⚛️┃physics

[pulsar canopy]

you mean there is no door? It's just a hole in the wall? Or do you mean the door has no model for now?

[winged brook]

door doesnt exsist

there are no doors

[pulsar canopy]

can you screenshjot the collider for the object? Cause you seem to have selected a door game object there, which implies there is a door

[winged brook]

its not a gameobject

its just highlighting around the mesh

when imoutside the object it doesnt hightlight

when im inside it, it does

its just highlighting the mesh

idk its just somin unitydoes

[pulsar canopy]

I see. If you're sure the collider is fine, it could be too low for the player to fit or it might be the bottom part and you can't step over it? It's weird that the bottom portion would be highlighted otherwise

[winged brook]

no it just wont let me through

i checked the height

[pulsar canopy]

but why are you saying the door is a plane? That's confusing me. Does it exist or does it not exist? The floor part of the door highlight seems to imply that the door exists

[winged brook]

i never said the door was a plane?

i said the room is just a plane?

like as in its thin

it doesnt have thickness

though i figured it out

[pulsar canopy]

you said it here

[winged brook]

mb i meant doorway

turns out it wasnt big enough of a hole, eventhough my player could clearly fit.

[noble dew]

this is my code that contains camera rotation, jumping, and player movement. since my player is a sphere, am i able to make the player have acceleration while horizontally moving, like a ball rolling?

[inner thistle]

https://learn.unity.com/course/roll-a-ball/tutorial/moving-the-player

in other words don't set the velocity directly, use AddForce

[mental garden]

i personally prefer setting the velocity directly, it gives me more control

[inner thistle]

They specifically wanted acceleration "like a ball rolling" which you get for free with AddForce, otherwise you'll have to calculate the acceleration, deceleration and dampening manually

[noble dew]

thanks, I’ll try that.

[frosty ore]

#💻┃code-beginner message
i can tinker with joints a little later.. (i'd love to help but unfortunately my physics joint experience requires me to have hands-on with the components)

[proud trellis]

Right, I'll move my question here. The below video is a simple example of a configurable joint I have, which is basically masquerading as a hinge joint.

So, as soon as the rigidbodies are enabled, the yellow brick starts rotating around the joint, which is expected behaviour, due to gravity and center of mass (0-3seconds)

I'd like a way to "Resist" this. I've tried

1. Zeroing out angular velocity on no input in FixedUpdate (it ends up slowly falling rather than falling with velocity)
2. Applying a countering force on no input. This would work, but its very very finnicky. As it is basically a stabilizing force, and it ends up generating velocity on its own, so hence creating osicllations
3. Playing around with configurableJoint.targetRotation. This yields the best results comparitively, but only when modifying values directly in the inspector. I set the joint's angularXDrive.positionSpring to an insanely high value, and then moving the X angle of target rotation in inspector. This seems to hold the joint very well. But when manipulating the same thing via code, its not great. I calculate the angle by basically doing Quaternion.Angle(initialRotation, currentRotation) of the yellow bricks rigidbody, and directly set it using
   configJoint.targetRotation = Quaternion.Euler(angle, 0,0)

the only other option that is in my handbook right now is, on no input, I would change the angular limits of the joint such that it isn't able to rotate at all. But that is not what I would like to do

[pulsar canopy]

Shouldn't it not fall if the target rotation is at 0,0,0 (or whichever value would keep it there?)

I have a case where my joint stands upright and doesn't fall because that's its target rotation.

When I try another type of joint for that, it just falls down

[proud trellis]

Ideally yes, and that works, but I'd like this to be a recurring behaviour, on any rotation

so I have something like this :

        void Awake()
         {
            initialRotation = rb.rotation;
            Freeze();
         }

        private void FixedUpdate()
        {
            if (!Input.GetKey(clockwiseKey) && !Input.GetKey(antiClockwiseKey) )
            {
                if (wasGivingInput)
                {
                    rb.angularVelocity = Vector3.zero;
                    wasGivingInput = false;
                    Freeze();
                }
                return;
            }

            if (!wasGivingInput)
            {
                wasGivingInput = true;
                configJoint.angularXDrive = new JointDrive()
                {
                    positionSpring = 0,
                    maximumForce = 0,
                    positionDamper = 0
                };
            }
            var isClockWise = Input.GetKey(clockwiseKey) ? 1 : -1;
            rb.AddTorque(attachPoint.up * (isClockWise * force));
        }
       private void Freeze()
        {
            configJoint.targetRotation = Quaternion.Euler(Quaternion.Angle(initialRotation, rb.rotation), 0, 0);
            configJoint.angularXDrive = new JointDrive()
            {
                positionSpring = 30000,
                maximumForce = 300000,
                positionDamper = 30000
            };
        }

This is resulting in some weird resting positions

The color is dark green whenever I'm applying inputs and hence forces, and white when no inputs are applied and it should stop right there immediately

sometimes its correct, sometimes its wrong

[pulsar canopy]

you are updating the target rotation which is the correct way but are the forces supposed to be at 0 when not giving input? That would make it fall without returning

maybe I'm misunsderstanding

[proud trellis]

the joint drive forces?

they're only 0 when input is being applied, and hence other forces are being applied

[pulsar canopy]

yeah, that's what I meant

was just wondering if that was the intended behavior

[proud trellis]

yes it is

[pulsar canopy]

idk then, I thought you were dfoing the correct thing

[proud trellis]

the rotation is wrong, that's all

[pulsar canopy]

Why in the fuck does my rigidbody rotate on the Z axis, even though its rotation is frozen?

I notice no difference in behavior when I uncheck that freeze

and yes, it rotates via torque, not transform

[timid dove]

show screenshots or a video and your code?

[pulsar canopy]

it's in prototyping so it's done with playmaker. I don't know if this will mean anything to you

[timid dove]

well you're modifying the rotation of the Transform directly right there

Which, for one, is a bad idea for a Rigidbody

[pulsar canopy]

what? How?

[pulsar canopy]

no

[timid dove]

For two - modifying the x and y rotations like that will result in "shadow" rotations on the z axis due to gimbal lock

[pulsar canopy]

that's not what that is

[timid dove]

it is a bad idea regardless for a dynamic Rigidbody to modify the Transform directly

[pulsar canopy]

I am just building a vector3 from the mouse x and mouse y, where Z remains zero

[pulsar canopy]

I am never doing that

[timid dove]

hence this

[pulsar canopy]

Yes, this is exactly what I was thinking

my thought exactly

[timid dove]

you are you just said you are lol

you're rotating the x and y directly via the Transform

[pulsar canopy]

however, shouldn't the freeze prevent that?

[timid dove]

from your mouse

it only prevents rotations from the physics engine

your direct changes will apply regardless

[pulsar canopy]

No I'm not

I'm just building a vector3

[timid dove]

oh I see I misinterpreted that. But yeah rotating on the x and y will get you z rotation through gimbal lock more or less.

What are you trying to make here

[pulsar canopy]

damn

I didn't know that

that's a little bit problematic

I may try to correct it in every frame

[pulsar canopy]

my player is a diver and can move and rotate in every direction. However, twisting in the Z is the only unhelpful axis

should stay centered, otherwise it gets confusing

are you telling me I need to use Quaternion axes directly? That's fucked. I don't even know how they work

[timid dove]

i am definitely not telling you that

Really what I'm telling you is that you should stop worrying about euler angles

[pulsar canopy]

yeah but the rigidbody freeze IS supposed to prevent that, right? Cause I'm rotating with torque

[timid dove]

the rigidbody freeze prevents rotation around the global z axis

it doesn't prevent your object from picking up any value in the apparent z euler angle in the inspector

euler angles are basically cancer

[pulsar canopy]

ok but what does this mean then? Aren't they the sole source of my worry?

[timid dove]

what you actually are worried about here I think is that your up axis direction remains along the plane whose normal is orthogonal to the player's forward direction.

[pulsar canopy]

okay

how exactly

I found this online:

"I think the issue is that you’re directly modifying the Euler angles - removing any benefit you get from having them internally stored as Quaternions. When you get the eulerAngles vector, or use Quaternion.Euler, you are essentially taking the representation in 3-dimensions and converting it to a quaternion. Much the same as if you convert a Vector2 to a Vector3, you don’t gain any information by doing this, and so you lose the benefits of the Quaternion.

Instead of modifying the Quaternion via the Euler angles, try using the Quaternion.RotateAround method. This allows you to rotate the Quaternion around a specified axis, without every converting it to the Euler representation (and thus avoid Gimbal Lock). If you call this 3 times, one for each spacial axis, you should get the effect you want."

Do you think this woould work?

this sounds legit to me

no, it's not legit. Can't do this with torque

[signal wyvern]

Has anyone here built there own physics/physics engine inside unity?

[frozen widget]

Nope, never even tried lol

Default Unity physics are mostly enough for me, however I have built some custom water physics, so could that count?

[unique cave]

I bet many of us haven't. If you have any specific issue or question, feel free to ask, someone of us might be able to help regardless. Also make sure to include short description of why you want to make such engine just in case there is easier way to achieve the same thing by other means, creating a physics engine is no trivial task

[unique cave]

Looking at your earlier message, it seems like it is related to synchronization issues in networking. As I said, making a physics engine is far from trivial task. Add fully deterministic to the requirement list and I would very highly recommend either using some other available physics engine or the default one with networking side solutions to synchronize things. I'm not the right person to talk on networking but by quick googling, it seems like unity has few solutions available already for networked physics (including for the ECS physics). Physics engines that are not fully deterministic are often used in networking environments, that much I know. Even a simple 2D engine like Box2D is tens of thousands of code lines. If you don't have a large team of developers working for you, I would very highly recommend not going for a custom engine. All the major physics engines also use C++ which makes them often faster than C# ever could be, making a decent physics engine that also performs decently well it very hard

[signal wyvern]

Well i dont know if id need an entire engine, I just would like some deterministic physics, But im using Gameobjects/Monobehaviour soo i cant just switch to ECS. Im attempting to run client prediction & re-simulation with a bunch of physical objects, but this is a problem because unitys base physics is non deterministic.

[upbeat sandal]

how do we get hit/ normal from a Collider2D collision?
I want to get hit from this collision

and its in OnTriggerEnter

[unique cave]

You basically don't. Triggers are not simulated the same in the physics system. Triggers pretty much only tell you if you have hit a collider or not. Is there a specific reason not to use regular colliders and OnCollisionEnter2D instead?

You could probably estimate the normal from the position of the collider but that is probably not very accurate

[upbeat sandal]

its because its a projectile (bullet) and i need to be trigger bc i dont use physics (except for collision and movement)
i need that my projectiles can go through obstacles and create a bounce sometimes (need to calculate new direction)

[unique cave]

On fast moving projectiles especially triggers are not often great because triggers do not support continuous collision detection so you may often run into tunneling issues where bullets go through objects without a notice

At least that is the way it works in the 3D engine, I'm not positive on the 2D one

[unique cave]

Have you ecountered any issues with unity physics or have you only heard that? PhysX like most physics engines are pretty deterministic between runs but not 100% deterministic on different systems so you cannot expect simulation on client and server side to yield exactly same results though very close on most cases. I assume only physics libraries that do fixed point math only can be 100% deterministic. That also has it's limitations and such engines are rare nowadays. I would assume that not to be a huge issue for networking though. You should probably simulate the physics on both client and server side and just make sure to sync any discrepancies in regular intervals from either direction. I would be surprised if the "nondeterminism" would be noticeable enough to cause any observable jittering between packages unless your game is very physics heavy. The people on #1390346492019212368 can probably help you more with that.

I haven't done a single multiplier game myself so I'm not the right person to tell you how to exactly implement that but even without knowing anything about your game, I would highly recommend going for other routes than writing your own physics engine. You would end up facing many much harder issues than that you are currently trying to fix most likely.

At least I'm yet to hear a good reason you need fully deterministic physics engine. If I have missed the reason, my bad

[signal wyvern]

Yes im trying it, Im attempting to run physics on both client & server, the discrepencys are small but they add up which causes for constant jittering for my player, as he will constantly re-conciliate to the correct point as the clients simulation is deviating.

[signal wyvern]

This is a little clip of my game, its a fps where players build and fight each-other on physical boats

[signal wyvern]

This clip of game play is using an old client authed approach, Im currently working on switching the game to a more server authed approach, With server authed movement & client prediction for several reasons.

[unique cave]

The only part that I don't understand is why the discrepancies would "add up". If you transferred the position, velocity, rotation and angular velocity from the server side "correct" simulation, the discrepancies would essentially reset each tick. My best guess is that the bigger issue is in your implementation of the applying of the server side data (which will come late from the server with potentially varying delays and even packet loss) on top of the client side simulation. I still cannot believe the difference between the simulation between the simulations over such short time span (one tick from server) would cause any issues. It should be doable to test that though by creating an identical test scene in server and client side and comparing them over time.

If by client prediction you should mean that you are not sending the data from the server to the client to continuously correct the discrepancies and instead just simulate it in both, then I fully understand how the two might and likely will get out of sync. What I'm trying to say is that this is probably more so an #1390346492019212368 issue than #⚛️┃physics issue. The physics system is very deterministic to an degree you wouldn't notice a difference in most cases. It is however not deterministic over long periods of time (on different hardware) meaning you have to correct the client side predictions preferably every tick I would assume. The client side prediction would only work as a way to make the game more responsive between ticks, not to simulate the game as a whole in parallel which would require level of determinism most physics engines do not offer due to performance and other reasons (for example it being not necessary for most games/applications).

[signal wyvern]

Maybe the term "Add up" isn't appropriate, the discrepancies will reset each tick but the discrepancies still exist and they are fairly noticable, my clients are running approx 20 frames ahead of my server atm, the clients simulation can deviate within that 20 frame period.

Im running the physics on both client & server, Im running my clients simulation ahead of my server to account for the latency, sending player inputs over to the server, Server buffers these inputs and replays them on the matching fixed frame, the server then sends periodic updates to the client containing the states & positions, client will then receive these stale updates and resimulate the remaining frames to catch back up to its current frame, and this is the point where the player will be warped to the corrected position.

Maybe i could reduce the lead the clients have, this may make the jitter less noticable.

[unique cave]

As I have said many times already, I'm not the right person to judge whether that is a good approach, idea though sounds about right to me. Have you tried debugging the code to be confident the implementation is correct at least? I'm trying to think of a flaw in that idea but my brain's not doing much anymore today. I'll get some sleep now

[signal wyvern]

No worries, Its fairly difficult thing to debug but my states match & reconciliation works perfectly when my player is moving on land without interacting with physical moving object, as soon as my player is moving on a boat with Rbs the discrepancy's occur, but thanks for attempting to help, its a difficult one to explain.

[unique cave]

No wonder it's hard to debug, there's a reason I haven't wanted to touch networking yet

[eternal kettle]

so im trying to use a navmesh agent's pathing to control a rigidbody

like figure out the direction with its path then use .AddForce() to push it in the right direction

im trying to figure it out from the comments from brolive on this reddit thread but im clearly doing something wrong
https://www.reddit.com/r/Unity3D/comments/uykn1j/how_to_utilize_navmesh_with_physics_if_at_all/

[severe moth]

I do that in my game and what I do is this:

In update I do this:
navAgent.nextPosition = rb.position;
navAgent.SetDestination(targetPosition);

and then I use navAgent.desiredVelocity to get the vector for applying force to the rigidbody

[severe moth]

although I have seperate functionality for the repathing so it doesn't calculate a new path every frame, but every x seconds. You can also use the corners like he did in the comments in the pastebin, but I found this way to be simpler

[eternal kettle]

thanks

it frustrating because I know what I wanna do but because I'm pretty new to using unity I don't know how to do it

[severe moth]

The pro of using desiredVelocity over the corners method is that I believe desired velocity is also affected by things like local avoidance from other agents

[severe moth]

Oh and remember to set
navAgent.updatePosition = false;
navAgent.updateRotation = false;

at start

[eternal kettle]

cool

[eternal kettle]

I read about update position but I still don't really get what it does

[severe moth]

its because the navagent will update its own position by default, so by disabling it we are taking over control explicitly

https://docs.unity3d.com/6000.2/Documentation/ScriptReference/AI.NavMeshAgent-updatePosition.html

[severe moth]

Has anyone tested using Temporal Gauss-seidel Solver instead of the default Projected Gauss-seidel solver?
I saw the talk that was done on Physx 4, and they demonstrated being able to have a stable long chain of segments, attached to a heavier, in order to demonstrate the common instability that happens with mass ratios (https://www.gdcvault.com/play/1026204/PhysX-4-Raising-the-Fidelity at 06:46). But I tried reproducing the same setup in Unity and I really do not see much of a difference between the two solvers, even at 4x higher solver iterations compared to the demonstrations. I'm not sure if I'm missing something to replicate it correctly

[naive rune]

Character controller getting stuck on diagonal movement while going up hugging walls. Does anyone have solution for this?

[unique cave]

Make sure to include relevant implementation details (such as code). Mine doesn't get stuck and I have no clue why yours would either. Gotta identify the issue before solving it

[naive rune]

I press forward and then call the move function. There is not much to share

        {
            CollisionFlags flags = controller.Move(new Vector3(
                motion.x * Time.deltaTime,
                applyAntiBump ? GetAntiBump : _gravity * Time.deltaTime,
                motion.y * Time.deltaTime));

            if (applyAnitBump)
                applyAnitBump = false;

            IsGrounded = (flags & CollisionFlags.Below) != 0;
            IsSideCollided = (flags & CollisionFlags.Sides) != 0;
            IsCeilingHit = (flags & CollisionFlags.Above) != 0;

            if(IsSideCollided)
                footIk.setFootPlacementWeight(0);
            else
                footIk.setFootPlacementWeight(1);

            d_grounded = IsGrounded;
            d_sided = IsSideCollided;
            d_ceilled = IsCeilingHit;
        } here ```

This is the only place its called

[tender vapor]

!code

[gloomy lodge]

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.

[upbeat sandal]

Hello
I'm struggling with my raycast. I want to get a hit point from my collision but my bullets are in trigger mode so there is no hit or contact from collision
I tried raycast2D but there is nothing hit by it and idk why
yes there is layers to ignore bullet, my collision worked before this passage (bullets destroyed when hit obstacle)
both are collider 2D, same z position (0)

    private void RecalculateDirectionAfetrBounce()
    {
        float distance = _objectStats.ProjectileSize + 0.1f;
        Vector2 direction = _rb.linearVelocity.normalized;

        RaycastHit2D[] hits = Physics2D.RaycastAll(transform.position, direction, distance, _ignoreBulletLayers);
        Debug.DrawRay(transform.position, direction * distance, Color.red, 2f);
        Debug.Log("Nombre de hits détectés : " + hits.Length);
        RaycastHit2D hit = default;
        foreach (var h in hits)
        {
            Debug.Log("Hit détecté : " + h.transform.name);
            if (h.collider != null && h.collider.gameObject != gameObject)
            {
                hit = h;
                break;
            }
        }
        if (hit.collider == null)
        {
            Debug.Log("Aucun hit valide trouvé.");
            return;
        }

        Debug.Log("Hit normal : " + hit.transform.gameObject.name);
        Debug.DrawRay(hit.point, hit.normal, Color.green, 2f);

        float angle = Vector3.Angle(direction, hit.normal);
        Debug.Log("Angle entre les vecteurs : " + angle + " degrés");
    }

[upbeat sandal]

hmmmmmm
I tried to shoot everywhere and i can hit something but only if it's also "is trigger" like my portals
Why ?

[timid dove]

does the object you are expecting to hit have a collider?

Can you show the collider if so?

[upbeat sandal]

they have
I think unity don't want raycast2D from isTrigger hit something with isTrigger=false
So I used :

private void RecalculateDirectionAfetrBounce(Collider2D collision)
    {
        Vector2 direction = _rb.linearVelocity.normalized;
        Vector2 hitPoint = collision.ClosestPoint(transform.position);   <--------------------- This to get a "point" from my collision
        Vector2 normal = (transform.position - (Vector3)hitPoint).normalized;
        float signedAngle = Vector2.SignedAngle(direction, normal);

        Vector2 newDirection = Vector2.Reflect(direction, normal);
        float angle = Mathf.Atan2(newDirection.y, newDirection.x) * Mathf.Rad2Deg - 90f;
        transform.rotation = Quaternion.Euler(0f, 0f, angle);
    }

[timid dove]

By default there is no problem with hitting non trigger colliders

you can use ContactFilter2D to control whether it hits triggers or not

but it always hits non triggers

there's some other issue you have

It could be your layermask for example

[upbeat sandal]

No layermask used, no edit in matrix

[timid dove]

The matrix is irrelevant here

you are indeed using a layermask:
RaycastHit2D[] hits = Physics2D.RaycastAll(transform.position, direction, distance, _ignoreBulletLayers);

[upbeat sandal]

Oh yeah this one is ok, it ignores bullet bc without this, i detect this.gameObject

[timid dove]

I understand what layer masks are for.
Perhaps it is excluding the wall as well

[upbeat sandal]

No

[timid dove]

Maybe if you shared some screenshots showing how you configured the layermask and the inspector of the object you're expecting the cast to hit

[wet junco]

Hey all, silly question. I'm procedurally generating a map with roads as shown in the screenshot. The dashed line and the road tile are two separate gameobjects, but they're played directly next to one another. They're actually made from a single prefab I found online, so they should be the exact same height as well. My player keeps clipping on these boundaries between gameobjects. If I use a box collider on the player, they'll get completely stuck on the border between the two objects for a while, but if I keep holding forward it'll sometimes make it over eventually. Using a capsule collider is a workaround, but it "ramps up" the edge and bumps the player into the air a bit.

Does anyone know how to prevent this?
Here's my movement code: https://pastebin.com/pge4na5m

[inner thistle]

Surely the dashed line doesn't need to have a collider

[eternal kettle]

any idea why the avoidance doesnt appear to be working?

ive got it set up so its

        agent.nextPosition = rb.position;
        agent.SetDestination(target.position);

        rb.AddForce(agent.desiredVelocity / speed, ForceMode.Impulse);

im probably doing something wrong

[severe moth]

Avoidance in terms of what? other agents?

[eternal kettle]

yeah

they seem to just run into each other

[severe moth]

Hmm it should work according to the documentation but i havent actually tested it with many agents taking same path. Are they the using the same nav agent?

Ill have to test my own too when I get home I guess

[wet junco]

If that line didn't have a collider then wouldn't the players capsule collider still ramp up since it'd fall in the gap momentarily? Also if any items or ragdolls fall on the dashed line they may fall through. I did also consider just making one collider for the entire road, but then I still have the same ramp up issue when transitioning from a grass plot to a road.

[tender vapor]

what do you mean by gaps

there's still a road under the line, no?

[inner thistle]

You said that the dashed line is a separate object from the road tile

Is the line on top of a road tile or is it between two road tiles?

[upbeat sandal]

Ok I'm back
I tried without layer and I hit Bullet (own transform).
I called this function in OnTriggerEnter2D and my log says "Raycast hit nothing" but as u can see, there is something AND this function is called bcause Bullet touched this wall
WHy my raycast didnt touch the wall?

        RaycastHit2D hit = Physics2D.Raycast(transform.position, _rb.linearVelocity.normalized, 5f, _ignoreBulletLayers);
        Debug.DrawRay(transform.position, _rb.linearVelocity.normalized * 5f, Color.red, 2f);
        if (hit.collider != null)
        {
            Debug.Log("Raycast hit something" + hit.collider.name);
        }
        else
        {
            Debug.Log("Raycast hit nothing");
        }

[timid dove]

You are using a layermask that is configured to only hit the Bullet layer

your wall is on the Wall layer

That is why the raycast isn't detecting the wall

[upbeat sandal]

No, I'm using NaugthAttribute and this layer is to ignore layer

like i said, without layer, i only touch bullet
with layer -> nothing

[timid dove]

layermasks are whitelists, not blacklist. The layers you select are the only layers they can hit

[upbeat sandal]

No 😅

[timid dove]

Yes 😅

[upbeat sandal]

using NaughtyAttributes;
[Layer, SerializeField]
private int _ignoreBulletLayers;

[timid dove]

yeah this is completely inappropriate

[Layer] is the wrong attribute entirely

that's for making a layer selector

the correct way to do this is just:

[SerializeField]
private LayerMask _ignoreBulletLayers;```

[Layer] gets you a layer index

a layer index is not what Raycast uses

Raycast uses a layermask which is a bitmask of the allowed layers for the raycast

That is your problem

[timid dove]

Switch to this and then set only Wall in the inspector for it

Let me emphasize again that a layer index and a layer mask are completely different and not interchangeable.

Notice even in the docs they call the property "layer index". It is not for making layermasks. https://dbrizov.github.io/na-docs/attributes/drawer_attributes/layer.html

[upbeat sandal]

Ok i'm so sry

you are right ofc

i hit the wall

1000 excuses to u man!

[hollow brook]

https://sergioabreu-g.medium.com/how-to-make-active-ragdolls-in-unity-35347dcb952d
I'm currently trying this active ragdoll and extending it a bit more.
I also like that it is not bloated.
is there a simple good one like it too?

[wet junco]

It's a separate object entirely. I realize it might have been smarter to just layer it on top of the road, but even if it did I would run into the same issue where my grass and road connect.

Also tagging @tender vapor since you asked the same question ^

[inner thistle]

That doesn't answer my question

[wet junco]

Yeah just came back to this and realized that haha

The road bits between the dashed line are part of the dashed line object

So it's a road tile, then a dashed line, then another road tile

It's not just the white lines, the asphalt inbetween the dashed lines is part of that object

[inner thistle]

right

Unless the entire walkable area is completely flat, you'll have to have some system to go over small obstacles like curbs etc so when you have that in place it'll solve this issue as well

If the entire area is completely flat, then you can either have one big box collider for the entire world or use ground/plane detection instead of colliders

[wet junco]

The whole area should be completely flat in theory, but there seems to be teeny tiny differences halting the player. I guess I'm more looking for an idea on how I could make a system to go over microscopic differences like this.

[wet junco]

Tried this, but it just defers the problem. If I make the whole road one collider, the issue is now between roads and grass. If I make an entire chunk one collider, then the issue is now between chunk and chunk.

Although I suppose I never considered making one collider that just extends as chunks render and despawn.

That could work if I can't find any better solution

[inner thistle]

And if the player never moves up and down then you can just freeze the Y axis and not bother with ground colliders at all

[wet junco]

They do move up and down unfortunately, so can't go that route

Do you have any ideas for a change I could make to my player controller to let the player step up these super tiny differences without notice? I wonder if Unity's character controller can handle this case and how it does it.

If not I'll probably try out the dynamic box collider approach over the whole map

[inner thistle]

I would just clamp the player's Y coordinate and be done with it

[wet junco]

Jumping is pretty important in this game

[rugged eagle]

K = m * s (this is kinetic energy formula if remember correctly)

[inner thistle]

yeah but you don't jump downwards right?

[wet junco]

Oh that's what you mean, I thought you meant lock it entirely

[inner thistle]

clamp it from one side yeah

[wet junco]

Is there a built in way to do that? The only way I can think of would be expensive

[rugged eagle]

So you want to jump. What is stopping you from applying forces upwards?

[wet junco]

Or rather, how can I prevent them from going below a certain Y in an efficient manner

[inner thistle]

If position.y < 0 then position.y = 0

[wet junco]

Oh wait I already move the player every frame

Yeah that would make sense haha

I'll give that a shot, thanks for the ideas!

[rugged eagle]

What if secret of fucking mimic in unity

[upbeat sandal]

It's me again 😅
Sometimes, idk why but my reflectedDir is hum... not in good direction
I saw that my normal isnt perpendicular, my normal is in opposite direction of my current direction (-180degree). Why? idk
I still hit the wall, never the bullet

       Vector2 direction = _rb.linearVelocity.normalized;

       RaycastHit2D hit = Physics2D.Raycast(transform.position, _rb.linearVelocity.normalized, 5f, ~_ignoreBulletLayers);
       Debug.DrawRay(transform.position, _rb.linearVelocity.normalized * 5f, Color.red, 2f);
       if (hit.collider != null)
       {
           Debug.DrawRay(hit.point, hit.normal, Color.red, 2f);
           Vector2 reflectedDir = Vector2.Reflect(direction, hit.normal).normalized;
           Debug.DrawRay(hit.point, reflectedDir, Color.blue, 2f);
           transform.up = reflectedDir;
       }

[timid dove]

because you are doing the raycast from inside the collider

that's not going to give you a good normal

because you're not actually hitting a surface

you could start the raycast from further back (for example, the bullet's previous position last FixedUpdate) to avoid this

[restive marsh]

Has anyone ever used joints for building things as a destruction system? Seems to be working for me in Unity Physics for DOTS.
https://www.youtube.com/watch?v=-c0HMbCCpO8

[upbeat sandal]

SO I can replace circle collider by a circle cast (+ racyast to detect wall) in fixedUpdate?
Or just add my raycast in fixed + circle collider ?

[hollow brook]

https://sergioabreu-g.medium.com/how-to-make-active-ragdolls-in-unity-35347dcb952d

So I got my active ragdoll to work properly and gave a couple actions, and happy with it.
Issue is: If I wanted to scale the ragdoll by 5x it fails to walk, I'd have adjust the stable joint, it works okayish but not ideal.
is it possible to scale the ragdoll without having to re-calculate the entire physics, like as if the ragdoll had a scale of 1x?

what other alternatives? scale down the world🥺

[timid dove]

If you're doing circle cast why do you need a regular Raycast

[hollow brook]

apparently, this is not worth using, unless you wanna ragdoll with ants.
anyway it was fun.

[boreal holly]

Hi all!
I need a bit of help figuring out a small issue with what seems to be my understanding of 2D rigidbodies and box colliders...
Keep in mind that this project is for the currently ongoing Game Jam so i dont have all the time in the world to try new things... but i am always of course still open to listen.

I am making a 2D platformer puzzle game where the player needs to get from point A to point B to beat the level.
To get to point B, the player needs to play the level multiple times in different ways, each time the player playes the level, they can loop back to the start, this will also create a clone of the player which is the exact playthrough the player just did, see the attached video to see the concept in action.

I rewrote the recorder just now to record player inputs instead of player positions, this is so that the player also push around the current replayers (replayer is what i call a recorded player loop).
Now that that is all set up, i re-enabled the rigedbody on the replayer prefab again as it should not also be simulated with the inputs its using.

This is where the issue comes in.
The way it worked before, the replayer prefab had 2 child objects that each had their own 2D box collider, one with the "Ground" tag and the other with the "Wall" tag.
This system worked great for the player to be able to jump on the replayers and bump into them and so on.
But now that the rigidbody is turned on again to the replayer prefab, this doesnt seem to have any effect anymore and the player does not regain their jump when jumping on a replayer.

Does anyone have any idea on how i could go about solving this?
if this doesnt work or takes to long, ill have to go back to the system i used before and that would be a shame.
Let me know if you need any of my code for context.

I thought it may fit better as a post in #1390346827005431951 so if you can help me and want to reply, please head over to that post: https://canary.discord.com/channels/489222168727519232/1436844502085271563

[twin slate]

Hello all, I have a 2d game and found that the hit collider in this picture is only triggering the first time. Here are my order of steps:

1. I turn on the attack collider
   2)It is within range of the collider it is hitting
2. I turn off attack collider after 20 or so frames

I have this picture attached, my quesstion is, why is it that when i enable this collider the first time it triggers OnTriggerEnter2D, but all the following tmes it won't?

How do people usually make attack colliders in 2d?
I was thinking of making an expanding one. over the 20 frames, which feels like it will work but a bit tedious.

[timid dove]

Do yourself a huge favor. Don't use a collider for this

Use Physics2D.OverlapBox

Or BoxCast

[twin slate]

omg, i completely forgot about this....ty, that will def just solve it

[tender vapor]

ive seen this recommendation quite a few times and i feel like i can't really wrap my head around this. what's the suggested workflow to designing the offset/size of the cast/query in worldspace?

when i'm figuring out hitboxes i usually feel like lining it up to the visuals well is a key part of that, and colliders make that pretty easy

[iron galleon]

So like, how does one manage the collisions of a character with ragdoll physics? Or a human character basically?

I'm currently using a single capsule, but how should I manage it if I were to add ragdoll? Or make the arms push objects off places?

[timid dove]

You could design a workflow with ScriptableObjects and editor tools if that's something you really need to do a lot.

But the main motivation is that a direct synchronous query is so much simpler than setting up a collider and waiting for a collision callback. You practically need a state machine for the collider based workflow

For a simple attack system it's easy enough to just tweak in code

[tender vapor]

ah... i already use a lot of state machine-inspired structures

[timid dove]

Are you doing any modification of the Rigidbidy in your code?

(showing the full stack trace may help)

[open storm]

i put code in code general

[eternal kettle]

let me know if you have anything more about this

[severe moth]

It def changes the desired velocity if there are two of the same agents on top of eachother. I tested by drawing a ray in the gizmo and then putting two agents on top of eachother, and if I disable one of the navagents, i can see a change in how the ray is pointing

just not maybe as much as you'd like though

Here you can see it tries to steer around it on the left, and on the right one if I disable the top ones navagent, you see it pointing straight ahead

Here's a video illustrating it better

[wet junco]

Hi all, I'm working on my movement code for my player and running into issues with external forces. Right now my player's movement code uses acceleration and is only running on frames where the player is providing input. I'm trying to add a pushing mechanic to make players able to push other players, but I'm having trouble figuring out how to apply that external force. If I apply that force and the player is standing still, it'll work great. If I apply it when the player is moving, then due to the movement code running every frame it'll override the push force and significantly dampen the effect. Here's my code: https://pastebin.com/LRrbw55c

Looking for tips or guidance on how to go about creating movement code that can be acted upon by external forces and incorporated into the main movement loop. If anyone has suggestions or samples to reference for a system like this please let me know 🙂

[dull shuttle]

i was recording trajectory of player's jump and noticed something which i don't understand
why does a jump on the same surface (red line) has longer horizontal movement than jump downwards (yellow line)?
idk maybe my brain just not braining if it's something obvious

[tender vapor]

that would depend on a lot of things
the air velocity, the gravity, etc

[hollow pumice]

I'm trying to set up a rope/wire using configureable joints. My goal is to have the top end be stationary/anchored and then be able to move the bottom end and have the rest of the rope follow fluidly like a rope normally would. However, whenever I try moving my bottom capsule it just keeps trying to snap back into place, which leads to a lot of jittering for the whole rope when I try to force it. I'm new to configureable joints and cant really find a straight answer online, does anyone know how to do this

[silver moss]

Is this in any way an issue in game or just when you try to move it in the editor via the transform handle?

Moving a transform will expectedly make the physics weird - you should generally move a rigidbody object using the rigidbody's methods and properties instead

(Fwiw, it might be the joint's projection settings doing the actual "snapping back")

[hollow pumice]

Would setting the handles transform position to be some other kinematic object in Update and then moving that second object fix things? Thats how I currently planned on moving it

[silver moss]

Nah, just don't use the transform when dealing with rigidbodies

You could use the ConfJoint's target position drive if you want it to move to a certain position

Or move it manually with AddForce/velocity

Or temporarily setting it to kinematic and using MovePosition/position to move it, I guess

[hollow pumice]

Yeah I'll probably end up doing that, end goal is being able to move it with a VR controller

[hollow pumice]

This worked, thanks!

[dull shuttle]

i use basic settings, jump velocity is also the same for each jump
im trying to add a trajectory visualisation for the player to see where they will land when using a skill

[tender vapor]

sounds like that would just be some basic kinematics then

[dull shuttle]

so why it happens? and how to show right trajectory to player?

[tender vapor]

hold on, are the lines you drew the result of the code that previews the trajectory, or a trace of the real position during the jump?

this is lacking quite a bit of context

[tidal yacht]

Hello everyone, hope you are doing well!

Does anyone know a good book or resources to learn how physics in Unity works?

[timid dove]

Kind of a broad topic, is there a specific thing you want to know? There are four different officially supported physics engines for different circumstances.

The most common one is NVidia PhysX, whch you can find documentation for here: https://nvidia-omniverse.github.io/PhysX/physx/5.6.1/

other supported engines are:

• Box2D
• Havok
• Unity Physics

[tidal yacht]

Yes it's a broad question. Mmmmmm

I am trying to learn how to build a 2D character and understand the principal ideas of physics inside Unity.

I hope I explained myself.

Thanks for sharing the resources.

[inner thistle]

"Principal ideas of physics" is just as vague. If you want to make a physics based 2D character controller there are plenty of resources to get started. E.g. https://learn.unity.com/course/2d-beginner-adventure-game or use Google

[livid sedge]

Honestly if you want to create a game and you are just starting out (I don't know what is your level of skills) I wouldn't focus on the underlying physics too much for now. For 2d platforming it is not that crazy for basic shapes, because it is mainly around gravity (and other forces) equations. It's much better to focus on how to use Rigidbody2D correctly and it abstracts the physics enough, then when you see how it works, what are the elements there like friction, forces, rotations etc. you can start looking into individual subjects, however depending on what you want to do you may need to brush up a bit on the mathematics as well (e.g. rotations require working with rotation matrices, applying varying forces on changing surfaces requries integrals and derivatives, so it is quite advanced math there)

[paper stratus]

Hello, I'm trying to add a magnetism ability to my 2d platformer. Basically, I want to add southern and northern polarities, and the ability to push or pull those with an activated ability. I've been looking online, but can't find anything, and I don't want to use ai, but I'm completely lost. Does anyone know how to go about that?

[timid dove]

what part are you confused or having trouble with

For any complicated mechanic or system in a game you need to break things down into smaller digestible parts and tackle them one at a time

the pushing and pulling just sound like adding forces

[paper stratus]

So far, im stuck on the whole adding variables, ive never really used Unity before

[timid dove]

Ok sounds like you're trying to run a marathon before you learn how to crawl

Start with the basic learning junior programmer pathway

!learn

[gloomy lodge]

:teacher: Unity Learn ↗

Over 750 hours of free live and on-demand learning content for all levels of experience!

[paper stratus]

Thanks

[rose pebble]

Hi, I need some help.
I have a lamp, chain and player, all with 2D colliders (non-trigger). The lamp and the chain are connected with Hinge Joint 2D.

I want the chain to react physically when the player collides with it, but without blocking the player's movement – the player should be able to pass through it. I tried many things but I still can’t achieve this.

What I’ve already tried (without success):

• Changing collision layers (but then the chain doesn’t react to the player at all),
• Making the chain a trigger and applying forces manually (caused a lot of bugs),
• Adding a separate collider to the player, still blocked movement,
• Switching the chain collider between trigger / non-trigger (movement looked bad).

So I’m looking for any idea or solution to make the chain react to the player, while the player can move through it freely. A tutorial link would also be appreciated.

I hope I chose the right chat to ask for help

[timid dove]

I think I know a hacky workaround that might work for you:

• Put the lamps on a layer, we'll call it "Aesthetics"
• Give the player a child object. Give that child object a kinematic Rigidbody2D, and a Collider2D of your choosing
• Put the player's child object also on the "Aesthetics" layer
• In Project Settings -> Physics 2D -> Layer Collision Matrix set it up so the Aesthetics layer only collides with itself and nothing else

[rose pebble]

Yo thanks. Changing the rigidbody 2d to kinematic actually worked

[timber depot]

quick question, I have a living ragdoll character, I have an invisible hand object connected to a lower arm (forearm) using a hinge joint 2D, what I want to happen is that the Forearm should be able to pull/push/move Hand effortlessly (like it's nothing) without making the Hand seem heavy
tried setting its mass to 0 but it's apparently not possible, how do I achieve this?

[timid dove]

Isn't this essentially the same exact question as the one above you?

[timber depot]

oh I forgot to mention

I tried that too but I'm also attaching weapons to that hand using joints

problem is that the weapon's force/weight should also affect the character, which won't work because the hand is kinematic

[timid dove]

I don't see how that changes things

You make a child object of the hand, and the child object is what has the kinematic rb

[timber depot]

but making it kinematic will just make weapons attached to it weightless for the stickman though

here's the kinematic setup I have (after trying to use joints to connect it to the lowerarm while not parented to it)

[timid dove]

No, it won't because the weapons would not be attached to the kinematic thing

Again, this would be a separate child object

[timid dove]

Yes now add a child object to Hand with a kinematic Rigidbody and a collider on a separate layer

[timber depot]

still not working

the weapon attached (using a joint) is still not able to affect the ragdoll

[eternal kettle]

they seem to avoid each other while moving, but when they actually get close to the target they dont anymore

any idea why that would be?

im guessing its that my target isnt an agent so they dont avoid it?

[severe moth]

I'm not exactly sure what you expect to happen here, if they all try to get to the same target then there is no way to avoid eachother if you just tell them to keep going to the target

[upbeat sandal]

Is circlecast more optimize than circle collider ? I just need to trigger colliders (entities and obstacles)?

[inner thistle]

They do different things so they're not comparable

If you mean overlapcircle then using the normal collider system is probably better because any manual checks come on top of the base physics system

Generally the answer to "is it more optimized to do manually something that the built-in physics already does" is almost always "no"

[upbeat sandal]

And to make bounce but with a trigger on my collider?
I have a bullet that calculate new direction to bounce because my collider is in trigger
And I saw with many many bullets, it takes a lot to CPU

[inner thistle]

If it needs to bounce shouldn't it be a normal collider?

[dusk kettle]

hi everybody! is thera a way to "draw" a collision layer over my tileset? reasoning:

• i set a layer of collidable objects
• everything drawn on that layer is collided with

which works just fine. the tilemap collider 2d though doesnt take stuff like behind trees into account or edges of houses where i'd like to walk behind. in godot we could paint a layer onto it - is there something like this for unity as well?

or if if not is the only option using another layer?

[timid dove]

the tilemap collider 2d though doesnt take stuff like behind trees into account or edges of houses where i'd like to walk behind
The collider shapes of your sprites can be customized from the sprite editor

[dusk kettle]

thats perfect will take a look at it, was looking at the wrong place then 😄

[timid dove]

https://docs.unity3d.com/6000.4/Documentation/Manual/sprite/sprite-editor/custom-physics-shape/custom-physics-shape-landing.html

[simple cloud]

I am trying to simulate the behaviour of a sail boat but I am struggling at the listing/heeling of the boat. I want that the lift(already calculated dont bother with it) pushes the boat down and the weight of the keel pushes it back up until it balances it out like in the picture. My idea:

    private void HandleListing(Vector3 lift)
    {
        Vector3 fresult = lift * Mathf.Abs(1 - transform.localRotation.eulerAngles.z/90) - listForceMulitplier * Mathf.Abs(transform.localRotation.eulerAngles.z/90) * lift.normalized;
        rb.AddForceAtPosition(fresult, tellTale.transform.position, ForceMode.Force);
    }

tellTale is the highest point of the boat

Heres a video of the current situation(its really weird and I don't know why it doesn't work (its wiggling too much)):

If anyone has an idea for the correct formular please tell me!

[timid dove]

I'm going to be honest your diagram and even your video are confusing me with the perspective and I'm not sure what is supposed to be what boat part. One thing I will say though is that relying on Transform.eulerAngles for your physics calculation is likely not to work well

[spark moat]

my character controlelr is bit too sticky?

and yeah i do have gravity added in character controller

[pulsar canopy]

Why don't configurable joint axis limits work? They only work for me when I make the min and max X 0.

Whichever other value I input, it seems to rotate freely as it pleases, whether I put min as negative and max as positive, or one of those as zero and the other as an either negative or positive number

[severe moth]

They work for me just fine and I'm using them everywhere in my game. Did you change any other setting, and are you sure you're editing the correct axis? You can visualize them by pressing the little button - only X has min and max, Y and Z has only one number.

[pulsar canopy]

Yeah I was. It might have been the friction cause it helped when I removed friction

But imagine a joint that works as if it were a head on a body

Even if the head'x x is restricted, when it hits a ceiling, it will rotate fully oout of the way

[severe moth]

Are the other axis limited?

[pulsar canopy]

That was at least partially due to friction and a sphere collider cause lowering friction helped but it still exists

[pulsar canopy]

All of them are limited

[severe moth]

Unity joints are not known for being the most stable, so if you forced them into a situation where they have no other out, then they will still rotate out of the way

especially if they are attached to a kinematic object that isnt being moved by applyign forces

[pulsar canopy]

It is being moved by forces and I have a low snapping limit (forgot the name but the distance where it snaps back) so it shouldn't happen but I made it work for my purpose

Not exactly as I wanted but what can you do

Joints are weird

[severe moth]

Whats the mass ratio for head to the rest?

Because that can make it even worse if the mass of the head is very small, then it doesnt take much to get into a situation where it's forced to move

[strong barn]

Hello people, I have an ArticulationBody hierarchy, and one of the nodes needs to be able to rotate within certain constraints, but any time the targets of the drives are bigger than approximately 120 degrees, the system becomes extremely unstable and starts spinning out of control.
Is this some known limitation of the articulation bodies? If so, is there any workaround? If not, is there something I could be doing wrong in the configuration of the articulation body?
(Unity 6.0)

[pulsar canopy]

it is small though. I have increased it since but that fucks with the rest of the calculations and forces so I changed it back

[strong barn]

I'm completely stumped with this. I don't know why "big" angles matter so much to it.
I made a new project in Unity 6.2, and I get the same exact situation...

[severe moth]

I'm not that familiar with articulation joints, but have you tried changing the drive type?

try changing it to Target if that exists

otherwise changing dampening might help too

[strong barn]

it does exist, but it disables stiffness and dampening, and I need this articulation body to react to environment physics (otherwise I could just use a hierarchy of kinematic rigidbodies and I wouldn't worry about articulation forces 😅 )

[untold shuttle]

Hello, I have an issue with my 2d top down game, I have a Tilemap Collider 2D set up on my wall tiles, works fine in editor, but in the build, they are passthrough, I can go through them, bullets too. I have looked everywhere to fix this, tried a bunch of things, verified build settings and player settings so that the layer matrix settings allows for the layers to interact, objects are static, the tile asset is read/write enabled, I am not sure what else to try anymore, anyone has any idea why they don't work? Normal colliders work both editor and build, but not tilemap collider. Hope this is the right thread

[timid dove]

Can you show the inspectors for your player and your tilemap as well as your player movement code?

[untold shuttle]

Hope this is enough, thanks for checking. This is happening in FixedUpdate, deltaTime is actually Time.fixedDeltaTime

[timid dove]

the full script would be better - but setting the velocity is almost certainly what you want instead of MovePosition

also, you should be using a CompositeCollider2D with that TilemapCollider

[untold shuttle]

I tried that as well, is there any other settings to add to it?

I'll add it now again to make sure and to have it

[timid dove]

you need the TlemapCollider2D's "Composite Operation" set to "Merge"

[untold shuttle]

oh I see, I'll make a build with this, won't take long and I'll see if it works, cheers

this should also fix a different issue where running into the wall wouldnt be smooth(because of the individual colliders), I have tried using merge or other settings, but I didn't know about the CompositeCollider2D component until today where I also saw it being mentioned in different posts

[untold shuttle]

drum rolls it works!, thank you very very much!

been bugging me since I started the project but now I needed to kinda show it so this saved it, thanks

[maiden wolf]

hi, i'm trying to make a VR multiplayer game with a focus on physics, what soft body physics addon would u guys recommend? i want one that lets me choose what polygons on the mesh can have soft body physics in the first place and how squishy/stretchy/elastic they can be, and of course i want it to interact with collision, ragdoll, dynamic bone, and fluid physics

[elfin latch]

obi soft body

very good asset

[halcyon sentinel]

Hey! I want to make a mesh follow a spline with momentum buildup, think how when a rollercoaster cart travels on a downwards slope, it speeds up, and the opposite for an uphill slope. I'm not sure where to start, I have no idea how I'd make this, which is why im here. Thanks

[timid dove]

You'd want to basically model this based on the equations for kinetic energy vs gravitational potential energy

In essence that can let you derive a pretty simple formula for velocity based on the current height (and initial height)

[halcyon sentinel]

ah so like a lerp?

[timid dove]

not really

[halcyon sentinel]

damn

i havent done kinetic or gravitational energy in school yet

[timid dove]

like a functino:

float GetCurrentVelocity(float currentHeight) { ... }```

the formula, which I used Gemini to derive, would be this

hmax being the "starting height" where your velocity was 0 and h being the current height

and g being the acceleration due to gravity (typically 9.8 m/s^2)

[halcyon sentinel]

so g is the gravity scale?

like the one you set in project settings > physics, right?

[timid dove]

It's the gravitational acceleration you set in project settings yeah

I mean it doesn't have to be what you set there

but it can be

[halcyon sentinel]

its like -9.81 in there right?

yeah ofc

[halcyon sentinel]

can you elaborate on this please?

im dumb as hell lol

[halcyon sentinel]

oh wait so itd be get current velocity, then it returns that equation?

[timid dove]

I'm just showing you an example function signature for you to write the math in

[timid dove]

the function would perform the calculation.

An AI tool would help you write the actual code if you want, and also to refine that formula to include things like a starting velocity, which you'd need to make it actually move at all

[halcyon sentinel]

nah no ai

im tryna learn all this stuff without ai

[timid dove]

Well you'd just do that math - but in code

but the formula does need to be adjusted for an initial velocity as I mentioned

The general formula with initial starting velocity and a starting height and current height is

[timid dove]

that's "v1 squared", i.e. "initial velocity squared" under the sqrt

[halcyon sentinel]

and then just apply it to the speed, right?

[timid dove]

well you need to feed in the current height and it will give you the speed out

[halcyon sentinel]

sorry didnt see this message

[timid dove]

and then you'd probably feed that into a spline follower to get it to move along the path

[halcyon sentinel]

yeah i forgot there was the currentHeight local var

and ive got a spline animate to move the playert

and then i guess i just set the speed according to the current vel?

[timid dove]

once you have the velocity from the formula then you'd have to multiply that by deltaTime to get the amount you expect to move this frame, then move along the spline that distance

[timid dove]

if your spline follower has a speed yeah

you'd adjust it every frame with the new calculated velocity based on the height

I recommend doing it in FixedUpdate or it won't be consistent

although IDK how SplineAnimate works

[halcyon sentinel]

so v1^2 = new velocity?

[timid dove]

#⚛️┃physics message

[halcyon sentinel]

well its either that or time.deltatime

i mean new velocity

v2 = new velocity

right?

[timid dove]

yea

[halcyon sentinel]

aight cool

[timid dove]

This all assumes zero friction/air resistance etc

you could add friction later if you want it

[halcyon sentinel]

yeah ofc

im not sure how this will play with my other ideas

because i want the player to be able to launch from the current rails and then snap to another rail

which would require a rigidbody or insane hardcoding

but thats a future me problem ig

so it kind of works, but it spazzes out when going uphill

grav = 9.81
_initialVelocity = 2

[timid dove]

I'm not sure how SplineAnimate and MaxSpeed works

SplineAnimate might not play well with the FixedUpdate stuff

in which case you'd either switch to Update (which will lose consistency) or do your own animation in Fixed

Just kinda guessing that might be the problem, not sure

[halcyon sentinel]

maybe

ill give it a try

nah still does it

idk man ill sort it out

thanks for your help

[thick creek]

Not 100% if this is the right spot or not to ask but Im making a 2D sprite based fighing game in a 3D space. The sprite's transform scaling was 0.2 so I scaled it up to 1 so it wasn't so small then it really messed up the forward and jump height. I ofc adjusted these accordingly but the jump fall speed is really slow like Im on the moon and no matter what I do, the walk and run speed dont seem to change at all. Does anyone know if they might know why this could be happening?

Was having the sprite's transform scaling be 0.2 the right thing to do and I chnaged it for no reason?

[timid dove]

your question is really vague

provide hard details - videos, screenshots, code

jump fall speed is really slow like Im on the moon and no matter what I do, the walk and run speed dont seem to change at all
These are almost certainly issues with your code

[thick creek]

Even if I did nothing to change the gravity or drag?

[timid dove]

Code can cause literally any imaginable behavior

This is almost certainly an issue with your code.

[spring vale]

Hello, does Unity Physics require knowledge in mathematics or physics?

[timid dove]

Really sort of a vague question. Knowledge of math and physics will improve your experience with Unity physics (or any physics engine) greatly.

does it "require" such knowledge? The code will run and do the thing it's meant to do regardless of your knowledge level.

[spring vale]

I have never experienced coding in a physics system because I do not know mathematical calculations. Seeing the Unity physics sample has made me curious and prompted me to ask these questions.

[timid dove]

Let me put it this way - anyone can walk up to an easel and paint something. But having a mastery of color theory, experience in arts and brushstrokes and lighting and shading will let you create a masterpiece.

It's the same way with a game engine/physics engine, etc...

The physics engine will simulate physics, realistically. You don't need any knowledge of physics to make it do that. But you will need some knowledge to make a game that interacts with the physics engine in a desired or gameplay-friendly way.

[spring vale]

That is good advice, really appreciate it.

[severe moth]

In my opinion, you will def. want to know about vectors and vector operations; cross product, dot product etc. At least I have used it in almost any heavily physics based game/project. They are also pretty easy to understand, so it's a good starting point and it opens up to a lot of things you can do with the unity physics engine, and it's also very useful outside of the physics engine.

[proud trellis]

if anyone is ever trying to figure out how to set target rotations for configurable joints, this is great 👍 https://gist.github.com/mstevenson/4958837

[normal girder]

It just seems like the only change here, was your perception of movement through space, and having code that took into account a minimal scale, that's now having to run on a bigger scale. Think like this, an ant jumping 10cm, it's a big jump, a person jumping that same amount, barely can be considered a jump.

[pulsar canopy]

Why don't I have a "connected body" field on a character joint? Wtf?

[severe moth]

What do you mean?

[thick creek]

Yeah someone sorta explained this after and I basically just reverted it all back to 0.2 scaling. I thought having it as 0.2 would come back to bite me in the ass or something so thats why I even made the scaling change despite everything working as intended

[normal girder]

Well, you could just have everything on scale 1, and multiply all the fixed values you were using by 5, that would make it so that you wouldn't be using micro movements throughout your code. But it all depends on what you're really going for.

[thick creek]

Idk if that couldve worked

I mean gravity was messing with too much so I just reverted everything

wait if I multiplied gravity which eas 9.81 with 5 would it just be exactly how it was when the rb scale was 0.2?

[normal girder]

kind of, but I wouldn't recommend that. The whole idea of keeping things at scale 1 as much as possible, is so that you can use default math, instead of having to adapt everything you add to your game, to having small values.

[torn sparrow]

does raycast hit the terrain if its below it?

[tender vapor]

it'll hit if the ray passes through it

[torn sparrow]

i think it doesnt, cause i replaced it with a raycast from above and it worked

[timid dove]

No

[mild saddle]

guys i have an issue when I hit a wall, my character gets pushed back a little when I turn the camera

is it because of character controller parameters or a controller script

its happnening not always

but sometimes

i just catched a moment of this

at the fifth second

[tender vapor]

are you using a rigidbody or something with the character controller?

[mild saddle]

only character controller

and scripts ofc

also im getting pushed back when i jump into the wall

[cursive hill]

Does unity have prebuilt options to work with water simulation?

I am trying to simulate realistic water

[timid dove]

This is pretty vague

That could mean anything from having a realistic looking body of water to a plumbing simulator

So what are you trying to actually accomplish?

[cursive hill]

I am actually doing both of what you said, small water as in cups and entire lakes

[timid dove]

But can you elaborate on the "realistic" thing?

Is it just a visual appearance thing? Are you going to have boats that need to behave properly on them or something?

[cursive hill]

It will be realistic as in the physics, that is what I am focused on

[gloomy lodge]

@paulcooper4432 muted

Reason: Too many messages in multiple channels
Duration: 29 minutes and 54 seconds

[upper vault]

trying to create Ragdoll in Unity 6.0 version and my character humanoid mesh is from mixamo and i have no idea how do i setup.

[runic torrent]

Given
•The total time of a jump (rise and descent) is 0.129s
•The gravity of the descent is 169.52m/s^2

How do I calculate the gravity of the ascent?

[tender vapor]

depends on the height or time of ascent or time of descent

[runic torrent]

That's just it. I don't know what those are

[tender vapor]

there are infinite ways to achieve what you want with just those 2 constraints

you need another constraint if you want a single value for the gravity of the ascent

[runic torrent]

All I know is the height is not above 4

@tender vapor If I told you that the horizontal length of a jump with these parameters was 1m would that help?

[tender vapor]

horizontal distance doesn't affect it at all, unfortunately

you need another constraint related to the vertical movement - either time or distance

or even starting vertical velocity

[runic torrent]

Oh I do have that one

[tender vapor]

i think that'd work, but it's kinda late for me

[runic torrent]

Starting velocity should be 11.8421m/s

[tender vapor]

hmm, might need another constraint...

t = 0.129 s
u = 11.8421 m/s
g2 = -169.52 m/s^2
solve for g1

t1 + t2 = t
h = ut1 + 1/2 g1 t1^2
h = - 1/2 g2 t2^2
ut1 + 1/2 g1 t1^2 = - 1/2 g2 t2^2

g1 = -(g2 t2^2 + 2ut1)/t1^2

needs either time of ascent or time of descent

or height would also work

so.. just back to the variables i said initially...

i didn't think that'd be 100% accurate lmao

hmm wait

u + g1t1 = 0

chucked it into wolfram alpha and got 2 solutions
g1 ≈ -188.636, h ≈ 0.371709
g1 ≈ -44.6738, h ≈ 1.56955
https://www.wolframalpha.com/input?i=solve+for+h+and+g_1%3B+t%3D0.129%3Bg_2+%3D+-169.52%3Bu+%3D+11.8421%3B+t_1+%2B+t_2+%3D+t%3B+ut_1+%2B+1%2F2+g_1+t_1^2+%3D+-+1%2F2+g_2+t_2^2%3B+u+%2B+g_1t_1+%3D+0%3Bh+%3D+ut_1+%2B+1%2F2+g_1+t_1^2

@runic torrent does that work for you

[runic torrent]

I'll try and double check the values later to see if they work

Thanks, though

[rose pebble]

Does anyone know how to make the cube land better on the button? What can I do with the collision, Rigidbody2D, or anything else so that cube lands properly on the button and doesn’t get stuck on small gaps in collisions or on the floor?
The cube and the button both have 2D collisions and are not set as triggers. The collisions and the situation are shown in the video and photos.

[tender vapor]

is the floor using a tile collider?

if so, have you applied a composite collider

[rose pebble]

floor have booth. But some "special" parts of floor have separatet box collider 2d

but those parts have the same size

[tender vapor]

make sure the tile collider is set to use the composite

[rose pebble]

literally pixel perfect

[tender vapor]

that might not matter. are they the parts catching the edges of the box?

[rose pebble]

no

[tender vapor]

alright, so not a concern for now

could you show the floor collider

the inspector for the tile collider, specifically

[rose pebble]

acutally i fixed the stopping on the flore. But there is still an issue with the button collision. Sometimes the cube slides smoothly onto the button. And sometimes it stops in front of it, as if the button were a wall.

[tender vapor]

probably because it's getting caught on its corner?

could consider making the button a trigger

[rose pebble]

Are there any values ​​in 2d unity that can change to make the cube better follow the curve collision?

[lucid storm]

Does anyone who understands how a climbing system works in 3D help me out? I'm following a tutorial by
Sahani Studio. My character is 2.5 units tall, and everything seems to work except the braced hop up animation, it either glitches me through the ground and ledge/wall, or it just doesn't hop up at all. I believe this to be an offset issue with my animations in correlation to my "Hop up offset" variable. Can someone please help, I have to finish this for a school project.

[tender vapor]

use mp4 to embed in discord

[runic torrent]

@tender vapor So what were the equations you used from yesterday?

[tender vapor]

kinematics
s = ut + 1/2 at^2 (once for the upwards portion, once for the downwards portion)
u + at = v = 0 (vertical velocity will be 0 at the apex of the jump)

[runic torrent]

So I remember u is supposed to be initial velocity. I take it t is total time? I'm not sure what the other variables are, though

[tender vapor]

i had 2 separate t values for the 2 segments

a is acceleration

[runic torrent]

How did you get those two t values, though?

[tender vapor]

i solved for them

had 4 equations, 4 unknowns

[runic torrent]

Ok but how?

[tender vapor]

unknowns: t1, t2, g1, h
knowns: t, u, g2
equations:
t1 + t2 = t
h = ut1 + 1/2 g1 t1^2
h = - 1/2 g2 t2^2
u + g1t1 = 0

[tender vapor]

well i did a bit by hand, and then i gave up and chucked them into wolframalpha

[runic torrent]

So you don't actually know how to isolate any of the unknowns?

[tender vapor]

well, i tried. but it was 3am and i didn't feel like working on it for very long

it's just math at that point

[runic torrent]

Well, I know it's at least possible now. I'm just not sure I know how to do it

I take it with one of the h equations I get one of the sides equal to (u +g1t1) and set it equal to zero?

[tender vapor]

i don't think the units work out there

h is in units of m, v is in units of m/s

you can't just look at the values

[runic torrent]

No, I mean I take the entire equation of (h = ut1 + 1/2 g1 t1^2), get one of the sides to be (u + g1t1), then replace that side with 0. But I'm not really sure how I would do that either

[tender vapor]

oh, no clue tbh

i eliminated h by equating the 2 halves

[runic torrent]

So ut1 + 1/2g1t1^2 = -1/2g2t2^2? What then? There are still like 3 unknowns in this

[tender vapor]

move stuff around until they cancel, i guess

you're basically asking me how to solve a system of equations

[runic torrent]

Do they even cancel?

[tender vapor]

apparently enough for wolframalpha to get answers

a non-dependant system of n equations with n unknowns generally can be solved

and it was degree 2, so it had 2 sets of answers

[runic torrent]

So when you say "cancel" you mean stuff like substituting (t1 + t2) with t when it crops up, right?

[tender vapor]

sure?

mate what are you trying to get from me

you can google for how to solve systems of equations if you want to know how to do it yourself

[bitter scarab]

Does anyone know how to fix this issue. For some reason when I put character joints on objects the physics are really slow for some reason idk why (sorry I can’t post images but I don’t think you need them as the settings are the default ones)

[unique cave]

Screen recording (video) of the problem would be really helpful in trying to see what is going on

[bitter scarab]

I’ll try in like 2 hours as I’m not near my pc rn

[unique cave]

Is there anything acting on the character parts? Like animations, scripts etc.? What is the scale of the character? I assume the character joint is configured by default for human sized characters. Smaller characters might require smaller damping etc.

[bitter scarab]

i increased the mass and thats so it's a bit faster

[unique cave]

From what I remember, the mass shouldn't matter much if at all. The only thing the mass often affects is the way it interacts with objects with different masses. What I'm trying to say is that often 2 objects interacting happens the same if both weigh 1 or 100 but if one weighs 1 and the other 100 it will make a difference

I'm not sure though if the character joint does something with the mass (spring and damping for example might be mass dependent)

[elfin latch]

do you have colliders on it?

it looks like a bad inertia tensor calculation

[meager valve]

why do I bounce with 0 bounce and bounce combine set to minimum?? why no friction with 0.99 frictionnnnnn

oh wait it was on multiply??

oh

nvm

oh nevermind IM STILL BOUNCING??

[meager valve]

is there a good way to apply friction on my own? with code? I want to make it so I slow down faster when I stop walking

[bitter scarab]

No

[elfin latch]

that’s why

it’s not able to calculate a inertia tensor since there’s nothing to calculate from

https://docs.unity3d.com/6000.2/Documentation/ScriptReference/Rigidbody-automaticInertiaTensor.html

[urban needle]

Hi, I got a stupid question.
If a rigid body is kinematic and I set the angular velocity to something not 0. Should it rotate ?

[timid dove]

3D rigidbodies will not
2D rigidbodies will

[urban needle]

I assume that means there is no universal truth to expect. I'm asking because in ECS Physics setting the velocity of a kinematic body will move it...

[timid dove]

Within each physics engine, it is consistent

but each physics engine does things its own way

[urban needle]

ok, I wonder how that's gonna play out with the swappable physics backend they announced 🤔

it also means I just can't rely on the velocity to move my kinematic platforms and I didn't even get to the nesting of kinematic bodies yet.

[timid dove]

Isn't that just the setting in project settings that lets you swap?

You will still need to design your game around the quirks of the engine you choose

[urban needle]

yes but I'm making a package and I can't assume the behavior of the physics engine since each may or may not behave coherently with my assumption.

[timid dove]

Yeah welcome to the hell of being a package/asset developer

You either:

• Painstakingly support all possibilities
• Add a disclaimer that you only support physics engine X/Y etc

[urban needle]

ok, thanks for clarifying these points.

[sick kayak]

https://discussions.unity.com/t/physics-overlapsphere-stops-working-after-about-an-hour/1697829

Any help would be appreciated

[full palm]

Any ideas how to implement this using Box2D API that Unity provides?

Seems like best solution of Ghost Collision issue in platformer with OnewayPlatforms, MovingPlatforms, etc..

https://briansemrau.github.io/dealing-with-ghost-collisions/

[timid dove]

Unity provides CompositeCollider2D for this

[full palm]

Yes, I understand that this can be solved with a Composite Collider in Outline mode. But I shared that post because the author gives an example where, in a platformer, you have a situations like that

Maybe I’m not aware of alternative options, but I specifically need to have two separate platform objects, each with its own collider, where one of them has a certain component attached and the other one does not

[timid dove]

I see - yeah I'm not sure.

[full palm]

I tried get contact area using _collider.GetContacts(filter, contacts); and calculate distance between.
But it works weird and requires a lot of exceptions. I think there must be better solution.

[timid dove]

well if it has contacts that means it already collided right? Which is too late to stop the collision I would think?

Or, I guess there will always be a collision it's about the impulse it creates

[full palm]

Yeah, so thats why Im looking for something like presolve func

[bold jetty]

So, in my game my character can move between two different elevation levels. (Top down view, picture jumping up onto a crate, or walking on the floor vs. walking on top of a bookshelf) When the character hops up or down, it would allow the character to walk on this new level, but the character would be blocked from walking on the previous level and vice versa. It seems like the way to do this is a bunch of colliders that turn on/off depending on what level the player is on. Is the best way to do this, adding a lot of colliders and have them turn on/off or is there a better method? A golden sun puzzle is a good comparison: https://www.youtube.com/watch?v=Wb1ASaKkMfU&t=190s

[timid dove]

Holy shit Golden Sun, what an amazing game that was

[short merlin]

My FirstPerson character uses a character controller. A box with box collider is parented to it. Is there a way to prevent my box going through other objects?

[silver moss]

I dont think character controller supports extra colliders

Probably better off using rigidbody physics

[timid dove]

CC doesn't support any extra colliders besides its own capsule shape

[short merlin]

Then I'm gonna have to move the logic to rb

[signal wyvern]

Can anyone help, Im trying to modify some contacts, as i have some specific collisions i want to modify, but im somehow unable to get the below ifs running, During runtime im adding some InstanceIds to a hashset & im comparing them to the modifiable contacts, but despite the collisions i still cant seem to get a match on my stored InstanceID & the instanceIDs from the contactPair.

` private void IgnoreInvalidForces(PhysicsScene scene, NativeArray<ModifiableContactPair> contactPairs)
{
for (int i = 0; i < contactPairs.Length; i++)
{
ModifiableContactPair contactPair = contactPairs[i];
//Debug.Log("main = " + contactPair.colliderInstanceID + " Other = " + contactPair.otherColliderInstanceID);
//LogIgnoredIDs();
if (PropManager.IgnoredRBIDs.Contains(contactPair.bodyInstanceID))
{
Debug.Log("Made it here as expected");

            if (PropManager.KinimaticPropIDs.Contains(contactPair.otherBodyInstanceID))
            {
                ModifiableMassProperties newProperties = new();

                newProperties.inverseMassScale = 1;
                newProperties.inverseInertiaScale = 1;

                contactPair.massProperties = newProperties;

                contactPairs[i] = contactPair;
            }

        }
        else if (PropManager.IgnoredRBIDs.Contains(contactPair.otherBodyInstanceID))
        {
            Debug.Log("Made it here2");
            if (PropManager.KinimaticPropIDs.Contains(contactPair.bodyInstanceID))
            {
                ModifiableMassProperties newProperties = new();

                newProperties.otherInverseMassScale = 1;
                newProperties.otherInverseInertiaScale = 1;

                contactPair.massProperties = newProperties;

                contactPairs[i] = contactPair;
            }
        }
    }
}`

Any physics gurus with experience with contact pairs here?

[timid dove]

!code

[gloomy lodge]

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.

[mental thistle]

Is it normal that collider ignores horizontal checks (I don't know how to properly form a question, it's a bit hard)?
I'm trying to make interaction system, allowing player to pick objects up and move them. Because I don't understand how to forbid player to push object through walls, I tried to make a "CheckBox" that would check for collisions in "carrying" area. Well, I wrote some OnTriggerEnter/Stay/Exit funcs, and for some reason, trigger won't detect when I collide it with the wall. With the floor it works (at least sometimes).
I think this problem is somehow linked with Physics Engine.
Here's code (idk why you would need that, it's the most simple one)

public GameObject CarriedObject { get; set; }
public bool CanDrop { get; private set; } = true;
private void OnTriggerEnter(Collider other)
{
    // if(CarriedObject is null) return;
    print("Enter");
    CanDrop = false;
}

private void OnTriggerStay(Collider other)
{
    // if(CarriedObject is null) return;
    print("Stay");
    CanDrop = false;
}

private void OnTriggerExit(Collider other)
{
    // if(CarriedObject is null) return;
    print("Exit");
    CanDrop = true;
}

[tender vapor]

Is it normal that collider ignores horizontal checks
no, that's probably not what's happening though

how are you moving it?

[mental thistle]

with the cam I guess

[mental thistle]

is that could be the issue, because I use CC too?

[tender vapor]

is it just moving as a child of the camera

[mental thistle]

yes

[tender vapor]

that'd basically be moving via transform, which doesn't work well with physics afaik.

actually, that's probably not the issue here. you'd be getting collisions but inconsistently

does the Cube have a rigidbody?

or the ObjectCarryPoint

[sand hamlet]

Yes

[mental thistle]

No

[tender vapor]

does ObjectCarryPoint have a rigidbody

[mental thistle]

No

[tender vapor]

so there's nothing to drive the collision

rigidbodies are what drive/detect collisions and triggers

[mental thistle]

oh

[tender vapor]

presumably your floor has one, that's why it can detect the trigger

[mental thistle]

I'm so sorry😭

[tender vapor]

https://unity.huh.how/physics-messages/trigger-matrix-3d

[mental thistle]

no, it has not

[tender vapor]

this is probably still something you should look into

[tender vapor]

does Room have a rigidbody

[mental thistle]

no

[tender vapor]

it's either Room, Plane, or there's another object with a rigidbody in that space

try searching t:Rigidbody in the hierarchy search bar

[mental thistle]

only three interactables have it

[tender vapor]

then it's hitting one of those

[mental thistle]

Thank you, thing is perfectly working now

[uncut ember]

Hello everyone ! I'm trying to figure out the best way to flip horizontally a Hinge Joint 2D without having it go crazy and also keeping the proper limits

This is a simple example.
If I try and rotate the gameobject (these are parented) it breaks up and if I try to do Scale X -1 , it works but the limits do not really follow (see 2nd picture)

Anyone has any ideas ?

(for some irrelevant reason I tried separating the sprite, i don't think it matters though)

(probably tried rotating only that but the hinge's limits are not symmetrical to just do that)

[signal wyvern]

Simple question: when using Addforce() in fixed update, is velocity changed immediately? Or does it wait until next fixed to update velocity?

[remote creek]

they are applied after subsequent physics update

[severe moth]

Should be in the same step, but you can easily veryify it by printing the velocity before and after

[timid dove]

In 2d it's immediate.
In 3d it gets stored until the physics update and you can read it from GetAccumulatedForce()

[tender vapor]

for each physics tick, FixedUpdate happens before the "Internal physics update", so effects should happen within the same tick

https://docs.unity3d.com/6000.2/Documentation/Manual/execution-order.html

[timid dove]

Yes within the same tick. But "immediately" implies to me that if you read .linearVelocity before and after the AddForce call it would change, which it doesn't

[tender vapor]

to me the question read as "this fixedupdate or next fixed update" given the last sentence, but 🤷

[signal wyvern]

This is helpful thanks, i done some testing to confirm anyway, I just wanted to know if i can read the velocity the line after adding the force

[timid dove]

yep - as someone who developed a unity asset for adding speed limits to Rigidbodies, I got very deep into the nuances of this.

[remote creek]

Someone knows why my car is doing that?

[noble crown]

what's the best way to dynamically fracture a mesh at runtime?

[noble crown]

seems like an issue with the joints, i'm not 100% sure though

[crude folio]

can i see how the setup is or how the wheels are parented in the hierarchy?

[crude folio]

Hmm, the wheel colliders are on the wheel meshes.

Try this way.

Create an empty game object for the wheels then create empty child game objects (one for each wheel) that would hold the wheel collider component, then make the wheel meshes the child of their respective wheel collider game objects.

[remote creek]

alr ill try that

[crude folio]

Alright.

Also, the wheel collider game objects should be where the respective tires should be (the position).

By the way, I can send the link to the kart I use in the scene of you want, so you can use it as reference.

[remote creek]

That would be so helpful

[crude folio]

Here it is.

[crude folio]

Cool, have you tested if it works?

[remote creek]

Does not work yet, the car goes brrrrr Going to fix it now will update soon

[crude folio]

Okay...

[crude folio]

@remote creek how's it going?
Figured it out yet?

[remote creek]

I actually closed the project for later. Right now I'm working on some ProBuilder modelling

[brave valve]

Hello, everyone. Can anyone advise or help me solve my physics problem? I am trying to create platformer physics and have added several functions for detecting the floor and obtaining its normal. Now I need to smooth the transition to the new normal so that it is not abrupt. I did that, but the problem is that when moving across the Terrain object, the script forces it to follow every normal, which is not very smooth in terms of degrees. This makes it seem like the player is ‘jittering’ when in fact they can still switch abruptly to the new normal. I tried Lerp/Slerp, Cross, Dot, but I couldn't connect any of them together to make it possible to walk smoothly on Terrain. By the way, here is my player hierarchy. Maybe someone knows how to solve this. I would be grateful.

[tender vapor]

see the "large code blocks" section below, and send your video as mp4 so it embeds in discord

[gloomy lodge]

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.

[quartz panther]

Hi, im trying to make a ship simulator, when the ship starts moving it causes weird oscillations, can anyone help to fix that?
I have added 11 float points across the ship symmetrically.
The ship doesn't interact with the waves currently, the water is just a plane object.
I will attach the buoyancy and shipController scripts below:
https://paste.mod.gg/vwtsryhsbxrm/1

[dark wren]

Testing out 6.3 low level physics. Anyone have any guesses why these circles are colliding with each other? I thought I copied over everything relevant from the ContactFiltering sample

[tender vapor]

what's calling OnContactFilter2D?

also, configure your ide

!ide

[gloomy lodge]

💡 IDE Configuration

If your IDE is not autocompleting code or underlining errors, please configure it.

Select one:

• Visual Studio (Installed via Unity Hub)
• Visual Studio (Installed manually)
• VS Code
• JetBrains Rider
• :question: Other/None

[dark wren]

in the sample it just goes through the physics system after setting it up in awake, nothing explicitly calls it

Oh, apparently there was a PhysicsCallbacks.IContactFilterCallback interface that they use, but after moving it it still doesn't work for me

this is the class I'm basing it off of

[tender vapor]

also i'm noticing that you're using OnContactFilter__2D__, and there's also a PhysicsShape__2D__, but you're using a PhysicsShape?

[tender vapor]

!code

[gloomy lodge]

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.

[dark wren]

Sorry

[tender vapor]

ah they're part of different namespaces

[dark wren]

In the sample code the shape assignments resolve to PhysicsShape

[tender vapor]

make sure you're using the right namespace

the 2d and 3d versions in lowlevel are differentiated by namespace name rather than class name

[dark wren]

I'm using UnityEngine.LowLevelPhysics2D

[dark wren]

also yeah I had it set up but it broke recently and I haven't been able to figure out why lol

[tender vapor]

ah, the method is part of an interface

[tender vapor]

check the bottom right for the C# {} thing, see if it has errors.
perhaps try checking the output tab for the .NET install tool

[tender vapor]

that first one seems about right

i'm probably misremembering something about how .NET install tool works

what about output for the c# extensions?

[dark wren]

ah, I guess I was out of date, I updated VSCode and it started working

[obsidian patio]

your damping is inside the foreach loop
so if 8 floaters are underwater thats 8x more damping than it should be, way too much
move it outside:

foreach (var f in floaters)
{
if (!f) continue;
Vector3 p = f.position;
float depth = waterLevel - p.y;
if (depth <= 0f) continue;

float sub = Mathf.Clamp01(depth / maxDepth);
float displacedVolume = floaterArea * depth;
float forceMag = waterDensity * g * displacedVolume * sub;

rb.AddForceAtPosition(Vector3.up * forceMag, p, ForceMode.Force);
}
// once, not 11 times
rb.AddForce(Vector3.down * rb.velocity.y * 5e5f, ForceMode.Force);
```or if you want per floater damping for better roll/pitch just lower the value:

Vector3 pointVel = rb.GetPointVelocity(p);
rb.AddForceAtPosition(Vector3.down * pointVel.y * 50000f, p, ForceMode.Force);

[dark wren]

any other ideas on this?

[tender vapor]

no, sorry

[dark wren]

blast

thanks for your help 🙏

[potent herald]

Hi, I am currently creating a game shown in the video where you rotatie a cube on the outside to get the blue cube on the inside out through the hole. Everything works usually correctly, but every time you use rapid movement or flick the mouse really fast to move the cube, the cube just clipps out of the cube and falls out without going through the hole (as shown in the video). I do have a failsafe check, but I feel that is kind of a temporary fix and would rather just solve the whole clipping situation in the first place. the cube object in the hierarchy image shows the parents and children and the rotating script is in the cube parent object. All of the planes are the children inside the body inside the cube have box colliders and a rigidbody shown in the picture that looks like the image of the inspector with the box colliders and the rigidbody. The blue cube in the middle has the interpolate setting for the rigidbody pictures out of the two of them. The script i use to rotate the cube is in the parent cube, which has no rigidbody or box colliders, and you can find the script in the message below. I would like to see if anyone has a solution, whether it is changing something in the script, or the settings in either rigidbodies, or something to do with the parent or children. I would appreciate the help, we can also talk in DMs. The Unity Version is 2019.4.40f1.

rotating cube script:

[safe reef]

the cube you're rotating should be rotated via its rigidbody to avoid it

[potent herald]

So should I just add a rigidbody to the parent instead of all of the walls?

[potent herald]

Alright, I did some tweaking with the colliders and the issue seeems to have been solved.

[quartz panther]

I changed the script like you said but it's still oscillating..

void FixedUpdate()
    {
        if (floaters == null || floaters.Length == 0) return;

        foreach (var f in floaters)
                {
                if (!f) continue;
                Vector3 p = f.position;
                float depth = waterLevel - p.y;
                if (depth <= 0f) continue;

                float sub = Mathf.Clamp01(depth / maxDepth);
                float displacedVolume = floaterArea * depth;
                float forceMag = waterDensity * g * displacedVolume * sub;

                rb.AddForceAtPosition(Vector3.up * forceMag, p, ForceMode.Force);
                }
                // once, not 11 times
                rb.AddForce(Vector3.down * rb.velocity.y * 5e5f, ForceMode.Force);
    }

[tender vapor]

have you tried checking what the added force for each floater is?

it also kinda seems like you're accounting for depth twice there through displacedVolume and sub

[obsidian patio]

That damping value (5e5f) is still way too high it overcorrects and bounces back. Try rb.mass * 5f instead Also you're multiplying depth twice (once in displacedVolume, once with sub) so it acts like a spring remove the * sub part

[old abyss]

i have a general question about determinism with physics and by determinism i mean the game plays out the same on android, iOS, and PC given that theres no player input.

If, say, uve already made the entire game on android and then out of nowhere u decide that u want to launch it on iOS and PC , how would u approach the cross-platform determinism problem cuz the physics across platforms is not deterministic

[bleak umbra]

multiple options: (1) Unity.Physics (the DOTS physics component, can be used without entities), or (2) make the game such that it doesn't require determinism, if its a cloud game, (3) make the physics run entirely on the cloud, (4) make a custom deterministic fixed-point physics layer and rewrite everything to use that

[old abyss]

ill give option 1 and 4 a try. 2 and 3 are not possible in my case.
Thanks for the guidance.

[quartz panther]

oscillation start as soon i apply thrust

[quartz panther]

void FixedUpdate()
    {
        if (floaters == null || floaters.Length == 0) return;

        foreach (var f in floaters)
        {
                if (!f) continue;
                Vector3 p = f.position;
                float depth = waterLevel - p.y;
                if (depth <= 0f) continue;

                //float sub = Mathf.Clamp01(depth / maxDepth);
                float displacedVolume = floaterArea * depth;
                float forceMag = waterDensity * g * displacedVolume;

                rb.AddForceAtPosition(Vector3.up * forceMag, p, ForceMode.Force);
                Debug.Log($"Floater {f.name}: Force = {forceMag:N0} N, Depth = {depth:F2}m", f);
        }
                // once, not 11 times
                rb.AddForce(Vector3.down * rb.velocity.y * (rb.mass * 5f), ForceMode.Force);
    }

Did the changes like you said, but they're still happening

[obsidian patio]

float depth = Mathf.Min(waterLevel - p.y, maxDepth);

Use maxDepth here, otherwise force goes infinite if the object sinks too deep

Try increase the 5f value try 10f, 15f, 20f and keep whichever stops the oscillation

You only shared FixedUpdate() need to know your FloaterArea and MaxDepth values

[cyan fossil]

Hey people! Can someone help with contact point impulse?
I'm sending 2 balls into each other with same speed and getting 2 collision events. Each event has one contact point but impulse is directioned in one direction. I was expecting 2 impulses with opposite direction, but maybe I don't understand something :/

[timid dove]

can you show the code?

[cyan fossil]

Ball Spawner

public class PhysicsTestSpawner : MonoBehaviour
{
    [SerializeField]
    private TestBall ballPrefab;

    [SerializeField]
    private float distance;
    
    [SerializeField]
    private float ballASpeed;
    
    [SerializeField]
    private float ballBSpeed;
    
    [SerializeField]
    private Vector3 ballADirection;
    
    [SerializeField]
    private Vector3 ballBDirection;

    private void Update()
    {
        if (Input.GetKeyDown(KeyCode.Space)) {
            SpawnBalls();
        }
    }

    private void SpawnBalls()
    {
        var ballA = Instantiate(ballPrefab, transform.position + transform.forward * distance / 2f, Quaternion.identity);
        var ballB = Instantiate(ballPrefab, transform.position - transform.forward * distance / 2f, Quaternion.identity);
        
        var ballARigidbody = ballA.GetComponent<Rigidbody>();
        var ballBRigidbody = ballB.GetComponent<Rigidbody>();
        
        ballARigidbody.linearVelocity = ballADirection * ballASpeed;
        ballBRigidbody.linearVelocity = ballBDirection * ballBSpeed;
    }
}

Ball

public class TestBall : MonoBehaviour
{
    [SerializeField]
    private Color debugColor;

    private void OnCollisionEnter(Collision other)
    {
        Debug.Log($"{name} impulse: {other.impulse}, relative velocity: {other.relativeVelocity}");
        
        foreach (var contact in other.contacts) {
            Debug.Log($"contact point: {contact.point}, impulse: {contact.impulse}, magnitude: {contact.impulse.magnitude}"); 
            Debug.DrawLine(contact.point, contact.point + contact.impulse, debugColor, 0.1f);
        }
        
        Debug.Log("----------------------------------------------------");
    }
}

Balls have disabled gravity to simplify testing

[timid dove]

I think what you need to be doing is lookiung at ContactPoint.collider and ContactPoint.otherCollider to understand from whose perspective the contact point is telling you the impulse

[cyan fossil]

I'll change the code to print it in console

[tender vapor]

also consider naming the 2 balls differently to distinguish them more easily

[cyan fossil]

yep, missed it in first iteration :D

[tender vapor]

the other thing i can think of would be if one were rotated, but that doesn't seem to be the case

[cyan fossil]

I guess no, they all have rotation zero

[timid dove]

this is weird 🤔

[cyan fossil]

I'm using Unity 6000.2.9f1 maybe will redo test with some older version to check if it's working the same

[tender vapor]

what about the normal of the collision?

[cyan fossil]

normals looks ok

[tender vapor]

well, if you scale the normal with the impulse, that'd go in opposite directions i guess?

not sure that's the intended way...

[cyan fossil]

hmm, maybe, let me add some debug lines to visualize it

[tender vapor]

physx contactpoint doesn't have an impulse field, just maxImpulse...

what if the 2 balls have different initial speeds?

[cyan fossil]

multiplying by normal looks somewhat ok, (left is Ball A with red line representing its impulse)

Ball A (left) had speed 5
Ball B (right) had speed 10

[tender vapor]

just to make sure we're on the same page, could you show the Debug.Logs you used for that (ie what each value is taken from)

[cyan fossil]

    private void OnCollisionEnter(Collision other)
    {
        Debug.Log($"{name} impulse: {other.impulse}, relative velocity: {other.relativeVelocity}");

        foreach (var contact in other.contacts) {
            var relatedImpulse = contact.impulse.magnitude * contact.normal;
            Debug.Log(
                $"contact point: {contact.point}, impulse: {contact.impulse}, related impulse: {relatedImpulse}, magnitude: {contact.impulse.magnitude}, normal:{contact.normal}, thisCollider: {contact.thisCollider.name}, otherCollider: {contact.otherCollider.name}");

            Debug.DrawLine(contact.point, contact.point + relatedImpulse / 1000f, debugColor);
        }

        Debug.Log("----------------------------------------------------");
    }

[tender vapor]

ah ok yeah, that relatedImpulse makes sense
i was initially thinking of scaling the impulse vector, but yeah considering it already considers the direction somewhat that wouldn't make sense

[cyan fossil]

Yeah, but I wanted to use data from collision to calculate damage on impact, but seems like there are no usefull data because there is only sum of impulses if I understood it correctrly. I can't apply different damage depending on impulse on hit. Maybe it is not intended to be used for that purpose, idk

Probably will do some calculations based on rigidbody mass * velocity and I will need to use somehow contact normals in that calculations but I haven't figured it out yet

There is problem that I have multiple colliders for different damage zones and because of that I have multiple OnCollisionEnter calls for each collider

Brain not braining :D

[timid dove]

I hate to give an ai link but this seems to clarify it to me pretty well and explains how to adjust it properly:
https://gemini.google.com/share/a08d3131e47c

[tender vapor]

Contact.normal - Usually points from otherCollider to thisCollider, but can be inconsistent depending on shape sorting.
this doesn't sound right to me

[cyan fossil]

Thanks! It does explain some things
need to test if direction is depending on object order in scene

And I'm not so sure about Don't rely on contact.normal pointing to "this" or "other" consistently. Instead, compare the impulse direction to the direction of the collision relative to your object. need to test it too

[timid dove]

Yeah I think the normals are actually consistent

pretty much using the normal directions should do it

[cyan fossil]

So impulse direction depends on order in scene
(i've spawned balls in different order)

[timid dove]

I think just using contact normal * impulse magnitude is the ticket

[cyan fossil]

One more question, if I have multiple colliders in child objects on collision I get multiple OnCollisionEnter calls. What is the best way to calculate something (damage for example) based on all collisions. My intuition says that I need to cache all collisions data and process them on end of physics update: use somehow yeild WaitForFixedUpdate or Awaitable.FixedUpdateAsync. Am I right?

[timid dove]

Yes you can sum all of the impulses up in OnCollisionEnter and do something with the results in a coroutine loop with WaitForFixedUpdate, or just in FixedUpdate.

[cyan fossil]

I'm trying to figure out how exactly should I use WaitForFixedUpdate
And if I will use FixedUpdate does it mean that I will process all impulses on the next physics "frame"?

[tender vapor]

yeah

[timid dove]

Either way it's going to happen between the physics frame that just happened and the next one

WaitForFixedUpdate means it will happen during this rendering frame, whereas FixedUpdate means it will happen during a future rendering frame.

The physics impacts are identical

As for how to use the coroutine you can just start one with an infinite loop and treat it like a special FixedUpdate:

var wait = new WaitForFixedUpdate():
while(true) {
  yield return wait;
  // Do your logic here
}```

[cyan fossil]

Yeah, you're right
Just need to keep in mind that fixed update is called before collisions, but it will process collisions that were before that

Thank you both!

[deft belfry]

Dont sure if this is a physics related question, but Ill ask it here anyway.

Something strange involving VSync: I currently have a rigidbody with interpolation on, and a script on it that slerps the rotation and position of the player in fixedUpdate. If VSync is turned on, things happen.

https://media.discordapp.net/attachments/493511037421879316/1448699234856603709/2dPlatformPrototype_2025-12-11_17-28-54.mp4?ex=693c35cd&is=693ae44d&hm=8442155d8ad08c46018a2ed70931055a78a457e088218f15416a4edd250bb88a&
VSync disabled (how the game should be)

https://media.discordapp.net/attachments/493511037421879316/1448699394374369361/2dPlatformPrototype_2025-12-11_17-29-55.mp4?ex=693c35f3&is=693ae473&hm=6e7441d79934b5c3688504c6e96facf43bab4dc3a1660255ffe35d87a1f6cc9f& VSync enabled for every V blank

https://media.discordapp.net/attachments/493511037421879316/1448699458945679522/2dPlatformPrototype_2025-12-11_17-30-46.mp4?ex=693c3602&is=693ae482&hm=4ed76eae4ce3ceb8a228fcee238c466b820634a79263f9de9aeef9aa73ea827d& VSync enabled for every second V blank

What is happening?

[inner thistle]

Presumably the code isn't framerate independent (slerping the rotation and position sounds suspicious.) You'll have to show the relevant code

[deft belfry]

    {
        processGround();
        if (isGrounded)
        {
            setLegCorrection();
            applyPositionRotation();
        } 
        applyVelocity(); 
        prevPos = transform.position;
        prevVelocity = rb.linearVelocity;
    } 

  //bodyPoint1 and bodyPoint2 are responsible for determing the player's position and rotation
    private void applyPositionRotation()
    {
        //position and rotation is applied using the perpendicular of the vector of the points;
        var perpendicular = -(Vector3)Vector2.Perpendicular(bodyPoint1 - bodyPoint2).normalized;
        transform.rotation = Quaternion.Slerp(transform.rotation, Quaternion.LookRotation(Vector3.forward, perpendicular), Time.deltaTime / rotTime);
        //an offset is put to make the body move above the ground
        transform.position = Vector3.Slerp(transform.position, (bodyPoint1 + bodyPoint2) / 2 + perpendicular * normalBodyOffset, Time.deltaTime / posTime);
    }```

[timid dove]

directly modifying the position and rotation of the Transform like this means you are rbeaking interpoliation

[inner thistle]

Yeah that's the "wrong (s)lerp" (https://unity.huh.how/lerp/wrong-lerp) and if they're rigidbody-based you shouldn't be changing the transform directly anyway

[timid dove]

double whammy because you're doing it in FixedUpdate so it's basically gauranteed to be jittery

[inner thistle]

oh right deltaTime should be the same in FixedUpdate

[deft belfry]

So, change the rigidbody's position and rotation instead?

[timid dove]

If you want interpolation to work properly you need to use velocity and angular velocity

or if it's kinematic you can use MovePosition and MoveRotation

[deft belfry]

How do I use velocity to keep the player above the ground?

Because the player isn't a typical character, and the body must be kept above the ground

[timid dove]

you do exactly what you're doing now, except achieve it by setting the velocity instead of directly setting the position.

set the character's velocity to the velocity that would be needed to reach the position you were hard assigning before

it's basically just:

• calculate that position
• calculate the diff from the current position to that position
• set the velocity to that diff / Time.fixedDeltaTime

[deft belfry]

What about the rotation?

[timid dove]

same deal just with angulkar velocity and rotation

[placid holly]

can someone help mw with making motorcycle physics?

[tender vapor]

is this at runtime?

[jolly crest]

yeah when i press play

[tender vapor]

and the rigidbody is just following gravity?

[jolly crest]

im trying to do ground check and noticed this gap is messing it up

[tender vapor]

it's probably not the gap messing it up

you haven't set your layermask

[jolly crest]

not sure about the following gravity thing

[tender vapor]

show the setup of your rigidbody

[jolly crest]

true im still missing pieces, but should this gap exist?

[tender vapor]

iirc, yeah, there's a "collider skin"

but i'm not super confident on that

[jolly crest]

hmm ookok

[tender vapor]

so yeah it's following gravity just fine

[jolly crest]

let me try putting the whole logic on this then

[tender vapor]

also unrelated - if you're using pixelart, make sure you have a pixel perfect camera

[jolly crest]

hmm I'll google that soon thanks

as a base im using 320x180

and 16x16 sprites

I do have the layer mask on the tilemap though the gap is messing up my gizmo line, i can just increase the line sure but it's a weird gap

i adjusted the collider hitbox seems to be a normal issue to prevent overflow between colliders https://discussions.unity.com/t/little-gap-between-colliders/791019/5

[tender vapor]

I do have the layer mask on the tilemap
do you mean layer

would it really be noticable? especially once you add a pixel perfect camera

also don't forget to set the game view resolution

[jolly crest]

think i have that

Thx for the tip, got the pixel camera now 🙂 didn't know about that extra

got the jumping working now and it aligns fine!

Thx a lot for the extra tips!!

The tutorial I watched didn't mention any of that 🙂

[deft belfry]

Strange behavior: I have a vector that stores the velocity needed for an interpolated rigidbody player to reach a specific point, by calculating the difference of the point and current rigidbody position. This vector is then added rb.linearVelocity. However, I've noticed something: If I use transform.position as the "current position", the player moves in the exact velocity I set it in, but if I use rb.position, the player moves slower. (Note, I do not modify transform.position, I only read the value) var target= (bodyPoint1 + bodyPoint2) / 2 + perpendicular * normalBodyOffset; var difference=target-(Vector2)transform.position; //positionVelocity is added to the rb's velocity in "applyVelocity" positionVelocity=difference/Time.deltaTime*posSpeed;

Why is it like this?

[tender vapor]

is this being done in update or fixedupdate?

[deft belfry]

FixedUpdate

[tender vapor]

the values of transform.position and rb.position are different (when you do this calculation), correct?

[deft belfry]

Yeah

Transform.position is slightly larger than rb.position

[tender vapor]

hm i don't know how interpolation stuff works under the hood enough to help, sorry

uh that doesn't really mean anything, they're positions

[deft belfry]

Still, shouldnt this work regularly with rb.position

[tender vapor]

hm just a sanity check, is the rb offset

[timid dove]

can you show the rest of the code? What is "added to the velociy applyVelocity" doing? What is happening with this positionVelocity value?

Assuming you are actually in FixedUpdate then Time.deltaTime should be identical to fixedDeltaTime but I would still use the latter to be explicit.

[deft belfry]

Yeah, Im in fixedupdate

[deft belfry]

            rb.linearVelocity = groundSpeed * (Vector2)rotateUpRightVec(Vector2.right) + positionVelocity+velocityOnPlatform;
            Debug.Log(rb.linearVelocity.magnitude);```

(velocityOnPlatform is (0,0))

[timid dove]

Not even seeing positionVelocity used here. Would you mind just sharing the full script?

[deft belfry]

    private void applyVelocity()
    {
        rb.gravityScale = isGrounded ? 0 : isFalling ? gravityFall : gravityJump;
        if (!canMove)
        {
            groundSpeed = 0f;
            rb.linearVelocity = Vector2.zero;
        }
        else if (isGrounded)
        {
            //check if on a wall if the angle is in the range
            //if on a wall, use the y input, else use the x input
            //because of the way Quaternion.Euler(0,0,rb.rotation)*Vector2.right behaves, inputs on a wall whose angle is below 0, or inputs on the ceiling, will be reversed
            // (as in pressing up on the wall will make the player go down, for example)
            //check for those cases, and invert the input
            //this is also done when calculating the ground speed after landing
            float ang = Mathf.Round(Vector2.SignedAngle(Quaternion.Euler(0,0,rb.rotation)*Vector2.up, Vector2.up)); float abs = Mathf.Abs(ang);
            float dir = (abs >= wallAngleMin && abs <= wallAngleMax)
            ? input.directionVector.y * (ang > 0 ? -1f : 1f)
            : input.directionVector.x * (abs < wallAngleMin ? 1f : -1f);
            groundSpeed = staticFuncs.accelDecel(groundSpeed, maxMoveSpeed, groundAccel, groundDecel, groundFriction, dir, Time.fixedDeltaTime, isGrounded);
            rb.linearVelocity = groundSpeed * (Vector2)rotateUpRightVec(Vector2.right) + positionVelocity+velocityOnPlatform;
            Debug.Log(rb.linearVelocity.magnitude);
        }
        else
        {
            var airSpeed = staticFuncs.accelDecel(rb.linearVelocityX, maxMoveSpeed, airAccel, airAccel, airAccel, input.directionVector.x, Time.deltaTime, false);
            rb.linearVelocity = new Vector2(airSpeed, rb.linearVelocityY);
        }
        velocity = rb.linearVelocity;
    }```

rb.linearVelocity = groundSpeed * (Vector2)rotateUpRightVec(Vector2.right) + positionVelocity+velocityOnPlatform;

[brazen dock]

Does disabling collision on Layer Collision Matrix also prevent OnCollisionEnter() from being executed? If so, how to make it happen while collision between such layers are disabled? Should I just be using OnTriggerEnter()?

[inner thistle]

Yes, there are no collision messages if there is no collision. And yes, if you want something to happen when colliders overlap but with no physical collision, use trigger colliders

[brazen dock]

I see. Thanks

[unique cave]

One note regarding triggers though. They don't support continuous collision detection to any degree so you may face issues with high moving objects. For things such as bullets, triggers are often not an option. One possible solution that I haven't tried myself would be to use contact modification (Physics.ContactModifyEventCCD) to remove all effects of the collision while still receiving CCD collision events.

[mild sparrow]

Hey guys, is there any way to make Cloth component interact with box colliders?

[timid dove]

it can only interact with the specific sphere and capsule colliders you assign in the inspector

[mild sparrow]

Alright, thank you)

[hidden hedge]

Hey, i'm working on a game heavily relying on physics. I have a large amount of rigidbodies in my scene (1000+). At the moment i have a spatial partitioning algorithm that makes everything far from the player to kinematic / sleep to avoid unecessary computation. But it appears that even if most of the rigidbodies are sleeping / kinematic, there's still a load remainig on the cpu for those objects.
I tested it in an empty scene with a thousand kinematic rigidbodies, and I have the same behaviour.

Does anyone knows a clever trick to avoid that ? Destroying and recreating rigidbodies is not a viable solution, as I have dependencies to the rigidbody component ( like joints). I thought of creating a scene with the physics update disabled and move the objects to this scene, but Unity is calling OnDisable / OnEnable on every component on the objects leaving/entering a scene, creating a terrible mess to manage :/

[bleak umbra]

Adding/removing the rigidbody (or disabling the object) is the only way to completely remove objects from the simulation if you care about having a continuous world. In a world made of discrete rooms, you can have them use separate physics scenes and simulate them manually. Lastly you can make your own solver that optimizes for your game’s specific shape.

Mind that colliders attached to rigid bodies will always be ‘active’ (depending on collision pair settings) and moving any rigid body with a collider is usually expensive when there are many active ones nearby which cannot be pruned by the broadphase.

[river moth]

Can't you just do isSimulated = false?

[river moth]

I think making rigidbody.isSimulated= false will solve your problem

[hidden hedge]

There's no isSimulated property on Rigidbody

[timid dove]

That's only for 2D

[bleak umbra]

its conceptually the same problemset, if you remove an entity from simulation and add it back later the cost is significant and proportional to the number of interactions that need to be considered. The point here is that having many objects is often not a problem of simulating the physics step (although that can be a bottleneck) but solving the interactions (collisions or simple triggers) compounded by the modification cost on the data-structures which make it work... for example a BVH is optimized for isolating interactions spatially, but if you have a large or fast moving entity or a long raycast, "the whole world" can be affected causing large rebuilds of internal data. Add/Remove of any object always needs to modify those data-structures in non-optimal ways. And the 'non-optimal' comes from the engine being generic and not knowing about the constraints in your particular use case that could be used for optimization. PhysX and Box2D are opinionated best guesses on what solves physics questions 'good enough' for ordinary problem sets that come up in games. At the boundaries of those we have to accept the nature of that tradeoff, or build our own engine. 1000s of simulated rigidbodies is one of these boundaries. Optimising that requires better algorithms (research subject) or design changes (GPU solvers etc.), often both. PhysX and Box2D are already pretty much optimal, in terms of implementation, at what they do/provide.

[river moth]

I am sorry it's only for 2D rigidbodies

[river moth]

Ye I forgot 😭 😭

[river moth]

Daymm holy moly
I think I understand it more now
Thanks a lot

[deft belfry]

I have a kinematic rigidbody platform that can rotate and move, and I want to make my player move and rotate alongside it. I know how to make the player move with the platform, but, how do I make it rotate with the platform as well?

(As in, making the platform carry the player as it rotates)

[silver moss]

How are you making it move with the platform currently? (And is it 3D?)

[deft belfry]

2D, with angular velocity and linear velocity

The player is a dynamic rigidbody that also moves with both of these

[silver moss]

I mean how are you currently making the character move with/follow the platform

[deft belfry]

Currently, if the player stands on the platform, the platform's linear velocity is added to the player's

[silver moss]

Been a while I did 2D but you can try tracking the angle difference of the platform each FixedUpdate and apply that to the character's rigidbody with MoveRotation

Or do it both with MovePositionAndRotation

[silver moss]

Unless doing this but with angularVelocity works?

[coral loom]

i made this simulation of around 40000 particles a few days ago, but im not happy with it

i followed a tutorial

and its good but the particles seperate easy

like they dont clump the way they do in sebastian lague's video and his simulation is much smoother than mine

i saw he implemented two things, position prediction and near density & pressure

but currently my simulation has a tendency to explode if i raise the pressure to be higher than 5

so far ive implemented just basic pressure and viscosity

what should be my next approach to improve the sim?

ik this is quite an open ended question, but im new and any help or advice will be very helpful

[jolly crest]

Hi all 🙂

why is my character sometimes falling through the ground?

I had all logic in update, but moved some to FixedUpdate() still not fixed the issue, any idea?

Collision Dectection in rigit body is set to discerete not continuous? could this afect that?

[timid dove]

Yes that's precisely what that affects

[inner thistle]

That said it shouldn't go through at those speeds even with discrete detection if the colliders are the same size as the visuals so there's likely something else going on

[jolly crest]

Not sure for now that seemed to work I'll keep an eye and see if it pop ups again the issue

Thanks for the help!

[worldly bison]

hi hi! not sure if this is the right channel for this question but basically I am doing the Essentials Pathway lesson and I tried adding some of my own shapes but for some reason the mesh collider of the arch isn't working as intended. The ball isn't going underneath it but hitting an invisible wall despite it's mesh also being set to arch. Does anyone know how to fix this?

[tender vapor]

sounds like the collider is convex there

[worldly bison]

oh yeah it is, when I uncheck it the object just falls through the ground and I'm unsure how to fix that. From the things I looked up I thought I needed convex to keep it up?

okay I just made the rigidbody kinematic which did the trick (I'm pretty sure I tried it before too but it just didn't work)

[unique cave]

Why does the object even have rigidbody if you don't need it to be dynamic?

Kinematic rigidbodies are pretty much only useful for temporarily disabling bodies and for moving/rotating platforms/obstacles that don't need to get influenced by other bodies (but others must get influenced by it)

[worldly bison]

idk I'm learning, I wanted to see if I could have some of the objects impact back into it and knock it back too or something

[unique cave]

If that is the case, you want the object to be non-kinematic, which rules out non-convex mesh colliders as an option. Pretty much the only way to achieve dynamic concave objects is to make the object out of multiple colliders (in your case likely either multiple convex mesh colliders or box colliders)

Setting a rigidbody to kinematic makes it static, pretty much disabling the whole rigidbody

[worldly bison]

ohhhh I see, thank you!
I did read about that option but it seemed like a bit too hard for me right now

[spare solstice]

hello everybody, merry christmas eve

so Im diving into modding for this first time with this game gladio mori, I made my first mod a few days ago, an armor pack that adds a few new pieces to the game, and Im trying to see about "cloth" physics in this game for the sake of armor

the idea is that the "cape" use the games characters flesh system to have some basic resistance to weapons, and would just be a second skin the character could have as a layer of protection\

I found a cape model online with a basic rig, but as far as implementing that into the SDK im a little lost

ig im asking what components would i need

[tender vapor]

sorry, we don't support modding here #📖┃code-of-conduct
both because of legal reasons, and because modding often brings about unique constraints that limit your workflow and tools, it's kinda outside the scope of the server

[spare solstice]

roger

[pulsar verge]

What is modding?

[tender vapor]

modification of an existing third-party game

[pulsar verge]

O Alr

[empty moat]

does anyone know why this line of code when applied to a linear damping of 10 perfectly returns a fixed movement speed of 16?

200 move speed and 10 damping somehow returns 16 when before i had issues of the player constantly flying away or something

[tender vapor]

and it's being called every FixedUpdate, correct?

[empty moat]

yh on fixed

i only know it returns 16 cus of this tiny line of text if thats of any help

[tender vapor]

i'm not quite sure what your question is

16 would be the expected value, according to the formula given in the docs for linearDamping

if you're curious about the math behind it, it's probably some integration stuff

[empty moat]

How do i see this

[tender vapor]

huh that's weird, the result seems to depend on deltaTime... i feel like im doing something wrong

[tender vapor]

i just opened up desmos and put in the formulas

[empty moat]

Never heard of it b4

[tender vapor]

desmos is a graphing calculator, a really powerful tool

[empty moat]

Well through throwing shit at the wall until it sticks i managed to make a fixed responsive speed that uses physics instead of clamping and stuff

I never knew you could create a cap with damping

But my wasd movement doesnt go over 16 so im just going to pray i never ever want to change that variable

[tender vapor]

it's basically applying an opposite force proportional to the current velocity, and the resulting steady velocity is basically equilibrium when the acceleration force and the damping force are equal

[empty moat]

Right

Well Ive got no idea how to actually do the calculation so im gonna ignore it and start working on modifying drag through different ways until its nice to me

[quartz panther]

Sorry for the late reply, i was very busy on another project and i restarted my unity project today.

These are the floater positions and other values:
There are total 13 floarters, their positions are:
front-left: (-10.5, -2, 110)
front-right: (10.5, -2, 110)
rear-right: (10.5,-2, -110)
rear-left: (-10.5, -2, -110)
center-right: (15,-2,0)
front-mid-right:(15, -2, 51)
rear-mid-right: (15, -2, -51)
center-left: (-15, -2, 0)
front-mid-left:(-15, -2, 51)
rear-mid-left: (-15, -2, -51)
center: (0,-2, 0)
bow: (0,-2, 139)
stear: (0, -2, -139)

ship mass: 1e+07
floater area: 577
max depth: 50
water density: 1000
ship centre of mass: (0,-20,0)

!code

[gloomy lodge]

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.

[quartz panther]

https://paste.ofcode.org/TeXw6ggHwxEhnaMdwGycUz
This is the ship movement controller script

[autumn mica]

Is there a way to learn what forces are currently actiong upon an object?

I have several different problems with my movement script or rather interactions with moving platforms or walls that I am struggling to find the source of. I think learning the origin of the forces causing the problems would help a lot.

[silver moss]

Not really. You'd need to search your codebase

You mentioned moving platforms, are you doing transform parenting or something?

More info would be welcome

[autumn mica]

Thats unfortunate but thanks for the info

[autumn mica]

Exactly that, I have the issue that when my platform stops moving, the player gets still thrown into the air and doesn't stop together with the platform.

Another issue is that when jumping into a wall straight on, the player gets stuck on the wall instead of falling down like it should.

[tender vapor]

sounds like you have friction

consider making either the player or the wall frictionless

[south granite]

why should it stop

[tender vapor]

if your platform stops instantaneously (for example) then the player might have leftover velocity carrying it into the air
you'd need to make it "stick" to the platform, i guess?

[autumn mica]

because the player is parented to the platform so any movement should be transferred

[autumn mica]

exactly

so the instant stop is the issue then?

[south granite]

what movement is it?

[autumn mica]

The platform and the player both move through ridigbodies and velocity changes if that is what you are refering to

[silver moss]

Do not use transform parenting with rigidbodies

Transform and rigidbody will both be confusing each other

[autumn mica]

oh okay

what would be a better alternative then?

[silver moss]

Usually you can just add the platform's velocity to the final velocity of the player

Assuming you calculate it every frame

This 3D or 2D?

[autumn mica]

I tried that but it didn't work for me, the player kept glitching into the platform on upward movement or floated off the platform on downward.

3D

[silver moss]

You tried without any parenting?

[autumn mica]

Yes, I had a method constantly applying the platforms velocity to the players RB

[silver moss]

Would need to see the actual code

You could also show your player's inspector with components visible, just to make sure there's nothing that conflicts with rigidbodry

Also you might want continuous collision detection

[autumn mica]

That is enabled, I don't remember if I did that before or after I abandoned the velocity transfer though... I will give it another try right now.

[autumn mica]

That pointed me to something, apparently another dev on the project implemented a custom drag script that is constantly acting without telling me... thanks a lot for your help until now, I will go and try the velocity once again.

[obsidian patio]

your main problem is this line:
transform.localEulerAngles = new Vector3(pitch, localEuler.y, roll);
you're setting transform directly while physics is also trying to control it. buoyancy pushes one way, this line forces another way they fight = oscillation
remove that line. if you want angle limits, do it with torque:

float rollError = Mathf.DeltaAngle(transform.eulerAngles.z, 0);```

```rb.AddTorque(transform.right * pitchError * 5e6f);
rb.AddTorque(transform.forward * rollError * 5e6f);```
also your ShipStabilizer pitch/roll damping is 2e+10 way too high try 1e+6

[tender vapor]

AddTorque shouldn't have deltaTime, it'll already account for that

[obsidian patio]

You're right edited

[signal wyvern]

If i add a joint between two objects during runtime does it clear the current velocity?

[timid dove]

Possibly. Very easy to test

[signal wyvern]

I need a real scientist to debug my Client prediction setup

[signal wyvern]

Ive tested adding does not clear velocity, but im trying to figure out if the act of Adding/Destroying a joint applys some kind of force? This is harder to test

[full flume]

hey guys quick question im having an issue when i build my 2d platformer game where in the editor its working fine but when i build the player cant seem to touch the ground or even touch he just falls through

[timid dove]

really hard to say what the problem is based on this vague description.

My wild guess would be that you have some framerate-dependent code of some kind.

[viral bone]

Did you write the physics by yourself or using colliders? Only thing i can think of from glance is error when comparing floor level with a == operator

[gaunt sinew]

Hey, I have a ragdoll for a human setup which is just rigidbodies, character joints and colliders on the bones of the armature.

The issue is that when I apply a force m_Pelvis.AddForce(direction * force, ForceMode.Impulse); the feet seem to stay planted, no matter how much force I apply. Any ideas of what might be causing this?

[timid dove]

probably some other script or component

an animator, an IK controller, etc.

[gaunt sinew]

That is what I thought too, but I am pretty sure I disabled everything or removed them and still happening... could be referenced by some other component on anther object I guess... ugh. Well at least it sounds like it isn't expected behavior

(I've almost never touched physics stuff)

[quartz panther]

I changed my decisions, I have spend a lot of time trying to fix the multiple floaters physics issue, and i also have to complete this project with mid february so i decided to go with simpler buoyancy physics, Now the ship movement works fine.

!code

[gloomy lodge]

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.

[elder lintel]

Why is my character falling so slowly?