⚛️┃physics

[silver moss]

So what if your capsule is like this?

The character up (yellow arrow) is more than 80 degrees from the right side normal (white arrow)

And the right ramp will get ignored because of the dot check

[indigo echo]

then this would'n be possible

the ramps aren't exactly 45 degrees, it's more like 30-ish

[silver moss]

Why are you using transform.up instead of world up direction though

Are you supposed to be able to run on ceilings and stuff?

[silver moss]

That makes the angle/dot difference even greater

[indigo echo]

yeah, exactly

something like this

[silver moss]

Alright

I think you just need to tweak the dot check, make it more forgiving

[indigo echo]

nah, doesn't explain what i showed earlier

also, i changed it to use RB.trasform.up on the check and it's still happening

[silver moss]

In this scenario (angled ramps on both sides), imagine this:
-Frame 1: You hit the Right ramp. You adjust your transform.up to align to the Right ramp
-Frame 2: You hit the Left ramp, while still colliding with the Right ramp. Your transform up is already aligned to Right ramp's normal (pretty much perpendicular to Left ramp) from the previous frame. Your dot product check for the Left ramp fails

Maybe use a combination of dot check with player up direction, and dot check of current ground normal?

Max of those maybe?

[indigo echo]

i get what you mean, but what i don't understand is why what's happening in the next screenshot is possible

[silver moss]

This one? What's wrong here exaclty

It looks to be aligned correctly - probably because it hit both surfaces at the same frame

[silver moss]

Anyway i'm pretty confident that this is a good fix

Make your minGroundDotProduct more forgiving but check the dot to the current ground normal (if already on ground)

[indigo echo]

i lowered the ramps, i don't know why queries say there's two hits and SC says there's one except when the character is on the exact middle

this is when there's only one brush

[silver moss]

Only reason I can think of is that your if statement evaluates to false and groundContactCount doesn't get incremented for one of the ramps

Have you normalized the vector that you put into Vector3.Dot?

[indigo echo]

i mean, it's only one vector at a time so it's already magnitude 1

and groundContactCount it's only to know if the character is on ground

[indigo echo]

here's another example, the closer ramp it's only one brush and the further one are two

[silver moss]

What number are you logging though? I haven't seed your recent code

[indigo echo]

it's just hitCount

static readonly RaycastHit[] hits = new RaycastHit[10];

void DetectSurfaces(){
  Vector3 sphereOrigin = RB.transform.position + RB.transform.up * 0.5f;

  int hitCount = Physics.SphereCastNonAlloc(sphereOrigin, 0.505f, -RB.transform.up, hits, 0.5f, layerMask, QueryTriggerInteraction.Ignore);

  for (int i = 0; i < hitCount; i++){
    if (Vector3.Dot(hits[i].normal, RB.transform.up) >= minGroundDotProduct){
      groundContactCount += 1;
      avgNormal += hits[i].normal;
    }
  }
  print(hitCount);
}

[silver moss]

And those magenta arrows are the hit normals shown by physics debugger (I haven't actually used it)?

[indigo echo]

yes, they are the queries you told me about

[silver moss]

Are you using any other physics queries apart from this? 👀

Specifically sphere checks

[indigo echo]

nope, just physics debuger

ah that, no

[silver moss]

Physics queries mean methods like Raycast, SphereCast etc. Those are queries

[indigo echo]

just SC

[silver moss]

Weird then, not sure why it would show two contacts(?) when it prints 1 🤔

Unless it registers the two contacts as one hit

Still weird

[indigo echo]

this is what runs after DetectSurfaces ```cs
bool isGrounded => groundContactCount > 0;

void StateUpdate(){
stepsSinceGrounded += stepsSinceGrounded < 2 ? 1 : 0;

if (isGrounded){
stepsSinceGrounded = 0;

RBNormal = avgNormal.normalized;
Debug.DrawRay(RB.transform.position, RBNormal, Color.blue);

}
else {
RBNormal = Vector3.RotateTowards(RBNormal, -gravity.normalized, RBNormalResetMaxRadDelta, 1);
}

RB.transform.up = RBNormal;

hVel = Vector3.ProjectOnPlane(RB.linearVelocity, RB.transform.up);
vVel = Vector3.Project(RB.linearVelocity, RB.transform.up);

groundInfo = new() //Update groundInfo
{
isGrounded = isGrounded,
RBNormal = RBNormal
};
}

this is what actually changes RB.transform.up

[young gate]

Is there a good method for ignoring collisions certain faces of a convex mesh collider?

I looked into using modifiable contacts for it but it seems to actually get the face of a contact isn't exactly simple as it's not part of the contact data

I could raycast to find it, but it seems like overkill especially for every contact every frame

[unique cave]

Why you want to do that? I think convex mesh colliders are considered as volumes and not triangles defining the boundary so that wouldn't really be possible with the physics system

So what is the context behind the question? What is it in your game that you want to fix?

[young gate]

So originally I was looking into it to deal with collisions on internal faces of bodies composed of multiple mesh colliders that would sometimes occur. Now I'm looking into it for using capsule colliders for articulationbody wheels while ignoring the hemispheres that would protrude from the wheel

[unique cave]

For the latter I believe contact modification would be the only way but I'm not familiar with the actual process of how it's done

[hidden snow]

Hi there, what is the best way to implement an enemy attack with firing arrows for example in the way that the object faces?

[unique cave]

Well, if cylinder collider is all you want, you could use mesh collider too. With the convex collider triangle limit the best quality you can get would be 64 vertice circles at both ends of the cylinder which is quite good but maybe not enough for a wheel

[young gate]

I haven't actually tested yet, but my intuition was that sphere/capsule or some custom circle would probably be better than cylinder

Not just for accuracy but for performance

[unique cave]

certainly would but excluding the extra parts would be difficult task

[young gate]

At least for the capsule I don't have to determine any face information, I can just exclude collisions based on normal or position

It was awful for the interior faces of composite mesh colliders though

[hidden snow]

Anyone get what I mean ?

[unique cave]

Some vague idea but don't really understand the whole picture of what you need help with

[shadow plinth]

Is it normal that OnCollisionEnter() sometimes just not being called when object collides with something? It just ignores collisions from time to time and I don't really know, how to counteract this :\

[silver moss]

First thing that comes to mind is that you are moving rigidbodies incorrectly and therefore they can't reliably detect collisions

For example moving the rb with transform, while you should move it with rb.AddForce/velocity/linearVelocity

[shadow plinth]

They are moving on their own fom gravity, in my case, box is just falling to the ground. Just in case: I don't use any custom system for gravity, everything is default rb systems

maybe my physics settings are incorrect or something

[silver moss]

What inspector is that 👀 Oh physics settings in unity 6(?)

[shadow plinth]

wait, it didn't exist in unity 5?

[silver moss]

What are your objects composed of and how have you confirmed that OnCollisionEnter is not getting called?

[silver moss]

It did, the styling is just different which threw me off

[shadow plinth]

I have a floor, which is part of room it have mesh collider, and rigidbody is cube with box collider. I am making script for playing sound on impact and in the end of OnCollisionEnter() I do Debug.Log() with relative velocity and position

[silver moss]

So you have no if-statements before the Debug.Log or anything?

[shadow plinth]

nah, debug is not in any if statements or something. In fact, it doesn't matter if I put it in the start or in the end of the function

here collisions with zero relative velocity are collisions from cube contacting with ceiling, but when it hits the floor, OnCollisionEnter() sometimes just ignores the collision entirely

[silver moss]

Okay I might have an idea

Is the whole room one mesh collider?

Like are the walls and the floor the same mesh

I noticed from the video that it doesn't play a sound when it slides down the wall

Which would hint that it is already colliding with the room (wall) when it hits the ground

[shadow plinth]

yes

[shadow plinth]

oh, I think you are right

[silver moss]

It counts as a single collision, i think unity produces only one collisionenter per collider

[shadow plinth]

Yeah, I see. OnCollisionEnter() is being called when entering collision with object, not creating new contact point

[silver moss]

Yeah so I suggest separating the room into smaller objects

Probably helps with rendering and physics performance too

[shadow plinth]

yeah, I think that would be a great idea

[shadow plinth]

I have investigated further and discovered, that bounciness of physics material applied to the object can trigger this issue. Basicly, objects with bounciness will not trigger "OnCollisionEnter()" or make "OnCollisionStay()" not provide new contacts for this collision sometimes

[shadow plinth]

and yeah, it helped but issue just became a little more rare

[silver moss]

Could show a video of the issue that is still happening

[shadow plinth]

it kinda detects the collision, but relative velocity is zero

so, setting higher bounciness helps it to correctly display velocity, but now impulse is zero

correct velocity and impulse calculation appears only when object is nearing the end of bouncing

[urban forge]

How can i build a wall run mechanic with a fp player

[unique cave]

I would assume iterating over all the contacts on OnCollisionStay and looking for the highest impact would also work. Is there some problem with that approach?

[shadow plinth]

Good idea! I will try it and write here about the results later

[shadow plinth]

every contact point have 0 impulse during first collision\bounce. Maybe the collision is being counteracted by bounciness too fast, so OnCollisionXXXX() functions just don't recive impulse and relative velocity?

[unique cave]

Is that with 100% bouncy collisions only?

[shadow plinth]

no, as shown here, setting bounciness so something higher than zero makes this issue possible

[unique cave]

That sounds very weird though, I don't think the bounce can be too fast to detect because the collision is obviously resolved. The impact shouldn't be too hard to calculate from the momentum change either so I can't really think of any reasons as to why that might happen

[shadow plinth]

yeah, that makes sense, but somehow this strange issue exists

[unique cave]

Now I'm just wondering whether OnCollisionExit would show the right impulse from the last frame in case the bouncy collision causes the collision to only happen for one frame

Kinda wild guess but I would check that too for good measure

[shadow plinth]

give me a sec, i will try it

[unique cave]

Since OnCollisionExit takes in Collision too it would make sense it would show the last step values or something like that because otherwise it would only return no contacts since it obviously wouldn't hit anymore

[shadow plinth]

yeah, it's wrong too

[unique cave]

Might also be there to get .gameObject, .artuculationBody etc. only though

Unfortunate

[shadow plinth]

I also tried slowing down the entire game, but it doesn't change anything

[unique cave]

If that's really the highest impulse of all contacts and not just of GetContact(0), I'm pretty much out of ideas

[shadow plinth]

it's the same for collision.impulse.magnitude and the highest impulse of all contacts... 💀

[unique cave]

Seems to have been issue for years from what I can see from the forums so it might as well be a problem with PhysX and not unity specifically. Couldn't find anything from UE thought which uses PhysX too so no idea what's the deal with that. Unfortunate no unity dev have confirmed the cause of the issue either on the forums at least that I can find. You could always file a bug report though there's a chance many have done so already on the same problem and they haven't fixed it to this day

[tight compass]

does anyone know why the size of the red colliders arent changing along with the radius etc of the collider components ? i'm using the physics debugger. my ragdoll seems to be influenced by the red ones and is giving wild results. ive set this one up manually after the ragdoll wizard was doing the same thing

also the colliders are not selectable with mouse select on

[silver moss]

Do you have animations playing that could be scaling the bones?

[tight compass]

no animations and ive applied scale in blender

also i have realised they are selectable but I have to actually click where the green colliders are located and not the red part

not sure if a screenshot of the rig helps but here it is. @silver moss

[tight compass]

if anyone else ever has this issue, it was a version bug. updating to 6.0.26f1 fixed it. https://discussions.unity.com/t/physics-debugger-display-issues-in-unity-6000-0-24f1/1543619

[low magnet]

Do you know why terrain made of tiles acts differently depending on the direction of movement when the player jumps while the terrain is above the player's head?
This problem does not occur in editor but only in webgl builds.

[gray coral]

Does anyone knows a better way to grabbing a physic object smoothly? I have tried using Lerp position and rotation, Fixed Joint and Spring Joint. The Lerp seems the more smoothly but still has glitches. My game is like Moving Out or Human Fall Flat but just the arms are controlled by sticks

[dull jungle]

Joints should be the smoothest built-in (with the right settings), you can also apply forces that change based on distance

If you want a physic-y feel you shouldnt lerp manually

[gray coral]

I used a Spring Joint and it was looking smooth but the object was damping around the hand sometimes, but I think it should work if I can fix the object around.

Is there a way to improve the physics engine to avoid glitches like when you grab the object against a wall? I twiked the rigidbody 'drag' and 'angular drag' but not sure if I am doing the right way

This is my game. Using the Lerp

[gray coral]

Seems is working better now with a fixed-joint and adjusting some parameters, but now the Rig of the right hand is moving weird, Any ideas, this is my right hand constraint, any ideas?

[spark grail]

How do I do water physics?

[soft prawn]

Hey all, so I'm working on a motocross style game and the physics of the bike are killing me. Right now I am trying to implement a raycast system that keeps the bike above my terrain but all that happens is it attempts to and the falls through the world (the box collider on the bike is keeping it from free falling). Any tips on how to keep this bike from constantly falling?

[pliant raft]

dm

[royal willow]

you should probably fix the runtime errors in your script

[idle estuary]

Hey guys

    {
        Vector3 forceX = orientationX * moveDir.x * currentAccelRate;
        Vector3 forceZ = orientationZ * moveDir.y * currentAccelRate;

        rb.AddForce(forceX + forceZ, ForceMode.VelocityChange);



        //Clamp
        Vector3 flatVelocity = new Vector3(rb.velocity.x,0,rb.velocity.z);
        if (flatVelocity.magnitude > currentMaxSpeed)
        {
            float fallspeed = rb.velocity.y;
            Vector3 n = flatVelocity.normalized * currentMaxSpeed;
            rb.velocity = new Vector3(n.x, fallspeed, n.z);
            if (Input.GetKeyDown(KeyCode.Space)) Debug.Log("Velocity " + rb.velocity);

        }
    }```

I have the code above called in FixedUpdate. While the included Debuglog does tell me the change of velocity was operated (maxSpeed set to 7), the player object still moves with a magnitude of approx 20 in game. Did I miss something ? In Midair (so without dampening) and when nothing is inputed, the intended effect is observed and the velocity is reduced to a magnitude of 7.  I don't understand why it does go beyond as clamp is done right after adding forces due to player input...

[royal willow]

does the rigidbody itself show a magnitude of 20 or does it show a magnitude of 7?

[idle estuary]

It shows a magnitude of 20 (and is moving at said speed) while the log display 7 at the end of the function BUT a log placed before the function shows a magnitude of 20...

There is no other place in the project that add speed to the rigidbody

[royal willow]

also you could try clamping the speed before you add any force and you can check this out to see if this helps
https://docs.unity3d.com/ScriptReference/Vector3.ClampMagnitude.html

[fresh palm]

Hiya :3 I am am having some issues adding collision to my spider thing rig:

[torpid hollow]

does anyone know how to make the object with spring joint more dominant then the connected body, right now I want to make a balloon rope system like GMod and i want the connected body to fly with the balloon but I do not know how to do it, anytips would be appreciated (got it working by changing the mass)

[red oriole]

I have two wheels with Hinge joints that are motorized, and both are attached to the same boggie
but, when i start the simulation, the boggie starts spinning around the first wheel very quickly?

[unique cave]

The wheels mass is 100 and the bogies 0.0000001, might that be related?

[loud ridge]

how can I make walking better I dont want this to happen

and these are my animations.

[hexed laurel]

That is so weard if someone gave you a solution tell me

[loud ridge]

I dont think anyone will.whenever I post in this server everyone ignores me.

[hexed laurel]

hmm they will believe me

them 🥹 🙂 me 🥹 them

[frigid pier]

To get an answer you need to post an answerable question.
Answer to

how can I make walking better I dont want this to happen
Is quite obvious one - set up animation and script it to work in a natural way and respond to movement.

No-one here will guide through hours of proper procedure of how to do that even if they know.
Your best bet is to find a tutorial that tackles this aspect and follow it. Or get an asset that already does that.

[frigid pier]

Don't even know what you are showing here even. Animations are not setup at all.

[silver moss]

I assume you are using joints for the legs. Make the animation animate the leg joint target angles?

[frigid pier]

Just adding legs shuffling around would work here.

That part should be scripted probably, you don't want rigid animation on that.

Simplest solution would probably be adding impulse force to joints one after another in the direction of movement.

[hexed laurel]

https://www.youtube.com/watch?v=LEwYmFT3xDk&ab_channel=CodeMonkey
This vedio will help you alot

[fresh palm]

I havent got a reponse either /:

[inner thistle]

because nobody knows what "some issues" is supposed to mean

[royal willow]

you have to actually provide details so we can help, we are not mind readers

[loud ridge]

What do you mean by not set up at all? If you took a look at the script you would understand also what I mean was very clear in the photo and the video.

[frigid pier]

What script?

[loud ridge]

sorry https://hatebin.com/rivpipcygk

I though I wrote it but didnt know I forgot

I posted this question on another channel before maybe thats why I forgot

[frigid pier]

Seems to have a lot of problems, it looks like you are triggering leg movement every frame when you need to pulse them from time to time when body shifts enough from them. No idea what animations there suppose to do.
You are pushing legs there with a force multiplied by an incredibly small deltaTime number which makes no sense there.

Like I mentioned previously

Simplest solution would probably be adding impulse force to joints one after another in the direction of movement.
Designate leg space where each of them will be at rest. When body shifts far enough from any of them push one with an impulse force diagonally top-right or top-left to make a step. (or just animate knee movement to that direction and release)

[fresh palm]

no one ever asked. and the issue is that the collision just dosent get added

[loud ridge]

those targetrotations on animations change rigidbody's rotations, top right-left may be a solution

[frigid pier]

You should start an infinitely running coroutine if you want to monitor legs constantly and have option to run a timer between each leg movement, and have it running in Physics mode awaiting for fixed update frame.

https://docs.unity3d.com/ScriptReference/WaitForFixedUpdate.html

[static spruce]

Hello.
I move both my player and the enemy using a Rigidbody and the AddForce function. If the player sticks to the wall and the enemy (who follows you) pushes, the player goes through the wall and ends up outside the room.

Is that enough information?

[timid dove]

Use continuous collision detection

[static spruce]

I don't want it to happen, of course

The player has collision detection already

[timid dove]

If course it has collision detection

I'm saying switch the mode to continuous

[static spruce]

Sorry, I meant the player has continuous collision detection already

[timid dove]

If it's going through the wall then you probably have some code that is not moving it via physics

or rather moving it via some other means than physics

What does "sticking to the wall" mean in your game exactly?

[static spruce]

Press up key and move upwards facing the wall

[timid dove]

Sharing videos, scripts, and screenshots might be useful

[timid dove]

and how does this work?

[static spruce]

Give me a sec

Selecting the code that I was going to show made me realize what was the problem

Thanks @timid dove

[frigid pier]

Get collisions coordinate, get tile https://docs.unity3d.com/ScriptReference/Tilemaps.Tilemap.GetTile.html

If your projectile goes too fast use https://docs.unity3d.com/Manual/ContinuousCollisionDetection.html

Composite collider is a single collider, if trigger contacts it, it generates https://docs.unity3d.com/ScriptReference/MonoBehaviour.OnTriggerEnter2D.html event

After that https://docs.unity3d.com/ScriptReference/MonoBehaviour.OnTriggerStay2D.html continuously

probe trigger stay event to track it

[frigid pier]

Already posted it

Up to you to remember what was hit already

use a dictionary with coordinates as keys

you can optimize by remembering last hit not to access dictionary constantly

you can optimize by checking coordinates yourself

[distant crater]

So I try to use a Platform Effector for one way platform and it turns out I can just walk between the top and bottom just fine. I just want to walk on top. How do I make the game ignore the bottom

[frigid pier]

When particle coordinate moves by a tile size

Tilemap is uniformly square, you can easily calculate tile coordinate yourself, but GetTile method probably light enough and does the same.

YourWorldCoordinate divided by tile size will get you a tile coordinate.

[frigid pier]

Platform Effector by design is for platforms so you can walk under them, use simple/tile collider.

You check only when you need to, when it stays collision

You check events only against collider you need

If it's a tilemap with a composite collider you are checking against, why would something overlap

what is overlap? If it's inside collider it triggers stay event, if it's not it's not

If your contact collider too large, use a small one then? Just for the point.

Collision coordinate is the coordinate of your bullet during collision event.

It's up to you to decide what point is a collision point on the object. If it's a fixed object you can just decide where it is manually, if it's a round object you can get point on outer radius opposite the direction it was travelling.

[distant crater]

I use tile collider already

[frigid pier]

Then you are not using physics methods to move and just pushing object inside collider.

You need to ask more descriptive questions. First one can be an empty game object on the bullet prefab placed on the point of the projectile, to easily get coordinates to check against. Second one is self explanatory.

[distant crater]

can we discuss this in dms?

[frigid pier]

There are plenty of tutorials that describe how to properly use physics system for movement. !learn

[flint portal]

:teacher: Unity Learn ↗

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

[wind meadow]

question, do triggers count as part of the non-trigger collider's shape

so like i have a capsule collider not set to trigger and then i have a sphere collider that is larger than the capsule collider but is set to trigger, does that sphere collider count as part of the capsule collider's shape for stuff like ontriggerenter?

[frigid pier]

Why would they be, they are different colliders with different properties. Something you can easily check as well.

[wind meadow]

eye sea, cause...that seemed to be the case at least (triggers count as part of collider shape)

[wind meadow]

is there some kind of collision matrix i'm missing or?

slowly providing screenshots

both capsule colliders (as pictured are NOT set to triggers)

but the sphere colliders are set to triggers

yet they trigger the OnTriggerStay response

so unsure what's going on

[timid dove]

The Rigidbody will receive messages for all its colliders

[wind meadow]

so rigidbody will call even if both triggers intersect?

[timid dove]

If you want OnTriggerXXX to work for only one collider, put that collider on a child object and put the script with the callback on that child only

You can also set such children to different layers and use the layer matrix

[wind meadow]

eye sea

guess i was surprised rigidbody allows trigger to interact with other triggers

given i thought the main feature/drawback was triggers cannot respond to each other

[timid dove]

Not sure where you got that idea

The main feature of a trigger is that it is not physical/solid

[wind meadow]

right

A trigger doesn't register a collision with an incoming Rigidbody. Instead, it sends OnTriggerEnter, OnTriggerExit and OnTriggerStay message when a rigidbody enters or exits the trigger volume.
I guess i misread this then

i guess the triggers are an extension of the rigidbody's bounds

[timid dove]

When they say register a collision they mean a physical collision/interaction

Where the motion of the Rigidbody is affecting

I.e. bouncing off

[wind meadow]

or blocking

i guess what i'm confused is

so the triggers if they're on the same gameobject as the rigidbody

count as an extension of the rigid body?

so if that trigger interacts with another trigger even if a non-trigger enters it, it will call the respective OnTrigger?

[timid dove]

All colliders on a Rigidbody or it's children define the Rigidbody's shape

All colliders on the Rigidbody or it's children are part of that body

[wind meadow]

trigger or not right?

[timid dove]

Yes

[wind meadow]

ok

now that makes sense

so if i move the trigger to a child of the gameobject

and that child has no rigidbody

then that trigger will only call its ontrigger functions when it collides with the capsule (defined as non-trigger)

assuming that other object with capsule ALSO has a trigger on child object and not the object with rigid?

[timid dove]

It will still be part of the Rigidbody but you will be able to put a script on that child object that will only get OnTrigger callbacks for that single collider

[wind meadow]

but basically

that trigger can still call ontrigger when up against another trigger then?

[timid dove]

Yes

[wind meadow]

if that trigger has a rigidbody either on itself or on teh parent

eye sea

well that's annoying but at least i understand why

question, how performant or non-performant would it be to call sphere overlap every fixed update?

[unique cave]

It all depends as with any performance related question. If it's only one object, doesn't sound a problem. If you have dozens of those calls or couple hundred we can start talking about the implications.

[stray charm]

i've never used unity physics before, does anyone have a good tutorial they can point me towards for getting a lantern in the player's hand to swing depending on the cameras movement?

reacting to wind too

https://www.youtube.com/watch?v=DOJAvA5uNSM&ab_channel=Createathing

assuming this is how i do it

[stray charm]

hmmm, doesn't seem to be working, not sure how to get the lantern to react to the camera and its parent hand turning/moving

[stray charm]

got it to work (that tutorial was wrong and incorrectly told me to assign the lantern as a child of the hand), but the physics are jittery... the lantern often lags behind then the hand moves and it doesn't look smooth. how could i fix that?

[wind meadow]

dozens of calls

cause it won't be just one gameobject

or really hundred

cause i plan on making an RTS and want to do a check if enemies are in range of the unit

[unique cave]

Physics check for figuring whether something is in range sounds very inefficient approach

[wind meadow]

fair, what would you say is the better approach?

[unique cave]

With couple dozen calls it might not make a performance difference but the physics system is just not designed for that

[wind meadow]

so should i move it to update then?

no that's worse

[unique cave]

Would be worse yes. Usually you would just keep list of the objects to check and iterate over it to check the distance. If you only need to know whether something is in range, you can use the squared distance too instead of the actual distance (Vector3.Distance is somewhat slow due to square root)

[wind meadow]

Usually you would just keep list of the objects to check and iterate over it to check the distance.
I guess for every player troop, iterate through them to see the distance between each player troop to all the enemy troops on the field?

[unique cave]

Basically you would replace each CheckSphere with the list iterated once. Depending there might be some better ways too and especially if there's a lot of enemies to check for, you may want to consider switching to quadtree/octree or similar data structure instead of list to speed up the search. The physics system uses similar data structures to accelerate the search but I wouldn't generally use the physics methods for checking something as simple as that

Profiling/benchmarking really is the only way to figure out any performance issue but I assume there are better ways than using a system that is not supposed to be used for that

[half elk]

Hey all, I'm working on a basic 3D mini golf game and I have all the typical things working, the course, the hole etc...

However, when applying force to the ball, it doesn't "feel" right

So between the ball, and the two physics materials

I'm wondering if there is anything obvious I'm missing

I know this is a little vague, but I'm happy to go into more detail

[timid dove]

What about it doesn't feel right?

[half elk]

@timid dove ( Ignoring the last hit where it goes off the world ) - I think maybe the friction is causing the ball to jump? Before I had no bounciness on the ball, but then I set it to 0.5. Maybe if I remove the bounciness on the ball and try to apply them to the wall colliders specifically?

Also, please pardon the design, I'm newer with asset creation :X

[coral mango]

Check when each thing is happening in the update cycle - you might need to move something to lateupdate or fixedupdate

[unique cave]

Unity does not simulate rolling resistance, only drag (air resistance). Look into the formula for rolling resistance and try to implement it yourself using AddForce. It looks very unnatural how slowly the ball stops when it's rolling on grass (and the speed decreses very fast at higher speed). drag should be set to something quite small

[royal willow]

Also for the bouncing, if you are using a mesh collider, check if it is actually flat as that might cause that issue

[woeful egret]

collider gets stuck inside ledges? Not sure whats happening here, i have a slippery material on both as well.

Note - if i make RB discrete detection this issue doesnt occur, however i cant have that due to fast moving preojectiles

[timid dove]

Likely due to some code you have somewhere

[woeful egret]

It’s possible but not likely. Seems to be pro builder mesh’s not being convex. Still investigating though

I’m physically passing through the mesh collider on impact and getting stuck

[stray charm]

thank u, turning interpolate on actually fixed it

i do have an issue now where if i move my camera quickly, the lantern fails to catch up and lags behind as seen here (where i'm spinning to the right)

there's only one hinge joint in use, not sure how to keep it so that lantern doesn't become seperated from its hand

how do i put a component into fixedupdate?

[royal willow]

use code

https://docs.unity3d.com/ScriptReference/MonoBehaviour.FixedUpdate.html

[unique cave]

component into fixedupdate? what does that even mean?

[stray charm]

i've been told to run the hinge joint's physics through fixedupdate as opposed to standard update and i'm as confused as you

[unique cave]

Hinge joint physics and all physics are always simulated in fixedupdate/separate physics loop

You cannot set whether hinge joint uses fixedUpdate or regular update, it uses the former regardless

[unique cave]

They likely meant that if you have your own code that moves the hinge parts or something like that, you should use fixedUpdate for that own code too

[stray charm]

gotcha, that makes sense, just a communication then

doesn't solve how to prevent my issue though hmmm

[unique cave]

If the player is rotated in regular update as it likely is, you could really simulate the lantern movements yourself and not rely on the physics system if you don't specifically need physics for it

[coral mango]

Unless it needs collision, I'd agree that using a springbone script is likely easier.

[soft prawn]

Has anyone here every really messed with motorcycle physics before? I've been working on some for a while now and there are some aspects I would like to discuss and see whats going wrong

[surreal oasis]

Anyone here having problems with Physx in Unity 6000.0.25f1 LTS that the moving object disappears from scene when colliding with other objects in 3D space?

[surreal oasis]

upgraded to 6000.0.27f1 LTS but still having the same disappering problem of the ball, even I don't have any script that destroys it 😦

[timid dove]

Physx wouldn't destroy a GameObject

It can only move things

[surreal oasis]

but is there any case where object destroys itself for no reason? Because I checked the scripts, there is nothing in there that Destroys it, downgraded it to 2022 LTS and still having the same problem. What should I do in physics settings?

the object that destroys itself, is a ball for a Pinball game

OMG, now i see that it was autodestruct enabled on trail renderer, omfg xD

yep, that fixed the problem, omfg xD xD

[timid dove]

yeah I was going to mention particle system as a possibility. TR is another one

[white hound]

anyone have settings for a humanoid ragdoll character joint?

[muted brook]

Anyone have ragdoll physics

[royal willow]

they're plenty tutorials the internet you can find and use to make them

[uncut anvil]

Hey guys - Do you know why when I try to get my IK hands to grab onto this Rigidbody object and connect them with a Fixed Joint why it keeps floating in front of me?

For more context: My left and right hands have the fixed joint on them. When I hold left click, essentially the arms move up and the IK target moves to the object and makes the "Connected Body" of the fixed joint for each hand the object that I want to grab

[runic warren]

Is there a way to set up spring joints such that a certain axis of the rigidbody always lines up with the joint anchor?

for example, in my scene using the VR AutoHand package, I'm trying to make a slingshot that'll launch objects, with the spring attached to the handle of the slingshot's "cradle"

right now, I can hold it like this

but I can also hold it like this

I don't wanna be able to do that lol

for context, objects launched with the slingshot would be placed in the white sphere collider next the handle

and the anchor of the slingshot is the orange dot on the right

[finite crescent]

What is the best way to handle rigid bodies getting stuck on walls. A code example would be great 🙂

[royal willow]

use either a physics material and calculate your own friction (or just add another collider solely for walls with the physics material on it) or shoot a raycast out in the direction of movement and if it hits a surface over a threshold AngleOfWall >= Threshold (or just use a wall layermask if you wish to do so) then just ignore inputs if that is true, the latter is the recommended option.

[finite crescent]

Is using physics materials frowned upon? Or is this a good practice to do?

[royal willow]

it is okay to use, but requires more effort or more setup to do than the latter in my previous message, Also an important thing I left out is you would want to do the AngleOfWall >= Threshold check whilst also checking if you are in the air as well, as it will ignore input's all the time even if you are on the ground. I'm also going off the notion that if you walk into a wall and keep on holding input whilst in the air, your rigidbody will get stuck as it is a common issue with movement

[timid dove]

Well the first thing to think about is why that's happening. In real life objects usually can't push themselves in the air magically.

So the realistic solution is to disable the ability to add horizontal force in the air, or at least as long as you're touching a wall

But there's nothing wrong with disabling friction with a physics material either

[finite crescent]

Thanks for the tips everyone, much appreciated

[serene ridge]

Is there something like a spring joint, but for compressive forces? Or am I missing something about how its set up?

🙂 Or maybe I've over-complicating my code... I'm doing something physics based where I have a flat surface that I want to be able to press down a bit on one side, and have it bounce back to the regular position as if there were springs at the corners (really, it's more long the lines of those little rubber legs on a boardgame that give a little but go back in position).

I feel like this ought to be something that min/max distance on the spring joint ought to handle, but the fact that both of them say that it will apply the force to bring them together if the distance exceeds that amount...

[finite crescent]

Hi everyone, me again. I have this speed control and my move speed is set to 5. however in my debug logs velocity is coming out at 5.5? Any suggestions?

I am also aware that my bools do nothing rn

nvm Im thick af

[royal willow]

log your rigidbody's velocity Debug.Log("Rb Velocity is " + rb.velocity) in the if statement in your picture

[finite crescent]

I want the player to shoot a shotgun and be propelled backwards. When I used the ForceMode.VelocityChange method the player just seems to jump backward not having any type of real smoothness. Obviously a shotgun shot is meant to be abrupt but not that abrupt. Any suggestions. Here is the current code:

{
    float angle = Vector3.Angle(GetPlayerMovementDirection(), GetPlayerLookDirection());
    float multiplier = (float)System.Math.Round(angle / 180, 2);

    if(GetPlayerMovementDirection().magnitude == 0)
    {
        multiplier = 1;
    }  

    if (OnSlope())
    {
        if (multiplier > 0.05)
        {
            maxSpeed = e.force;
            rb.AddForce(Vector3.ProjectOnPlane(-1f * GetPlayerLookDirection(), slopeHit.normal).normalized * e.force * multiplier * 5f, ForceMode.VelocityChange);
        }
    }
    else
    {
        if(multiplier > 0.05)
        {
            maxSpeed = e.force;
            rb.AddForce(-1f * GetPlayerLookDirection().normalized * e.force * multiplier * 5f, ForceMode.VelocityChange);
        }
    }
}

[royal willow]

you could try changing forcemode.velocitychange to forcemode.impulse

[finite crescent]

I tried that and it went ballistic

[royal willow]

what do you mean by ballistic

[finite crescent]

It just goes so fast that the player wont know whats happening

[royal willow]

try lowering the force you add to it

[finite crescent]

I have then it doesnt move at all it is very odd

[royal willow]

what did you lower it too

and what was it before

[finite crescent]

it was 20 lowered to about 5

it might be to do with my player mass as its currently 1

that did the trick

[royal willow]

yeah I was gonna ask about the mass of your rigidbody, glad you fixed it

[finite crescent]

what should the mass be realistically?

[royal willow]

the default mass should be fine 🤔 (you can play around with the mass if you want) Were you absolutely sure the force being applied on the rigidbody was 20 or 5 (using debug logs), or were you basing that off of the multiplier variable you have?

[idle estuary]

Hi there,

I have the following code called once each FixedUpdate.

        RaycastHit rayHit;
        if (Physics.Raycast(transform.position, Vector3.up, out rayHit, yVel * Time.fixedDeltaTime + buffer, wallNGroundLayers))
        {
            Debug.DrawRay(transform.position, Vector3.up * (yVel * Time.fixedDeltaTime + buffer), Color.yellow, 5);

        }
        else
        {
            Debug.DrawRay(transform.position, Vector3.up * (yVel * Time.fixedDeltaTime + buffer), Color.green, 5);

        }```
The resulting behavior is presented in the image below.
As you can see, the ray doesn't register the collision with the ground. he layerMask has been correctly selected and the layer correctly applied to ground. No red errors. Any idea on what to check next ?

[timid dove]

he layerMask has been correctly selected and the layer correctly applied to ground
Show us how you set up the layermask and also show us the inspector for the ground object

Also - is this a regular Unity Plane? It will only have one-sided collision afaik.

You're shooting the ray from below

[idle estuary]

It is a cube actually. I'm shooting the ray from the bottom of a "Player", in a upwards direction but with a vertical velocity that is negative. So in the situation on the screenshot, the transform position of said player if above the ground (outside any given cube). The layermask is set up as Serialized Field in code. The LayerMask is then assigned in editor (Image below) and all the cubes composing the "ground" per se have their layer set to "Ground" (image below).

[timid dove]

Raycasts that originate inside a collider will not hit that collider. It says as much in the documentation for Raycast

[timid dove]

Show the full inspectors please

[idle estuary]

I am aware but the player object (and its children of course) doesn't have a collider ! Here are screenshots of the terrain which each individual cube having the setup shown in the second screenshot.

[timid dove]

The Raycast is starting inside the ground collider isn't it?

Since it's shooting up

And rereading this is looks like you're trying to use a negative maxDistance? Raycast doesn't work that way

[idle estuary]

Ah so I just have to take the absolute and cast it downwards

It was my understanding that draw ray and raycast behave the same

my bad

Yep this was the issue 🙂 thx @timid dove

[uncut anvil]

Hey everyone, do you know why I'm unable to connect my hand's rigidbody to this Fire Extinguisher's configurable joint?

It clearly adds it (through script) to the connected body, but when I pull my player away, it doesn't stick to the hand

Trying it with a fixed joint works better, but not ideally, so I'm trying a configurable joint. Maybe a setting is wrong?

[coral mango]

How are you moving the player?

[uncut anvil]

Sorry I didn't see this, man. I use this:

float moveHorizontal = Input.GetAxis("Horizontal");
float moveVertical = Input.GetAxis("Vertical");

movement = new Vector3(moveHorizontal, 0f, moveVertical);

// Clamps movement so it doesn't go too high when going diagonal
movement = Vector3.ClampMagnitude(movement, 1);

[kind mauve]

I gave up having a one rigidbody on a player, now each subcollider have a kinematic one
The problem is
The main parent have a dynamic rigidbody with no colliders, it's the player basically, it's layer is arbitrary as I hope it doesn't matter
The child with a collider and a kinematic rigidbody
which meant to be colliding with the floor
(the layers of the floor and a child collider GameObject are meant to collide in the matrix)
Is not doing it's job of detecting the collision
any advices?

I guess kinematic colliders don't mean to collide?

this is.... infuriating

I was going to use that with a combination with Platform Effector 2D to only get collisions I want

So, everything is 2D
Long story short I want a player to have different colliders for colliding with different things
How do I meant to achieve that?

[coral mango]

That doesn't actually involve any code that moves the player, that is just the inputs.

[kind mauve]

why do I see it twice in 2 days
#💻┃code-beginner message

also can someone explain to me how does layer overrides work on child objects having only a collider but not a rigidbody (parent have one)

do they work but assuming the rigidbody layer?

do they even ever work?

[coral mango]

Unless there's a rigidbody, the colliders are all treated as part of the parent with a RB.

[kind mauve]

so I guess the work but assuming the rigidbody layer

I ve spent to many time going back and forth trying to figure out how to make such usual thing

Should I make a parent body having a no collisions layer and give it the collision managing script and include layers I need on childs with no rbs?

[uncut anvil]

Lol sorry, add that onto this 😄

// Movement
rb.MovePosition(rb.position + moveSpeed * Time.deltaTime * movement);

[tight dock]

Hello

I've been dealing with a problem with ragdolls, that is, I can't seem to get them to be the least bit good looking

Like what the fuck is this

New project, ubiquitous avatar, default ragdoll wizard settings, and it freaks out horribly

I'm using unity 6 but I doubt that would mess up physics

So many videos online use this same avatar and ragdoll wizard look great, or at least playable

here and here

They pretty much just do this and it looks infinitely better

Oh cool they killed physics in unity 6 https://discussions.unity.com/t/ragdoll-is-broken-in-unity-version-6/947075/50

That's my lesson to never use a fucking pre-release ever again

[timid dove]

Looks like it's been fixed no?

[tight dock]

Yeah I updated to .28f1 and it seems to be okay now

Still gonna avoid pre-releases from now on, I just thought that they were problematic for large studios, not the everyday user

[timid dove]

Unity 6 isn't a prerelease since October 17th

[mossy swan]

Hi, figured this might be the best place for this. For those that have implemented raycast bullets that move over time (not instant), did you ever encounter the issue where you could "miss" thin or fast moving objects, due to the fact that the raycasts aren't part of the main physics update? So on Frame N, your bullet would move forwards and end just in front of a thin object moving in the other direction. Internal physics update then runs and moves that object forward, putting it "behind" your bullet... and then on Frame N+1, your raycast now starts ahead of the object and therefore "misses" it. Are there any clever solutions for this? I'm currently thinking of just starting the raycast from the previous starting position and basically make the raycast twice as long. It's still not a perfect solution mind.

[inner thistle]

do the raycasts on FixedUpdate

[mossy swan]

I am. That's not the issue. It's the fact that they aren't part of the main internal physics update, so one still happens after the other.

[inner thistle]

It won't cause the issue the way you're describing it then

[mossy swan]

It really does. Note this is not an "infinite" one frame raycast... this bullet is moving over time.

[inner thistle]

I don't doubt there's an issue, but the internal physics update runs in the same step as FixedUpdate so your code can't be out of sync if it's also in FixedUpdate

[mossy swan]

I'll get a diagram together, it's probably easier to show that way.

[inner thistle]

Is there a reason it's using raycasting instead of physics collision detection? Setting collision detection mode to continuous would solve the problem. With raycasting you'd basically have to re-implement continuous collision detection manually

[mossy swan]

Yeah, these bullets are not rigid bodies. They are just positions and velocities basically. Don't want the overhead of making them full rigid bodies. It's pretty common for bullets to be implemented as raycasts, so hoping others have encountered this issue.

[silver moss]

Maybe in the second frame you could raycast the first frame's path again?

Thats something i have thought but gotta experiment with a bit. It should also make hits more forgiving so sniping moving targets should get easier

Another scenario you have to consider is when the bullet ends up inside the enemy between updates. Raycast wont detect colliders that it starts inside of, so an OverlapSphere before raycasting can be used to fix that

[lost current]

random question fellas

so I'm trying to remake a couple player movement controllers I made for a sidescrolling 2D platformer & I want them to like, not constantly set the rb's velocity to a specific value (as most player controllers do from what I've seen online)

is there a way to do that without having it feeling weird?

[mossy swan]

Yeah that's where I'm currently at. Thinking about just making the raycast start from the bullets previous starting position. It's not perfect because the other object could be moving even faster and "jump" that entire raycast too... But at the same time it should be possible to roughly know what the fastest speed an object in your game can travel.

[mossy swan]

Yeah, hoping to avoid doing that, and just have a solution with the raycasts. But there are definitely cases where what you describe can happen. For example, the object could be moving perpendicular to the bullets path. No change to the raycasts length will fix that if the object moves in from the side and ends up on the raycasts starting position.

[idle estuary]

    {
        if (depth == 0)
        {
            return (startingPostion, initialVelocity); 
        }
        RaycastHit wallHit;
        Vector3 targetPosition = startingPostion + initialVelocity * Time.deltaTime;

        if (Physics.CapsuleCast(startingPostion + new Vector3(0, GameData.playerThickness, 0),
                        startingPostion + new Vector3(0, GameData.playerHeight - GameData.playerThickness, 0),
                        GameData.playerThickness - GameData.collisionBuffer,
                        initialVelocity.normalized,
                        out wallHit,
                        initialVelocity.magnitude * Time.deltaTime + GameData.collisionBuffer,
                        wallNGroundLayers
                        ))
        {
            Vector3 newTargetPosition = startingPostion;
            float distanceToNormal = Mathf.Sin(90 - Vector3.Angle(-initialVelocity, wallHit.normal)) * wallHit.distance;
            if (distanceToNormal > GameData.collisionBuffer) //the player is not presently against the wall so we snap them to it.
            {
                newTargetPosition = startingPostion + initialVelocity.normalized * wallHit.distance /*+ wallHit.normal * GameData.collisionBuffer*/;
            }

            Vector3 vectorToProject = targetPosition - newTargetPosition; //the remaining velocity vector that we have to project
            Vector3 newVelocity = Vector3.ProjectOnPlane(vectorToProject, wallHit.normal) / Time.deltaTime * GameData.wallSlidingDeccel;


            return CollideAndSlide(newTargetPosition, newVelocity, depth - 1);
        }

        //we hit nothing
        return (startingPostion + initialVelocity * Time.deltaTime , initialVelocity);
    }```
I'm looking for critics/ review for my code for Collide and Slide algorithm. Things works only in very specific conditions and I don't know where I went wrong here.

ScreenShot is clearer

[stuck bay]

Actually, is it okay to constantly manipulate this? Physics2D.queriesStartInColliders = false;

[tender vapor]

@cunning wigeon is this the kind of behavior you're talking about?

there's no wavedash code in here at all, just a dash, and the ground collision handles the sliding

you can tune it with drag, friction, bounce, etc

[cunning wigeon]

no i don't think so, if you look up celeste wavedashing or wavedashing in pathfinder or wavedashing in risk of rain returns you can see what i mean

idk

it doesn't look like wavedashing to me but i might be wrong ig

[tender vapor]

im basing it off what i see from ssb

the basic logic is the same, vertical speed is transferred into horizontal speed, celeste's technique seems to incorporate a jump into that to get airtime

[abstract lily]

Okay so does anyone have a tutorial to understand how character joint work... Cause one thing that made me fear making physics game is the character joint, The arrow that should explain don't match up the limit i set. And it just confuses me so badly

I don't know why it like this

it just horrible

[elder peak]

HI, i need help with a small probleme i have, my player collider right now use a capsule collider, the issue with this is when i walk on narrow edge the player will slide of the edge. what type of collider can i use to be able to walk on a pixel wide edge? im trying to mimic the movement of csgo kz game mode. thank you

i tried to use raycast to detect the edge of the ground but it does weird thing like making the player jump 3 time as high as normal and other stuff

[timid dove]

box collider or changing your controller code

[timid dove]

raycasts do not do this. Your code does this.

[elder peak]

thank you, i think pretty much everything is possible in unity but how hard would it be to make a slope physic using a box collider because they all use capsule

[timid dove]

Depends how good you are

[elder peak]

right, that kind of was a dumb question of me XD

but its possible to make it smooth right

[brittle crown]

So i have a ragdoll where the Bones are parented to each other, and this causes bugs when making a child bone Kinematic, if i unparent all of them: Would the Rig still be the same?

[timid dove]

and this causes bugs when making a child bone Kinematic
Such as?

[brittle crown]

So if i make a child Kinematic and move a parent, it bugs like ALOT and starts spinning and going crazy, ill make a screenshot

Yeah the ragdoll just goes crazy

[timid dove]

how are you moving the parent

[brittle crown]

I just checked and it also happens when just not moving the parent

but all i really wanna know is if the Rig Stays the same if i unparent the Bones?

[timid dove]

No - the skinned mesh renderer depends on the hierarchy

[brittle crown]

Alright, Thanks

[uncut anvil]

Hey guys - Do you know how to close the gap between a Fixed Joint and its connected body? When I attach mine to the hand of my guy (left hand in screenshot), it always has a gap between the object and the hand.

I just add the fixed joint when it hits a small trigger (sphere) around the hand, which has a radius of .25

[timid dove]

how is the hand moving

[uncut anvil]

Code for attaching Fixed Joint

// Check if the collider is interactable
if (collider.CompareTag("Interactable"))
{
    grabbedObj = collider.GetComponent<Rigidbody>();

    grabbedObj.AddComponent<FixedJoint>().connectedBody = lHand.GetComponentInParent<Rigidbody>();
}

[uncut anvil]

When I hold left click, it moves the arms towards a target using animation rigging

If the arm swings by anything interactable, it grabs it

[timid dove]

If it's not moving via physics, bad stuff is going to happen.

As for the positions they're determined via https://docs.unity3d.com/ScriptReference/Joint-anchor.html and https://docs.unity3d.com/ScriptReference/Joint-connectedAnchor.html

[uncut anvil]

Yeah it seems to be that way sometimes..

Idk how I would do that with physics though. Raising the arms, that is

Right now I have a rigidbody ragdoll that I use to move my player. Everything else is moving via physics

I'll try messing around with the anchors though. Thanks again Praetor

[stuck bay]

I assume theres no way to simulate physics at different rates for different objects?

• this can be done with physicsScenes but then they cant interact with each other iirc. And i need that functionality

[exotic coyote]

Anyone here done car collisions before with rigidbodies i am having sticky and rotation issues

[unique cave]

Someone likely has. Make sure to post screen recordings, relevant code etc. to make it possible to help

[exotic coyote]

Will do once i get home 🙂

[timid dove]

I assume theres no way to simulate physics at different rates for different objects?
To what end?

The physics simulation is one big self-contained thing.

[vast rose]

Hey guys, a question, I made a script to throw objects in a certain direction calculated by a raycast from the camera, everything works perfectly except when the character looks down, the rigidbody can't get enough force horizontally to reach the point, why ?

        Vector3 forceDirection = (itemHolder.transform.position - targetPosition).normalized;
        Vector3 horizontalForce = forceDirection * throwForce;

        // Add a fixed vertical force to create the arc
        Vector3 verticalForce = Vector3.up * throwUpwardForce;

        // Final force to be applied
        Vector3 forceToAdd = horizontalForce + verticalForce;

        // Apply force to Rigidbody
        itemRigidbody.AddForce(forceToAdd, ForceMode.Impulse);

(the target position in this case is the camera raycast)

the item holder is positioned to the right in the player's view

[timid dove]

first off:
(itemHolder.transform.position - targetPosition) this is almost certainly backwards

A - B gives you the direction from B to A

Second - you're basically just fudging things here

I don't know exactly what the end goal is here

but if your goal is a perfect parabolic arc that lands at the target, you need to actually do the math to calculate the ballistic/parabolic trajectory to give the correct initial velocity.

[vast rose]

Yes, I started doing this because I also think it is the solution, but I was in doubt, should I use AddForce or just apply the necessary linearVelocity directly?

[timid dove]

You might as well just apply it directly, but it makes no difference

the only effect of AddForce is to change the velocity.

So you should do whatever is simpler for you

which is likely assigning the velocity.

[vast rose]

Perfect, thank you very much for your help!

[glass bramble]

hi
i am begginer and i was creating my first level using probuilder.
but when i hit walls the charcter randomly goes up and stop usally and the top of the level
i tried and when i put a norrmal 3d cube it does not do that
also my walls and ground are all the same object
what should i do?

[woeful egret]

when i slide, i set collider size to be smaller, for some reason the collider goes through the floor for 1 to 2 frames.

(This also causes me to be not grounded which ends slide.)

Anyways just wondering is this a bug with capsule colliders? Or a known issue? The collider im standing on is just a box collider.

[vast rose]

Hello again! I implemented my function but the problem remains the same, it seems to be a problem with the rigidbody, I really don't know what is happening, when I look down too much, the rigidbody just doesn't go where it should

(the error starts after 12 seconds)

[timid dove]

Show your code?

[vast rose]

        Vector3 displacement = targetPosition - itemHolder.transform.position;
        float distanceHorizontal = new Vector3(displacement.x, 0, displacement.z).magnitude;
        float flightTime = Mathf.Sqrt(2 * distanceHorizontal / gravity);

        Vector3 horizontalVelocity = new Vector3(displacement.x, 0, displacement.z).normalized * (distanceHorizontal / flightTime);
        float verticalVelocity = (displacement.y + 0.5f * gravity * flightTime * flightTime) / flightTime;

        Vector3 finalVelocity = horizontalVelocity + Vector3.up * verticalVelocity;

        itemRigidbody.linearVelocity = finalVelocity;```

[timid dove]

float flightTime = Mathf.Sqrt(2 * distanceHorizontal / gravity);
this isn't quite correct... 🤔

[vast rose]

t = sqr(2h / g) no ?

I researched and this is the formula I found

I'm using the same formula to draw the trajectory, and it seems correct, but the rigibody doesn't seem to obey

[timid dove]

but you're using the wrong v

that's the formula for calculating how long something will be in the air for a given starting vertical velocity

the horizontal velocity doesn't make sense there

or actually it might even be the formula for calculating the maximum height achieved?

Either way it doesn't really make sense there

[vast rose]

Nevermind, I just found out that the reason it wasn't working was because the rigidbody was colliding with the player's body 🤦‍♂️

Man, I hate game dev sometimes

As I said, my formula is not wrong because I am using the same formula to draw the path, and the path is normal, it had to be something to do with the rigidbody

[hoary kelp]

Not sure in which channel to ask this but I'll ask here since It's about physics. So I run into a "problem" that when a characeter jumps near a wall, then it would have friction, I added a 0 friction material to my character (which is only a capsule for now). Will I run into problems later? I'm still learning unity

[timid dove]

you will run into problems later if you have a need for the player to have nonzero friction

if you don't, you won't.

[hoary kelp]

well, I'm not seeing where I would need that, but for example, let's say I make a terrain where the players needs friction. I could just add an if that adds friction to it when on that terrain no?
But I'm thinking super ahead right now

[timid dove]

Well it would be more than an "if", but yeah it's not worth worrying about right now

[hoary kelp]

yeah, I understand! thanks

[gray sentinel]

As far as physics are concerned, submarines would effectively use the same rough model as aircraft with the additional effects of buoyancy right?

[royal willow]

If you mean are the physics of an aircraft and submarine the same, no they are not. If you are talking about would they look the same, no they would not. I'm not entirely sure what you mean by "same rough model" as they would be made to fit the physics of each vehicle

[gray sentinel]

Dang ok.

I haven't played with physics much, but I was looking at putting together some physics assets from the store so I could have ground vehicles, boats, submarines, planes, blimps ect and everything in-between. Not looking for flight sim levels of detail here but just trying to get an idea of how that would work and how to think about it.

Have to identify the discrete systems and they have to interact with each other in a functional way.

[royal willow]

sure, just look up the physics of each vehicle and what they use (e.g hydrodynamics for submarines, aerodynamics for aircraft)

Each system is its own thing, they may be "similar" but they will be entirely different on how it actually works (e.g how the foils on a sub are "similar" to airfoils)

[gray sentinel]

I'm not going for super realistic either, so in the approximation, the fluid dynamics of a subs dive plane wouldn't be much different than that of an aircrafts elevator right?

You end up just applying a directional/rotational force based on the deflection of the control surface which is then multiplied by some coefficient of it's velocity?

[royal willow]

well air acts like a fluid so yes they act the same

they try and produce the same effect, I'm an aviation guy so I don't know a deep dive into submarines

[gray sentinel]

" deep dive into submarines" har har

[royal willow]

I was not even planning that lmao

[gray sentinel]

excluding them as a control surface, I woudl think that a submarines dive plane or any of it's static planes would be a symmetrical airfoil with no angle of attack

[royal willow]

if any of its foils are level, then sure I guess you could say it has no AOA (AKA 0)

[gray sentinel]

no AOA relative to the shape of the vessel I guess

[royal willow]

well AOA is the chord line to the relative to the oncoming wind, but I guess you could change that to oncoming water? I have no clue

They probably have some new terminology for it

[gray sentinel]

I think there is a bit less to worry about when it comes to water, as flow seperation isn't much of a thing?

and I don't want to deal with cavitation

So what I need is an asset for some good driving physics with tires/tracks. a decent asset for aerodynamics that I can use a simplified version for water, and an asset to address bouyancy

ohhh.. a blimp would also use bouyancy

in order to make blimps work you need air density at different altitudes

[royal willow]

sorry, I was doing something. there are tons of driving physics out there and there is this github page which has some good aircraft physics https://github.com/brihernandez/SimpleWings

I used this before and it was quite good, there are also other github assets out there but I have not tested them

as for buoyancy there are some unity tutorials that have wave buoyancy (if you want waves that is, recommended for semi realism/arcady to actual realism)

[gray sentinel]

I'm not 100% sure I want waves

[royal willow]

you want like a flat plane?

maybe with a shader on it to look like water

[gray sentinel]

Yeah a shader, I'm just wondering if the gameplay impact would be beneficial or detremental

Turrets on a ship would have a harder time in rough sees, but something on land shooting at a ship wouldn't

it would be a heavy environmental impact on balance that is assymetric between land/sea and air

I guess if I have the sea's be always consistent it would be easier to balance against it

That SimpleWings looks good

The other asset on the store I was looking at went a bit too in depth on airfoils so this might be better

[royal willow]

oooh you are making a battle scenario using land vehicles/aircraft/watercraft? If you make waves happen in the open sea and not near the shores it would be harder to shoot other enemies, if you just make it a flat plane but a lot of terrain it would heavily depend on how big the terrain is and the placement of it

you would have to test though as it would depend also on the ships speed, fire range and among other things

[gray sentinel]

Yeah, something similar I guess to From The Depths, but without it's nightmarish building system.

[royal willow]

if you do go the flat plane route, the buoyancy is much easier to implement as you do not have to do any calculations for the current wave height to move the object up and down

[gray sentinel]

That's true

I'm thinking of going with something in the asset store for bouyancy and see if it can fit my needs

[royal willow]

yeah the asset store or github

github has some gems you can find

[gray sentinel]

hmm I'll take a look there

I haven't done too much 3d, so this project is probably out of scope for me, but I'll see how far I get I guess

[royal willow]

just don't burn yourself out, Also try not to get stuck on a certain thing for to long

[gray sentinel]

I mean, that's what's kililed the last 3-4 projects I've done

There is always something complicated I try to setup and I get burned out on it

I haven't scoped this project much, I'll probably make a prototype and realize that it's too much and go back to making something in 2d lol

Anyways, thanks for the info

[royal willow]

if you space things out accordingly, don't get burned out and use resources such as documentation, articles and this discord then you should be fine.

[wraith berry]

Hi guys im trying to make a fishing line with joints and collision to realistically simulate it, spring joints are just giving me grief is there any other joints people might reccomend

[bleak umbra]

you may want to consider making a custom simulation. game physics are generally not suitable to produce realistic results.

[wraith berry]

thats for the advice il have a look into it

[wind charm]

i'm building an ar drone game where user need to fly drone passing though some obstacles. here's my collission setup, the problem is that. when the drone touch the sphere in the middle of the obstacle, it just don't move maybe because of the rigid body, but still it should trigger ontriggerenter since i use isTrigger for the sphere collission but it didn't do anything, i add log but there's no log

it only trigger when the edge/line of drone box collider touch obstacle's edge/line of the sphere collider. why is that?

[silver moss]

So the drone cant go through the ring? Make sure the ring colliders are not blocking the hole. Uncheck "Convex" in the mesh colliders

[pale mist]

I was reading that unity's physics is deterministic, yet theres a difference in the physics between the different platforms

the one on the left is for PC while the one on the right is for mobile

[silver moss]

Where did you read that it's deterministic?

My understanding is that PhysX is not deterministic across devices

Anyway, the difference in your video is pretty massive so I don't think it's just a determinism issue

[timid dove]

Unity offers deterministic physics engine but the default one is not it.

[silver moss]

Perhaps your code is framerate dependent

[timid dove]

But yeah most likely it's that ^^

I.e. adding forces in Update

[pale mist]

The only other deterministic physics package that I saw linked was last updated almost a decade ago

dphysics

[silver moss]

I know that there's DOTS physics, Havok, and Bullet available

Those are either deterministic or have the possibility to be

In any case could you show the code you use for jumping before we talk different physics engines 😄

[pale mist]

Ooh I heard of bullet sdk, I thought it was only for python or self-built for c projects

Ill look into it

I'll look at my other code again though

i might have to git blame someone today

thanks for helping me

[silver moss]

Is canJump continuous or does it last for one frame?

Like can you hold down the button and it applies it every frame?

[pale mist]

It was the rb.addforce, told someone else to increase the gravity of the player but i didnt think that it would be done this way

and since it's called in update it did become framerate dependent

[silver moss]

Right, that's the issue

You should add the continious force in FixedUpdate

[pale mist]

yep ill do that 🙏

[silver moss]

How is the stick following the cursor?

What components are you using and what does the code look like?

[tranquil bridge]

Is there a way to make a 2D physics material not multiplicative, but a maximum instead like the 3D ones? I'm trying to let a player collider with 0 friction to interact with a slope collider that should be stopping it completely

nevermind ig 2D doesn't have that option, guess I'll have to implement that myself

[coral mango]

Or just add an invisible wall

[loud ridge]

do you know how can I orevent this? That stick follows my mouse cursor.it should just rotate without lifting my character.But also I should be able to lift myself by holding sword down so mass hsould be bigger.
I use a hingejoint2D and a script like this.
https://hatebin.com/lyodtprcpb

[timid dove]

well this is the natural consequence of such movement - Newtons's second law

so to cancel it out, you could calculate exactly what the force would be that you're exerting on the player and apply an opposite force manually.

[zealous violet]

Hiiii, I am a unity beginner, i am confused how to make two object collide with each other

[lone frost]

hey so i got a project file which converts 3d model/animtion into pixel art sprite sheet
here is the link:-https://www.youtube.com/watch?v=kALXAWSDYEo&t=5s

but i need to use legacy rig for this to work and when i convert my fbx into legacy rig it becomes invisble like shown in the images but is visible in any other rig mode
how can i fix this please?

[timid dove]

What's wrong with what you have?

[zealous violet]

that pink cube falling through plane, i want it to collide with plane

[frigid jetty]

does the plane have a collider attached?

[timid dove]

Make sure they both have colliders
Make sure the Rigidbody of the cube is using Continuous collision detection

[zealous violet]

okay thanks i will check that

[rain citrus]

Hi. I'm using Unity Physics 1.3.8 and when I disable Automatic Tensor on the rigidbody, I get fields to enter a custom tensor. But whatever I enter there, nothing changes in the simulation. I can set it to zeroes and it requires a lot of force to start rotating and I can set it to 10000s and it still behaves the same. With Automatic Tensor enabled, the spheres can be rotated with much less force already. Is it expected that these fields have no effect?

[frigid jetty]

There was a point in time when these constraints were not implemented yet, I don't know if that has changed, let me see if I can find out

[rain citrus]

I just saw that the Inertia Tensor on the baked PhysicsCollider component is always set to 0.1s no matter if I disable Automatic Tensor. I would guess it's currently not converted yet?

[frigid jetty]

rather than a value try typing the word 'infinity' into your tensor value

if it's still rotating then it's not implemented

[wind quail]

anyone knows why my cars are still colliding with each other?

i have assigned the children of the prefab to that layer too

[rain citrus]

Yes, it behaves like any other value I put there. Thanks for looking into it!

[timid dove]

Is that the Physics 3D settings or 2D settings?

[wind quail]

oh wati

wait

[timid dove]

since it says cloth, I assume it's 3d

[wind quail]

yes

thanks haha

[stuck pollen]

Yo im trying to make a dash and since nothing I do seems to be working lol i decided to ask chat and it came up with this that also doesnt work(its supposed to dash on z):

Any helpers in chat?

[timid dove]

Most likely the problem is due to your other movement code overwriting the velocity

[runic warren]

What's the most proper method of applying counter-rotational forces? For example, if I have a object meant to act as a motor spinning some blades around in a circle, I want the motor itself to also be trying to rotate the opposite direction

[timid dove]

AddTorque

[runic warren]

Follow up question, is there a way to lock the rotation axis of those blades to the object they're parented under without using joints? freeze rotation is locking the rotations relative to the world instead of locally

[timid dove]

nope.

If this is a machine like a helicopter, you're going to have a lot of trouble accurately simulating it in Unity

You might want to look into ArticulationBody instead of Rigidbody for machine simulations: https://docs.unity3d.com/6000.0/Documentation/Manual/class-ArticulationBody.html

[runic warren]

The asset I'm using for object interaction requires that it has a rigidbody, as well as that it doesn't have any other components reliant on said rigidbody

at least for what I'm trying to do with it

[latent oar]

Hello, I've been messing around with unity for around a month now and I was wondering if there's a generalized best practice when it comes to gravity.

Should I use the inbuilt gravity system attached to rigidbody component, or would it be better to simulate gravity using my own system?

For context, this is regarding a 2D platformer

[royal willow]

you can use the built in gravity system unless you need a custom one (i.e. custom gravity directions)

[timid dove]

What would you hope to gain by creating your own gravity system?

General best practice is to use the built in gravity unless and until it fails to meet the needs of your game.

[narrow quiver]

every time i joint many objects together they start flying

[silver moss]

How did you join them, show the objects

[narrow quiver]

ok so firstly i have set of objects with rigidbodies that in runtime generate themself shape to pick

then i scan for everything that touches and assign connector between them, which is 2 configurable joints with all motion types locked

the more pieces i connect the more destructive it gets

[silver moss]

Having 30+ interconnected joints is probably never a good idea

Is this for the destructable car you mentioned? Why not just have them as children (without rigidbodies) and add the rigidbody to the part when it detaches?

Like I suggested before

Also note that rigidbodies should not be children of other rigidbodies unless the child is kinematic

[narrow quiver]

how i would make wheel without children?

also i probably can make on parent but then joint break force wont trigger

[silver moss]

What does that mean?

Are you using actual physical, rotating wheels or what?

In any case a non-kinematic rigidbody should not be a child of another rigidbody.

[narrow quiver]

yes

ones that start spinning and move car along when adding force

wheel colliders bad since they only work on raycast down

[silver moss]

So how is that related to what i said

Why do you expect that the wheel has to be a child

[narrow quiver]

since i cant really make wheel prefab otherwise

i should keep all parts loose somewhere?

[silver moss]

They can have an empty parent

Like actually empty, no rigidbodies or anything

And do not move that empty parent

[narrow quiver]

i assume so since constraint does all work moving them already

didnt know parenting changes that much

[silver moss]

Parent-child relation moves the child's transform and this will conflict with the rigidbody trying to also move it

[silver moss]

I would just use non-rotating wheels and add the forces manually, depending on if the wheel is on ground or not

But if you can get physically rotating wheels to work reliably then great

[narrow quiver]

they worked at the point where i was jointing everything to single rigidbody

now instead im connecting between 2 different ones each time so i need redo wheel mechanic once again

ok but now

that gives me networking issue

how would i allow rigidbody physics to be networked then

if im syncing x gameobject position

another constraint for every object to hold onto these objects?

[grizzled tundra]

hey guys...so my wheels run away
https://pastebin.com/KtkS7M0h - Script that causes this

[frosty ore]

its most https://docs.unity3d.com/Manual/WheelColliderTutorial.html likely ur hiearchy setup..

Should be something like

• Bike (w/ the rigidbody)
  • Bike Graphics
  • Front Wheel Graphics
  • Rear Wheel Graphics

[grizzled tundra]

ah okay thanks

[frosty ore]

wheel collider and wheel mesh should end up being on the same level

the fact that ur colliders are nested inside hte graphics may be the issue

havent ever done a bike.. but im guessing its pretty similar to the =same issue

[silver moss]

That's probably it

[frosty ore]

always makes wheels YEET

[silver moss]

The wheel collider modifiers the world pose -> you assign the world pose to the graphics -> wheel collider is child of graphics

So it is a feedback loop

[frosty ore]

thats the technical jargon i lack ^

[grizzled tundra]

ah i see, wheel go yeet cos of recursive

[frosty ore]

bet!

makes a good clip tho 🤣

[grizzled tundra]

🤣

i suppose i have another issue now....the wheel dont spin right

this is all for the rear wheel, the front is the same issue

[frosty ore]

now it appears the wheels offset doesnt match the graphics

[grizzled tundra]

yeah i cant win here

[frosty ore]

heres what u may have to do.. (wheel colliders work best with objects that are scaled 1:1:1 with the pivot in the center)

you may need to make a gameobject as the wheel (an empty) and adjust the graphics soo that the wheel is centered..

then that empty gameobject acting as the wheels main object now has as scale of 1:1:1 u can use it instead of the actual mesh

[grizzled tundra]

ah okay

[frosty ore]

alot of models have shitty wheels..

artist dont take the time to make sure the pivots are correct

[grizzled tundra]

i am said artist rn 😅

[frosty ore]

lol.. no worries 😛

thats even better tbh..

u can pull it back into ur modelling software and edit the pivots a bit better

also apply the scale to zero it out.. (1:1:1)

[grizzled tundra]

oh boy a new thing i gotta learn rq 😅

[frosty ore]

its easy do u use blender?

[grizzled tundra]

yeah

[frosty ore]

i can show u real quick..

[grizzled tundra]

please

[frosty ore]

for the pivot u can just select the wheel and set origin to geometry.. If its a true circle.. it should get pretty close to center

for the applying the scale.. u just select the object in object mode..
Ctrl + A then Apply -> Scale

then when u re-export the wheel should have a scale of 1:1:1

if its not.. it may be because its nested inside the whole object (bike) and it may need its scale applied so it becomes 1:1:1

funfact: (always make backups)

u can check visually w/ the Info tab on the right...

should show u both the Scale u currently have... and the actual dimensions below it

[grizzled tundra]

i think ive done it

done it to both

[frosty ore]

yup.. says 1:1:1 now..

and u should be able to rotate it in the scene perfectly if its origin matches the center

[grizzled tundra]

u are a life saver honestly

so glad i asked for help now

[frosty ore]

remember to utilize these too

u can change what origin ur using to rotate and whatnot.. and next to it <--u can change from global to local.. to make sure stuff looks as it should

[grizzled tundra]

gotcha

i use those quite a bit

[frosty ore]

good, good.. welp good luck 🍀

gotta get to work myself

[grizzled tundra]

thanks a bunch

[grizzled tundra]

hey guys what do i need to adjust to stop the bounciness of the suspension?

also bounces when my character isnt mounted

[silver moss]

Did you play with the suspension settings?
https://docs.unity3d.com/6000.0/Documentation/ScriptReference/WheelCollider.html

[spare valve]

is there a way to make a 2d collider from a mesh?

[edgy hamlet]

does two object collide without rigidbody? i just want them to stop when they hit

[spare valve]

i thinnk collider without rigidbody is ment to be static

[edgy hamlet]

i see, so same apllies for multiplayer i guess

i need to setup with physics enabled

[silver moss]

What does the mesh look like?

I suppose you could use PolygonCollider2D: https://docs.unity3d.com/ScriptReference/PolygonCollider2D.html
But you need to somehow determine what vertices form the 2D edge

[spare valve]

thats what i was doing but i was wondering if there was a nother way

[grizzled tundra]

Yeah i had to do some maths to determine the spring and damper. Managed to work it out in the end

[twin nebula]

I'm assuming that unity just adds the velocity onto the position in the next physics step right? Assuming there's no drag and stuff

I'm trying to predict where my rigidbody controller will end up because it keeps hopping up and down on slopes

[grizzled tundra]

is there a way to redefine where the suspension shoudl react from? My bike looks ridiculous with the back wheel just moving about off the axles and forks (replied to message to show video)

[timid dove]

assuming NO drag and NO collisions and NO gravity and NO AddForce calls, you can predict the objects' position next physics frame with:

Vector3 nextPosition = rb.position + (rb.velocity * Time.fixedDeltaTime);```

[twin nebula]

Thank you 🙏

[orchid sinew]

Hey guys, I'm trying to set the player on fire when they touch the red object but it throws itself out of the sphere when it touches the first time. Anybody know why?

[timid dove]

Character controllers are designed to not overlap physical objects

[loud ridge]

how can I orevent it from happening.I mean that vibrating effect.

I cant make weapons trigger because I want them to collide with shields I have and othr weapons.

[narrow quiver]

if car is heavy and im pushing it with 1 force will i be able reach same force (for example 10) and car will be slower while force remain the same?

[timid dove]

This question doesn't make sense. Are you confusing forces with velocity?

[narrow quiver]

im using addforce()

perhaps i should be using something for velocity instead

[timid dove]

No, you conceptually are confusing the two

they are related concepts but your question makes no sense

so I have no idea what you're asking

[narrow quiver]

im not sure to which one im refering

but what i was asking for is if car is heavier does it have to go with slower speed to reach it

[royal willow]

yes that is how physics works, F = M * A (newtons second law) so the heavier the car is the longer it will take to reach top speed

if you set the velocity directly, it will immediately reach the desired speed which is unrealistic (unless of course you ease the value to the desired speed but this is more of an arcady approach)

[narrow quiver]

no i wanted to do something like

im applying constant 1 force from behind

but i noticed it doesnt do it consistently

instead it accelerates

so i wanted limit it to for example 10 force max

but also wanted to limit it based on how heavy car is

so if its heavier then max speed is lowewr

[royal willow]

so all you want is a constant force but incrementally increase that force until it reaches a force of 10 (or based off of the weight of the vehicle)?

or do you want add a force to the car and clamp the max speed of the vehicle to 10?

[timid dove]

slwoer speed to reach what?

[timid dove]

Yes, consistent acceleration is what a force does.

[timid dove]

Force is not something you limit. Don't get confused between limiting the -force- and limiting the velocity

increase that force until it reaches a force of 10
This is a completely meaningless statement. Force does not "reach" anything, nor does it build up over time

it is something you apply immediately with the AddForce call.

[royal willow]

yes I know, I was asking if they wanted to increase the force that is currently being applied to the vehicle over time, the question they are asking does not make sense

[narrow quiver]

max speed of a part that gives speed

so if rest of car is heavier the limit is lower

[royal willow]

so all you want is to do is to make the max speed of the vehicle based off of the weight of the vehicle?

[narrow quiver]

i simply want acceleration curve to fit in 1 second

it takes me like 10 secs to reach max speed

[timid dove]

then just use ForceMode.Acceleration

[royal willow]

you could have just said that

[timid dove]

and mass will be factored out of the equation entirely

[narrow quiver]

i thought that isnt what i want because it ignores mass

[timid dove]

that's exactly waht you want

it doesn't ignore mass

it handles the mass part for you

so that you can ignore mass

internally it is multiplying the mass into the force it applies to cancel the mass out.

[narrow quiver]

well i think it works better now

[compact rampart]

Hey, does anybody know how to make a physics material for ice in 3D? (I'm using Unity 2023.1 LTS btw, not Unity 6)

[silver moss]

Wouldnt you just use low friction values?

[sonic jetty]

I can't seem to figure out what Unity changed the "Used by Composite" option to in Unity 6 on the Tilemap Collider Object. Has anyone been able to figure out what the same functionality was moved to? Would it just be the "merge" operation?

[compact rampart]

I'm sorry if I seem dumb, I'm completely new to Unity and game dev itself.

[silver moss]

@compact rampart And probably Friction combine: Min

[compact rampart]

Okay, I'll try that

[silver moss]

Set dynamic and static friction to 0

On the physic material

[twin nebula]

I have a configurable joint setup and I was wondering how I could let a rigidbody move freely but have the joint keep it a minimum distance away from the connected body

So for example if I wanted a max distance of 5 units the rigidbody could move as close as it wanted but no further than 5 units

I tried linear limit but the limit seems to do literally nothing

Either it does nothing or it's extremely inaccurate

I also want it to be a hard limit not some sort of springy stuff (although it can be a tad springy)

[distant crater]

So I'm having issues with slopes and I'm going nuts about it.

[silver moss]

Show the inspector of the joint that you tried, with the limits etc. visible

And how are you moving the rigidbodies?

[twin nebula]

The orange capsule is the main collider which the hand is attached to. That's moved with AddForce

[twin nebula]

There's not really a way to fix what you want in a good way which is why I made my capsule float above the ground and adjust itself depending on the floor below

If you want I'm happy to share the movement code as a base

[silver moss]

Yeah show the code

[twin nebula]

Alr give me a mo

https://pastebin.com/GgHDgASA

Actual movement of the collider is between line 57 and 66

[silver moss]

The issue might be that your hand is a child of the player rigidbody. Is that the case?

[twin nebula]

Player is just the container main collider is the rigidbody

Linear drive does.. something? It's like the hand becomes limp on the floor and will randomly decide to go to where it should be

[silver moss]

I can't see anything wrong from a quick glance:
-Joint limits seem ok (although I didn't work with joints in a while)
-Code seems ok - you are moving the rb with with forces

It's worth to try adjusting Mass scale/Connected mass scale

[silver moss]

Linear drive is another way of controlling the position yeah.

[twin nebula]

Dang mass scale was a good call

I've changed the limit to 0.1 and changed the mass scale to 10 (so the player isn't dragged around by the hand as much) and it looks a lot more stable

[silver moss]

Oh nice

[twin nebula]

But unfortunately it just crumbles the minute use gravity is enabled

I probably should've tried that before sending that message

[silver moss]

Let's step back a bit, what are you trying to accomplish here?

Like what game mechanic

[twin nebula]

I'm trying to have the left hand (the ball) attach to the main collider and act as a floating physical hand that I can attach to objects

Instead of it being hard locked at a certain distance from the main collider I want it to retract but have linear limit sort of cap how far it can extend out

[silver moss]

I see.
The "Player" object is not moving at all, correct?

[twin nebula]

Yep

If worse comes to worst it wouldn't be a bad idea to write my own sort of cap instead of using a joint I guess. That way I could keep the hand completely separate from the main collider and decide whenever the hand can influence the body

[silver moss]

Yeah, controlling the hand rigidbody manually is a good option too

It's definitely doable with joints but maybe a bit overkill

And can be tricky as you noticed...

[twin nebula]

Yeah lol I hate joints sometimes but they can be awesome

Well thank you for the insight and I appreciate the help

[distant crater]

but do you think it'll help me with how I launch myself off slopes sometimes?

[twin nebula]

The movement I have? Yeah

It snaps up and down steps / stairs and wont shoot off slopes

I didn't give everything so that movement isn't just drag and drop (and you'll probably have to get rid of a few network related things in it)

[distant crater]

ok, i'll check. thanks

[twin nebula]

Was just to kind of give an idea but if you need help setting it up lmk

No worries#

[honest river]

Trigger collides and calls OnTriggerEnter2D with another trigger of a layer that I disabled in project settings.

Character GameObject: Rigidbody2D, BoxCollider2D(Trigger enabled), Layer: Character
Gate Goal GameObject: Rigidbody2D, BoxCollider2D(Trigger enabled), Layer: Gate Goal.

In the physics Character and Gate Goal are disabled. Yet, it collides and generates events.

[timid dove]

Make sure you changed 2D settings not 3D

Also make sure there are no other colliders on the objects

[honest river]

Ah I see yeah I was changing 3D settings

Thanks

So there is no way to have the same collider to intersect with some layers and block other? I'll have to setup different colliders to block and intersect and make them the child game objects of the main game object?

[uncut anvil]

Hey guys question from a failed attempt by me -

How would you guys go about attaching a rigidbody to the hand of a player character?

I've tried using a trigger on the hand and "on enter" creating a fixed joint on the item that attaches the hand as the connected body. However, whenever I do that, the object is always flying around unpredictably.

I'm open to any ideas you guys may have.

[serene flower]

much simpler, make it a child of the player hand and make it kinematic until its released

making it kinematic, forces, collisions or joints will not affect the rigidbody anymore. So it wont be flying around

[uncut anvil]

Ok, that's a much simpler solution, I agree. Do you think there's any way to have gravity effect it at that point? As in, heavy object feel heavy?

[serene flower]

Idk maybe make a simple animation making the hand go down to create the feeling that it's heavy, are you making a vr game?

Depends on ur game and how do you want it to feel

[uncut anvil]

It's a top down game, but if we're carrying around a heavy object, it would be nice for it to look and feel heavy

Rather than a person effortlessly carrying around something without weight, regardless of the RB mass

[honest river]

Is it not possible to have parent game object have a trigger collider and a child game object have a blocking collider?

Right now whatever is on he parent game object works and the child game object is just ignored.

[maiden belfry]

Define "works" here

You can absolutely have a non-trigger collider on a child object

and it will be an obstacle for objects with a rigidbody and a collider

[short bolt]

No option in unity to make a joint connected to an object not controlled by physics?
I am trying to have something attached to a weapon the player hold, and have it move as the player walks and moves the weapon. The issue when the player walks, the Character Joint seem to stretch. There doesn't seem to have an option for a joinb where the joint cannot stretch outside of the socket?
There is no way to make a joint rotate only?

[maiden belfry]

This seems like two unrelated questions

Lock the position if you don't want the position to change

I'm not sure how this relates to "an object not controlled by physics"

[short bolt]

I mean... locking the position lock it's position globally. So it's not useful if it suppose to lock onto another object

From what I understand... all joints in Unity physics have springiness, you can somehow force it not to spring. But making it work is kind if weird

I might try to implement my own small physics like code

Just for this specific thing

And use the collision by unity

[maiden belfry]

Only if you haven't actually connected the joint to a rigidbody

Note that you can use a kinematic rigidbody if you want the joint to stick to something that isn't affected by physics

[honest river]

Well, this is the behavior I get:
If I have a trigger on parent and blocking colllider on child, only the trigger works.
If I have a blocking on parent and trigger on child, only the blocking works.

[honest river]

I tried all combinations with rigid body 2D: none, on either and both of them.

[frail fossil]

@shadow heath let me give more context

[shadow heath]

alright

[frail fossil]

so I the top part that read chest roation info

[shadow heath]

you said your hand wasn't a separate rigidbody..?

[frail fossil]

and position info

[frail fossil]

those are pointers for the ik constarints

here are the hands

[shadow heath]

so what do you want exactly? the hands to grab the gun?

[frail fossil]

All this mess started with multi-aim control not working if the target was on the camera

[frail fossil]

yes

I have a parent object that hold the guns position but that gun lags

[shadow heath]

then you would need to create animation states and toggle them via scripts

you can't make dynamic changes to your gameobjects during runtime without scripting

[frail fossil]

I have animation states for basic animations

the object is visually lagging behind when I turn but the cords are not

the whole point of the aim and ik constraints from what I know is that they do not need states

[shadow heath]

then use a script to toggle between them?

[frail fossil]

nessacrily

[shadow heath]

it's not that hard lol

[frail fossil]

I have that?? already

but wont help this issue

[shadow heath]

then rework your set up because that's exactly what you need for that set up.

[frail fossil]

I dont have any states that would interfere with the hand holding the gun?

wait I just had an idea

what if I parent the hand bone to the gun

that might do what I want and eliminate the scripts causing the lag

[shadow heath]

try it out

[frail fossil]

that worked

The issue with implementing scripts the way you are saying is that since I just learned unity I would have to completely tear apart this project which I sadly lack the time for

you are Otherwise completely correct that I need to completely rework my project

but I rn just needed a quick solution

still doesnt make sense why the cordinate rely scripts cause it to lag when parented to the bone constaint pointers but not to the bones themselves, even thought the bones rquire information from the same scripts

[shadow heath]

completely understandable. glad you got it to work. it's a pain to negotiate between these small details when I don't entirely know what you've set up behind the scenes. by the way, how much scripting was involved?

[frail fossil]

a decent amount

So i have an animation handler

a fake parent

a fake parent rotator

input manager

mouse look

Player position changer

[shadow heath]

I was under the impression you had none of that which is why I suggested to manipulate the physics a little, lol

[frail fossil]

its fine

its late for me rn

And I have been debuggin random issues constantly

[frail fossil]

this usually breaks eveyrthing but this time it didnt and it just works

https://www.youtube.com/watch?v=nVqcxarP9J4

I would need a specified gun script if I were to read player input and then if they are turning at a rotation have the gun purposefully lag behind a bit as if it had mass

[shadow heath]

the reason it lags behind is I believe there is some physics acting on it as constraints don't fully block physics interactions. this could be coming from a script or if it isn't a physics issue it could be overlapping of different rotational movements from the orientation. it's happening when your character moves/turns right? or when standing still as well?

[frail fossil]

the object had no physics impacting it?

it would more jitter

[shadow heath]

if your rigidbody uses constraints, physics still act on it

[frail fossil]

well my gun doesnt have a rigid body

if I add a rigidy body the jitterness disappears but then physics come

[shadow heath]

you just set it up? did you remove it

[frail fossil]

I removed it due to it causing more problems

I should very much use a rigidbody

but that would again require a project rework

[shadow heath]

well you might need to rework it to get it to work.

[frail fossil]

I mean it just works now so

Im happy

at the very least I didnt spawn any topology monsters during the creation of this

[short bolt]

It is actually already connected to a rigidboy that is kinematic

[heady latch]

The upper leg is using a motor and the lower legs are using Rigidbody2D.MoveRotation. the lower legs have started to distance themselves from the joints and snap back. Any ideas on ways to avoid this would be apreesh

[grizzled tundra]

Hey guys, im making my own wheel collider as per recommended but i am getting this bouncing effect from it all
Here is the code to what ive done so far https://pastebin.com/KTZppARe

i have mesh colliders on both wheels with convex enabled on them also if that helps

[dapper lichen]

can't give a SPECIFIC suggestion, but I do know that there are a few options for how it computes collision, in the inspector for the collider- I'd start playing with those.

doh! a whole day late- oops

[grizzled tundra]

np xd

im nowhere near i want

[silver moss]

From a quick look at the code, I noticed that you are modifying the object with transform:cs transform.position = targetPosition;
Andcs transform.Rotate(0f, steering * turnSpeed * Time.deltaTime, 0f);
You should never modify a rigidbody's transform directly, it will lead to incorrect physics, bad collision detection, just overall glitching.
Use the rigidbody methods/properties to move and rotate it instead.

Also looks like your character has a CharacterController..? Did you make it a child of the bike? Show more info about your setup

No idea if character controller works should be a child of a rigidbody. You should at least disable its collisions with the bike

[hidden trout]

Hey all, how can i prevent my player from rotating when they collide with a surface? I know i can constrain rotation, but the thing is i still want to be able to rotate in all directions (in 0G), and still not bounce all over the place when they hit a wall.

        if (isColliding) rb.constraints = RigidbodyConstraints.FreezeRotation;
        else rb.constraints = RigidbodyConstraints.None;

I've tried doing this at the start of fixed update, however it still doesn't seem to work. isColliding is set by the onCollisionEnter and OnCollisionExit functions:

    private void OnCollisionEnter(Collision collision) => isColliding = true;
    private void OnCollisionExit(Collision collision) => isColliding = false;

[timid dove]

If you're doing it in response to OnCollisionEnter it's already too late

[hidden trout]

Yeah, I thought that was the case, but I don't know how else to go about doing it without somehow predicting it's movement

[timid dove]

You can probably do this with the contact modification API:

https://docs.unity3d.com/6000.0/Documentation/ScriptReference/Physics.ContactModifyEvent.html

[timid dove]

somehow predicting its movement
This could also be done with raycasts/capsulecasts etc or Physics.SweepTest

[hidden trout]

I'll check it out, thanks

[winter rose]

im using configurable joints for my rope but when and object collides with it above snail speed it freaks out and starts clipping and flying any idea on how to fix this?
either that or it ignores collsion

[winter rose]

unreleated to my first question. but what 2d joint would you use for a pulley system

[bleak umbra]

joints probably wont help you implementing actual (non-goofy) mechanisms. it’s usually best to make a custom simulation for those.

[quick rock]

Hi
If i check for input through the new input system and need to do some physics calculations as reaction to the input it doesnt work. Is there a correct way to do this instead? I tried setting a flag that input for jump has been pressed and in fixed update i wait for the flag to be true and then run the calculation, when i try to print the current velocity in there it prints (0,0,0) even tho in rigidbody component i can see it isnt that.

any idea what to do ?

void FixedUpdate() {
    if (_physicsJumpFlag) {
        // Jump logic, cant have its own function or doesnt work
        _physicsJumpFlag = false;

        // Log velocity before and after applying force
        print("Velocity before jump: " + m_rb.velocity);

        Vector3 jumpDir = m_rb.velocity + Vector3.up * m_Stats.jumpForce.currentValue;
        m_rb.AddForce(jumpDir, ForceMode.Impulse);

        // Log velocity after adding force
        print("Velocity after jump: " + m_rb.velocity);
    }
}


// Receive jump input
public void Jump() {
    // Register jump input
    StartCoroutine(JumpBufferCoroutine());
    // Attempt to execute jump immediately if conditions allow
    TryJump();
}

void TryJump() {
    if (!_canJump || !(m_Stats.isGrounded || _coyoteTimeActive) || !_bufferedJump) return;
    ExecuteJump();
}

void ExecuteJump() {
    _canJump = false;
    _bufferedJump = false;

    m_Crouching?.StopCrouching();
    _physicsJumpFlag = true;
}

when i print the velocity in fixed update outseide of the flag it has correct values

I feel like there is something weird with the physics ticks and the input being offset off of them

[timid dove]

Miossing a lot of info here

JumpBufferCoroutine()
What's this?

Also what are these and what else interacts with them?

    _canJump = false;
    _bufferedJump = false;```

[timid dove]

Also:

when i try to print the current velocity in there it prints (0,0,0)
AddForce doesn't update the velocity of the Rigidbody right away. It doesn't happen until the physics simulation

Also missing from here is how the rest of your movement code works.

[quick rock]

using System.Collections;
using UnityEngine;

[RequireComponent(typeof(Rigidbody))]
[RequireComponent(typeof(FPMPlayerStats))]
public class FPMJump : MonoBehaviour {

    [Header("Jump Settings")]
    public float coyoteTime = 0.2f;
    public float jumpBufferTime = 0.2f;

    FPMPlayerStats m_Stats;
    FPMCrouch m_Crouching;
    Rigidbody m_rb;

    bool _canJump = true;
    bool _bufferedJump = false;
    bool _coyoteTimeActive = false;
    bool _physicsJumpFlag = false; // Physics calculations must be in fixed update or it doesnt work so i have to do this workaround

    void Awake() {
        m_rb = GetComponent<Rigidbody>();
        m_Stats = GetComponent<FPMPlayerStats>();
        m_Crouching = GetComponent<FPMCrouch>();
    }

    void FixedUpdate() {
        print(m_rb.velocity); // This prints correctly
        if (_physicsJumpFlag) {
            // Jump logic, cant have its own function or doesnt work
            _physicsJumpFlag = false;

            print("Velocity before jump: " + m_rb.velocity); // This prints (0,0,0)

            Vector3 jumpDir = m_rb.velocity + Vector3.up * m_Stats.jumpForce.currentValue;
            m_rb.AddForce(jumpDir, ForceMode.Impulse);
        }
    }

    // Receive jump input
    public void Jump() {
        // Register jump input
        StartCoroutine(JumpBufferCoroutine());
        // Attempt to execute jump immediately if conditions allow
        TryJump();
    }

    void TryJump() {
        if (!_canJump || !(m_Stats.isGrounded || _coyoteTimeActive) || !_bufferedJump) return;
        ExecuteJump();
    }

    void ExecuteJump() {
        _canJump = false;
        _bufferedJump = false;

        m_Crouching?.StopCrouching();
        _physicsJumpFlag = true;
    }

    void OnGroundedChanged(bool isGrounded) {
        if (isGrounded) {
            _canJump = true;
            // Execute the buffered jump if it exists
            if (_bufferedJump) {
                ExecuteJump();
            }
        } else {
            // Start coyote time if leaving the ground without jumping
            if (_canJump) {
                StartCoroutine(CoyoteTimeRoutine());
            }
        }
    }

    #region Coyote and buffer coroutines

    private IEnumerator CoyoteTimeRoutine() {
        _coyoteTimeActive = true;
        yield return new WaitForSeconds(coyoteTime);
        _coyoteTimeActive = false;
    }

    private IEnumerator JumpBufferCoroutine() {
        _bufferedJump = true;
        yield return new WaitForSeconds(jumpBufferTime);
        _bufferedJump = false;
    }

    #endregion

    #region AddGroundedListener

    void OnEnable() {
        m_Stats.OnGroundedStatusChanged.AddListener(OnGroundedChanged);
    }

    void OnDisable() {
        m_Stats.OnGroundedStatusChanged.RemoveListener(OnGroundedChanged);
    }

    #endregion
}

Here is the entire class for context.

the rest of the movement isnt really important. what bugs me is how can the velocity be printed correctly in the fixed update if it is outside of the physics jump flag

The Jump() is what the input class calls. The input works via the new input system so it just reacts to the event caused by new input

[timid dove]

The rest of the movement is very important

Adding forces to jump like this only works if the movement code doesn't interfere with that

[quick rock]

oh yea but what im askin and kinda need to know how is it possible that

void FixedUpdate() {
        print(m_rb.velocity); // This prints correct values 
        if (_physicsJumpFlag) {
            print("Velocity before jump: " + m_rb.velocity); // This prints (0,0,0)
        }
    }

because it does not make any sense to me

its both in the same update function

[timid dove]

That's not possible.

If you're seeing that it's likely two different frames or two different instances of the script

void FixedUpdate() {
        print($"Velocity is {m_rb.velocity} at {Time.fixedTime} on {GetInstanceID()}");
        if (_physicsJumpFlag) {
            print($"Velocity before jump: {m_rb.velocity} at {Time.fixedTime} on {GetInstanceID()}");
        }
    }```
Try this more informative logging

[quick rock]

the fixed time is different..