⚛️┃physics

[rugged kindle]

crazy. the points are already in the vector, that is wild to me.

explains why i cant find much online about it

[silver moss]

Ofc you can write your own tool that parses the SVG (xml?) into points/shapes and create polygoncollider2D's out of those

[timid dove]

Why is that crazy or wild?

Unity uses Box2d for 2d physics. Here are the things it supports: https://box2d.org/documentation/md_collision.html

[rugged kindle]

the thought crossed my mind. then i thought surely someone else did this and now everyone uses that

[timid dove]

convex polygons and line segments are the clost thing

[tender vapor]

(svg is based on xml)

[timid dove]

It's not trivial to just support arbitrary curves for collisions

[rugged kindle]

That makes it all click. duh. thank you for that

[tender vapor]

svg supports bezier curves, text, css, animation, transparency, gradients, masks, a bit much to decide a conversion from all that to some shapes

[rugged kindle]

I do have mostly straight lines but the curves are what require me to up the sampling when generating the colliders in unity. so maybe i do the straight lines in one sprite and the curves in another

Thank you all for explaining this. I feel like I understand why what I expected was a lot to expect

[rugged kindle]

Edge colliders seem to be way better for what I am doing and I was able to whip up a script that reads the vector path data from the SVG file (had to give it a txt extension and get a lil help from chatgpt) and converts each straight-line path into a scaled, flipped set of Unity edgeCollider points.

[kind mauve]

Is this fine to not use rigidbodies for animated characters hitboxes?

I mean they are just there for bullets and raycasts to collide into them

[rugged kindle]

I am pretty new, but all signs point to yes that is fine. Are your bullets physical objects with rigidbodies? If the bullet has one that the character doesn't need one. just colliders is fine if you ask me. if there is no bullet object (just raycasts) then you dont need it on either. Still would like to see if a pro agrees.

however, i also spent a few hours last night trying to composite edge colliders only to find out (it seems) you cannot composite them 🙂

[kind mauve]

I wonder what are possible edgecases where I used to use collision.contacts[0].point as a place to spawn something

at what cases I would get multiple contact points?

weird U shaped collider hitting a wall?

I guess for an explosive with a more or less potato collider using [0] point to spawn explosion nothing can go wrong

But for weird complex colliders... I guess I would have to calculate average?

[silver moss]

@kind mauve Use the first contact when you know it doesn't have to be super precise.
There are many cases where you get multiple contacts:
-Two colliders touching at some corners, each point gets its own contact (like your U shape example)
-Two collider edges are touching, it generates a few contacts along that edge
-^Same but with faces, for example a box resting on a plane

[kind mauve]

can I at least expect at latter cases that those contacts are more or less symmetrical?

[silver moss]

No idea, go and test 🤔

Like a Debug.Break + some Debug.DrawRay for each contact on collision

[timid dove]

in some cases you might want to spawn something at all the points

[thick monolith]

Anyone know if there is something equivalent to ConfigurableJoint damping/spring in 2D?

This specifically:

joint.angularYZDrive = new JointDrive
{
    positionSpring = Module.Data.JointStrength,
    positionDamper = Module.Data.JointDamper,
    maximumForce = Mathf.Infinity,
};

I'm using ConfigurableJoint to create a tree-like structure, but since my game is 2D, I have all of my 3D objects locked on z-axis. I'm wondering if I could use 2D physics and gain some performance and stability. Right now I'm testing HingeJoint2D but am not seeing any spring force options

Example of what I need - notice the springyness (3D version)

Example of what happens in 2D with hinge joints (currently converted my code to 2D with hinges, don't have a video with this tree using ConfigurableJoints, but trust me its stable :D)

[kind mauve]

I just want to admit that this looks cool

[plush plinth]

Any reason why updating configurable joint angular drive values in editor doesn't affect them unless i do that while not play mode?

[silver moss]

Not that i know of. You mean inspector or via code?

[plush plinth]

inspector

[silver moss]

Pretty sure the configurable joint inspector is a bit buggy

[plush plinth]

its not a big issue, just annoying having to stop and run the game

ah, got it

[dense marlin]

Is there a good way to get this to work with? I’ve been trying with animations to limiting success. I can’t get the speed I want.

[silver moss]

Not enough context. Are you making a pinball game or what

[dense marlin]

Yes

[silver moss]

What speed? Speed of the ball?

[dense marlin]

Yeah

It just kind of slowly pushes it off even if I have it slammed down

[silver moss]

You shouldn't use actual physics collisions for the launch mechanism (unless you are making some ultra detailed simulation)

Just set the velocity of the ball's rigidbody

[dense marlin]

Oh ok I will add that into the collision code

I do need to clean up the model though

[dense marlin]

I was reading the documentation and it says to use add force

[timid dove]

yes do that

[silver moss]

What mechanism is this?

I thought its what launches the ball but apparently not

If its supposed to bump the ball on collision then yeah addforce

[timid dove]

it's a bumper from the looks of it

(the pink and white part at the top is the actually visible part in the game)

[dapper crescent]

My code looks like this, but my car is shaking a lot. I’d really appreciate it if you could help me.

[silver moss]

Is this AI generated code?

[dapper crescent]

yes,I got some help

[south granite]

Please let people know in advance when you’ve used ai

[timid dove]

Have you tried asking the AI about why it's shaking and how to fix it?

[dapper crescent]

yes,I asked many times

[dapper crescent]

ok,sorry

[timid dove]

Do you have interpolation enabled on the Rigidbody?

[dapper crescent]

yes

[stuck saddle]

happy to provide mor information if necessary, im unsure what category this goes in, but I was redirected here after asking in general. My player object interacts with the white object by walking near/over it to "pick it up" (opens UI canvas) but for some reason when i walk away (which is set to disable the canvas and does so successfully) the player gets stuck (no moving, including jumping, camera still can rotate) for some reason . been getting stuck between houses and in other spots too, with no obvious explanation, its been driving me mad. the colliders are slightly oversized but, the sphere collider has "slippery" physics so shuldnt be getting stuck to anything and there's not even a boundary where the letter is. theres a collider on it but its set to "is trigger" and its not even the envelope that it gets stuck on, but a second or two after. I tried disabling the script that turns on the canvas and it still didnt fix the issue so its not a scripting issue. I dont know if its related to how Ive been getting stuck in other nearby areas or not, but its immensely frustrating!

[timid dove]

It'll almost certainly be related to your code or the scene setup. There's not enough information here to diagnose it exactly. One good thing you should do is pause the game in the editor when the issue happens and double check everything in the scene - the movement script, the colliders, etc.

Actually - now that I think of it

This may be the biggest clue:

the player gets stuck (no moving, including jumping, camera still can rotate) for some reason

This is pretty typical of what happens when you have set Time.timeScale to 0. The camera rotation often still works because it's not multiplied by Time.deltaTime

Are you setting the time scale to 0 when you open this menu @stuck saddle ?

[stuck saddle]

i feell ike i am stupid because i have no idea where to find this information 😭 i am doing this in for a class and they have only taught us visual scripting, im not even sure how to access the C# code

but i dont think its connected to any visual script i made so i think it must be a setting or something, maybe a setting that has altered that

could it be related to the particle system coming out from the white object (particles not pictured here)? even when i disable that it happens, though

[timid dove]

In your code

this is something you would be doing in your code

[timid dove]

I think it is in your code yes

show your code for when you open the menu

[timid dove]

YOu can show the visual script

[stuck saddle]

where is this located? i havent done this before

well thats the thing

[timid dove]

there is no C# code if you did a visual script

[stuck saddle]

the visual script is off

so i dont understand how its happening

i disabled it

[timid dove]

Then how are you showing the menu at all

show me some screenshots

[stuck saddle]

im not opening a menu, i disabled the script to test if it still did it when i turned the canvas off

[timid dove]

BTW if this is based don like an OnTriggerEnter or something, it doesn't matter if you disabled the script

[stuck saddle]

ohhh

[timid dove]

it still runs

[stuck saddle]

ok yeah its probably that then

let me show it

wait its not on trigger enter i thoughti t was for some reason

its on update

[timid dove]

How many scripts are in your project

[stuck saddle]

four graphs but there are a looot of embeds

most are on the player though

but this issue specifically wasnt happening in that area til i added the envelope

(white object)

[timid dove]

Just check all of them and see if you're setting time scale anywhere

(this is part of why visual scripting is not great)

[stuck saddle]

ok yes, i am, but only when the pause menu is activated
i noticed though i actually can move very slightly after walking over the object if i hold down a move key long enough. like veeeeery slightly

so it has to be an issue with that object specifically i just have no idea what

[silver moss]

Paste screenshots of the scripts here

[stuck saddle]

sorry i am not sure which scripts you mean, the only script i have pertaining to this object is disabed, the on update one
i can still send it if you think its relevant

i mean maybe its a player movement issue i guess? but i havent had the issue before with other stuff

[silver moss]

Don't have enough context to ask for a specific script, so just paste them all I guess.

Faster than us asking tons of questions in this instance

Especially the player movement script

[stuck saddle]

ah

i think i figured it out

i made a script to make jumping more realistic to ground my player and it is pushing down on my player while im walking too

so that must be why the friction when i walk on the letter makes me stuck

yup, that was it. i genuinely wouldnt have figured that out if you guys didnt prompt me to check that script so i really appreciate the help :)

[stuck saddle]

im wondering if this could be a physics issue, what do you guys think? https://discord.com/channels/489222168727519232/1477422337610744039

[gentle holly]

Hi all, I am trying to setup wheel colliders for a motorbike with a sidecar but getting all kinds of balance issues. What do you pros do in this case?

it is steering in random direction even when i just drive forward or back

i have only rear wheel is for motor

front wheel is not a "steering" wheel per se, as its pivot is elsewhere because that's how bikes work

and the sidecar just has a wheel collider but it doesnt perform any special function, throttle or steering

i am not sure how to setup collider for chassis in this case, or where centre of mass should be exactly, i assume "at the centre" but apparently not

[craggy raven]

I'm working on crouch and wall running mechanics for an upcoming horror series!

[lilac quest]

Hello ! I have a question regarding Fixed Joint. I have a serie of connected rigidbodies. But the structure never settles and wobbles/runs away. The problem seems to be that the collider overlaps. But Fixed Joint has an "Enable Collision" that I have obviously unchecked.
My issue is that my game will create such structures at runtime and I can't assure there won't be overlaps, so I need to support it. Any ideas ? I've skimmed internet without any luck at the moment : (

I have tried using configurable joint and I can get the structure rock solid. But it still tries to "run away" ^^'

[unique cave]

I have heard the fixed joint being quite unstable. Do you need the breaking behavior or why do you use them instead of parenting all the parts within one rigidbody?

[lilac quest]

Fixed Joint Vs Collider Overlaps

I'm using Mirror for multiplayer and it does not like reparenting

[exotic estuary]

hello, how do i make the object i grab be more attached to the magnet? I used a fixedjoint for the object, and i cant make that the magnet will have more force because it has similiar results, and changing linear damping doesent work either

[timid dove]

Can you explain what exactly you're trying to do here and what's going wrong?

[exotic estuary]

Im trying to make a good magnet, but when I catch something the object doesent stay in the magnet very well

And another thing, I had to disable collision with the object i catch with the player but it doesent look good, i did this to prevent the object to move the player unvoluntarily but is there a way to just make the object slide off the player? Same thing with the magnet

The matrix thing doesent work

[old abyss]

1. make the player a kinematic body. that solves the problem of the object being attracted hitting the magnet and moving it involuntarily.
2. how are you implementing the attraction/repulsion? im guessing the red object in ur video is the magnet and the white object is being replused and the gray rectange is being attracted to the magnet. Are you computing the distance between the magnet and the objects to attract/repulse or do u have something like a field around each object (a trigger col or something) and objects react to the magnet when they come in contact with each others fields?

[exotic estuary]

i will just show the code, but how to show the code not as a txt file?

[timid dove]

!code

[gloomy lodge]

Posting code

📃 Large Code Blocks

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

📃 Inline Code

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

[pale swan]

Magnets have more influence the closer an object is. Have the forces applied be relative to the distance the object is from the magnet, as opposed to a flat value.

[pale swan]

o.O

We'd prefer using pastebin, etc over downloading files.

[exotic estuary]

so how would you recommend me to attach an object smoothly in my magnet?

[tender vapor]

you were shown above how to send files properly

use the pastebin sites

share links, not files

[exotic estuary]

and btw i cant put the player in kinematic or it will push objects trough the wall

[pale swan]

If input && distance <= maxDistance, get direction and addForce in that direction towards the magnet.
addForce relative to distance, ie, addforce * distance^-1. Or * (1/distance)

[exotic estuary]

https://paste.myst.rs/vrgleea8/history/0

[exotic estuary]

this good?

[exotic estuary]

i dont get it sorry

[pale swan]

https://pastebin.com/yvJ3XpR7

[gentle holly]

So..... the way i got my motorbike and sidecar to work with wheel colliders was just to ignore the sidecar specifics, and just have 4 wheels total, just like a car, 2 of these are legitimate and present on the motorcycle, 2 are just placed invisibly to the side of that.

It's fake as hell, but whatever, it gets me passed my task

[echo forge]

What would be the best way to light the inside of this room

[pale swan]

Kinda depends on the goal, day/night, lamps, screen light, etc.

[timid dove]

Not a physics question^

[exotic estuary]

oh okay, but what will this improve here?

nvm i got it, thats actually cool, i didnt think about this

[exotic estuary]

but does anyine have any idea of fixing the problem with attached objects?

[exotic estuary]

why is making a magnet so hard... is my game idea too hard for being my first real game? Do you recommend me anything else or am i just not good

[quick arrow]

Hello everyone, I have an issue with the rigid body in my vr game and to be more specific on what issue Im having I have a mazebox and the player is suppose to grab and when the player grabs it it activated the phsyics for the ball and the player is suppose to tilt the mazebox until the ball goes to a hall.
but the issue is the ball dropts all of the sudden or jumbs off the maze box
I made both the box and the ball continues in their rigid body but still problem is not fixed does anyone have any reccomendation to how to fix it

[timid dove]

well... this kind of thing is always pretty finicky. A few things i'd do:

1. Make sure the maze box is actually rotating via the physics engine. This means if it's kinematic it should be using Rigidbody.MoveRotation. If it's dynamic it should be using angular velocity and/or torque
2. Make sure the maze box has continuous speculative as the collision detection mode - which is the best for rotating objects
3. Try to keep the rotations slower/less jarring. Too sudden of a movement has the potential to be too much for the engine to handle in one frame

[jagged canopy]

Something that absolutely baffles me about Unity is that doing rb.MoveRotation ignores rb interpolation. I set my fixed timestep to 0.05 (so 20hz) just to make it apparent when interpolation isn't working. Interpolation is on, i rotate my character, and the rotation is insanely choppy. Positional movement is fine though. I move my character by setting rb.velocity, and positionally, they interpolate as intended. but doing rb.MoveRotation completely bypasses rotational interpolation. I did find out that setting rb.angularVelocity and also rb.AddTorque don't bypass it, which is cool, except using those to turn a character is super impractical in my case. It'd be like a balancing act to get my character not to topple over and summersault since you have to have the rb rotation constraints freed up. I just wanna set my rotation without my character's rotation teleporting man...

[timid dove]

MoveRotation only obeys interpolation for kinematic bodies. Both MovePosition and MoveRotation in 3D are really intended for kinematic bodies only

[jagged canopy]

I see. what's the solution to trying to rotate a non-kinematic rigid body then? Like a player. For example, if I want to align my character to a ground normal for example.

[timid dove]

It depends - do you need physical realism for this rotation?
Does your character ever actually rotate? Is this like a game where you're on the surface of a planet? Or are you just talking about a world where the ground is always the same orientation?

[jagged canopy]

Nothing crazy, i'm just rotating their rigidbody gameObject, which is the parent object of the entire player. So just matching the transform.up to the ground normal.

Quaternion toTargetUp = Quaternion.FromToRotation(rb.transform.up, groundNormal) * rb.rotation;
nextRotation = Quaternion.Slerp(rb.rotation, toTargetUp, alignSpeed * Time.deltaTime);

rb.MoveRotation(nextRotation);

should i instead manipulate the rb game object's rotation in Update? For example: transform.rotation = X

I actually just tested it. So instead of MoveRotation I set the transform.rotation directly in Update. It was rotating at 20 hz so that's a no go.

My hack right now is to use rb.MoveRotation and perform it in Update so it isn't choppy.

[timid dove]

that doesn't fix or change anything

MoveRotation just queues it up to happen in the physics update

[timid dove]

this doesn't actually answer any of my questions

no - never modify the Transform directly

[jagged canopy]

It did fix it because even at 20 hz, it's moving at my framerate instead of at 20 hz

[timid dove]

you can do rb.rotation = nextRotation; in Update

directly

but it won't have physical realism

hence why I asked about if you need physical realism for it

(i.e. collisions etc)

[jagged canopy]

I am taking a capsule that has a rigid body on it, and rotating it from upright, to tilting on a slope with some lerp time

So far I've been doing it via rb.MoveRotation but that doesn't interpolate.

Ground is not the same orientation. there are slopes the player should rotate to align to

the player also rotates on their local y axis with left/right input. the camera is fixed behind them. It's here where you an also see the 20 hz rotation.

[timid dove]

the player also rotates on their local y axis with left/right input

And how are you doing that?

if you're modifying the Transform directly, that will break interpolation

[jagged canopy]

All MoveRotation

[timid dove]

and if you're doing MoveRotation there, that will conflict with this MoveRotatio

MoveRotation cues up the rotation until the next physics update

[jagged canopy]

And it doesn't interpolate with rb interpolation right?

[timid dove]

having two different scripts doing it at once will not work well

[timid dove]

AFAIK it only respects interpolation for kinematic bodies

[jagged canopy]

I see, which mine isn't. What method would you go about rotating the rb on its local y. In order to spin on that axis via input.

[timid dove]

if you don't need physical realism, which you don't if it's a capsule, again you can directly do rb.rotation = in Update

[jagged canopy]

I see, I'll give this a try, ty!

[jagged canopy]

I tried setting rb.rotation directly and it made it so that both rotation AND position interpolation was bypassed. I think if you directly edit the transform of a rb it bypasses all interpolation.

[timid dove]

The RB Is not the Transform.

Correct, modifying the Transform breaks interpolation

Sharing your code might help

[jagged canopy]

rb.transform.rotation, same result

[jagged canopy]

This is a simplification of what I have now

In Update():

float turnInputSmoothing = Mathf.Lerp(turnLerp, horizontalInput, 3f * Time.deltaTime);
float turnSpeed = 120f;
float deltaDegrees = turnInputSmoothing * turnSpeed * Time.deltaTime;
Vector3 localUp = rb.transform.up;
rb.transform.rotation = Quaternion.AngleAxis(deltaDegrees, localUp) * rb.transform.rotation;

You can recreate this by making the fixed timestep 0.05 to make jitter more obvious. It'll make things that don't interpolate look like they're running at 20fps. Anyways, then take an object and rotate it around its y axis. I have a cinemachine follow camera behind the player at all times, CM Brain set to LateUpdate. You'll see the rotation is laggy. And if you move the player during the rotation (I almost always move by setting rb.velocity), the movement will also be laggy. But move without rotating and your movement will be interpolated correctly.

It seems directly modifying the rb.transform will bypass any interpolation, position OR rotation. However, when I do MoveRotation(Quaternion.AngleAxis(deltaDegrees, localUp) * rb.transform.rotation) instead, it doesn't seem to bypass the interpolation on either position or rotation.

[tender vapor]

yes, that's the transform.. you aren't supposed to modify the transform
try rb.rotation

rb.MoveRotation only interpolates for kinematic rbs
rb.rotation doesn't interpolate, it's a teleport
for dynamic rbs you would generally use angularVelocity or addTorque

[jagged canopy]

I did try rb.rotation

[jagged canopy]

right here

[tender vapor]

well yeah rotation and position are teleports

[jagged canopy]

Then why are they being suggested lmao

[tender vapor]

idk what the original question was lol

could you link to it

[jagged canopy]

All good lol, I'm basically asking how to rotate a rigidbody in a way that doesn't break rb interpolate

MoveRotation breaks it, setting rb rotation breaks it

[tender vapor]

is it dynamic or kinematic

[jagged canopy]

Dynamic. Setting angularVelocity and doing AddTorque doesn't break it. However i don't see those as really being good ways to precisely control my player's rotation. For example. I want to align their transform.up to the ground normal, doing AddTorque/angularVelocity would be insane

[tender vapor]

welp angularVelocity and AddTorque are the ways with interpolation then

for that task you'd probably want angular velocity

you could do what MoveRotation would do and calculate the needed angularVelocity to get to the right rotation (assuming your player will already be close to the correct rotation)

(if not, it might be jarring and you'd want to make that happen over multiple ticks, probably)

[jagged canopy]

doesn't angular velocity have momentum and stuff? you'd need to like, counterbalance it constantly? also in order for angularVelocity to work, you need to have your rb rotation constraints disabled. so my character would be rolling and doing summersaults lol

[tender vapor]

you could just have angularVelocity = Quaternion.identity in the case they're already upright (which should result from the math without having to add a special case)

for the latter, i'm not sure, i don't know what order the simulation is applied in. i'd assume setting angularVelocity directly with some specific rb setup could get there but i'm not confident

[jagged canopy]

I pretty much just want to smoothly lerp my rb towards an orientation and have rb interpolation that rotation so it doesn't rotate at 50 hz since the default timestep is 0.02.

Unity not having a way to do this is insane

[tender vapor]

this isn't really a unity thing

this is what physx provides

(i agree it's kinda weird, just saying to blame the right person lol)

i don't have the time to test this stuff right now, so that's about the limit of what i can recommend

[oblique abyss]

I used to be able to use the unity physics debugger to show all colliders, but now its simply not working.... did something break recently, running 6000.0.58f2

I have gizmos on, I reset all physics debugger windows then selected show all

but still nothing

[old abyss]

you can
https://docs.unity3d.com/6000.3/Documentation/ScriptReference/Rigidbody.MoveRotation.html
Says in the docs:
"Use Rigidbody.MoveRotation to rotate a Rigidbody, complying with the Rigidbody's interpolation setting.

If Rigidbody interpolation is enabled on the Rigidbody, calling Rigidbody.MoveRotation will resulting in a smooth transition between the two rotations in any intermediate frames rendered. This should be used if you want to continuously rotate a rigidbody in each FixedUpdate."

ive personally tried it as well. lerping with interpolation off may or may not give jittery visual but using rb.movepos/moverot with interpolation smooths out the jitter

[jagged canopy]

MoveRotation breaks interpolation. You can test it by turning the physics step to 0.5 and you’ll see it hard jumps with zero interpolation

[old abyss]

a physics step of 0.5 is an extreme case. it never occured to me to test it with such values

[jagged canopy]

Just like how MovePosition also breaks interpolation

[jagged canopy]

It’s just to make it obvious when interpolation is breaking. Pretty much for testing purposes only

I spent 16 hours yesterday testing what breaks it or not lol

Actually I lied I was testing with 0.2 which is 5hz

[old abyss]

you should prolly compile your findings and post them

[unique cave]

MovePosition doesn't break interpolation, it just doesn't work for non kinematic bodies, it's meant for kinematic

[unique cave]

Same MoveRotation call for both having interpolation on with physics step of 0.2. MovePosition and MoveRotation are only really meant for kinematic bodies but the documentation is all but clear about it.

[jagged canopy]

Exactly, that aligns with my findings

In this context I mentioned I have a non kinematic character

[unique cave]

In that case you should not be using MoveRotation

[jagged canopy]

So my question is, what is the alternative to MoveRotation for rotating a non kinematic rb object while keeping interpolation. Let’s say I want to lerp its transform.up towards Vector3.down.

[timid dove]

as mentioned many times:

• angular velocity
• manually setting rb.rotation in Update

[jagged canopy]

Second point is false

[timid dove]

It's true and I do it all the time.

[jagged canopy]

Directly setting rb.rotation interrupts interpolation. You’re just brute force teleporting the rotation

[unique cave]

I would also consider combining the two. Using angular velocity to get the interpolation and use rb.rotation to make it always stay in phase no matter what

[timid dove]

but you're doing it in Update so you get perfectly smooth rotation anyway

[jagged canopy]

It makes it so position stops getting interpolated too though

[timid dove]

not in my experience

Maybe if you've turned on a setting

like "automatically sync Transforms" or something

I know this works because I've solved dozens of peoples' issues with jittering in their first person controllers by having them rotate via rb.rotation = instead of rotating the Transform

and they get smooth motion and rotation

[jagged canopy]

I can check for a setting but in my tests I did yesterday, when you are actively setting rb.rotation, interpolation for position stops

I have pretty much all default physics and rigidbody settings in my project. Other than 0.2 timestep for stress testing

And besides that, angular velocity is probably the most insane way to control an objects orientation

My solution was to use MoveRotation, which, like setting rb.rotation, it’s a brute force solution that teleports the rotation every frame. However, unlike setting rb.rotation, using MoveRotation doesn’t seem to break the positional interpolation.

[tender vapor]

doesn't MoveRotation also work on the fixed time step

it sets angular velocity

[timid dove]

you can just use angularvelocity to "set" the rotation directly like this.

// In FixedUpdate:
Quaternion delta = targetRotation * Quaternion.Inverse(rb.rotation);
delta.ToAngleAxis(out float angleDeg, out Vector3 axis);
// Remap from [0, 360] to [-180, 180] so we always take the short path
if (angleDeg > 180f) angleDeg -= 360f;
rb.angularVelocity = axis * (angleDeg * Mathf.Deg2Rad / Time.fixedDeltaTime);```

[tender vapor]

or maybe that's 2d physics actually...

[unique cave]

In 3D it only sets the angular velocity for kinematic bodies, for non-kinematic it just teleports like rb.rotation. For 2D they work for non-kinematic too afaik yeah

[golden ruin]

I am having this issue where when my legs that have ragdoll properties hit the ground they jitter/bug out. I want them to lay smooth on the ground.

[vestal jay]

Hey guys what is the formula for the corrective angle to level out a turret to the horizon (or any other angle really) if the entire platform is not on level ground?

[timid dove]

Project the direction vector on whatever plane you want it aligned with

[dire merlin]

i have a trigger collider that is causing physical collisions with another collider. anyone know why this could be?

additional info: i have a character controller with a capsule collider & an attack animation, when that animation is played, an attack hitbox is set to active, this is a box collider set to be a trigger. it is quite clearly intersecting with he capsule collider. i have set in the physics settings of the project that the layer the attack hitbox is on not to affect the player layer but it's still causing collisions. the only thing i can think is that the attack hitbox is a child of the player (not the capsule collider though)

[tender vapor]

the capsule collider on the CC being the attack hitbox or the player hitbox?

a CC is already a collider

[dire merlin]

it's a rigidbody based movement system, not the character controller component. the attack hitbox is a seperate thing, let me get a screenshot

[tender vapor]

ah, gotcha.

[dire merlin]

This is the collider that is spawned when attacking

this is the capsule collider

[tender vapor]

what do you mean by "it's causing collisions" exactly?

[dire merlin]

when i play the attack animation, the player is stopped while running as though bumping into something or if standing still and attacking, moved slightly backwards

[tender vapor]

are you sure it's this collider? are you getting collision messages with that collider?

[dire merlin]

If I delete this collider entirely, the issue goes away

I've just edited the collider substantially so that there's no chance it can collide with the capsule collider of the player and the issue has also gone away now, i don't understand why "Is Trigger" isn't working

[jade crypt]

How do I get an object to move across surfaces without being projected into the air if the surfaces are not perfectly aligned? I am using a sphere collider (that might be the issue). You can see in the video I hit a small bump and fly up. I am using forces and not translation (more context).

[timid dove]

the best thing to do would be to replace the colliders at those points to be one continuous mesh collider instead of two separate colliders

[jade crypt]

Okay, so I need to update level/3d design, rather than a physics problem

[halcyon sentinel]

im having an issue where the colliders are affecting animation. any way to fix this?

essentially walking onto a new tile will sometimes make the vacuum jump between the inAir animation and the grounded animation

[timid dove]

sounds like an issue with your code that detects if you're grounded

[unique cave]

Probably the case of the classic literal edge case where the collider collides with the edge of two colliders even though there shouldn't be an accessible edge. It can make the character bumb a little and be airborne for a very short duration. It's a limitation of the discrete physics engine which probably can't be fully prevented without making the whole floor out of single continuous collider. You can prevent that in your ground check by adding little margin that allows the player to be slightly off the ground. You can also use different ground checks for the movement and the animation if you want

[halcyon sentinel]

thats the issue.
Ill try out your fixes shortly!

Thanks!

[halcyon sentinel]

SOLUTION!!!
im dumb as hell

I was using OnColliderStay() for groundchecks

i was meaning to swap it over to a casting approach but Ive been too lazy and forgot about it

but now its a Physics.CheckSphere() function and it works amazing.

[supple sapphire]

Im working on a movement controller

using unity ECS

im doing a drag and acceleration-based approach

whenever I jump towards a corner like the one provided on that cube, it tends to bounce me upwards higher than a jump normally would

Im not quite sure what to do

I did set up the physics material

here's the movement system's onUpdate

I'm not quite sure what I can do to mitigate this issue

theoretically I could just make it not actually use a rigid body velocity and instead just cast in the direction of the calculated velocity and move it based on that

but I have a feeling that that'd have some unintended side effects

and it would make it so that I cant interact with other physics objects properly

which isn't the end of the world, I dont need them to interact necessarily

but it's nice to keep that option open if possible

[halcyon sentinel]

!code

[gloomy lodge]

Posting code

📃 Large Code Blocks

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

📃 Inline Code

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

[primal tapir]

hi, i am doing 6. from https://learn.unity.com/pathway/junior-programmer/unit/player-control/tutorial/1-2-move-the-vehicle-with-your-first-line-of-c?version=6.0

however even though i have the rigid body on both the car and the object, the car still goes through the obstacle in a way

[timid dove]

IIRC that tutorial is terrible and has you move a Rigidbody with transform.Translate

Basically just throw that tutorial in the trash. Dynamic Rigidbodies need to be moved via the Rigidbody if you want physical realism.

[chilly spade]

Does anyone have any ideas about this? ONLY the second assert is failing.

(Single collider on the raycasting transform)

Given the Physics2D Raycast Docs say: Notes: Raycasts will not detect Colliders for which the Raycast origin is inside the Collider do you have any hints or ideas as to why the last assert would fail?

RaycastHit2D hit = Physics2D.Raycast(transform.position, Vector2.up, rayLength);

Debug.Assert(GetComponent<BoxCollider2D>().bounds.Contains(transform.position), $"Position {transform.position} is outside of its collider bounds {GetComponent<BoxCollider2D>().bounds}");

Debug.Assert(hit.collider != GetComponent<BoxCollider2D>(), "Raycast hit own/origin-containing collider, which should not happen.");

Thanks!

[silver moss]

You linked the Physics doc, not Physics2D doc!

Physics2D is capable of detecting a query starting inside a collider

[chilly spade]

URGH

thank you ❤️

[silver moss]

https://docs.unity3d.com/6000.3/Documentation/ScriptReference/Physics2D-queriesStartInColliders.html

It can be toggled, at least globally^

[chilly spade]

I was literally staring at the types thinking I might have something which isn't 2d at one point but I never considered I was reading the wrong docs

such a classic bit of oversight haha

[tender vapor]

the type that wasn't 2d was in your browser lmao

[chilly spade]

Suppose I might be cheeky and ask if there is a reason for that? (Quite new to 3d programming)

[silver moss]

I can only assume that it's computationally simpler to check if the point is inside a collider in 2D than in 3D

Like testing a PolygonCollider2D would be way simpler than testing a MeshCollider

Could just also be that that's how it comes out of the box with the 2D engine (Box2D) vs. thet 3D engine (PhysX)

[dense marlin]

void OnTriggerEnter(Collider other)
{
if(other.gameObject.CompareTag("Bumper"))
{

      b_Rigidbody.AddForce(-other.contacts[0].normal,ForceMode.Impulse);
        Debug.Log("added force");
    }

}

what am I doing wrong

contacts is an error

[inner thistle]

What exactly do you want to happen? That would add force towards the normal of the contact which is unlikely to be what you want

[dense marlin]

I want it to be pushed in the other direction

[tender vapor]

triggers don't have collisions, so they don't have contacts

[dense marlin]

How would I change it?

[inner thistle]

Other direction related to what?

[dense marlin]

of the other object hit

[inner thistle]

If you want a normal physical collision, just don't make it a trigger and it'll automatically bounce normally

[dense marlin]

I'm trying to have it like launch in the opposite direction of the collision

[tender vapor]

so, normal collision behavior?

[dense marlin]

I wanted to be very forceful because it's supposed to be a pinball bumper

[tender vapor]

i think you could also achieve that by setting a really hi bounce coefficient

[inner thistle]

Make it a normal non-trigger collider and then you can add extra force towards Vector3.Reflect(b_Rigidbody.velocity, other.contacts[0].normal)

[dense marlin]

Is there a way to compare tag or something? I'd like to make sure it's the right object that it is hitting

[tender vapor]

yeah, by comparing tag

you could also use layers, which is generally recommended

[mossy jay]

do i say stuff about scripts in here

because its for movement?

[dense marlin]

Yeah, I got that working but other does not exist

It also does not work if it's just velocity

[supple sapphire]

https://paste.ofcode.org/VwCzLmVtikBFfeFRaLZEei

[supple sapphire]

I do in fact still need help finding a way to mitigate this...

[supple sapphire]

(please ping reply)

[halcyon sentinel]

it appears to me youre hitting the edge of the cube with your current velocity and having said velocity transferred into upward momentum?

im not expert on physics but it seems like that could be an issue

[dense marlin]

still need help

[inner thistle]

You have to give some details. Show the code and the errors, and be more specific than "does not work"

[dense marlin]

Ok

[supple sapphire]

Thats what I was thinking but I dont know if theres any good way to mitigate that

[dense marlin]

void OnCollisionEnter(Collision collision)
{
if(gameObject.CompareTag("Bumper"))
{

      Vector3.Reflect(b_Rigidbody.linearVelocity, other.contacts[0].normal);
        Debug.Log("added force");
    }

}

[silver moss]

Does that Debug.Log work when you collide?

Also Vector3.Reflect returns the new value, you are currently not doing anything with it.
It should likely be b_Rigidbody.linearVelocity = Vector3.Reflect(...)

Oh this code wouldn't even compile

[inner thistle]

You changed the parameter to collision so you have to use that instead of other

[dense marlin]

it now compiels but does nod do debug.log

not

I got it working but no force is added

[inner thistle]

because you're not adding any force

[dense marlin]

how do I? with this

[timid dove]

you're also still comparing the tag on the wrong object afaict

if you still have gameObject.CompareTag("Bumper")

you're checking your own tag, not the object you collided with

[dense marlin]

this is the code for the ball

[timid dove]

right, so you're checking if the ball's tag is "Bumper"

which doesn't make sense

[inner thistle]

Vector3 direction = Vector3.Reflect(b_Rigidbody.linearVelocity, other.contacts[0].normal);
float speed = 10;
b_Rigidbody.AddForce(direction * speed, ForceMode.Impulse);

[dense marlin]

how do I check the other?

[timid dove]

by using the parameter in your OnCollisionEnter

[dense marlin]

void OnCollisionEnter(Collision collision)
{
Debug.Log("aaaaa");
if(collision.gameObject.CompareTag("Bumper"))
{

      Vector3.Reflect(b_Rigidbody.linearVelocity, collision.contacts[0].normal);
      Vector3 direction = Vector3.Reflect(b_Rigidbody.linearVelocity, collision.contacts[0].normal);

float speed = 10;
b_Rigidbody.AddForce(direction * speed, ForceMode.Impulse);
Debug.Log("added force");
}

}

[dense marlin]

it works thx

[timid dove]

That first Reflect line is pointless and can be removed

The second one is doing all the work

[dense marlin]

ok thx

[timid dove]

Also small nitpick - collision.GetContact(0) would be slightly more efficient.

[silver moss]

Yep, most unity properties that return an array usually allocate a new array every time you call it

There's usually a newer method replacement GetX

[distant crater]

In my Unity 2d project, my player uses a capsule collider to navigate an uneven environment.
I on the other hand, slide off the side of platforms because of the capsule collider.
How do I stop the player from sliding off edges while still being able to move up and down slopes with the capsule collider

[silver moss]

Probably the simplest fix is to disable gravity when grounded

[solar elm]

Can one develop on mobile?

[tender vapor]

technically yes, realistically/practically no

[cinder lake]

Anyone know how I would make something like infinity from jjk like red and blue

[south granite]

A sphere

[plush plinth]

configurable joint target rotation isn't the same as transform rotation?

are they in different spaces?

[timid dove]

Yes

Transform.rotation is world space

[dusky arrow]

So I want doors double doors to open when a player touches them like you hold the controller and have to physically open the door

I’ve been struggling the past 3 hours

[south granite]

wheres the hinge

and you need to take screenshots

[dusky arrow]

I don’t know how to take screenshots on my laptop

[south granite]

i believe in you

you can figure it out 🙏

[dusky arrow]

Guess I’ll have to google it bc there’s no screenshot option on the keyboard

Google did not help

I have to snip every image I can’t just screenshot

Gah

my hinge is right there to the right where the rotate is

[hard fiber]

Is there a way to smooth sliding collisions with Unity Physics? I'm porting my game from Godot to Unity and while it was buttery smooth in Godot in Unity it seems to hit weird "bumps" regardless whether my tube is a box collider or a mesh collider, and it always seems to hit "bumps" in the same spots

[bleak umbra]

in physX, sliding a box across a mesh collider can produce glitches at triangle boundaries that may look like jumps you are seeing. A sphere-like shape has fewer issues with that. You can play around with contact offset and bounce thresholds to maybe fix it for your particular situation. Otherwise you can make your own physics by moving the rigid body kinematically and ray casting.

[hard fiber]

Thanks Anikki! your response makes sense and helped me find https://docs.unity3d.com/Packages/com.unity.physics%401.4/manual/ghost-collisions.html

I’ll try a few strategies out, thank you!

[urban needle]

Be careful, the documentation you refer to is for Unity.Physics, which is the physics engine for ECS (Entity Component System). If you are using Monibehaviors, you are using PhysX which is a different engine. While there may be some common ground to the behavior you are seeing and solution to them, I just wanted to highlight the distinction.

[sweet owl]

Hello! I am fairly new to unity, this is from one of my "dev logs", I was just wondering if anyone knew why this was happening? I saw a video that said using convex for multiple items is bad for performance in a game cause it uses a lot of computer power, I was just wondering if there could be an easier solution for this or a work around. I don't want to learn the wrong way to do this. It works as expected but i think there could be something else the pros do that i'm missing

[timid dove]

You have it completely backwards. It's concave colliders that are bad for performance.

Convex is much simpler

Anyway Unity's physics don't support dynamic Rigidbodies with concave mesh colliders, precisely for that performance reason.

If the Rigidbody is dynamic, the collider must be convex

[sweet owl]

oh that's interesting. Thank you for clarifying

[sweet owl]

so this should be completely fine?

[timid dove]

fine in what sense?

If you're ok with those collider shapes, it's fine yeah

if you want a more realistic collider, you could use two or more smaller colliders, including convex meshes and primitives

[sweet owl]

Fine, performance wise. Using collisions for each pin

[sweet owl]

Okay thank you. I’ll look into that

[timid dove]

There's no general answer to this. If there's only ten pins in the game, it almost certainly doesn't matter what kind of collider you use. If there are 1000 pins, then it makes more of a difference.

[sweet owl]

I see. Appreciate this 🤝

[viral elbow]

is it possible to enable/disable rigidbody?

[inner thistle]

Depends what the purpose of doing that is. Usually making it kinematic (3d) or setting simulated to false (2d) is enough. Otherwise you'll have to remove or add the component

[viral field]

If during Runtime, Can it be removed, aside from being fully destroyed, and then fully re-added? Oh.. i think that is what you are saying, nm

@viral elbow you have a response ^^

[viral elbow]

im using navmesh, so enemy can follow player. The enemy can also dash, however it uses rigidbody to do that and rigidbody and navmesh dont work well together. So I wanted to see if i can enable/diable rigidbody for dash, as I will be disabling navmesh agent during that action

movement-navmesh
dash-rigidbody
but only one must be active/enabled at a time

[supple sapphire]

Why are you using a rigidbody in the first place?

[jagged canyon]

Oh yeah forgot I need to learn physic

[viral elbow]

... to dash?

dashing is awkward with navmesh, so i use rigidbody. follow player, rigidbody disabled-> start dashing, rigidbody enabled and navmesh disabled->after dash, rigidbody disabled, navmesh agent updated and enabled

[supple sapphire]

ok hear me out

collider cast

[viral elbow]

elaborate

[viral elbow]

collider cast needs a rigidbody too...

[supple sapphire]

it doesn't

just use a capsule cast or whatever for your hitbox

its a function you can call on Physics

for a dash, if it's teleport-y, just find the direction you want it to go in, cast forward and see if there's something in your way, if there is, use the distance value to put it against the thing or whatever

and if you want it to happen over time, just do some logic where every frame it does that protocol for part of the distance

[viral elbow]

ill try it out, thanks

[supple sapphire]

np

[gleaming flint]

capsulecast? its called spherecast

also its not collider cast its raycast

[timid dove]

Both CapsuleCast and SphereCast exist

[gleaming flint]

then what does capsule cast do?

is it just spherecast but with different x and z widths?

thats my assumption

[timid dove]

same thing as a spherecast or boxcast but with a capsule shape instead of a sphere or box

[gleaming flint]

that is so usefull thank you

(sarcasm)

[timid dove]

I mean... not sure how else to describe it

[gleaming flint]

a capsule is the same as a sphere in the y axis

[timid dove]

Physics casts cast a shape through the scene

[gleaming flint]

but different widths in x and z

so is it like a capsule lying on its side?

the raycast i mean

[timid dove]

It can be in any orientation you want

A capsule is like if you cut a sphere in half and extruded it

[gleaming flint]

dumah i know what a capsule is 💀🙏

[timid dove]

now take that and cast it through the scene in a straight line

[gleaming flint]

and you can define the orientation and scale?

[timid dove]

your description of a capsule as a "sphere with different widths in the x and z" made it seem like you didn't

[timid dove]

https://docs.unity3d.com/6000.3/Documentation/ScriptReference/Physics.CapsuleCast.html

[gleaming flint]

okay maybe i did a bad job at explaing\

[timid dove]

the capsule is defined by the two focus points and a radius, so yes any orientation and size you want

[gleaming flint]

the 2 focus points are the same point is it was a sphere

[timid dove]

if you make a degenerate capsule where the two focus points are the same you get a sphere yes

this diagram shows the two points - A and B, and the radius

you can put the points anywhere and use any radius

[gleaming flint]

bruh i aint that stupid

[south granite]

then google it lmfao

[gleaming flint]

i have bad experience with raycasts

[cinder lake]

can someone help me make a force like an magnet

[tender vapor]

what issues are you encountering?

that kind of attractive force is just a force that changes with distance

[cinder lake]

I've been learning as I'm going but I just want to make basically an attractive force like thing infinity from jjk

[tender vapor]

have you tried googling that kind of stuff, as a start

[cinder lake]

thought this would be a better place to look sorry

[tender vapor]

in general you should look to existing resources first, because they'll likely have known working solutions, and it'd be faster and give more results if one doesn't work
general questions will yield general answers - if you have more specific questions, or you can't find info about something, feel free to ask

[gleaming flint]

lol its just like those ppl on reddit who never do a google search and jump striaght to asking the unemployed

but at least that way, the responses will show up in future google searches whereas here it wouldnt

[mystic mica]

Hey, I'm trying to make a level where you throw a cyllinder around but I have a problem where my cyllinder hit the jointure of my floor parts, I was told in #💻┃code-beginner that I can use V to snap vertexes together, it's a really cool tool but I still somehow have the issue afterward (and i can't really use that for turns), would anyone please know how I can avoid that?

[timid dove]

this is a common/well known phenomenon called "ghost collisions"

not a solution to your problem but something to research

[mystic mica]

My "temporary" solution for now is to use ProBuilder and use the merge object tool

[crude condor]

I can give you a formula but it's not very accurate and definetly not the same as it works irl but if you need something simple it will be perfect

[timid dove]

yes merging the meshes into one collider is one option

[gleaming flint]

how do yall simulate fluids?

i mean im making a planet and stuff and right now, the player is shooting a raycast downward and if it hits the ocean floor, it applies water physics to it

but that only works for a static ocean

and doesnt actually behave like fluid

so i am curious to know how some other ppl here do it

[timid dove]

this is an incredibly broad and deep topic, with many different answers, and it's going to depend heavily on what aspects of "fluid simulation" you're actually interested in.

[gleaming flint]

well i have experience with minecraft, roblox, cities skylines, and kerbal space program

ksp sort of works how mine works now

just a static sphere

minecraft is not volume preserved so its wierd

roblox water doesnt even have any physics itself

it only has physics for things interracting with it

and cities skylines water is i just lwk dont understand it

[timid dove]

ok well the real question is what you are looking to do in your game

[gleaming flint]

im honestly not too sure yet

right now, its just purely aadventure and exploration

but im planning on adding other aspects to the game

like actual progression

[timid dove]

Doesn't sound like anything that needs a realistic fluid simulation

[gleaming flint]

its a mini planet where you walk around

however i want to add like digging and placing stuyff to it

like deforming the terrain mesh when you interact with it

[timid dove]

That's a completely different subject

[gleaming flint]

but then the water wouldnt work

[timid dove]

For terrain deformation you probably want marching cubes

[gleaming flint]

marching cubes?

[timid dove]

Look it up

That's how they make e.g. deep space galactic

And astroneer

[gleaming flint]

typing "k" is not allowed 💀

"you need more context"

[south granite]

reactions a common alternative

just k comes off dimissive anyway 😛

[gleaming flint]

since when were you here

[south granite]

its a public discord

[gleaming flint]

also @timid dove your name keeps reminding me of percy jackson. like praetor raina avila ramírez arellano

lol

[frigid pier]

@gleaming flint Keep it on-topic, please.

[gleaming flint]

sorry mb

[gleaming flint]

OHHHHHHHH ive seen this before

but most of the results are for terrain

right now, it takes in an input seed, hashes it, then uses the seed to generate a 2d perlin noise map, and applies vertex offset based on the values of the noisemap

thats for the terrain

but idk about the water

[sacred bay]

Howdy doodles, would anyone have good knowledge on the configurable joint and the behavior between the joint and it's connection if made during play? I can't for the life of me get it to a result that works.

[abstract lily]

Hey small question is it right if i'm using the mesh collider with a none rendered mesh to be used to make custom physics collider for more complex collider? or should i'd rather script my own collider?

[unique cave]

I don't know what you mean by "to make custom physics collider for more complex collider" but it is totally fine (and often recommended) to use different meshes for rendering and physics and also to have a collider without rendering anything. Rendering and physics are separate systems and don't require each other to work. Mesh Collider makes collision shape out of a mesh and Mesh Renderer renders a mesh, there's no inherent connection between the two (other than that of mesh colliders picking the same mesh that is rendered if nothing else is specified). I don't know what you mean by "script your own collider" either, but you likely don't need to do that regardless.

[abstract lily]

My bad i miss spoke. I knew they were two different things and that even if i don't have a mesh rendered i could make a custom collider but i was wondering if doing something this way was not recommended cause it could cause some problem.

Also what i mean't by script your own collider is actually making a custom collider system, long time ago i was looking how to do a cylinder collider for the bottom of a trashcan but on the unity forum they were saying there isn't a way to do that since it's not supported by Physx and that i should make a custom collider system with a script. I just discovered how to do that with the mesh collider cause i was playing with the mesh collider. that's why i was wondering if it was a good idea and wouldn't lead to future problem!

[unique cave]

It won't cause any problems in itself no. In this case there no need to do that but just in case you are wondering, optimized realtime physics engines are complicated and math heavy libraries. It's far from trivial to do one yourself. Don't know what custom collider system would refer to other than a whole physics engine. PhysX being open source your could probably take their source code and add some custom stuff to it but making unity work with that would probably still mean a lot of work.

[abstract lily]

well that was pretty much the plan if i wanted to do custom collider. Though i thought i could try to communicate with the physics engine to make custom collider using code but probably not possible like that

[unique cave]

I'm at least not aware of any ways to directly modify the physics engine from within unity. I have heard contact modification could be used to do some custom collision shapes but in this case I don't even know what the point would be in messing with the engine. The primitives, mesh colliders and combinations of those are usually all it takes for most games

[abstract lily]

yeah still thanks for the answer ^^

[cobalt hearth]

why is OnTriggerStay2D() not running on the Player Attack script? neither collider is excluding the other's layer and they're set up to collide with each other in the collision matrix. im 100% sure the method isn't running because i have it set to log if it runs, which is not happening.

im aware the attack's collider is disabled, but it gets enabled temporarily through script when im trying to get them to collide, so i dont think that's the issue?

[timid dove]

1. I don;'t see any Rigidbodies here

[cobalt hearth]

theyre in the parents

[timid dove]

Rigidbodies are required for all physics callbacks

[timid dove]

2. you said "temporarily"
   how temporary are we talking about?

[cobalt hearth]

about 0.15 seconds

[timid dove]

What's the goal here

like a melee attack?

[cobalt hearth]

yeah

[timid dove]

I wouldn't use trigger colliders for that at all, that's a really clunky way to do it because you have this weird async thing going on where you enable it then wait for a callback.

You should use a direct physics query

you will get an immediate answer right in your attack function

[cobalt hearth]

alright i will probably try that

[timid dove]

For example: https://docs.unity3d.com/6000.3/Documentation/ScriptReference/Physics2D.OverlapBox.html

[cobalt hearth]

still though do you have any idea why this isnt working?

[timid dove]

might be that it's not enabled long enough (a physics simulation frame needs to run). Mught aslo be that the scripts need to be on the Rigidbody objects not the colliders to receive the callbacks

[cobalt hearth]

ohh interesting thank you

[drifting wasp]

Is it possible to create a mesh collider that isn't convex and also is a trigger?

[unique cave]

It seems like no. If you show what kind of mesh you need the trigger for and why, we can maybe give some advice as to how to best achieve the same in other ways

[drifting wasp]

just a simple torus. I need to make a shockwave.

[unique cave]

What kind of shockwave? Why you need trigger for that?

[drifting wasp]

a torus scaling up, similar to ULTRAKILL. I need it to deal damage

[timid dove]

I would do something like:

• An OverlapSphere and exclude anything within a minimum distance (the "inner radius" of the torus)
• Two OverlapSpheres and exclude anything that shows up in the smaller one as well (the inner radius)

[unique cave]

Distance check in code could be fine too

[deft belfry]

I have a beginning position and a target position. How do I shoot a projectile in an arc so it would land on the target?

[tender vapor]

figure out what that arc is, mathematically
if it's falling due to gravity, it'd be a parabola, for example

[deft belfry]

Yeah, there is gravity

[tender vapor]

then it'd be a parabola

figure out another constraint, like the travel time or horizontal speed

from there it's kinematics to find the rest of the variables

[deft belfry]

Lets say travel time is 1 second

[tender vapor]

so you need another constraint actually

so travel time constrains horizontal speed
you need additionally, one of:
max height
vertical speed
overall speed
launch angle

[deft belfry]

Lets say: gravity, travel time and launch angle

[unique cave]

That might actually be too many constraints. I think given gravity and launch angle, there are only one or two launch speeds which would end up in the correct target position. That would mean the travel time would not be a free variable there. Travel time and gravity would definitely be enough to produce one trajectory, my intuition is having hard time understanding whether there are finite or infinite trajectories to choose from, my best guess would be the former. The travel time definitely defines the horizontal speed but also has relation to the vertical component (if you shoot higher up, it will take longer to drop back).

Instead of picking random constraints, it would be much better if you explained what you are actually trying to do and for what it's needed.

Most cannons for example don't have any control over the launch velocity and can only change the angle to hit the target

[silver moss]

Does travel time here mean a duration before the gravity kicks in? So it flies straight for that amount of time?

[deft belfry]

Travel time is how long it takes for the objects to hit the target

[silver moss]

Ah right of course

Are you going for super accurate aiming/perfect math formula here?

Or is it going to be "humanized" somehow

[deft belfry]

Accurate

We already know what the distance between the start position and the target position is, and travel time is already set

Gravity is also already set (Physics.gravity.y*rb.gravityScale)

[unique cave]

But now the launch angle isn't set right?

Assuming the projectile is in free fall without any forces acting on it (like drag), calculating the horizontal speed is trivial. The only thing left would be the vertical speed (which would then define the angle at the same time) which you should be able to calculate from the acceleration (gravity) and the travel time (and the y levels of the start and end). If you need drag or something like that, then the problem suddenly becomes much harder

That should actually give you only one possible starting velocity meaning the travel time and gravity should be enough constraints to unambiguously define the trajectory

Let me know if anything that I said is wrong, it is quite some time sice I last time used these formulas

[deft belfry]

Is there an equation for the vertical velocity

[unique cave]

That is the equation for the vertical velocity. Well, you have to get v0 into the left side alone, but that is quite basic level stuff (move some terms and divide by t)

[deft belfry]

It says Vx?

Wouldmt it be Vy?

[unique cave]

That doesn't matter. It's formula for position in constant acceleration in one axis, might as well be Z

You have acceleration along the Y axis, so you can just imagine there's y in the place of x in that formula and life is good. The original source of the image for more information: https://www.physicsthisweek.com/topic/position-kinematics-equation/

[deft belfry]

So:
Target position=Starting positionprojectile velocitytime+0.5gravitytime²?

[tender vapor]

seems like you wrote a * where it should be a +

the first one

[unique cave]

yup, from there you need to solve for projectile velocity

[tender vapor]

you could also say vt = 1/2 g t^2 if the starting/ending Y pos are the same and gravity is downwards (simplified from the above)

then v = 1/2 g t

[unique cave]

True, we don't even know whether the start and end are at the same level. That's why I asked for more information eralier with little success

[safe flare]

Does the physics world really get rebuilt every fixed update if you move a collider without a rigid body on it?

[tender vapor]

idk about the entire physics world, but at least that collider gets rebuilt, iirc

[silver moss]

What does rebuilding mean here 🤔

It probably needs to at least update the collider's AABB and move it to the correct position in the spatial structure

If it would recalculate everything then it wouldn't be that efficient

[safe flare]

im assuming rebuilt means it's cached and that gets recalculated idk

I've seen a few things saying something along the lines of the physics world / part of the physics world gets rebuilt when you move a collider without a rigid body on it.

That colliders without rigid bodies are implicitly static and treated as such by performance optimizations, which cause extra overhead when moving colliders that don't have a rigid body on them.

[unique cave]

Ignore whatever you have heard. That indeed was a thing some 10 years ago, unity specifically addressed that when they updated the PhysX version in unity 5 or 2017 or something. I can try to find you the article. Basically that's a thing from the past you don't need to worry about no more. Moving objects that do not have a rigidbody have been more than fine since then

[safe flare]

I saw a unity staff/dev say this for 2D physics back in 2022, so I looked it up and people say the same thing for 3D physics.

I would actually prefer a way to check myself

What call would it show in the profiler?

if it was rebuilding parts of the physics world

[unique cave]

I have no clue about 2D. I will try to find you the article about it, last time I checked, all the stuff regarding the PhysX update were not up on the webpage

[safe flare]

yea I couldnt find anything about this on the API or manual.

[unique cave]

This is it https://unity.com/blog/technology/high-performance-physics-in-unity-5

Some of the other articles are not up anymore and I couldn't find them in wayback machine either (or I just can't use the thing)

So assuming it hasn't changed since, all colliders are nowadays treated as dynamic because people never understood the system and ended up in performance trouble. I don't even know how the rebuilding would look in the profiler, I don't think I have ever even used unity version where that would have been a thing.

If people online say it's still a thing, I'm not surprised, people are really slow to find about updates and expect things to work as they did 10 years ago. I have personally never faced such performance issue with "static colliders" and would love to see the hard evidence about it if that was still a thing

[tender vapor]

docs contradict this part

all colliders are nowadays treated as dynamic

even in latest docs
https://docs.unity3d.com/6000.5/Documentation/Manual/collider-types-introduction.html

The Unity documentation refers to the following collider types:

• Static colliders: The GameObject has a collider but no physics body (Rigidbody or ArticulationBody).
  and then it describes how static colliders are different from other types

[unique cave]

In the article I linked the "treated as dynamic" refers to the underlying collision tree structure only. The docs page doesn't talk anything about performance. It is no surprise moving via transform is bad for the simulation like illustrated by the example of sleeping body not waking up when object is teleported inside it via transform. Combining rigidbodies and moving static colliders isn't great but the point was that there is no "recalculation" of any heavy structure happening (as far as I'm aware).

[tender vapor]

couldve been more specific in your conclusion, but fair. haven't gotten that far into it yet

[unique cave]

Sure, I honestly didn't know "static colliders" were still used as a term, "treated as dynamic" was definitely imprecise wording

[drifting wasp]

Can I move a collider via animation? Not the gameobject, but edit the component values.

[timid dove]

You can do whatever you like in your animation with serialized values but note that changing the collider shape in an animation won't necessarily have realistic physical results. For example rapidly upscaling a box won't realistically launch objects resting on top of it.

[drifting wasp]

when I try animating the values in the collider, they don't actually keyframe. They change as if I wasn't trying to animate them.
So, If I were to set keyframes:
size 1,1,1
size 2,2,2
the end result will be 2,2,2 as if the animation keyframes never happened.
and yes, I'm sure I'm adding keyframes and the background on the value turns red.

trying to change it here doesn't allow me. I can type it in, but the value doesn't change

[silver moss]

Are you doing this in play mode?

[timid dove]

when you say they don't "keyframe" do you mean they don't interpolate between keyframes?

or they don't change at all

[drifting wasp]

No, scene.

[drifting wasp]

They don't interpolate, or snap. They just set and change the keyframes behind them

[timid dove]

If you're trying to move a box collider it's probably better and more correct to give it a kinematic Rigidbody and turn on "animate physics" on your animator

and then actually move the position of it

rather than the collider center

[pure socket]

Anyone have any insight for an approach to make instantiated objects account for lagging behind its source object? For added context I have physical bullets shooting from a gun, but when rotating at higher speeds the bullet's spawn position tends to lag behind a little bit. I guess this counts as a physics problem? If not I can ask in a different channel.

[timid dove]

You would really need to show your code. Whatever you're seeing is an interaction between your code and the scene

[pure socket]

Alright will do

private void Shoot()
{
    Ray ray = new Ray(targetCamera.transform.position, targetCamera.transform.forward);

    GameObject bullet = Instantiate(bulletPrefab, bulletSpawn.transform.position, bulletSpawn.transform.rotation);
    bullet.GetComponent<Rigidbody>().AddForce(targetCamera.transform.forward * bulletForce, ForceMode.Impulse);

    Destroy(bullet, bulletLife);
}```
This is the function that instantiates the bullet at *bulletSpawn*'s transform position, which is here, parented to the root bone of the gun's rig.

I can get a clip of the object lag if needed

it's only visible for like a frame (before the bullet transforms and travels forward) and I'll probably make the mesh renderer disabled ultimately, but I worry the lag will cause some collision/trigger inconsistencies for like if you drag shoot right beside something** **

[timid dove]

need more context for this code

show the full script

i.e. when does Shoot() run

is it in FixedUpdate? is it in Update?

If it's in Update then you need to remember that physics don't run every frame. You may get:
Frame 1:

• Spwawn the bullet
• rotate the camera a little
  Frame 2:
• rotate the camera some more
• physics finally runs

So it would look like the bullet is sarting to move from a point away from the bullet spawn poiint because it only starts moving a whole frame later after you turned a bit already.

[pure socket]

well see I thought about that and I think I'm going to need to rework how it executes because I tried processing the inputs and stuff in FixedUpdate instead and it doesn't quite work as expected

that's probably why though

I suppose I could use like a listener variable that gets checked in FixedUpdate and then executes the function there

[pure socket]

I think this idea worked

[daring mesa]

Im trying to make a script for dragging objects in 2D but once they are let go off they will be "thrown" in a direction where the mouse was moving but for some reason once I let go off the object it either nearly completely loses horizontal momentum and starts falling or the object moves quite slow
https://pastebin.com/WXt6a3kh
I tried chaning a few different things and trying to do it with TargetJoint but it doesn't seem to help. (This could be the wrong place to ask or me just being stupid but I hope someone could help)

[silver moss]

Did you try dragging an object that just has a collider, a rigidbody and this script and nothing else?

Because it sort of sounds like something else could be modifying its velocity

Also I can sense some framerate dependency issues. You are updating prevPos in Update but it should be done in FixedUpdate so the final launch velocity would not depend on frame timings.

Is the idea that moving the mouse faster would result in a greater velocity when releasing?

[daring mesa]

Ehh.. didnt decide on whether mouse speed shpuld depend, but i do need RB so its affected by gravity + i will try putting it in fixedupdate

[timid dove]

This AddForce method isn't really right for this application either

For this sort of thing I would be measuring the velocity (or average velocity) from the final frame of holding (or averaged over the last several frames, potentially dropping outliers), and setting that velocity directly

[elfin latch]

try using a PID controller

[past cosmos]

Hello all, I have a problem and would like some assistance.

The situation: Two balls in a box.
The problem: I would like the balls to bounce off each other, but not transfer any velocity. I would like gravity to still take effect and such, as if the collisions were perfectly elastic.
What I have tried: I have set the physics material on both objects to the settings as seen in the picture.

Thank you all!

[timid dove]

If you want to customize a collision you pretty much want to go with https://docs.unity3d.com/6000.3/Documentation/ScriptReference/Physics.ContactModifyEvent.html

[past cosmos]

forgot to mention, unsure if this would change anything, but i am using universal2D, not 3D
but anyways, i am pretty new to this, so how exactly does this work?

[timid dove]

This doesn't work in 2D, no

[tender vapor]

(btw, universal just refers to the rendering pipeline, the physics is just 2d)

[past cosmos]

what do you recommend i do than?

[past cosmos]

i see

[timid dove]

probably just track your velocity for each object beforehand (in fixedupdate) and in OnCollisionEnter2D just set their new velocities to exactly their old velocities reflected on the collision normal

[worthy arrow]

Hey, i have a problem with crouching when using built-in CharacterController. My character has a height of 1.7 but even though, im standing in front of 1.85 empty space i cant get benath it, i have to crouch. ChatGPT told me its because of the step offset so i set it to 0 temporarily for a test and this bug really disappeared. Earlier i was using rigidbody and i never had any issues with crouching, it was "pixel-perfect", but how do i solve the crouch issue with this built-in CharacterController? is there a solid fix? im thinking of making my character ~ 1.55(it seems like reducing 0.15 fixes it) but i feel like it might cause other bugs later that i cant foresee at the moment

[timid dove]

CharacterController has a skin width

check what yours is set to

and then double it because there's a "skin" on top and bottom

[worthy arrow]

my skin width is set to 0.01

[timid dove]

The height is also not inclusive of 2x the radius for the endcaps

IIRC

The height is the distance between the centers of the two spheres

[worthy arrow]

ohhhh

[timid dove]

So to get the total actual height of the cc it's like height + 2*radius + 2*skinWidth

Double check this though

I could totally be wrong

[worthy arrow]

im going to test it now

so my character height is 1.7 + 0.4*2 + 0.01*2 = 2.52
i put a ceiling-block 2.5 above the ground level and im moving freely, at 2.2 still freely, starting at 2.1 im getting slowed for a short time, also, like i said, setting step offset to 0 solves the issue entirely. I think it has to be something else here

step offset is 0.46

[timid dove]

This is over my head then (pun intended), sorry

I think I was wrong about the radius

The total height is probably just height + 2*skinWidth

[worthy arrow]

thank you for trying to help tho 🙂

https://discussions.unity.com/t/character-controller-step-offset-and-moving-through-gaps/515689

found a few topics like this, 13 years later, still not fixed 😢

[fleet cargo]

could someone please help me figure out why my wheel rbs absolutely freak out when they're connected to other rbs with a fixedjoint?

The wheel top-level parent is made of an empty with an rb and hingejoint. It has a sphere collider rb as its child, and set as the hingejoint's connected body. I've set the layers to exclude other placed parts (i.e the grey cubes you see me place).

Both the wheel rbs are set to mass of 5 (for a total of 10), and each cube has a mass of 10.

oh

had to turn off preprocessing on the hinge joint

spend hours debugging ts and then figure it out within minutes of writing a discord message 😭

[proper pike]

how do i get limbs to stop drifting like this (making togglable ragdoll)

the structure is the armature has rigid bodies, colliders, and joints, and colliders are turned on when ragdolled

i tried fixing it by turning on isKinematic on the limb rigidbodies but it just ruins the player physics n starts making me fly

it also sometimes snaps back into place

[gilded breach]

I’m making a pong-like game, but for some reason the ball will start spinning when hit at certain angles when it should be still. How should I fix that?

[tender vapor]

should it spin at all? if not, just constrain the rotation

[gilded breach]

Is there an option to do that for the box collider?

[timid dove]

no it's part of the Rigidbody

[gilded breach]

Is there an option on there to do that?

[timid dove]

That's what we just said

[gilded breach]

Oh sorry

[kind mauve]

what does Max Collision Shapes really means for particles?

given I want accurate results using particles to deal damage, would it make sense to increase it from default 256?

[reef stratus]

Can someone help? The wheel physics dont seem to work sometimes, i have no idea why

[molten zephyr]

i dont know but add a sphere collider and rigid body to the wheels.

[timid dove]

we would need more details

[reef stratus]

What details?

The car uses 4 Wheel Colliders, which sometimes dont colide with the box colider i have on the object i show on video

[timid dove]

like how the car prefab is set up. Screenshots etc. Code. Whatever scripts are on it.

[reef stratus]

The car itself has a Rigidbody + trigger box collider, on a "Vehicle" layer which shouldnt colide with anything
Then i have BottomCollider and TopCollier on VehicleColision layer for actual colision

Then there are 4 wheels on default layer, which sometimes do not colide with the platform

I can show anything else

Here is the script that instantiates the car

[timid dove]

You're putting the car on the UI layer?

[reef stratus]

yes so UI camera can catch it

I remember trying to make it a different layer which didnt help, the issue persisted

[timid dove]

i would take a look at the car in the in the hierarchy window when the wrong thing happens

see if anything is different

dig through all the physocs components etc

[signal wyvern]

Anyone got any tips on how to create a fairly stable and accurate ragdoll? im trying to use the ragdol wizard with some simple synty characters but im just getting some strange results where arms twist Un-naturally I assumed implementing ragdolls would be simple...

[sacred dock]

Bone axes are probably wrong—Unity expects limbs to bend on specific directions, and Synty rigs often dont match that. The ragdoll wizard is just a base, so you need to manually fix joint limits (especially twist). Also make sure colliders and mass distribution are reasonable, or physics will behave weirdly. Basically, the twisting comes from bad joint setup, not the ragdoll system itself.

[signal wyvern]

Okay thanks but how do I know how to configure my joint considering the character mesh? The character join component has soo many values to configure.

[sacred dock]

Focus on Axis and Swing Axis so the joint matches the bones natural twist. Set Twist Limits small for forearms/calves and bigger for upper arms/thighs. Align the Anchor to the bone pivot and adjust Swing Limits so limbs don’t bend through the mesh. Then test in Play mode and tweak iteratively.

[signal wyvern]

Can you just explain how i would configure this one joint so i can learn and apply this to the others? i really just want to know what the axis and swing axis should be for each one?

[sacred dock]

The main thing that causes the weird arm twisting is the joint setup:
Axis should point along the bone (shoulder to elbow)
Swing axis should be perpendicular to that
If that’s wrong, the arm will twist instead of bend.

Big tip: elbows shouldn’t behave like ball joints. Lock them down a lot:
Very small swing
Almost no twist
So they act more like a hinge
Also the ragdoll wizard isn’t perfect — it gives you a starting point, but you have to tweak:
Axis
Limits
Sometimes anchors

One more thing: keep your masses reasonable (heavier torso, lighter limbs), or it’ll get jittery.
That’s basically it — once the axis is correct, most of the “weird” behavior disappears.

[sacred dock]

For specifically this, Your arm is clearly going sideways (left ↔ right)
So the bone direction is along X axis

Fix it like this:
Axis: (1, 0, 0) ← along the arm
Swing Axis: (0, 1, 0) ← perpendicular

[signal wyvern]

So for the left arm

Axis: (1, 0, 0) ← along the arm
Swing Axis: (0, 1, 0) ← perpendicular

& right arm

Axis: (1, 0, 0) ← along the arm
Swing Axis: (0, -1, 0) ← perpendicular

?

[sacred dock]

Almost but a small correction, use the same values:
Axis: (1, 0, 0)
Swing Axis: (0, 1, 0)

Because they are still using the same local bone orientation. Unity joints use local space not world space

Start with identical settings on both sides and
only change if one side behaves wrong

[trim bison]

I have some enemies that keeps clipping out of walls (mesh colliders) even with dynamic continous rb
The most confusing part is that it only happens in the builded form (more fps)
has anyone came accross a similiar problem?

[random palm]

In 2D I had a problem of rb getting stuck in the wall on collissions sometimes. Had to change some default physics settings values, and believe it was because my world scale was smaller.

I first believed it was impossible to fix. Took me 1 year to realize. Using PolygonCollider2D on rb and EdgeCollider2D on walls.

[trim bison]

I see

seems this is just a reoccuring issue with meshcollider

[silver moss]

How are you moving the enemies, code-wise, could anything be modifying its transform directly? Does it have an animator? What's the collider setup for the characters?
And are you sure that the wall meshes have valid geometry? "Watertight" geometry is usually the best.

Also, make sure to not have rigidbodies on the walls, not even kinematic

[trim bison]

They have a navmesh which should be disabled during the noclip
What does "watertight" mean? I only know that convex normally do not have these problems compared to normal concave

[silver moss]

It means no holes/open edges in the mesh

Convex makes it watertight though

But anyway, the most important question was how are you moving the enemies?

[timid dove]

#🏃┃animation

[lean sable]

ty

[wet flame]

Hello,

I'm trying to use ECS DOTS to do multiplayer game. I try to keep my project simple as I can.

Here I'm strugguling :

• I create a static object cube in a subscene

When I start the game I cannot see this gameobject in "Physics Debugger Tabs"

[pure socket]

Any insight someone can give for making an NPC moved with Physics.AddForce not slide around like they're on ice, as well as an efficient way to clamp the velocity so their speed doesn't just increase exponentially the whole time?

[timid dove]

#1062393052863414313

[tender vapor]

add linear drag and/or friction

[amber hawk]

How hard is it to make realistic water?

[timid dove]

define "realistic"

[amber hawk]

By realistic I mean water that looks believable and behaves believable. Stuff like reflections, depth, waves, foam, shoreline interaction, and decent movement physics.

[timid dove]

All of that stuff is pretty vague

the answer is - hard

depending on how realisitic you want - harder

"decent movement physics" < what does this actually mean? You want a real time fluid simulation? That's stuff people do for their PHD theses

There are assets of course that do it out of the box that you can find and/or pay for. But they might not fully meet your exact needs.

[amber hawk]

kinda

[timid dove]

Obi Fluid for example

[pulsar canopy]

Very vague question. Purely visually, water that looks good? Hard, but there are assets. For example, hdrp comes with its in built water system that looks good.

However, if you're talking about an actual fluid simulation as Praetor says, I really wouldn't do that if I were you.

HDRP has all that reflection, refraction, tessellation displacement and a billion other things built in, but it's not physically accurate, nor will it respond to someone's movement in the water

[upper helm]

how would i calculate the reflection direction of a ball bouncing off a surface

[timid dove]

Vector3.Reflect

[fervent flicker]

hi guys i need a quick suggestion from you, i'm working on XR simulation of Transformers and i need to show the EMF ( Electromagnetic Field visuals) around the transformer coil. what should i do with that? any suggestion will help, i have tried the SPline but i don;t think it is working as it should be, i need arrows to animate according to show the directiion of flow also. what do you think?

[bleak umbra]

Probably getting a line-drawing asset (or DIY’ing one) is your best option. You can try your luck with line renderer or splines, but, out-of-the-box, those aren’t exactly made for that use-case and will be somewhat annoying.

[timid dove]

Shapes

This asset maybe? Using PolyLine:

https://acegikmo.com/shapes/docs/#shapes

[fervent flicker]

Thanks for the suggestion but i solved it using the Torus shape with shader graph showing arrows and making them animated using time Node. And its worked well and also optimized solution i guess. 🙃

[barren minnow]

Hello .
I'm new to Unity and to Game creation in general . I've started a month ago and i'm working on a tiny project to learn the step of the creation of a Game . I'm make a 2d survival isometric game and today I tried to make item drop of a tree when it's cut . I want to make the item pop out and fall on the ground , but I don't know how to simulate a gravity for an isometric world . If I add a Collider to the Tilemap , nothing can move , not even my player . And if I don't the Item just fall into void . Someone knows a way to make that ?

[inner thistle]

Physics simulation is overkill for that. Just make an animation

[dry shell]

I'm working on the weaver "making bonehead" tutorial to help me understand procedural animations by animating a gecko.

So I need to make sure the axes are oriented correctly and if not I need to use "offsets" for each bone.

I've found two different methods and I want to know which one is generally considered cleaner since since the game I'm making will be mostly procedural animations..

One method was to create an empty game object and parent it to the bone that is not oriented properly and rotate the parent.

The other method was using quaternions in a script. Any help is appreciated!

[timid dove]

"using quaternions in a script" is pretty vague. I can think of about 1000 different ways to use Quaternions in a script.

[dry shell]

Sorry about that, I worded it wrong .Specifically the method talked about applying an offset into the rotation.

[frigid pier]

Don't cross-post. Pick one channel.

[ember heart]

Um, guys

I was tryna learn the joints and stuff

But dis happened

Why did the stick go up just because i added a hinge joint to it?

And the yellow cylinder is spinning

[tender vapor]

consider using screen recordings instead of videos of screens

[tender vapor]

please don't crosspost

[elfin echo]

Isnt fluid mechanics bachelor stuff

[timid dove]

It's PhD stuff

Or it was at one time

People building on the works of older PhDs may now do it at the baccalaureate level, not sure. I've been out of academia for 15 years.

[elfin echo]

Ill have fluid mechanics next semester in my ee class

[timid dove]

there's a difference between learning the fundamental principles and building a convincing realtime simulation.

[elfin echo]

I hope so

I hope its not too hard

[regal flower]

does anyone know how to fix the dress? I guess I should ask IS there a way to fix it?

[silver moss]

The dress should have proper skin weights so it moves with the leg bones, and/or use Cloth physics with capsule colliders on the legs

[regal flower]

yeah I added capsule colliders and have been messing with the settings, what do you mean about skin weights though?

[silver moss]

Bone weights/skinning/weight painting

You'd use a 3D app (like blender) to weight paint the dress so that certain parts follow certain bones

Just like the character model's legs follow the leg bones. You'd do the same for the dress

It looks like the lower part of the dress currently just follows the torso/has no weights

IF you get the cloth to be stable then this might not be needed though.

[regal flower]

im just gonna move the left hand to the wall now lol

[gentle cape]

Does anyone here know what this is?
At low speeds and slightly of center the discs keep pushing on instead of bouncing of/reacting

[silver moss]

How are you moving the disc, is it just a single addforce/velocity set?

Try enabling continuous collision detection on the rigidbody

I doubt that's it though

[gentle cape]

Yeah it's just an impulse where I add force

[inner thistle]

How is that wiggling effect done?

[gentle cape]

Anyway figured it out, it was a DOTween function I set up to slow the discs down that interfered with the physics engine ... it now works fine again
https://youtu.be/aJZ3dYU2_WU

[misty rapids]

how do i dampen my physbones make them move slower but not like for the return just when moving arround

is it cause my bones are really small?

[paper stratus]

Hello, I'm making a magnet-based game, and I'm having an issue where if the player is on the edge of a platform, using the magnetism power, it moves toward them slower than it would if they were in the middle, and doesn't push very well. I think the problem is because I'm using AddForce, and the x and z positions are locked, but i don't know how to fix it.

[timid dove]

!code

[gloomy lodge]

Posting code

📃 Large Code Blocks

Use links to services like:
https://pastes.dev/
https://paste.yunohost.org/
https://share.sidia.net/
https://paste.ofcode.org/
https://paste.myst.rs/

📃 Inline Code

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

[timid dove]

based on your description the x component of motion is locked for one of the bodies, but you're applying a force with x and y components

The x component is basically gogin to be ignored

so you're left with just the y component. And the y component will be smallest the farther away from the center you are, since you're normalizing the vector

Think about two right triangles, both with hypotenuse length of 1 - but with different shapes like so. THink about what Y will be for each of them

the one with the larger x (ignored in your case) will have a smaller y, and vice versa

[paper stratus]

Would it be better to make a separate function, and use addforce just on the one axis?

[timid dove]

the separate funciton is probably a good idea from an organizational perspective but it doesn't really make a difference to the logical issue here

Since you want this to only work "up and down" (you constrained the x axis) the real thing you need to be doing is just normalizing the y component - which means just finding if Y is positive or negative and using 1 or -1 times your force constant.

[paper stratus]

Would it be something like "if(player transform position x > transform position x)
Addforce up"?

[timid dove]

we don't care about the x position, as far as I understand

[paper stratus]

Oh, sorry, i got the two confused

But, for the y position?

I will test at home when I can

[timid dove]

you're freezing the x position you said, so I'm assuming whatever object this is only moves up and down

[paper stratus]

Yes, i am planning to do one for up and down, and a different for left and right

[paper stratus]

Hello again, and Thank you so much! I changed the vector to ///Vector2 directionToPush = (Player.transform.position.normalized - transform.position.normalized).normalized;///, and now, it pulls without slowing. only problem now, It reaches the player height, then goes up and down, and flies off Edit: nevermind, speed was too high, though it is still going up and down

[glass sand]

Hello, I tried to make a spaceship to go right, left, up and down by using PID controller. But when I press left, my ship will go left and up at the same time, and vice versa. I know that there is problems on my FixedUpdate, but I don't know the type of problems. Can anyone help me?

[silver moss]

What does controller.Update do?

[glass sand]

I don't know. 😅 I'm a newbie who learn Unity for 2-3 months I think, that script is a reverse engineered-style from me. I change the horizontal.cs from (https://github.com/vazgriz/PID_Controller/blob/master/Assets/Scripts/Horizontal.cs) to apply into the spaceship. I tried to script the vertical movement (up, down) and the spaceship's movement went wrong

[timid dove]

show your full script.

!code

[gloomy lodge]

Posting code

📃 Large Code Blocks

Use links to services like:
https://pastes.dev/
https://paste.yunohost.org/
https://share.sidia.net/
https://paste.ofcode.org/
https://paste.myst.rs/

📃 Inline Code

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

[glass sand]

https://paste.ofcode.org/JgXSJpZjuA6L84xs8GE8Tb

[glass sand]

hello, here it is. The code are messy because i reverse engineered the Horizontal.cs script and try to experiment with the new input system and others

[timid dove]

This link doesn't work

[glass sand]

Eh, it work to me.

[tender vapor]

does work for me still

[tender vapor]

btw, seems like your ide is not configured

!ide

[gloomy lodge]

💡 IDE Configuration

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

Select one:

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

[tender vapor]

PIDController is stateful, so you can't use the same one to handle both axes separately

you could have 2 separate controllers for each axis, though not sure how that'd go - you might want to consider making your own that supports Vector2s or something like that

[glass sand]

Alright, I will try

[glass sand]

MY MAN, THANK YOU, ITS WORKING. Your explanation are better than AI. You should be proud of yourself.

I use AI to check for error if I don't understand how to fix the problem and explain how the script work. For script, I follow the YT tutorial and look at other scripts in the Github.

[tender vapor]

"better than AI" is not a very high bar

[tender vapor]

you'd probably want to do this to make sure distances/velocities are handled well (and for ease of use), idk if it'd work as expected if you control the axes separately

[timid dove]

it's working now. weird

[craggy raven]

Im making a character controller and when I raycast to a wedge downwards it goes through the wedge and hits the part below it. Any idea why this is happening? I tried enabling convex on the mesh and disabling

[timid dove]

Either you aren't Raycasting in the place you think you are, or the distance is wrong, or there's an issue with layermasks, or your collider isn't where you think it is. You need to do some debugging

[craggy raven]

How would you check the collider on a mesh collider?

I already checked the collisions groups

Plus I debugged and it hits the wedge but then skips it and goes to the part under it.

[timid dove]

What do you mean by "check the collider"?

[timid dove]

And what do you mean by this exactly? What debugging steps did you take? How is it hitting more than one thing?

[maiden wolf]

hey i wanna mess around with physics, i already have my eye on Obi fluid, soft body, and cloth physics and now i just need to find a good destruction physics system that preserves textures, any recommendations? i want it to work with character/skinned meshes too of course

[craggy raven]

I figured it out. I was using an raycastresult list and wasn't checking the distance of the objects passed and now it works

[timid dove]

Is there a good reason you're doing a RaycastAll if you just want the nearest object?

[craggy raven]

Windows and enemy's

I use it for alot of different things. I am updating it tomorrow to use tagging to get enemy and thin objects like windows and fences

Plus for single objects if I get the closest to origin I can get the most close object whilst having data like if an object is being supported by another.

Then use the closest for floor placement and the other to manipulate the environment

[abstract lily]

Little question how could i set the velocity of a rigid body without moving it. For example while the rigidbody would be kinematic i would set is velocity to be = what it move in an animation and when i switch off iskinematic it would become a physics object and move in the direction i set

[inner thistle]

Calculate the velocity, set the rigidbody velocity after setting it non-kinematic. The velocity of an object is (position this frame - position previous frame) / deltatime

[abstract lily]

i mean my guess was pretty much that but i would have loved if you could directly put velocity while kinematic, guess not. still thank you for the velocity tip, save me some research

[inner thistle]

You don't gain anything from setting the velocity while it's kinematic because it doesn't do anything

[unique cave]

Well it does actually afaik. For one, it makes other bodys colliding with it know the velocity and therefore makes the collision much more natural. It also enables interpolation. Haven't actually tried it this way but MovePosition does set the velocity which makes me believe that it would be what does the trick

I should try that

[unique cave]

Oh, it gives the "Setting linear velocity of a kinematic body is not supported" message and does nothing, does not seem to work then

Would have been nice for conveyor belts and such

[maiden wolf]

anyone?

[rough helm]

Hi, I'm trying to move a ship using AddForceAtPoint and have the acceleration be gradual. Does anyone have any idea how to do it?

[inner thistle]

You'll have to tell what problems you're having with that and show your setup for anyone to help

[snow surge]

A little car physics type question regarding traction. The car is flying through the air, and has just landed one tire on a stone block. I want the tire in this situation to "anchor" its position by applying a torque and force to the vehicle (using Acceleration/VelocityChange). I think I'm supposed to take the cross product of the red and blue vector in order to get the torque, but for some reason, the torque it gives me is way too much.

(Purple circle is center of mass and purple line is linear velocity. Blue line is the point velocity of the tire and red is the offset from center of mass)

My in game version of this is giving the green line (predicted torque to cancel out rotation) that is much longer than the yellow line (angular velocity of the rigidbody at this frame)

[timid dove]

You can just use AddForceAtPosition to apply the friction and the torque will come naturally

[snow surge]

The issue with AddForceAtPosition is that I need to apply the torque based on all four tires, not just one. I need the torque required to completely anchor a tire in one frame.

[timid dove]

You can call it once for each tire

Using the appropriate friction calculations for the materials each tire is touching

Assuming this one has significantly higher friction, you'll see a net torque

[snow surge]

Wait, does AddForceAtPosition take into account previous calls when calculating the torque?

I imagined I'd be wiggling since I'd be overcorrecting with four tires

Like if the car is spinning, and I use AddForceAtPosition() for each tire, wouldn't I be applying the red force (enough to stop it with one tire), when I want to instead use enough force for four tires (blue)

[snow surge]

 Vector3 planarWheelVel = Vector3.ProjectOnPlane(rb.GetPointVelocity(GetWheelPosGlobal(i)), rb.rotation * Vector3.up);

 rb.AddForceAtPosition(-planarWheelVel, GetWheelPosGlobal(i), ForceMode.VelocityChange);

This does in fact make the cars freak out instead of stopping on a dime

[timid dove]

they're all additive. The total reaction of the car takes into account all the calls. They are independent from one another though

[pulsar canopy]

Physx doesn't understand pressure points, right?

For example, it makes no difference to the engine whether a rigidbody sword hits something with the blade edge of blade face?

[unique cave]

What exactly is pressure point and why does it matter? What difference should it make whether the blade or the face hits something?

[pulsar canopy]

Let's say, for example, you have a joint somewhere with a break force.

Is the same amount of force appllied to it regardless of which part hits it? I.e. do rigidbodies care about pressure in any way?

[unique cave]

Pressure is not part of the equation. PhysX like most physics engines only operate on rigidbodies which do not deform or bend. If there was no deforming/bending, the pressure wouldn't make any difference in the real world either (I think?). PhysX also operates on contact points, not areas of contact. Therefore you could argue the pressure to always be infinite. You could think of it as if everything in he PhysX world was made of diamond, extremely hard material which never gives up. Forces (and torques) are what matters, not pressure

[pulsar canopy]

Yea that's pretty much all I wanted to know

[stray belfry]

How can I check if a Collider2D’s exact shape (including things like PolygonCollider2D) would overlap with other colliders at an arbitrary position, without actually moving or instantiating the object in the scene? I’m working on nav/path generation and simulating positions in code, so what I need is essentially to “place” a collider shape at a given position and test if it would intersect with the world. Using bounds isn’t accurate enough since it’s just an AABB, and the built-in OverlapBox/Circle/Capsule methods only work for primitive shapes, not arbitrary collider types. Collider2D.Cast also doesn’t quite fit since it’s a sweep, not a pure overlap check at a specific position. Is there any way in Unity to query overlap using the actual collider geometry at a given position, or is this not supported?

[stray belfry]

currently im thinking to instantiating the object in the scene anyway as a part of the nav graph building process then destroy them after the process is done, easier to follow this process too that way, but feels a tad bit inefficient

[timid dove]

you could just move the collider to the place you want and use https://docs.unity3d.com/6000.3/Documentation/ScriptReference/Collider2D.Overlap.html

then put it back when you're done

[stray belfry]

Moving one collider doesn’t really solve my case, because I’m not testing one obstacle shape at one position.
During navgraph generation I need the obstacle prefab colliders to exist at all their tile/world positions at the same time, because the enemy box can overlap multiple obstacle tiles at once.
Example: if two obstacle tiles are stacked/adjacent and each has a unique partial collider, checking only one moved collider can say “no overlap,” even though the enemy box would actually hit the other collider at the same time.

[tender vapor]

you would be moving the enemy collider (or rather, a copy) to the point you want to check and then Overlap would tell you all the colliders that would overlap

it's not testing whether 2 given colliders overlap

[timid dove]

What Chris said. It checks against all colliders in the scene. I was referring to the enemy collider

[stray belfry]

I already know how to check using the enemy collider, I use it's bounds, I don't mind being limited to a box shape for the enemys collider, but in the scene, where I have obstacles tiles, which spawns prefabs, these prefabs are stored in the tile data, and do not exist in the scene because we do this during scene time, not playing time. There are no instances to check against when placing an overlap at the enemies calculated position. And it is that which I want to find an effecient solution for.

I have a method that can get all the tiles that we would collide in, and get their data, so if we collide with 3 tiles, 2 of which are obstacles, I'd be able to get their prefabs and spawn them in the center of their respective tiles, but each jump calculation can do up to 45 * (6/0.05) * 2 calculations/checks and for each frame of those jumps I place an overlap the size of the enemy so it would quite expensive to always instantiate a prefab, so yea, I need help with figuring out a more effective way of getting a unique collider shape to be placed where the tile prefab would spawn during run time, and then when checking each jump frame I do an overlap to see if we touched any of those unique colliders that we would be placing

[silver moss]

You can reuse the instantiated object instead of instantiating it once per check

For me personally it would help if you show screenshots of the game/map to better understand your project's needs.

[stray belfry]

so this is a screenshot from the project, ive edited the picture to a hypothetical polygon collider for the obstacle tiles (the 2 spikes), and a blue outline for the "enemy" which is just a gray box with an arrow towards the direction its facing (a sort of whitebox for testing).

a singular instantiated object instead of instantiating it once per check wont work since in this example, the enemy "jumped" (we dont "place" an enemy when trying to find an optimal jump, we use its width and height from the "enemy data" SO to create an overlap box) in this instance, this would be one of the simulated frames (see image 2 for a real simulated path of the enemy, basically to find out if a trajectory is accepted is if for each physics frame, we place an enemy sized overlap box at the calculated position at that frame, in the second image, the enemy is in an arbitrary example position unrelated to the trajectory). so as you can see, in one of these simulated frames, the enemy might be touching 2 obstacle tiles of the same type, meaning if we only check using one, say we check the first/closest obstacle, you would see that we dont touch the collider (green shape) and that would then make the trajectorycheck think this position is reachable/okay, when in fact its not, because as u can see, it touches the second obstacles hitbox.

thats the reason one obstacle prefab that gets moved around, does not work.

and simply doing an overlap box does not work either because this is a navigation graph bake which is run before the game starts, not during run team, meaning those instances of those obstacle tile prefabs are not instantiated yet

we would simply instantiate all of them in the beginning as a part of the bake process, and then do our usual checks as the system currently does (it already relies on overlap box but as you know by now that didnt work for this particular situation) which now would work because we do instantiate the prefabs, which is sort of my leading solution

and then at the end of the process, delete those instances

but that seems like it would end up being costly to instantiate so many obstacles, becuase while i understand this process would take lots of times, especially due to how i calculate jumps, i dont want it to take forever, and if i can find a way to optamize this, that would be nice

[stray belfry]

frankly this seems like more trouble than its worth, usually no one uses such complex colliders for such things anyway

[silver cave]

Is anyone able to give me some insight on my game? I've made these big rigidbody meshs made out of cubes, I give each cube its own box collider. every box collider for the mesh is stored in one GameObject to keep the editor clean. To keep the amount of active colliders down I instantiate the colliders with the collider disabled(boxCollider.enable = false) and as my player gets closer to each cube, the collider enables. This system works but when they enable, the rigidbody moves without me touching it. Are my colliders colliding with each other? they all definitely touch each other as their coordinates are always whole numbers. Is the solution really to give each collider its own empty GameObject? That is all I can find online.

The gameobject that holds all the box colliders is a child object of the parent. the parent itself has no colliders

[tender vapor]

does each hitbox really need to be managed on its own? would enabling/disabling them all at the same time be viable? i feel like that'd make more sense

you could use a composite collider for that i believe, though im not sure it's the typical approach

[unique cave]

Have you profiled to make sure this optimization is required and works? Often when people do these kinds of optimizations, they end up being much slower than the optimizations unity has in place automatically. Enabling and disabling stuff on chunk basis (when chunks are used) can be very effective. Especially if every object has their own script instance attatched which checks for distances, that can get very slow very quickly

If we are talking of game similar to Minecraft, having a GameObject for each cube would alone be really inefficient

[timid dove]

What kind of game is this

[warped light]

Hey everyone,
I’m working on a pirate game project, but I’m having a lot of trouble fixing player jitter while standing or walking on the ship in the ocean.
Right now, the character feels like it’s constantly shaking a little, and walking around on the ship feels pretty unstable. It gets even harder because the game is multiplayer, and I’m using FishNet for networking.
How would you approach this kind of system? What would be a good way to make walking on a moving ship feel smoother and more stable for all players?
I’m mainly looking for ideas or implementation approaches to reduce jitter and make the movement feel cleaner.

[pulsar canopy]

There is too much here to comment on. For starters, is the character a rigidbody? Does it need to be a rigidbody? If it does, how does it move and with what settings? What kind of movement are you trying to achieve?

For example, character controller is much simpler and has no issues with jittering. It all depends on what you're trying to achieve

[foggy oar]

im not too sure where i should ask this but, i have an object, that when rotated, stretches weirdly and uncontrollably while the collider stays the same. How can I fix this?(I tried searching on how to fix this but I cant find anything)

the size of the parent is 8.506345, 35, 1

[foggy oar]

i tried making an invisible parent (empty object) but this still happens

By the way, it happens to the little squares too.

[silver cave]

@tender vapor @timid dove @unique cave

Chris- my game is a slop voxel game made of cubes. Think of Minecraft but spaceships instead of houses. My ship of 1000+ cubes isn't too much on the cpu, but i know when i have 100-150 players in a server its gonna get expensive real quick with all those box colliders unless everyone can only make a ship of like 900 blocks total. I have currently not tried just enabling/disabling all the colliders at once. I feel like this would have to be done with a coroutine to avoid a lag spike and would probably end up with the same result. Is there a difference between a compound collider and a composite collider? is there any difference when each hitbox is attached to its own empty object instead of them all attached to one?

AleksiH- The optimizations are more then required for what I want to do. My Spaceships are dynamic rigidbodys with a mesh. The mesh resembles cubes like Minecraft and is split up amoungst a couple sub-meshs for handling different materials. Basically, there is one ship parent object that does the checking if something is close to a block and turns on the correct hitbox, its a 5 radius circle of activation around the player. The script only starts checking when a player enters its activation zone, which is a giant circle trigger collider around the ship.

praetorblue- I am ripping off a game called Starmade. it came out a while ago and died fairly quick. It is a voxel game based in space. You build space-ships out of blocks like Minecraft, mine asteroids with your ships. and use those materials to make other things so on and so fourth. I would have just used a Mesh Collider for the ships but I wanted a dynamic rigidbody that reacts to forces instead of doing it kinetically. I might just have to bite the bullet and emulate the physics with a kinematic Rigidbody instead.

[hard fiber]

I’ve been racking my brain trying try to understand the sled physics in https://youtube.com/shorts/IX_whJmxnDI?si=4QGLjWs_V_8OXaz4

i have been working on a sledding game for a long time but its just riddled with ghost collisions I can’t get rid of

Does anyone have ideas for how to implement smooth sliding physics - a mesh ground and mesh sled? Something else?

[timid dove]

I think if you really want to do this efficiently you might want an algorithm that can dynamically convert your set of voxels into a minimal set of box colliders. E.g. if you have a whole section of the ship that's a contiguous rectangular prism, it should be able to use one single BoxCollider for that whole section

[tender vapor]

yeah apparently composite colliders are a 2d thing so they don't apply here, mb

[silver cave]

@timid dove So I was thinking about this. Based on the geometry of the ships that players will build, their ships could still end up with anywhere from 150-500 Box colliders. Would this really be that more efficient than like 50 non-convex mesh colliders attached to a single kinematic rigidbody? I say 50 because each chunk is its own mesh. I did this to avoid re-drawing the entire mesh of the whole ship each time a block needs to be added or destroyed. Edit: I also feel that with how fast i'll need blocks to be able to be destroyed, having an intermediate step that calculates, destroys, and re-adds box colliders every time will slow down the performance. With a mesh collider, it is destroyed and re-drawn without my shittly coded algorithim finding the optimal amount. Edit2: I see now that I cant have non convex trigger colliders, It seems my only option is to try your suggestion. ty.

[timid dove]

You'll definitely want to use object pooling for the colliders too

[quasi lion]

Im having some issues with phsyics. I am making a pinball machine, but even tho i have a hinge limiting the flippers movement sometimes it just glitches above the ball or something and rotates in a way that shouldnt even be allowed by the joint.
These are my settings. I feel like in general the physics are just doing weird stuff sometimes with these flippers but this is the only thing i can relaly place (i feel like it sometimes doesnt give as much power but that might be also because of this issue)

[frigid pier]

Make sure your ball is driven by physics forces and not being overriden

[quasi lion]

i dont have any code for the ball its just a rigidbody with gravity

[frigid pier]

Also for fast moving objects you want at least https://docs.unity3d.com/6000.4/Documentation/Manual/ContinuousCollisionDetection.html

[quasi lion]

I already have that enabled

maybe ill try extending the colliders lower

[frigid pier]

Check the joint manual as well, make sure it's properly limited.

[quasi lion]

i mean i have the axis, and its just moving outside of that axis

i dont think any other settings should allow for that to happen

[proper dove]

Hello there!
I'm making my first game right now, and there is this issue I have with slopes.

So, I want my character to glide off slopes at a certain angle, but I tried a few things and none of them are working.
I'm doing a lot with ChatGPT, but the things it recommended didn't work, so maybe someone here knows what I should try.
Btw, the shown script uses the Input Action asset to get, well… the inputs. I'm using a CharacterController, by the way.

https://pastes.dev/2TwwwKUl6Z

[tender vapor]

!code

[gloomy lodge]

Posting code

📃 Large Code Blocks

Use links to services like:
https://pastes.dev/
https://paste.yunohost.org/
https://share.sidia.net/
https://paste.ofcode.org/
https://paste.myst.rs/

📃 Inline Code

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

[tender vapor]

CC.Move should only be called once per frame, btw.

oh the other Move has a return after it

[proper dove]

oh my bad

[quasi lion]

man its really hard to not just complain about these physics when stuff is literally glitching through eachother even though i have stuff set as continues

[tender vapor]

make sure it's set to interpolate and you're actually moving using physics as well

[quasi lion]

interpolate?

[quasi lion]

i use a hinge motor

[tender vapor]

interpolation: interpolate

[quasi lion]

idk if that uses physics

[quasi lion]

ill try that i have none rn

on both the ball and the flipper?

[tender vapor]

mm i'm not familiar with that, but as a rule of thumb you have to make sure stuff isn't fighting over the transform
eg rigidbody and CC are mutually exclusive, if you have either then you shouldn't modify the transform, etc

[quasi lion]

i mean the only thing im using to move it is a motor which i think uses a rigidbody. as the whole joint needs one

[quasi lion]

thx this seems to have fixed it so far or at least improved it

ye it seems better

[silver moss]

Interpolation is purely visual and doesnt affect the physics simulation, so if that really fixed it then the issue was just cosmetic

The Physics settings has some settings to improve joint stability (like Solver Iterations iirc), they are pretty low by default

A couple of other things that might help with stability are to use a smaller fixed timestep in the Time settings and to disable Enable Preprocessing on the joint

[quasi lion]

weird cuz it does seem way less now so maybe it was just visual then

ye i did bump up the solver iterations for some other bugts but it didnt help for this one

[quasi lion]

oooh this seemed to make it better too

idk all these things combined made it behave way more deterministic

so thats good

[silver cave]

is it possible to make my own kind of primitive collider say a ramp?

[silver moss]

Not a primitive but you can always use a meshcollider

(A convex one probably)

[tropic girder]

hey, sorry in advance if im asking in the wrong channel,

is it possible to calculate bullet drop and target leading in one equation?

im basically coding a turret that tracks targets and shoots at them, adjusting for the bullet drop (from gravity) and adjusting to the targets velocity.

the way im doing it right now is im using a simple ballistics formula from wikipedia which calculates the angle at which you have to shoot to account for bullet drop, and once i have that i calculate the time to target and then the bullet lead (predicted target position), and i iterate this cycle like 2-3 times

and it works pretty well but i was wondering if its possible to do it in one calculation?

is it even worth bothering to calculate it in one go?

[inner thistle]

I doubt you'd win much if anything by doing it all in one go. Unless you need pinpoint accuracy or the targets are very fast compared to the bullet speed, it might be enough to estimate the flight time (or use a separate formula for that) and predict the lead first, and then doing the ballistics calculation once

[silver moss]

Yep, doing it iteratively might feel "dirty" but its often good enough for games

[tropic girder]

hmmm yeah all the stuff i read online is pointing towards a quartic formula, which looks like moon runes to me, so i think im gonna implement a dynamic iteration count depending on flight time

[unique cave]

I would think that such iterative algorithm would converge really fast towards the right solution. By managing the iteration count you can find a good balance between performance and accuracy

[tropic girder]

i consistently struggle with ocd lol 😅

[unique cave]

Would definitely set up a test to measure the accuracy. I would bet even with very little iterations, you would get so close it doesn't even matter if it isn't mathematically perfect

[tropic girder]

just did that, it seems first calculating the bullet drop, then the lead, and then drop again gives pinpoint precision with the least calculations, tho doing drop once then lead was very close in accuracy, so you were right to say it converges really fast

[steel notch]

I am trying to upload a avatar to VRC and i cant becuase i keep geting this error? I do not know how to modle its my frist time just even coloring an avatar and i need help to know how to fix this!!

(also idk where to post this im trying my best!)

[timid dove]

!vrchat

[gloomy lodge]

Join the growing VRChat community as you explore, play, and create the future of social VR! https://discord.com/invite/vrchat

[carmine needle]

I have a question about the Low Level Physics 2D api. What is the expected way to move a Kinematic PhysicsBody? Is there a Rigidbody2D.MovePosition equivalent? Should I just directly set the velocity?

[tender vapor]

Kinematic: Unity applies no forces or gravity, but allows collisions with Dynamic bodies. Kinematic bodies move only when you manually reposition them or set their velocity.
appears so

MovePosition works by calculating the velocity to get there anyways

[noble oar]

  private int weight;
  private Ray ray;
  private Vector3 rayPos;
  private bool isGrounded;

  private bool _CheckForGround() // Sends out 4 raycasts around the player to check if he is grounded, only sends a bool for now
  {
      isGrounded = false;

      for (int i = 0; i <= 3; i++)
      {
          if (i == 0)
          {
              rayPos = new Vector3((transform.position.x + 0.5f), transform.position.y, transform.position.z);
          }
          if (i == 1)
          {
              rayPos = new Vector3((transform.position.x - 0.5f), transform.position.y, transform.position.z);
          }
          if (i == 2)
          {
              rayPos = new Vector3(transform.position.x, transform.position.y, (transform.position.z + 0.5f));
          }
          if (i == 3)
          {
              rayPos = new Vector3(transform.position.x, transform.position.y, (transform.position.z - 0.5f));
          }


          ray = new Ray(rayPos, new Vector3(0, -1, 0));

          if (Physics.Raycast(ray, out RaycastHit hit, 1.1f))
          {
              isGrounded = true;
          }
      }
      return isGrounded;
  }

  void FixedUpdate()
  {
      if (_CheckForGround())
      {
          // Make player jump if he wants
          // depending on material can add footstep sounds
      }
      else
      {
          rb.AddForce(-transform.up * weight, ForceMode.Acceleration);  // Pulling character down, simulating gravity
      }


      Vector3 LocalMovement = (transform.right * direction.x) + (transform.forward * direction.y);
      rb.AddForce(LocalMovement * speed, ForceMode.VelocityChange);

      rb.transform.localRotation = Quaternion.Euler(0, mousePos.x * Sensitivity, 0);
  }

Hey guys, just wondering if coding a certain weight to a singular object is nessessary or is there a attribute system for gameobject's "weight" that I'm not informed of? If there isn't then here is my code and wondering how you guys would improve on it?

[inner thistle]

Gravity is famously the same for all objects regardless of their weight

Also rigidbodies have a built in gravity simulation, you don't need to make one yourself

[noble oar]

Am I able to change the gravity for a singular rigid body from components or is that something I have to hard code in?

At first I thought changing mass would make it go down faster but I was wrong

It would be useful to know because lets say if the player goes into the water, being able to change it's weight to be lighter would be nice if weight is already an attribute for the rigidbody like Mass is. But I have to hard code the weight in the script above

[inner thistle]

Weight doesn't affect gravity. In water, what makes things "lighter" is buoyancy which depends on density

and in air what makes things fall slower is drag, which can be simulated with linear damping

[noble oar]

Ahhh ok

I did a quick search and saw that messing around with linear damping could get what I want

Tysm for the help

[hushed shoal]

To do it in a more realistic way you could apply an upward force to the object, with the length of the vector calculated from the depth of the object as well as the density. This could even be expanded further by taking multiple points on the object, and applying forces at those points, so it would bob up and down, and rotate when waves hit.

[unique cave]

How deep underwater something is doesn't matter actually, only how much of the object is underwater (volume). Is that what you mean? If something is fully submerged, it doesn't matter if it is 1m or 100m deep, the buoyancy force will be equal

[hushed shoal]

Good point!

[pulsar canopy]

That's not quite true. Buoyancy is lower the deeper you are