a Windows build too, so it must be something editor -> build general issue. My vehicle prototype uses only Rigidbody and WheelColliders and runs and seems to ouput sensible values for the various forces (torque, rpm, slip). However, when I build and run independent of editor the forces are massively different such that the vehicle does not have enough torque to even move (my calculations are giving <10 motorTorque on the WheelCollider, rather than 500+), rpms look similar at a glance, but wheel slip is much higher. Everything is run within FixedUpdate, except for updating wheel position/rotation from GetWorldPose and the HUD (run in Update). Are there specific player/build settings I should look at to identify the difference in physics calculations?

it does seem to be framerate dependent as changing vsync in editor also breaks behaviour.... will check further

heh, ok so it was the wheels in Update being the problem and running the editor in vsync

[sand hamlet]

anyway to stop the jittering when the character reaches desired speed?

im just stopping the AddForce when it reaches the speed

[timid dove]

Unfortunately this isn't really sufficient to maintain a perfect speed limit because Addforce doesn't affect the velocity until after the simulation step runs. Essentially your velocity measurement is one frame behind

For the correct way to handle this, you could check out the source code of my asset which has a speed limit feature.
Basically you need to look at the AccumulatedForce on the body and cancel it out after clamping

[sand hamlet]

` switch (limit.Directionality) {
case Directionality.Omnidirectional:
Vector3 clampedVelocity = Vector3.ClampMagnitude(expectedNewVelocity, limit.ScalarLimit);

                // We're hard clamping so remove all accumulated force
                _rb.AddForce(-accumulatedNewtons);
                _rb.SetLinearVelocity(clampedVelocity);
                break;`

does this hard cap the speed? becuz i want the character to be able to gain speed above the running max speed

setting the speed to clampedVelocity

[timid dove]

That code is for a hard cap yes

My asset also handles "soft" caps which allow the object to go faster than the limit if it gets hit by another object for example

but not with AddForce

You could consider just downloading the asset (it's free on the Unity asset store) and seeing if it meets your needs

It's not clear to me from your description exactly what you are trying to accomplish

[sand hamlet]

i think im just gonna go along with the jitter

its nothing major i think

probably

hopefully

[rare carbon]

Thank you!

[sand hamlet]

` if (moveDirection != Vector3.zero && !crouchSlide.isCrouching && !Strafing && !crouchSlide.isSliding && isGrounded)
{
float maxmaxSpeed = maxSpeed + 1f;
if (speedometer.XZspeed < maxSpeed)
{
rb.AddForce(moveDirection, ForceMode.Force);
Debug.Log("Adding Speed");
}
if (speedometer.XZspeed >= maxSpeed && speedometer.XZspeed < maxmaxSpeed &&!_jumpBuffered)
{
rb.linearVelocity = moveDirection.normalized * maxSpeed;
Debug.Log("Speed Max locked");
if (angle > StrafeMaxAngle && angle < StrafeMaxAngle2)
{
rb.linearVelocity = moveDirection.normalized * maxSpeed / StrafeReduction;
}

        }`

this part of the script is behaving differently if i recompile the script mid testing

the jitter is removed if i recompile for some reason

why would that be?

[inner thistle]

Recompiling during play can cause all sorts of issues, it doesn't mean anything by itself. It's usually better to switch it off and set it to only compile when the game is stopped

[sand hamlet]

i want it to not jitter and recompiling is the way to fix it seems like

[timid dove]

probably a script execution order issue or something .

Recompiling isn't a fix, it's a temporary bandaid that hides the issue.

[sand hamlet]

yeah i wanna fix it

i just dont understand what the issue is

[inner thistle]

This isn't an efficient way of finding the issue, you can't really know what the actual difference is between fresh compilation and compiling during play. Debug it "normally" with a debugger or log messages

Also is it really a problem if the max speed fluctuates a little? It doesn't seem to cause any issues visually

[sand hamlet]

its not a big problem yeah but it just irks me a little

also if i add the speed meter to the finished product

seing it jitter like that would be really weird

[inner thistle]

It probably always rounds up to exactly 13 if you round it to one decimal point

[spiral spindle]

Apologies for the repost, but I thought I might have some more luck in the Physics channel. I want the player object to stick to a mesh as it curves. Currently I'm adding rigidbody force to the object and rotating as it goes, but when the player is upside down it breaks and starts floating freely. Any ideas how I could achieve this?

Like in mario kart when the map does a loop-de-loop, the car is still driving as if the ground is it's directional force, that's what I want to achieve

[sand hamlet]

float maxmaxSpeed = maxSpeed + 1f; float minmaxSpeed = maxSpeed - 1f; if (speedometer.XZspeed >= minmaxSpeed && !_jumpBuffered && speedometer.XZspeed < maxmaxSpeed) { rb.linearVelocity = moveDirection.normalized * maxSpeed; if (angle > StrafeMaxAngle && angle < StrafeMaxAngle2) { rb.linearVelocity = moveDirection.normalized * maxSpeed / StrafeReduction; } } else if (speedometer.XZspeed < maxSpeed) { rb.AddForce(moveDirection, ForceMode.Force); Debug.Log("Adding Speed"); }
fixed the jittering with this script but rb.linearVelocity is getting set 0.21 lower than specified

why would it be 0.21 lower?

rb.linearVelocity = moveDirection.normalized * maxSpeed;
if i set the max speed 10 it gives me 9.79

[timid dove]

do you have any drag on the Rigidbody?

also known as "linear damping"

[sand hamlet]

no

[timid dove]

is the object touching any other objects?

Such as the ground?

Also which branch of code is running here exactly?

Is it the if or the else if?

and where does this speedometer.XZspeed come from?

[sand hamlet]

touching ground yes

[sand hamlet]

Vector3(rb.linearVelocity.x, 0, rb.linearVelocity.z).magnitude

[sand hamlet]

its the if

[timid dove]

friction

friction reduces velocity

[sand hamlet]

shouldnt rb.linearVelocity set the velocity to watever this is regardless of external circumstances?

yeah it is the friction

[timid dove]

Yes, after which the friction will take effect during the simulation phase

[sand hamlet]

got it thanks

[frigid pier]

...generally you use joints/hinges to have multiple rigidbodies be part of one system.

[warm dock]

how do I make the player (that has character controller component) to be able to push the enemy but without the enemy being able to push the player (but can push the other enemies)?

[timid dove]

2d or 3d

[warm dock]

3d

[timid dove]

Not with CC but with rigidbodies you could use a selective kinematics asset like this https://assetstore.unity.com/packages/tools/physics/betterphysics-selective-kinematics-244370?srsltid=AfmBOoqmthNV5Rbe7qqA_tKh7AmtikBK_r6qerjZv3UuJcTqfqz825Ng

[warm dock]

thanks, I'll try that

[cold socket]

could play around with making kinematic objects that follow the player, while only interacting with specific layers using a layermask.

program the object to disable itself without input and so on.

[signal wyvern]

Does anyone know of any Rigidbody replacements which produce similar physics which is deterministic?

[timid dove]

"deterministic" is pretty vague. Physx is already determinstic on the same hardware. Can you be more clear about what your exact requirements are?

[signal wyvern]

For networking purposes RBs are not ideal as the results are non-deterministic, not all my users will be using the same hardware so can’t really rely on rbs. Just wondering if anyone has made a script which makes an object physical without the use of RBs?

[timid dove]

Plenty of games use networked rigidbodies with server authority and client prediction and are fine. It's pretty rare to have a significant desync based on hardware precision and on those rare occasions it can be handled relatively gracefully with rollbacks and resimulation techniques.

"a script which makes an object physical without the use of RBs" - you're describing essentially a custom physics engine here. And yes, plenty of other physics engines are used in unity networking contexts including:

• Havok
• photon fusion physics

[signal wyvern]

The networking unity discord seems to somewhat disagree with you, floating point precision between different hardware will cause discrepancy's between client & server but i guess if its infrequent enough it shouldnt matter much.

I wouldn't be able to use any other engine as im using NGO and which isnt compatible with the whole DOTS ECS stuff

[wise geyser]

Hey boys.

I trying to make a roulette wheel, which i have created in blender. To accommodate the complex geometry im using Mesh collider without Convex and a rigid body. This partially works, if the wheel isnt turning. But i've created a script to rotate the inner wheel on the Y axis. This results in the sphere clipping through the model. Any bright people that has some ideas how I would fix it.?

[timid dove]

This is won't work with concave colliders.

Any moving object is going to need to consist of solely primitive or convex colliders if you want proper collisions

1. Rebuild the collider from primitives and convex meshes
2. Make sure the object is rotating via MoveRotation in FixedUpdate or angular velocity

[unique cave]

Actually MoveRotation similarly to MovePosition isn't any better than setting rb.rotation for non-kinematic bodies (apart from interpolation which doesn't affect physics). Angular velocity and addTorque should be preferred to avoid tunneling issues

[unique cave]

One thing to try though would be to set the rigidbody to kinematic and then rotate via MoveRotation which in that case would be slightly better than teleporting the rotation

[timid dove]

Yes I was suggesting MoveRotation for a kinematic body

Which cannot use angular velocity

[unique cave]

ah right. What I'm wondering now is whether they would be able to use the concave mesh collider with the kinematic + MoveRotation though?

[timid dove]

No

The collisions won't work

[unique cave]

Good to know

[unique cave]

Now that I tried, mine seems to work totally reasonably in Unity 6 at least

Compared to Rigidbody.rotation at least, it is much much better

Don't know how recommended that is but it looks fine and there are no error or anything

[dapper eagle]

I am writing a custom suspension implementation. I have noticed a difference between RaycastNonAlloc and Raycast. I think it's probably working correctly but my code requires an adjustment if i want to use NonAlloc.

Raycast Implementation

            _isGrounded = Physics.Raycast(_position, -transform.up, out _hitResult, _suspensionLength + _wheelRadius, Physics.AllLayers, QueryTriggerInteraction.Ignore);
            _previousSuspensionLength = _currentSuspensionLength;
            _currentSuspensionLength = _isGrounded ? _hitResult.distance - _wheelRadius : _suspensionLength;
            _load = ((_suspensionLength - _currentSuspensionLength) * _suspensionSpringRate) + (((_previousSuspensionLength - _currentSuspensionLength) / Time.fixedDeltaTime) * _suspensionDamperRate);
            _load = _load > 0.0f ? _load : 0.0f;
            _rigidbody.AddForceAtPosition(_hitResult.normal * _load, _position);

RaycastNonAlloc implementation

_isGrounded = 1 >= Physics.RaycastNonAlloc(_position, -transform.up, _hit, _suspensionLength + _wheelRadius, Physics.AllLayers, QueryTriggerInteraction.Ignore);
// Left out to save characters
_currentSuspensionLength = _isGrounded ? _hit[0].distance - _wheelRadius : _suspensionLength;
// Left out to save characters
// Left out to save characters
_rigidbody.AddForceAtPosition(_hit[0].normal * _load, _position);

The difference is that when i drag my car object into the ground and the suspension generates an opposite force the suspension seems stuck in the car itself.

Is it possible raycast non alloc does not account for colliders onto the same object? the collider is in a parent object

I am trying to debug it, but i can't quite deduct the difference

[timid dove]

Wdym by "same object" here exactly?

Raycast doesn't have any sense of an object from which the ray "originates". It's a static method.

Your code looks incorrect to me though. That 1 >= Raycast condition seems backwards.

That seems to be saying you're grounded only if the Raycast doesn't return any hits.

[dapper eagle]

The wheel script is on a child object of the object with the rigidbody and box collider. The raycast if i remember correctly ignores colliders it starts in unless some specific object. I think you are right about the comparator it should be 1 <= hits

Maybe that’s the issue, when i can i will check

[timid dove]

In my mind the most straightforward way to write it is if (hits > 0)

[dapper eagle]

It was the expression that was wrong, i tested it now with 1 <= Physics.RaycastNonAlloc() and that works.

[hollow echo]

What difference 👀

[dapper eagle]

It’s already solved, i thought maybe i needed to do some extra code but it was the 1 >= that needed to be <=

[hollow echo]

Sure, but what difference between Raycast and RaycastNonAlloc are you talking about?

[dapper eagle]

I thought there was a difference but it was just that line wrong on my end.

[cedar oar]

Im having the issue where cubes can be held inside walls and its kinda jittery. I know why but i dont know how to fix it. whis is my current code and im jsut pushing an object to a point

    public void UpdateTransform(Transform Target)
    {
        rb.transform.rotation = Target.parent.parent.rotation;

        Vector3 previousPos = rb.position;
        rb.MovePosition(Vector3.Lerp(rb.position, Target.position, Time.deltaTime * atractiveForce));
        Vector3 delta = rb.position - previousPos;
        rb.linearVelocity = new Vector3(delta.x, 0, delta.z);
    }

[timid dove]

You can't be doing both MovePosition and setting the velocity

you should pick only one - and that one should be setting the velocity because MovePosition does not respect collisions

you should also not be rotating the object via its Transform

you should basically never touch the Transform of a Rigidbody if you want it to behave realistically

[cedar oar]

how do i lerp with setting velocity?

[timid dove]

Just lerp the target position

basically replace MovePosition with just calculating a position to go to

        Vector3 previousPos = rb.position;
        Vector3 targetPos = Vector3.Lerp(rb.position, Target.position, Time.deltaTime * atractiveForce);
        Vector3 delta = targetPos - previousPos;
        rb.linearVelocity = new Vector3(delta.x, 0, delta.z);```

[cedar oar]

it seems to be moving slwoly regardless of atractiveForce

[timid dove]

this:
Vector3 delta = targetPos - previousPos;
needs to be:
Vector3 delta = (targetPos - previousPos) / Time.fixedDeltaTime;

[cedar oar]

that works. Should i turn off the cubes gravity when held so it can be well held?

[timid dove]

Well - it seemed like you were purposely excluding the y axis from the holding behavior

is that not on purpose?

If it's not then you want to simplify like:

Vector3 previousPos = rb.position;
Vector3 targetPos = Vector3.Lerp(rb.position, Target.position, Time.deltaTime * atractiveForce);
Vector3 delta = targetPos - previousPos;
rb.linearVelocity = delta / Time.fixedDeltaTime;```

[cedar oar]

ah it wasnt. ill add it

[cedar oar]

what should i do so the face of the cube faces the player?

that was the intent around the idea of changing rotation transform

[timid dove]

a similar thing you did with velocity, but for angular velocity

get a target rotation (the player's rotation probably), then calculate the rotation difference (Quaternion.Inverse is useful for this), then set the angular velocity similarly

[cedar oar]

how do i do the last part of delta/time.fixedDeltaTime?

        Quaternion previousRot = rb.rotation;
        Quaternion.Inverse(previousRot);
        Quaternion targetRot = Quaternion.Lerp(rb.rotation, Target.rotation, Time.deltaTime * atractiveForce);
        Quaternion deltaRot = targetRot * previousRot;
        rb.angularVelocity = deltaRot / Time.fixedDeltaTime;

my code is this

[timid dove]

You're not doing anything with the inverse line

It should be delta = target * Inverse(previous)

[cedar oar]

what about the last line?

[cedar oar]

        Quaternion previousRot = rb.rotation;
        Quaternion targetRot = Quaternion.Lerp(rb.rotation, Target.rotation, Time.deltaTime * atractiveForce);
        Quaternion deltaRot = targetRot * Quaternion.Inverse(previousRot);
        rb.rotation = deltaRot / Time.fixedDeltaTime;

code looks like this and the last line says i cant do devision with quatieron to flaot

[timid dove]

1. Angular velocity, not rotation
2. Convert it to euler angles

[finite linden]

Can anyone help me with joints? I have it working correctly when the rock has a low mass (video 1). The plane moves around the rock until the hinge joints reach their angle limit, after which the rock is pulled by the ship. However, when I increase the mass + damping on the rock (video 2), instead of the plane's velocity being stopped because it can't pull the rock, the hinge ignores its limits and also comes off. Essentially, I want a stronger hinge. Does anyone know what the problem is and how I can fix it? I've been debugging this for hours 😭

[timid dove]

what code are you using to move these objects?

[finite linden]

The ship is the only thing with code. This is the only code that matters:

Vector2 direction = new Vector2(Mathf.Cos(transform.eulerAngles.z * Mathf.Deg2Rad),
Mathf.Sin(transform.eulerAngles.z * Mathf.Deg2Rad));
  rb.AddForce(direction * moveSpeed, ForceMode2D.Force);```

Is there a better way to format code?

[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]

though there's nothing wrong with what you did

other than you should put ```cs so it gets color coded

Vector2 direction = new Vector2(Mathf.Cos(transform.eulerAngles.z * Mathf.Deg2Rad),
Mathf.Sin(transform.eulerAngles.z * Mathf.Deg2Rad))```
Instead of this couldn't you just do `transform.right` btw?

Have you tried changing the settings on the hinge joint?

[finite linden]

i've messed around a little

[short yoke]

Given 10 or 20 bodies, is it possible to predict their positions in 1 year?
Like I want to predict orbits for my planets and later even spaceships. How is that possible? The planets don't realy interact with my spaceships, only with other planets. Same with my spaceships, they only interact with planets, not other spaceships

My idea I have currently is to just simulate it until that time. But that is kinda difficult, because I want to predict it 10-20 years in the future, and making that lag free is not realy possible, especially with low end CPUs

[timid dove]

Use a second physics scene and simulate.

But yes it's going to be computationally restricted.

If this is for orbital mechanics you would use orbital mechanics equations instead to predict the position

[short yoke]

It's for a map screen and maneuver nodes like seen in Kerbal Space Program

[timid dove]

I gathered as much yeah

[cedar oar]

a note KSP does use rails for planets at times so you can still speed up without issues

[short yoke]

Hm yeah. My current system is pretty dynamic, because even the sun is changing it's position due to the 1 planet orbiting it

I'm afraid of landing rockets on planets because they're always gonna move lol

[cedar oar]

Im trying to make a box share the same rotation as the player but it seems to not be doing anything

        Quaternion previousRot = rb.rotation;
        Quaternion targetRot = Quaternion.Lerp(rb.rotation, Target.rotation, Time.deltaTime * atractiveForce);
        Quaternion deltaRot = targetRot * Quaternion.Inverse(previousRot);
        deltaRot = Quaternion.Euler(deltaRot.x, deltaRot.y, deltaRot.z);
        rb.angularVelocity = new Vector3 (deltaRot.x / Time.fixedDeltaTime, deltaRot.y / Time.fixedDeltaTime, deltaRot.z / Time.fixedDeltaTime);

[short yoke]

Why don't you just

vector3 player_rotation = player.transform.rotation;
obj.transform.rotation = player_rotation;

[cedar oar]

i was told not to

and i also like lerping

[short yoke]

yeah alr idk lol

[cedar oar]

i just strugalling a lot with rotation based stuff in genral

[short yoke]

I remember when I started with unity 2 years ago... It was a horror so I immeadiatly stopped just because of rotation

[cedar oar]

ive been using unity for a few years but never programing to the extent i am now

[short yoke]

Ah flip, I'll have to code my own line renderer

[frail crater]

Absolute Noob here. I am trying to add a wheel collider to a wheel but the wheel collider is running "against the tyres grain" if that makes sense, along the incorrect axis. I found the wheel as a free asset. Hopefully an easy fix?

[whole tusk]

put the visual components on a different object

[timid dove]

First thing's first. Press the ` key and make sure the "Tool Settings" overlay is visible in your scene view:

Actually I see it in the bottom right of your scene view in your screenshot

You need to set that to "Pivot" and "Local"

Then you are actually looking at the object's real position and orientation

Wheel Colliders are always aligned with the object's forward direction, which is the blue arrow (once you're in local mode)

As not jordan said, a good way to fix this is to move the MeshRenderer to a separate GameObject which can be a sibling of the wheel collider in the hierarchy

[cedar oar]

But I still haven’t fixed this

[inner thistle]

That starts out well but everything after the first line it's just nonsense

Try something like this:

Quaternion deltaRot = Target.rotation * Quaternion.Inverse(rb.rotation);
deltaRot.ToAngleAxis(out float angle, out Vector3 axis);

if (angle > 180f)
{
    angle -= 360f;
}

if (Mathf.Abs(angle) > 0.01f)
{
    rb.angularVelocity = angle * Mathf.Deg2Rad * speed * axis.normalized;
}
else
{
    rb.angularVelocity = Vector3.zero;
}

[cedar oar]

Yeah quaternions elude me very much

[inner thistle]

Quaternion's x, y and z (and w) components don't have anything to do with Euler angles so you can't treat them as such

[cedar oar]

how can i make it so the cube dosent tilt up and down with the player looking?

what i tried didnt work

Normally I’d just change all angels but the one I wanted to stay and make it 0 and while it kinda worked it didnt fully

[inner thistle]

Usually there's a player model or object that has the correct rotation that the movement code uses so change Target to point to that instead

[cedar oar]

Okay it’s a little late for me so I’ll try tomorrow

Thanks for the help. Luckily I do already have a point like that to use

[balmy hollow]

mesh collider doesnt take the shape of the mesh

i require assistance idk why its doing this

[timid dove]

what setings do you have on the MeshCollider?

[timid dove]

do you intend for it to be convex?

[balmy hollow]

yeah unless theres another way to make a collider for a hollow shape

[timid dove]

Convex is the opposite of hollow

you will not be able to run inside it if it's convex

[balmy hollow]

oh alr

[timid dove]

It sounds like you want it to be concave, which means you should uncheck that box

[balmy hollow]

omg thank you

those collider lines werent showing up when i had convex off so i thought it just want doing anything

but now it works

thank you again

[cedar oar]

so how can i restrict player speed to 1: allow stuff like launch pads to work but also 2: prevent the player from launching themselves on accidental steep slops created by cubes and 3: still have falling not be painfully slow

[coral mango]

That is because when convex is turned off it just uses the existing mesh.

[alpine crypt]

I have a 3D game that uses vertical and horizontal knockback. It basically sets the linearVelocity of the rigidbody of the victim to the knockback Vector3.

But i have a problem where if a move has both positive horizontal and vertical knockback, but the victim is very close to a wall, what should happen is the character should go up and not horizontal. But instead what happens is all the momentum dies and the character falls right away.

How can i get the character to go up when there's a wall very close to the character and both horizontal and vertical forces are being applied?

nvm i figured it out using physics materials and messing with the friction

i'm not sure if it's applying the full vertical force, but at the very least it's pushing the character up

if someone has another solution please lmk

[timid dove]

i mean physically that's going to be the most realistic result.

any other solution would just be trying to recreate that interaction

[alpine crypt]

through code, right?

[rustic parrot]

My player position start at 0.5 for Y, but once i add a Rigidbody 2D, it changes the position to 0.51f, why ?
It doesn't look nice, the player seems to be levitating instead of being grounded

Found a fix by adding an Y offset on the tilemapCollider but still curious to know why this happens

[timid dove]

In physics 2d settings there's a thing called "default contact offset"

this is a necessary part of the physics engine which results in a small offset between colliding objects

[rustic parrot]

alright thank you, so this offset is mandatory i guess, and we need other offset to fix this ?

[timid dove]

You will most likely find that in your game at full resolution such an offset will not be noticeable to the player anyway. But yes you can offset the sprite to compensate

[rustic parrot]

I see, maybe i'm focusing too much on details

[zenith temple]

Can anyone help me with this. I have no friction, i am setting the angular/velocity from the platform. The platform and player have no friction, no drag, and no mass. However, it appears there is still some inertia? Centripical force? IDK

Both are rigidbodies and usint Unity 2022.3 basic physics.

Same behaviour no matter the interpolate, kinematic, and constraints. I have tried constraining rotation on both, I have tried boosting past the velocity difference (overshooting) and it still happens!

rb.velocity = getPlatform.rb.GetPointVelocity(rb.position);

Please don't recommend fixed joints etc. or move position unless you have some magic solution to remove jittering involved in both methods.

[unique cave]

Don't know what jittering you mean. MovePosition/Rotation + kinematic rigidbody is often the way to go for moving platforms. It is up to the character controller to handle that correctly (using the point velocity might not be sufficient since that makes the object slowly drift outwards)

[zenith temple]

I'm gathering its the point velocity now, thank you.

[zenith temple]

The jittering for moveposition when on top of a physics platform looks like motion blur at a high frame rate. no motion blur on

Do you mean switching the player to kinematic at least when on physics platform?

[unique cave]

No I mean the platform should be kinematic and moved via the Move methods, is that what you do already?

[timid dove]

When you configure everything correctly you will not get jittering

[unique cave]

This might visualize the point. In the image the platform is moving in circular path but if we take the point velocities at each point (the vectors) and move towards that velocity, in discrete simulation, we will inevitably divert from the actual path of the platform. I wouldn't know for sure but this could be one of the problems you have. One way to minimize this would be to predict the next position of the platform and derive the required velocity from there

[zenith temple]

I see. Yes the platform is kinematic in the video. Getting current linear velocity is the problem. Any pointers for making MovePosition look smooth?

[unique cave]

So what you are moving with MovePosition and what jitters? Do you have interpolation enabled for the jittering object?

[zenith temple]

The platform is interpolating to be smooth, same as the player. But adding the platform velocity to the player body pos in MovePostion makes it motion blur

[unique cave]

I don't know what "motion blur" would look in this case but If you want to keep the player non kinematic during the ride, you should probably consider moving it directly via velocity

MovePosition/Rotation are mostly useful for kinematic bodies

[zenith temple]

the point of moving platforms is to inherit their velocity to solve puzzles yeah

but moveposition does seem to give it velocity

[unique cave]

For kinematic bodies it surely does but if I remember correctly, it just teleports the body for non-kinematic, I could be wrong though

[zenith temple]

yes yes, you are right, i was mistaken

I don't know if you will see it in a compressed video:

[unique cave]

So that would probably not count for the interpolation. Is settings the velocity of the player directly an option?

[zenith temple]

the player jitters. continuous, interpolated, all that jaz. yes, i set the velocity and it behaves exactly as adding the difference.

it would be nice to know the velocity is being updated on pllatform before player but i don't know enought about unity physics

rb.velocity = getPlatform.rb.GetPointVelocity(rb.position);

[unique cave]

So was there an issue besides the drifting with this approach?

[zenith temple]

no, exactly the same as a velocity += difference

so i prefer the latter because other objects can interact and pass velocity additions

[unique cave]

don't know what difference refers to here but if there is no jittering, you should probably go for it. Avoiding the drifting should be "as easy as" not using GetPointVelocity and figuring out the correct velocity yourself

[zenith temple]

I suppose I could just pass the motion to the player when the platform moves as a velocity change. oh the differnce is say, the player could not be knocked off the platform by hazard if its setting its velocity to platform.

I think I know what to do, thanks so much Aleksi for confirming what it is.

[unique cave]

Np. feel free to ask if you have any questions later on

[normal leaf]

Hello, I'm making a brawler-type game where each player is driven through active ragdolls. The way I have it setup is using configurable joints and making the target pose for each joint the joint of a reference animation. (Shown in blue). The ragdoll also has a force on the head pulling it upwards and a force keeping the hips straight up. Movement is done by adding force to the hips. ** I'm having trouble making the ragdoll more stable and making controlling the camera not so wobbly. Also maybe not having the ragdoll's feet dragging on the floor when it walks. Any tips?**

[zenith temple]

I highly suggest driving motion from the feet and going up. I'm not sure why you'd put all that work into the character and not have a 3rd person camera but you'll always wobble with the camera attached to the head like that. If I was you, I would be moving the hands, feet and head manually and the rest follow. Might need to have a lot of dynamic configuration though.

It might be you end up with a hybrid of ragdoll and IK rig

[normal leaf]

by moving the limbs manually do you mean by adding forces to them or using IK? do you think that i should push the feet forward and make the rest of the body align with the feet? also the model is off mixamo lmao

[wraith junco]

How would you guys simulate a suspension-less wheel? Like in Garry's Mod the ones you can place on objects with the tool gun.
Maybe just decrease the friction on the wheel object a bunch, and then make the model rotate proportional to the parent object's velocity? Or is there a better way?

https://i.ytimg.com/vi/JTOtGpZDw98/maxresdefault.jpg

[warm dock]

how do I make it so the player cannot phase through object?

the character is using rigidbody and for the object is box collider

[inner thistle]

Does the character have a collider and are the colliders 3d

[warm dock]

yea

[inner thistle]

Is the movement done with forces or velocity?

[warm dock]

it's moveposition I believe

rb.MovePosition(rb.position + moveDirection * movementSpeed * Time.deltaTime);

you're referring to this right?

[warm dock]

do I need to change it to either those two? does using moveposition ignore the object's collider

[inner thistle]

It might cause that if the colliders are very thin

[warm dock]

I thought of that too, but it does the same thing when running towards bench and the throne which has a thick box collider

[cedar oar]

Im wanting to fix an issue with my cubes where no matter the mass of the cube, it can push any other object. What would be the best way to fix it? my current code for it is this:

        Vector3 previousPos = rb.position;
        Vector3 targetPos = Vector3.Lerp(rb.position, Target.position, Time.deltaTime * atractiveForce);
        Vector3 delta = targetPos - previousPos;
        rb.linearVelocity = delta / Time.fixedDeltaTime;

[timid dove]

how do you want it to behave exactly?

[cedar oar]

preferably+ how pushing an object into another would in real life. It may move a little depending on the mass but not to this extent

[crisp owl]

I have a simple problem but I dont have an easy fix

lets say I have some rigidbody thats moving with some mass. I have a “wall” object that the rigidbody needs to collide with, but instead of hitting it and stopping, it should break through.

the wall cannot be a trigger, it has to be a solid object

I just don’t want it to stop the momentum of some objects

i also want to limit changes to layering and the wall itself

im also not in unity 6 im in 2021

any simple fix? i feel like this should be possible

[warm dock]

what the hell

the thing that you want is exactly the thing that I'm trying to fix

[crisp owl]

the “walls” have rigidbody set to kinematic

[warm dock]

Like this?

[crisp owl]

lmao well join in

nah thats different

[warm dock]

ou

[crisp owl]

I have breakable objects I want the player to collide with, but bosses to just break through

[warm dock]

Ohh okay I get it now

[cedar oar]

Is there a way to do what I want?

[hasty cedar]

I would have 2 colliders on separate physics layers, one collider is trigger on a layer that only interacts with the player, and the other non trigger that collides with everything else it should appear solid to

[crisp owl]

yeah that's the best I could think of too

kind of a sucky workaround imo

[hasty cedar]

I think it works pretty well, because it's never meant to be solid to the player, so it feels like an appropriate use of layers

[hasty cedar]

other option is maybe to have an animation of the player breaking it, and applying physics after the collision

[crisp owl]

its annoying to need two colliders

such is life with unitys physics

trying to wrangle a realistic physics sim into working for a game

[hasty cedar]

In my 2d game I have a bunch of zones that deal damage over time. I use OnTriggerStay2D for it to deal damage to the player.
When the player stands perfectly still while inside the zone, OnTriggerStay2D no longer gets called, and it resumes immediately after the player moves.
I want OnTriggerStay2D to be called every cycle the colliders overlap, is there an easy way to do this?

[crisp owl]

I do it in update

track the overlapped objects by OnTriggerEnter2D and OnTriggerExit2D

[hasty cedar]

I found that to be a bit unsafe especially with coroutines, because things can be destroyed after Enter and before Exit

[crisp owl]

or fixedupdate rather

I do think disabling an object inside of a trigger will not call OnTriggerExit2D which… I think its really dumb

[hasty cedar]

I'm pretty sure if the zone spawns on top of player, OnTriggerEnter isn't called either, so there needs to be a manual check in Awake

[crisp owl]

I have a very similar script

I dont use stay

[hasty cedar]

yeah I could probably do it your way, just have to be safe with some checks, I was hoping there's some setting to just force it to call OnTriggerStay

[crisp owl]

for that reason

I think I simply check if the object is disabled or destroyed

[hasty cedar]

that makes sense, thanks

[crisp owl]

You could even make it its own component just to track objects in a trigger

[warm dock]

how do I fix this ;(_ _

the character uses rigidbody (3D) with moveposition for movement and the object all uses box collider

[random palm]

Maybe this can help you: https://discussions.unity.com/t/why-does-rigidbody-moveposition-ignore-collisions/876471/3

[timid dove]

Show your Rigidbody inspector

[warm dock]

thanks, I'll check it

[timid dove]

and can you show the inspector of one of the objects that you are trying to collide with?

(also the capsule collider here)

[timid dove]

can you turn on gizmos in your scene view so we can see the outlines of the colliders? Your box collider is both tiny and very high up according to those settings

so we should make sure they're actually in the position we expect

[warm dock]

it is there, also it doesn't phase like if it were have no collider at all. it kind of let you phase if you run continuously, and if you stop it pushes you away

I can take record another video if you want

[timid dove]

why do you have a scale of 1400

[warm dock]

when I import the 3d object it was very small so I upscale it hha

[warm dock]

aghh I've been trying to fix this problem for over 2 days rn, if anyone know the solution please I'd be very happy

[safe reef]

Moveposition is the issue, change it to forces or velocities instead, because moveposition can go inside colliders no problem

[safe reef]

This sounds like when you're moving it pierces through and when you let go physics does its job and pushes you out the collider so it makes sense to make sure you're moving correctly

[inner thistle]

Have you tried fixing the scale already? It's possible that a massively scaled down collider works wonky even though the final size looks correct in the scene. If you can't fix the model you can add a parent to it, scale the model and add a non-scaled collider to the parent

Or at least test if it works with just the default non-scaled cube with a collider, to confirm it's those particular objects at fault and not something else

[warm dock]

oh my you're right, it did worked thanks!! although I do like the mvoement of the moveposition better

[warm dock]

I thought of that too, but I've tried remaking the room with smaller scale and it did not work. but thank you nonetheless

[sand hamlet]

how do i add force that only affects the direction of movement and doesnt add any speed

[inner thistle]

Change the velocity directly. rb.linearVelocity = direction * rb.linearVelocity.magnitude

[sand hamlet]

but then its gonna be pretty annoyin to make the turn natural

leftRightDirection = Vector3.Cross(rb.linearVelocity, Vector3.up).normalized;
i tried this but

this doesnt seem to be taking vertical movement into consideration

if im applying force to the true left right of the direction im going in then i think it should work

[warm dock]

how do I manually apply gravity to a kinematic player with moveposition movement, I've tried various way but it doesn't work. lot of times it just overrides the movement

[timid dove]

if you're using MovePosition you are completely in control of the motion of the object

so you'll need to simulate gravity and everything else yourself

The basic idea is to keep track of the velocity, or at least the vertical (y axis) velocity, and accelerate it each physics frame.

[warm dock]

yea I just learned about it, thanks

I was able to make it work tho

[viral cove]

So, I am currently writing a fluid simulation based on this video
https://www.youtube.com/watch?v=rSKMYc1CQHE

And I did most of the things that were in here (I just have yet to move it to a compute shader and I haven't implemented ability to interact with the fluid via mouse)

But I'd like to know if anyone here could have any idea how could I modify it so it could interact with other gameobjects

[timid crater]

Hello, i made a code that has 2 states between a normal animated player, and a ragdoll player, the problem is this chest collider wont fall correctly and will stay upright all the time which causes alot of issues, is there any fixes to this ?

[viral cove]

Is there really no one here who could point me in the right direction with this?

[inner thistle]

It's not the kind of topic that could be summarized in a useful way. Google for "fluid simulation algorithm with obstacles" and try to adapt them to your use case

[twin nebula]

How do limits work? It still absolutely bewilders me that you guys understand joint limits...

You would THINK that this would allow the cube in question to rotate about 180 degrees in total, and it does, but for some arbitrary reason the limit is applying perpendicular to the limit