⚛️┃physics

[simple grove]

lol

the "balance" I understand that is what the player will do right?

[viral ginkgo]

Eric Fakos, lead developer of Proton Quantus

[simple grove]

lol

Just one thing: I'm not lead. fholm is

transform.rotation = Quaternion.Slerp(transform.rotation, Quaternion.Euler(0, 0, -Input.acceleration.z * 90), 0.2f);
This will rotate the character... you are also interpolating here, which should already smooth a bit...

I assume you are running from FixedUpdate, right?

So no need to consider Time.DeltaTime

so there's an error

that 0.2f

you should read from Time.DeltaTime * a confirgurable interpolation

Because unity's update will NOT work constantly at 50Hz like that 0.2 assumes

that's why

The reason it is smooth is because you are interpolating A LOT

You are taking almost 1 full second to rotate it...

Maybe you want it to react faster, etc... It's all doable, but you need to got your code right

[pure mist]

Regarding Joints. I have a fixed joint on camera, and it connects to a rigidbody (to simulate dragging). Works great. One issue is now that I want to set the object's "depth" (near or far from camara) and to do that I need to modify the joint's connected anchor. However, I am getting some weird readings. I understand it's perhaps due to the anchor's orientation following a different convention than the global orientation? If this is so, how would I go about converting that over and reading it in relation to world position?

Here is my debug code:

Debug.Log($"Connected Anchor: <b>{Mount.connectedAnchor}</b> | Cam Pos: <b>{Camera.main.transform.position}</b> | From Camera: <b>{Mount.connectedAnchor + Camera.main.transform.position}</b>");

And the result:
Connected Anchor: (0.9, -9.7, -6.5) | Cam Pos: (0.0, 7.0, -6.7) | From Camera: (0.9, -2.7, -13.2)
And the actual position of the connected rb (in world coords): (0.15, 0.743, 3.05)

Please help 🙂

[stuck bay]

Yo can i have help setting up foot ik on 2d character
i have used hinge joints for the ragdoll

[wet bluff]

hey guys

I just found a script on youtube where it converts my polygon collider to an edge collider (making the collider shape hollow)

Is there anyway to save what the script does so that when I stop playing the game the changes made in the game are permanent?

Is there a way to copy the points data of the polygon 2d component and paste it in the edfe collider component in the unity editor?

[pure mist]

Whew this took a while to figure out, but I did, I can now set the object depth (while dragging it) with any joint by only modifying the connected anchor point. Works awesome 🙂 If anyone else had issue with this, let me know.

[wet bluff]

I figured it out

On runtime, you can copy the updated component

And then paste that component in the editor

[karmic yoke]

Whats best practice for retrieving a smooth velocity from a ridgitbody?

Is this normal for a rigid body that is not moving? https://im6.ezgif.com/tmp/ezgif-6-bbb196b4acf5.gif

[north egret]

use SmoothDamp

[pure mist]

I have this unwanted behavior, was wondering if anyone else had this issue? I am using a Joint (doesn't matter which one, this applies to all of them). My script adjusts the connectedAnchor point so that the connectedBody moves in relation to this joint. Now this works perfectly if the host joint is either in motion or is rotating. However, if the host joint gameObject is NOT moving or rotating and the script is adjusting the connectedAnchor, the connectedAnchor actually updates and moves (verified by OnDrawGizmos), but the connectedBody doesn't move and won't update it's movement UNTIL the host joint gameObject moves or rotates, to which then the joint updates the connectedBody correctly. If this is confusing a requires a visual demonstration, let me know I will make one.

As a dirty workaround, I just have the host joint object transform backwards by 0.001 to make use the update occurs correctly. If anyone can suggest something more cleaner. I'm all ears 🙂

[flat pendant]

For Ragdoll, is there a way to bake transforms of a ragdoll to use as static? e.g. running a ragdoll simulation in editor and freezing transforms for runtime?

[karmic yoke]

@north egret How would I use that to pull a clean velocity from the ridgit body?

[viral ginkgo]

@karmic yoke i dont know what you are doing but from the question you are asking, it soulnds like youre doing things weirdly

why dont you tell us why you need this "clean" velocity in the first place?

[sweet flicker]

Im making a 3 omniwheel platform in Unity. I want it to move north east by powering some motors. With side friction completely turned off, all directions except North East and North West, work fine. Now im testing it with adjusting several variables, but no matter what I do... the result is the same. Anyone having a good explanation of what the variables exactly do? When I look it up online, everyone says something different.

The picture below shows the traveled path starting from the center. It always starts correctly, but at a specific point is basically starts rotating another direction.

[loud moon]

Anyone have a good understanding of what preprocessing does for joints?

[stuck bay]

I have simple a cuboid, with 2 raycast wheel collider at start and end center of shorter edge. Which acts like suspension of bike. Now what is the best way to balance it? Moving C.O.M or adding torque to balance?

[karmic yoke]

            var targetRotation = Quaternion.LookRotation(targetpos - transform.position, Vector3.up);
            transform.rotation = Quaternion.Slerp(transform.rotation, targetRotation, Time.deltaTime * rotationSpeed);
            rb.AddForce(transform.forward * Thrust);```

I'm using this to control a ridged body to a target position But it keeps over shooting its target by a decent amount. Does anyone have an example of a pid controller to make it a little more accurate?

[viral ginkgo]

@karmic yoke

Vector3 pos;
Vector3 velocity;

Vector3 targetPos;
void Update(){
  Vector3 delta = targetPos - pos;
  Vector3 pForce = delta * pMultiplier * deltaTime;
  Vector3 dForce = -velocity * dMultiplier * deltaTime;
  velocity += pForce + dForce;
  
  pos += velocity * deltaTime;
}

position pd

read and edit velocity from rigidbody, read position from transform or rb.position

and dont use pos+= . . . line in the end

this example just makes its own physics in plain c# you know

@karmic yoke you should also clamp force before applying it
so that character doesnt apply impossible forces when either distance is too big or velocity is too big

[opaque vine]

Hi, I'm trying to do player movement using transform.Translate but currently when the player tries to run against a wall he sort of glitches in and out of the wall. I would like to prevent that by just moving the player to the side. I tried using hitInfo.normal to compute that vector but that doesn't seem to work for cubes not aligned to 90 degrees to the player (the normal vector is weird). Any ideas how I can compute the green line?

[viral ginkgo]

@opaque vine
green = Vector3.ProjectOnPlane(red, surfaceNormal)

[opaque vine]

Woah thats easy! Thanks a lot !!

[viral ginkgo]

np

[ocean pivot]

Oh neat, I never knew that was a thing

[viral ginkgo]

do the ctrl+space now and then

[ocean pivot]

I mean, still means I have to think that Vector3 has that 😂

[opaque vine]

Yeah I really wasnt expecting that

and I feel like its a weird place to put that function

[viral ginkgo]

Vector3 math is just in Vector3 static functions

You also got Vector3.Lerp in there

And you also have Color.Lerp for colors

[opaque vine]

I guess

[stuck bay]

Hey, I think it fits here: How do I get the position where a particle collides with an object? I'm using OnParticleCollide.

[exotic flax]

Uhh

Maybe

Oh wait

Nvm

Sorry idk

[mighty sluice]

it's like browning motion, but with lamprey

https://i.imgur.com/0D7OFae.mp4

[stuck bay]

Can anyone please help? should I use pid with one of those?

[supple sparrow]

PID maybe, also did you try with a spring joint ?

[stuck bay]

I'll check spring joint, should I use PID with moving center of mass or with torque to balance?

I am planning to use it with network, so avoiding default unity wheel collider, my ray cast wheel is very similar to the concept in this video. Have no idea how spring joint would work with networking

[supple sparrow]

Multiple ways of doing it, depending on the accuracy you need:

• have your server authoritative for physics (calculations done server side only) then sync transforms (pos & rot) on the client
• have the server send relevant variables to the clients so they can apply their own physics (spoiler: unity Physics are not deterministic so only works for simple games where collisions are not the main focus of gameplay)
• implement your own physics system, not relying on Unity's one

[tropic wigeon]

Hello. (I am not sure where do I post this)
I am learning DOTs and one of the problem I have is my Entity (sphere) sinking in even though I I applied the Collison filter correctly. Not sure what is causing that.

[opaque vine]

SweepTest does not seems to work, when you are already "in" that object even by a tiny tiny bit. So if you run against a wall with high speed you glitch into the wall. Is there any workout besides adding extra range to the sweeptest?

[supple sparrow]

Increase simulation steps maybe? Would probs put a heavier load on the computation though

[shrewd shoal]

Fish must consume ball

[lapis plaza]

@stuck bay if you actually use physics in way that bike can rotate in that axis then PID would be most common way to solve that, yes

you could use ML to train to correct it too I suppose

[obsidian whale]

i have a capsule collider and rigidbody attached to my player, and when a enemy projectile with a sphere collider/rigidbody hits the player, sometimes the player just goes flying up into the air with a high velocity? anyone know how to troubleshoot

[tender gulch]

@obsidian whale how do you move the player/enemy projectile?

[obsidian whale]

enemy projectile is just flying through the air from its initial AddForce the enemy gives it (arc parabola), and the player is standing completely still ontop of a moving vehicle when it happens

[tender gulch]

How is it standing on it? Is it parented?

[obsidian whale]

no, it is just standing ontop. the force of gravity pulling the player down onto the surface of the car is what allows the player to stay perfectly still ontop

the player and car are two seperate gameobjects, and the car is moving using wheelcollider torque

[tender gulch]

hmmm

might be because of lag..?

try artificially reducing fps and see if that occurs more often.

[obsidian whale]

i increased solver velocity iterations to 32 (from 1) and it hasnt happened since, but ill keeptesting a bit 🤷‍♂️

[tender gulch]

that might have a huge impact on performance.

Did you try setting collision detection on your object rbs to continuous dynamic?

[karmic yoke]

What would be the best way to add some noise to to a projectile traveling to a target. below is what I have for a missile. rb.velocity = transform.forward * Thrust; var targetRotation = Quaternion.LookRotation(targetObject.transform.position - transform.position); rb.MoveRotation(Quaternion.RotateTowards(transform.rotation, targetRotation, rotationSpeed));

[north egret]

i have a question, why wont OnTriggerEnter2D() owrk even if i have a 2d trigger?

[tender gulch]

Do you also have something that enters the trigger?😁

[proven lance]

Do you have a rigidbody xd

[random leaf]

Hey guys, I don't use complex / realistic physics in unity much but in the case of a floating airship eg. blimp - in 3d not 2d (with player able to walk around while in motion)
Would the right approach be to give the base object an accurate level of mass, and have sufficient constant upward force (eg. at 6 points for stability) to keep it afloat?

Would I need multiple base objects instead of one (connected by joints?) with accurate mass to give better stability?

Are there any tutorials that go over general more complex physics use cases and best practices? (Not just tutorial on how to use spring joint for example)

Cheeers 🙂

[viral ginkgo]

blimp huh
sounds like you'd want a center of mass
and a center of lift
@random leaf

you should be able to change center of mass through the scripts

for center of lift, you somehow would need to figure out center of volume
and add upwards force depending on the volume every frame

[random leaf]

that sounds like a good approach

[mental kelp]

QUESTION:
I am working with a hinge joint on a moving object. The hinge joint is child of that moving object and the joint moves that parent object by moving the hinge joint as a lever. Now i see that moving the parent object without rotation works just fine. But when i try to rotate the parent object, the hinge joint/lever gets unwanted movement. I thought parenting would fix that? Is there some special "Connected anchor" setup for this?

[mental kelp]

TLDR: A hinge joint setup that represents a lever rotating in x axis has false intput when rotating the whole joint along y or z axis. x axis is fine though. Why?

[random leaf]

Crappy models to simplify the idea:
Say this is my airship, how would I have all the parts taken into account for buoyancy in the air, how would I keep my upward forces balanced (ie. able to tip slightly but not immediately tip down like a helicopter), all collision objects considered when bumping into other objects, but again so that a collision (esp. at slow speed) wouldn't send it careening off slowly towards the ground?

I know these are complex questions but I'm struggling to figure out what I need to google even to figure out the answers

Thanks for any help / suggestions 🙂

[mental kelp]

@random leaf How about u connect the top and bottom with chains/ropes implemented with maybe hinge joints. Then give the top a RB with constant up forces just like gravity but up

then apply RB force movement to the bottom part and the chains and top part should follow accordingly

it will be quite hard to stabalize it for floating because rising force and gravity must cancel each other perfectly.

[mental kelp]

or apply force movement depending on where you want to have your driving thruster/propeller

i guessed the bottom part cause of the propeller like thing in the back

[stuck bay]

Oh okay, thanks you so much!

[stuck bay]

I'll try ML as well

[random leaf]

Yeah it's pretty simplified really, I'll see what I can figure out 🙂

[random leaf]

Found this which was a great help for stability

https://answers.unity.com/questions/10425/how-to-stabilize-angular-motion-alignment-of-hover.html

[oak isle]

2D Code

Error:
Simply, I am making a multiplayer game in 2D. Currently using BoxCollider2D. I want the player to be able to collide with walls, and ECT. When two players collide, as in one running into one. The player that got pushed is vibrating, or getting pushed away from the other player. This only happens in the client side that pushed the character and not the server. I know this is mainly because the player position can only be changed within the player itself through my code. But instead of doing anything with this, I would just like to ignore collision on players. I tried using the Physics2D and disabling the layer that uses the Player to stop colliding. It did not help.

[random leaf]

What sort of joint setup would you use for a spinning propeller? One that spins up from 0, but also with no forces upon it will spin back down to 0?

The propeller doesn't need to have any impact on the rest of the physics, more of a visual thing but I was hoping to leverage the angular drag to get it to spin itself down

ah figured it out

configurable joint

[oak isle]

For My Proble:
It seems that my Prefab didn't change it's layer. And I just checked it to make sure. I found a fix to it Lol.

[random leaf]

Got some airshipping going on 🙂

[viral ginkgo]

@random leaf
1- center of lift is higher than center of mass
2- make upwards force from center of lift
3- make downwards force from center off mass
4- make sure you have some angular drag on these or it will oscillate

and thats it, your airship will be stable

[gilded sparrow]

What's a simple formula for steering? To not overshoot a homing target?
I'm only doing a velocity += dirToTarget * speed * deltatime, and it's overshooting
I'm supposed to do like, dir + something based on current velocity or something..

[pseudo peak]

I have a problem guys, OnTriggerEnter() doesn't get called if it happens outside the camera's view.

[gilded sparrow]

What are u trying to get trigger?
Maybe something moved by Animator?

[pseudo peak]

That's exactly what I'm trying to do

It doesn't register when the player hits an enemy or enemy hits the player.

If you're wondering how the enemy is outside the camera view in that case, it's because it's a multiplayer game

And everything is done through host, and when the host isn't watching, client's hits doesn't count

I've tested it in offline mode too by blocking the camera, so I'm sure it's not a networking issue

[gilded sparrow]

U gotta set the animator to AlwaysAnimate

[pseudo peak]

I owe you a drink, it works!

Thank you! 😁

[gilded sparrow]

Alright

So yeah about my homing formula
This is i guess some rocket science

Vector3 dir = target - p.position;
dir = dir.normalized;
dir *= homingSpeed; // speed towards target
dir -= p.velocity; // the "balance" force

p.velocity += dir * Time.fixedDeltaTime;
p.velocity = Vector3.ClampMagnitude(p.velocity, homingSpeed);

But this still overshoots
My use case is, the particles explodes outwards, but overtime they're "steered" towards the target position. So it looks like a missile bursting outwards first then straight to the target

If the target is too close, or the burst outwards is too strong, then it'll overshoot. So i guess the "speed towards target" part also needs to be balanced, or even reversed, depending on some factor. I just dont know what this factor is..

[kind kestrel]

just curious what everyone’s process is for creating accurate and inexpensive colliders in Unity? I’m trying to import a friend’s 3d art and apply a rididbody but every method i’ve used to create colliders is inaccurate or too expensive on physics. i’ve tried techies collider creator and a couple of other paid and non paid assets for collider creators but nothing is working smoothly.

[stark girder]

Guys. When an object falls and collides with my another object, the object which was at rest, squishes a little. How can I prevent it?

it is 2D btw

[random leaf]

How would you implement a controllable character able to walk around on a moving "vehicle" with semi-complex mesh?

Needs to be physics based to gel with the physics of the moving thing & be able to navigate it

Parenting to the physics object is definitely important for the local physics but I'm still getting issues with sliding

[hexed horizon]

is jobified/burst physx casting possible?

[viral ginkgo]

@random leaf Could parhaps duplicate the vehicle physics components in a static environment and have the character simulate physics there

You'd make a capsule with rigidbody for player

Duplicate vehicle physics mesh

Put them in a different layer

And attach the oncollision callbacks and stuff

And then do some matrix transformations to transform the transform of that capsule in that envoronment in to your visual character

[random leaf]

ahh interesting idea

[upbeat dirge]

hey guys

i wanna ask before going to buy Magica cloth being on sale, is there any hope that unity cloth 2020 will be as good as Magica cloth or Obi cloth ? any roadmap for it ?

[gilded sparrow]

Is it a good idea to pool many rigidbody colliders?
I'm thinking around 1000 on average. Pooling means setting it in/active often, which i hear is a perf hit
What should i do instead for pool? Just "teleport" them out of the map when not in use?

[burnt wedge]

Hi, I'm having a problem setting up collision detection on custom mesh which is 10k triangles.
All I want to do is to get what points from my custom mesh are currently inside a box collider , then I will deform them to be outside of box
I'm trying to achieve something like it:
https://i.stack.imgur.com/kQfRv.gif

But I'm having a problem if there should be two rigid bodies or just one? And they shouldn't move (IsKinematic = true, which also disables detecton? 😛 )

[stuck bay]

@gilded sparrow maybe consider using DOTS for the collisions?

[gilded sparrow]

Cant.. and 1000 is still alright

[stuck bay]

Alright - pooling is usually trading memory for performance.

Or maybe something more hacky - run a force sleep script on rigibodies not in use. That would save the collision checks.

[gilded sparrow]

That could work too

I still need to teleport it tho right? Bcoz after they're forced sleep, they can still be sleeping in the middle of the road and have cars crashing on them (just an analogy)

[stuck bay]

Good question. I would try it out without teleporting first (guessing you are hiding the mesh) - if it works and is performant no reason to move them around.

[gilded sparrow]

Theyre just invisible triggers

Yea definitely need to profile stuff

[burnt wedge]

anyone this?

[noble notch]

trying to make a fully physics based tank tread, here's my current progress

any tips to make the simulation more stable?

I've tried increase physics simulation iterations with no real difference

[quiet ivy]

quick math problem, how do i make player not be able to go over speed limit (but also not slow down when pressing the key in direction of velocity) but also slow down when pressing in the other direction or change direction when pressing the key to go to the side

(i am using addforce)

[tough path]

So this is builds off of @quiet ivy's question but slightly better explained:
So I'm developing a first person game, and I want to make realistic player movement. To move the player, all I have to do is call one line of code AddForce(), in which I pass in a value that acts as the amount of force being applied. The problem with this is that if I add a constant amount of force, then the player takes a really long time to accelerate and it moves really fast, so what I have to do is change how much force is being applied based on the velocity of the player and the friction and gravity but I don't know exactly how to calculate that. Is there maybe a function that I can use that takes some variables about the player's velocity, drag, etc. and determine how much force should be applied to the player?

[quiet ivy]

i got most of the walking fine

even keeping the velocity of the object i am standing on

but in the air it just feels not right

i want to have slight control in air

[tough path]

What's your drag set to Foxyz?

[quiet ivy]

0

[tough path]

good lol

[quiet ivy]

the problem is

when i am jumping off a moving object

and holding w for example

in the direction im moving

it slows me down

and i don't want that

if i set forcemode.force then it's gonna go over the limit

[tough path]

So what you do is you add the moving object's velocity to your velocity

[quiet ivy]

basically

it works wonders but as i said i want little air control and so i wouldn't get slowed down while pressing the key in the direction im flying to

but still being able to slow down if i press in the opposite direction

from what i've got it's just a vector math problem but it doesn't seem that i can figure that out by myself

ping when answering me cause i'm going sleep

[distant crater]

I have a weird issue, basically i try to stomp on enemies in my 2d platformers but sometimes the stomping function happens far away from said enemy when i'm above them

[dusty verge]

I need some help :/
for some reason if I enable any line that has references to velocity.x or z I can no longer fall.

basically all im doing is taking my rotation dividing it by the absolute value of itself to get its direction.

IEnumerator rebound ()
{
velocity.y = Mathf.Sqrt(jumpHeight * -2f * gravity);
float absRotationx = Mathf.Abs(Rotationx);
float absRotationz = Mathf.Abs(Rotationz);
velocity.x += Rotationx / absRotationx * 50;
//velocity.z += Rotationz / absRotationz * 50;
yield return new WaitForSeconds (0.1f);
//velocity.x -= Rotationx / absRotationx * 50;
//velocity.z -= Rotationz / absRotationz * 50;

I just tested it again I cant move vertically at all

[stuck bay]

You might be misinterpreting the value out of context when calling for it. Consider making the velocity call points individually (though may take longer) instead of calling a variable (in this case, rotation or velocity). If you have difficulty reaching out to the rotation or velocity reference, there is most likely a problem of how your variable was stored. Good luck @dusty verge

[dusty verge]

hmm alright, thanks

[quiet ivy]

I'm up again and ready to negotiate

[dusty verge]

This line of code is supposed to output the rotation of the y axis but it gives me a decimal between 1 and 0 instead of the actual angle of rotation listed in the inspector is there a way I could get the rotation listed in the inspector using C
Rotation = FPP.transform.rotation.y;

[dusty verge]

fixed it had to use eulerAngles

[raven grotto]

hey guys, I'm trying to use Articulation Bodies to make a robot arm's tool follow a target. I want the movement of the end of the robot arm to naturally adjust the position of all the other joints, but I'm struggling to get the robot to move at all. Here's a pic of the types of motions I want to simulate and how I've set up my joints (they're indicated in light blue) -

lastly, this is my current hierarchy for the robot. The "tool" is the smallest attachment at the end of the robot. This hierarchy makes the components move naturally when the component below them is moved or rotated, but for Articulation Bodies to work, I understand the "tool" component needs to be the parent?

[supple sparrow]

I never used articulation bodies (i would just go the IK route) but isn't this force limit to high ? Maybe that's what is preventing motion ??

[raven grotto]

hey, thanks for the response! I tried several lower force limit values, but nothing changed :/ I believe there's something wrong with the hierarchy or the joint configurations I made, but I can't figure out what. If I try to move the end of the arm, the entire robot just vibrates rapidly but doesn't move

[supple sparrow]

Just to make sure, can you try zero as values for Force Limit AND Damping ?

Otherwise I'll let answer someone who already worked with that

[raven grotto]

hey, thanks again! I tried to zero both values and I can move the joints now, but I can't push/pull the tool joint and cause the entire robot arm to move in response. I assume I need to use inverse kinematics in combination with articulation bodies to get that sort of movement?

To clarify what my goal is, I'm making a HoloLens 2 application where the user can pull a robot arm with hand gestures in whatever positions they wish for validation purposes. I want to be able to drag the end of the robot arm and have the rest of the arm respond realistically to that motion.

[slim olive]

I'm doing a physics2d.overlaparea
how would I get a count of how many colliders are in contact

Collider2D[] collider2Ds = new Collider2D[2];
                ContactFilter2D contactFilter2D = new ContactFilter2D();
                cast.OverlapCollider(contactFilter2D.NoFilter(), collider2Ds);```
This was my attempt after reading the scripting API but it doesn't work

[supple sparrow]

Nice ! Yeah IK purpose is exactly what you describe. Have a look at the "Animation Rigging" package, they have a simple 2 bones solver easy to implement. It should even bé enough on its own, but I'm not sure why you're using Articulation body so maybe you still need it. And also I dont know how they would behave together as I never used it

[raven grotto]

thanks for the pointers! I wanted to use articulation bodies to get access to real simulation physics with target velocity, joint friction etc. to make the robot behave like in real life, especially when it's following a target. I'm not sure how realistic the behavior will be with IK but I will give it a shot!

[supple sparrow]

It just solves position on n-1 bone. But no joint friction, so presumably lower than your expectations. Happy coding !

[outer bone]

Hya! I'm working on a Glider Plane in my game. It uses a rigidbody and edits to the rigidbody.velocity to move.

When I add a collider though the rigidbody / object behaves completely different then without; Direction changes, the object moves way faster, etc

Why is that, am I using a wrong approach?

Ive tried using a standard 3D Cube with the same script, also getting weird results

What specifically changes when I add a collider?

I'm using the same Velocity vars

Oh and if I disable the Box Collider of the cube, it flies just fine too

Anyone any idea what Im doing wrong?

[tranquil onyx]

maybe the center of mass?

@outer bone try setting rigidbody.centerOfMass to what it is when you just have one collider. I also know that just having larger colliders can affect the behavior of rigibodies, so you may need to adjust your force amounts

i believe you can only change the COM in code btw

[outer bone]

I wasnt using Force, instead calculation velocity (which im thinking of getting rid now again :P)

Im unsure what exactly caused the behaviour, but I re-coded the scripts and rebuild the gameObjects and it works well now

I did find some articles / posts saying things change since you add a center of Mass that that could have had something to do with it

others recommended setting the inertiaTensor

but that is Definately above my paygrade 😛 Im glad I got it working now, controlling in-air collisions with physics materials, works pretty well!

[gilded sparrow]

HI im just wondering
The ParticleSystem's Collision module said it uses the Physics world. Is there a way to make use of this kinda of "physics" object, say without using the particle system?

I mean, i can handle 30k particles total with the Collision module, but i definitely can't do 30k rigidbody + collider gameobjects

[stuck bay]

How can i fix this?

[stuck bay]

Im using Rigidbody2D

no code or anything

[tough path]

@stuck bay set the collision detection to continuous

[stuck bay]

i did

[tough path]

Oh dang

[stuck bay]

wait

for the walls aswell?

[tough path]

Oh the walls have a rigidbody?

Why?

[stuck bay]

no i was asking if they need to have it also

[tough path]

Try it

[ruby frigate]

I have a player with CapsuleCollider and RigidBody.
He is walking on a floor with a Box Collider.

Floor's BoxCollider has a physics material with very high friction.

Yet, player's movement is not slowed at all (just as if there was no friction). Does anyone have an idea what I'm doing wrong?

[stuck bay]

ok

[tough path]

Copy and paste?

[stuck bay]

no still doesn't work

[tough path]

@ruby frigate let me see a screenshot of your physics material

@stuck bay I'm sorry but I dont think unity has the capability to handle collisions that well

Wait 4 years for havok and dots to come out of alpha lol

[stuck bay]

its top down, i see games that have top down collisions?

[tough path]

Then your problem will be solved

Yeah of course

[stuck bay]

how do they do that

[tough path]

But, from what I can tell from the video, you're forcing the object into another

[ruby frigate]

@tough path

[tough path]

Set the friction combine to multiply

Idk if that'll work

[ruby frigate]

Sadly didn't change the problem.
It seems the friction does work, but stops at some point.
So the player slows down quickly, but will then keep moving forever, at maybe 20% of original speed (instead of slowing down to 0 due to friction)

[tough path]

@ruby frigate thats not a friction problem

Thats a rigidbody addforce problem

[ruby frigate]

Is addforce not working with the friction settings in PhysicsMaterial?

[tough path]

No not what I meant

So I assume you're doing add force on fixed update and you're adding it based on Input.GetAxis for player input, right?

[ruby frigate]

Doing it with Input.GetAxis, but in Update() and multiplying by DeltaTime.

[tough path]

Okay well first of all

Do it in fixed update

You should always do physics related stuff in fixed update

Second of all, how physics works is if you add the same amount of force over a period of time, you don't stop accelerating (unless there's friction involved)

So what you have to do is somehow calculate how much force you should add based on the player's velocity

I've been trying to figure out how to do that and I've had a hard time

I'm gonna email my physics teacher tomorrow lol

[ruby frigate]

But shouldn't the friction slow down the player? I apply force while pressing W. Then I stop pressing -> shouldn't the friction force be applied until the player eventually stops

[tough path]

It is slowing down the player

But in this case, friction is a constant

So the same effect I was talking about would still happen, but it would just happen slower (based on how much friction you have)

[ruby frigate]

Well there definately is something wrong with my setup.
Because no matter how you look at it, this behaviour is weirds:

• pressing w for 3 seconds speeds player up to max speed.
• releasing w
• player speed quickly is decreased from 100% to about 20%
• then player keeps moving forever at 20% of max speed

[tough path]

Yeah unity doesn't handle physics very well

I've said this before and ill say it again: just wait four years for havok and dots to come out of alpha testing then you'll be set lol

Simple enough, right?

[ruby frigate]

Ok now it seems to work, but removing all PhysicsMaterials, setting the friction directly in the player's collider, and setting friction to 10 -_-

[novel pasture]

I'm trying to use the GraphicRaycaster to see if my click hits any UI button, however I can't seem to restrict the layers that can be hit. .My buttons have a specific layer (UIButton) but my raycasts are also hitting all the other UI elements (which I don't want to detect).

Is something wrong with my GraphicRaycaster settings ? I tried with "All" and "None", no luck

[mellow idol]

Does anyone know if ConfigurableJoint has a limit for Angular Y&Z limits? I cannot seem to set any value below 3 (aside from 0). Angular X Limit can be set to anything though 🤷‍♂️

[umbral vector]

sticking one more question in here because I'm not sure how to go about this or what to look up (2/3 through calc 3)

i'm revamping my grappling hook system and it doesn't use any rigidbodies, where in this problem (a,b,c) is my position and (d,e,f) is my projected position

i've been directing player velocity directly from (a,b,c) to (d,e,f), and while it works, it's causing some odd bouncing issues because of approximations

the sphere is mapped by x^2 + y^2 + z^2 = grappleLength

[viral ginkgo]

first you need to define a function for that particular curve
in parametric format as such:
fx(t) = ...
fy(t) = ...
fz(t) = ...
where fx is x position, fy is y position, fz is z positon
at given "t"

and then, you take the time derivatives of each component and obtain:
f'x(t), f'y(t), f'z(t)
and then, these are your x,y,z components of your swing vector (for your time derivative "t")
@umbral vector

But you shouldn't be needing any of that calculus to move an object in a curve like that

Vector3.Slerp is for exactly these situations

[umbral vector]

I'm still really new to unity, I was going to use the unit directional derivative to direct the player's velocity to the sphere's surface

oh my god

never used slerp before but it looks like exactly what I needed

thank you ❤️

[viral ginkgo]

no probs

@umbral vector

you have a third param in lerp and slerp, thats the "t" of the curve

@umbral vector But this is also probably not the best way

Wanna hear how i'd do it?

[umbral vector]

absolutely, i've gotten no guidance on the best way to do it so far

[viral ginkgo]

1- apply external forces or gravity
2-look at all the forces applied last frame, update velocity
3-remove velocity component on rope direction
4-update position from the velocity
5-translate the player in rope direction so distance between rope pivot and player doesnt change
@umbral vector

This will give you a stick instead of a rope tho

But it can easily be made like rope

[umbral vector]

the way I'm doing it now goes like this, and it only applies when the player is outside or on the grapple hook's length:

1. find where player is expected to be from current velocity
2. draw a line from the hook to the expected location
3. clamp the line to the grapple hook length
4. find direction from current location to clamped expected location
5. direct player velocity in expected that direction

i had teleports but ended up removing them

there's two problems with this approach though

the player will move in circles at the bottom because gravity is always pulling the player down, then that force gets turned into sideways movement

as of changing some things in my code the player is constantly gaining velocity

as in every swing they'll go a bit higher

[viral ginkgo]

multiply velocity with 0.95f every frame to make air resistance

[umbral vector]

that's what i've been using so far

[viral ginkgo]

that doesnt sound good

[umbral vector]

one second

the comments are really outdated

[viral ginkgo]

whats with the expected position and stuff

[umbral vector]

expected position is where the player will be next frame

a vector is made between the hook and that expected position, then clamped to the rope's length

[viral ginkgo]

you should be thinking about what velocity should be only
and position should figure out itself
and afterwards, you just translate the pos a little bit to maintain rope length

[umbral vector]

it's only used to determine what direction the player will move in

there's no teleportation with these scripts, the expected location is just a value

[viral ginkgo]

i see okay

[umbral vector]

translateDirection will need to make use of slerp

right now it just uses homemade lerp

[viral ginkgo]

translate direction shouldn't be necessary

you dont have gravity?

you dont have force/acceleration

[stuck bay]

Hello

[viral ginkgo]

you should be doing velocity += gravity

[umbral vector]

oops, the player doesn't feel gravity on frames when being redirected right now

one moment

[viral ginkgo]

and that should be the energy to start all the movement

or some initial velocity

[umbral vector]

so it turns out the player was having gravity applied on rope physics frames, just after the redirecting happened, definitely the wrong place

however i also found the increasing velocity problem

this dumb peice of code that was going to make the bottom of the swing a magnet which one of my friends suggested i try to stop the perpetual circular movement at the bottom

[viral ginkgo]

thats just gravity

and that shouldn't stop the circular motion, air friction should

bottom is a fixed point?

and thats not gravity?

thats a magnet then as you said

and that wont stop the swing neither

[umbral vector]

basically it would pull the player towards the bottom of the sphere

[viral ginkgo]

it will just make it swing faster

[umbral vector]

actively pull them that is

this was another idea i had for a solution to the circling problem

it didn't work at all, unless i messed up the code

[viral ginkgo]

you seem to be making it complicated, why dont you think it over

1- apply external forces or gravity
2-look at all the forces applied last frame, update velocity
3-remove velocity component on rope direction
4-update position from the velocity
5-translate the player in rope direction so distance between rope pivot and player doesnt change

[umbral vector]

thank you for helping me out

i think i need to take a look at all the stuff that unity can do on its own with vectors too

[viral ginkgo]

give me your code and i see if i can do it how about that

some spoonfeeding for you

[umbral vector]

there's a few different objects at play but i'll send it

[viral ginkgo]

just send me your latest piece

[umbral vector]

also my movement is totally messed up

[viral ginkgo]

send it here

doesnt matter

[umbral vector]

https://hatebin.com/itjgbdsqkw

[viral ginkgo]

thats a fatass script

[umbral vector]

[viral ginkgo]

i make you one simple

just a transform

and rope pivot

[umbral vector]

https://youtu.be/xHihYlu_E3U

this is what it looks like now

assumed lerp problems at ~0:14

[viral ginkgo]

using UnityEngine;
public class Swingalang : MonoBehaviour 
{
    public Vector3 ropePivot;
    float ropeLen;
    Vector3 velocity;
    Vector3 forcesApplied;
    void Start(){
        ropeLen = (ropePivot - transform.position).magnitude;
    }
    void FixedUpdate(){
        // add external forces
        forcesApplied += Vector3.down * 9.8f * 0.02f;
        // forces get added to velocity every frame
        velocity += forcesApplied;
        // change velocity so the component on ropedir is always zero
        Vector3 velocityOnRopeDir = Vector3.Project(velocity, (ropePivot - transform.position));
        velocity -= velocityOnRopeDir;
        
        // add velocity to position every frame         
        Vector3 newPos = transform.position + velocity;
        
        // change position so that rope length stays the same, you already did this part didn't you?
        Vector3 dir = (newPos - ropePivot).normalized;
        newPos = dir * ropeLen;
        transform.position = newPos;
        
        // these forces are already applied, we want to reset this variable
        forcesApplied = Vector3.zero;
        // air friction
        velocity *= 0.96f;
    }
}

@umbral vector

make a new scene, put this script on a empty game object

give object position -10,0,0

and hit play

[umbral vector]

computer is being odd, one minute

alas it is unmoving

[viral ginkgo]

hmm

you set object position -10 ,0,0 in inspector?

@umbral vector

[umbral vector]

i did that, then forgot to add the script

[viral ginkgo]

add the script?

[umbral vector]

it's not moving still, i'm gonna check if i might have messed up something

[viral ginkgo]

i did not test this code

i am on ubuntu now and i dont have unity

make sure fixedupdate runs

put a print in there

make sure there are no errors in unity console

[umbral vector]

it works perfectly

well shit

[viral ginkgo]

so you understand the code?

[umbral vector]

honestly i don't at all

[viral ginkgo]

lemme add some comments then

[umbral vector]

what does project do in this situation?

[viral ginkgo]

projects a vector on a direction

i used that to get the velocity components on rope direction

using UnityEngine;
public class Swingalang : MonoBehaviour 
{
    public Vector3 ropePivot;
    float ropeLen;
    Vector3 velocity;
    Vector3 forcesApplied;
    void Start(){
        ropeLen = (ropePivot - transform.position).magnitude;
    }
    void FixedUpdate(){
        // add external forces
        forcesApplied += Vector3.down * 9.8f * 0.02f;
        // forces get added to velocity every frame
        velocity += forcesApplied;
        // change velocity so the component on ropedir is always zero
        Vector3 velocityOnRopeDir = Vector3.Project(velocity, (ropePivot - transform.position));
        velocity -= velocityOnRopeDir;
        
        // add velocity to position every frame         
        Vector3 newPos = transform.position + velocity;
        
        // change position so that rope length stays the same, you already did this part didn't you?
        Vector3 dir = (newPos - ropePivot).normalized;
        newPos = dir * ropeLen;
        transform.position = newPos;
        
        // these forces are already applied, we want to reset this variable
        forcesApplied = Vector3.zero;
        // air friction
        velocity *= 0.96f;
    }
}

@umbral vector

https://youtu.be/20GkBnhQqY0?t=61

[umbral vector]

thank you for sending all this, but i still can't understand how your code "Vector3 velocityOnRopeDir = Vector3.Project(velocity, (ropePivot - transform.position))" works

it's way simpler than everything i created though

[viral ginkgo]

You understand Vector3.Project?

@umbral vector

You could do the same with dot operation

a1 = Vector3.Project(a,b)

[umbral vector]

i don't understand projection, especially in the context of this code :(

i thought i did and i was very wrong

i can tell sort of what it's doing with visuals online

[viral ginkgo]

we are just trying make sure that velocity doesn't try to extend/contract the rope

we are trying to keep the velocity on swing direction thats the purpose

you probably got confused trying to do the same thing

this is how i do it

[umbral vector]

so velocity is being projected onto the player's position, right?

[viral ginkgo]

velocity is projected on this
(ropePivot - transform.position)

which is rope direction

which is "b" in the image

velocity is "a"

[umbral vector]

i'm hitting a brick wall here, i still don't understand

wouldn't velocityOnRopeDir be 0 in this situation?

[viral ginkgo]

exactly

but we apply gravity on that force you know

if you dont do that operation im doing, your velocity wont look like this

look at how i update velocity

it does not know about rope until that part of code

velocity += gravity
this is what happens to velocity

think of this as

"rope tension"

a force in rope direction

[umbral vector]

that's exactly what i'm missing from my code

[viral ginkgo]

velocity can be is blue by default

what i do is

blue - green

so velocity will become red

.
or in other words, i figure out green
and apply its reverse as force

you could say thats rope tension force

[umbral vector]

where is the magnitude of that force coming from?

[viral ginkgo]

@umbral vector blue?

thats just velocity

it can be anything you want

[umbral vector]

the rope tension force, sorry

[viral ginkgo]

if you pull against rope

with some force

that makes tension

rope will pull back in opposite direction

physicswise

that works with forces

in physics

but i did that with velocity

[umbral vector]

i'm actively trying to figure it out, sorry if it doesn't seem like i'm making any progress

[viral ginkgo]

thats not how physics works, thats not how highschool physics questions are solved an all

[umbral vector]

i know you've given me all the information needed to understand it

i just need to think about it longer, i'm sorry for making you wait

tension is at a maximum at the bottom and top of the loop

it's 0 at the middle

[viral ginkgo]

main entry point is:
rope velocity cant have a magnitude in "green" direction
and we somehow need to eliminate it
start there

dont think about tension

[umbral vector]

i just don't understand what vector3.project is doing at all

[viral ginkgo]

i used that to find green

https://cdn.discordapp.com/attachments/497874196274348032/781286729045770270/Screenshot_2020-11-26_AutoDraw.png

to find green from blue(velocity)

[umbral vector]

i'm seeing some light, one moment

it's finding the distance gravity goes away from the pivot point

[viral ginkgo]

blue is not gravity
its what gravity did to velocity if there wasnt any rope

you are getting close i think

[umbral vector]

gravity is always going to make velocity go more downwards, isn't it?

wouldn't blue always just go down?

[viral ginkgo]

yes

it would do down yes

if there was no rope

[umbral vector]

sorry, i worded myself wrongly

[viral ginkgo]

and right now, we are discussing what the rope does on the velocity

[umbral vector]

blue is a constant downwards direction, right?

[viral ginkgo]

no

[umbral vector]

if you want to go, i've kept you here for an hour and a half already

you've given me everything i'd need to figure this problem out on my own

[viral ginkgo]

another example

[umbral vector]

i just don't want to keep you here, i really thank you for all the help you've given me

[viral ginkgo]

allright i'll go after i explain this

1- lets say blue was your velocity right, no rope or anything

2- and then you shoot grapple and there is rope

3- at that moment, green is calculated and subtracted from velocity(blue)
4- and your new velocity is red

[umbral vector]

blue is velocity directly after gravity is applied to it then?

[viral ginkgo]

assume there is no gravity for this example

[umbral vector]

i can't thank you enough for the help, i'm going to review everything you've said until i understand what's going on

[viral ginkgo]

@umbral vector if you solved some of these questions back in the day, you should find it familiar

[umbral vector]

it took me an hour and a shower but i get what's going on here

i stuck myself in a thought box and couldn't get out of it

thanks again, rewriting the grappling hook!

[fossil merlin]

hey i have multiple cars with a mesh collider in the same spot. and i dont want them to fly around because they collide with each other. How can i make them ignore each other? i added a check in my OnCollisionEnter function but this does not prevent it

the meshcollider is attached to one of the car's childs

and this happend right now

[outer ocean]

@fossil merlin set them to a same specific layer and then set in Project Settings-Physics that the 2 layers dont collide

[fossil merlin]

solved it ty

[limpid quartz]

Hey peeps. Does anyone know why my rigidbody is clipping through changes in a surface's geometry sometimes?

I've set it to speculative continuous, as that's what I'd expect from it. It's non-kinematic. I have three colliders attached to my player. I affect it in code using rb.velocity = velocity. However, I also tried using rb.AddForce(), but that hasn't helped the issue at all. I used to use a character controller, and no clipping occurred, but now it seems to be fairly common

The documentation also highlights the problem of clipping on gaps

https://docs.unity3d.com/Manual/ContinuousCollisionDetection.html

but then mentions no solution whatsoever

[limpid quartz]

I hate to badmouth the Unity team, but are rigidbodies just totally broken? Everywhere I look online it's nothing but complaints and bug reports

[silver swan]

So, what's the deal with wheel colliders and its problem about not moving in a straight line? "Car" keeps turning left or right (depending on speed and settings) on its own. I've found a lot of pages on unity forum about that but no correct solution, most of the posts were made 5+ years ago and said that it is just a unity version bug. But it is still a thing in 2020 unity! Any ideas?

[tender gulch]

Not enough info on your problem...

[viral ginkgo]

@limpid quartz changing collider meshes and rigidbody interaction is probably not something physics engine is expecting

you could always do a raycast from character, downwards direction
see where it hits, reposition the character
and make sure this happens after the terrain/mesh edit

[barren bough]

i have just a cylinder with multiple bones, how can i make it soft? like a rope, or a snake

with joints and rigidbody?

[silver swan]

When using the wheel colliders and moving forwards, it starts to steer left or right by itself when accelerating

[tender gulch]

Still not enough. What's your setup in the scene? The hierarchy of the objects involving the wheel colliders? Wheel colliders inspector values? The vehicle GO and it's inspector? Your code?

[dusty verge]

is it possible to simulate acceleration on a character controller I'm trying to figure out how to make a smooth transition from a quick push off of a wall from my character to slowly decelerating from air friction but I can only find examples with rigid bodies.

[limpid quartz]

Would it not be fairly normal to assume that most games would at some point have two bounding boxes attached together though?

Such as tiles?

[viral ginkgo]

You want tile collision?

[limpid quartz]

Or a seam, such as a door in a wall? It seems like this is an incredibly basic piece of functionality that's present in almost all video games

Tile collision or being able to handle a new surface along an object if possible 🙂

[viral ginkgo]

you want a door, and thats about it then

when you close the door, you are making a new tile collider there

and that glitches the character if a character was in the doorway

@limpid quartz am i understanding correctly?

if thats the case, you can either:
-prevent door from closing when the tile is occupied with a character
-or use a very thin collider for the door so it very easily pushes out the character

[limpid quartz]

This is essentially what's happening

[viral ginkgo]

asking for number 1:
could the ball's grounded code be detecting two objects and add two times more force?

if you did a raycastall in ball's movement code you might've done that mistake

[limpid quartz]

Yeah, that's the thing. It's fairly complex movement code that I have at the moment

It's a FPS game with momentum, wall jumping, sliding, etc

The player uses a rigidbody, 3 separate colliders for torso, feet and upper body, and I also fire raycasts out in 4 horizontal directions and also downwards

[viral ginkgo]

.
for number 2, i dont see a very nice way to handle that
you could look at velocity, cast a ray from that direction, reposition the object depending on where it hits, as i said before

[limpid quartz]

I already do that, and it seems to change direction most of the time

[viral ginkgo]

it would only change position

[limpid quartz]

but every now and again I'll just fall through the floor and out of the scene 😭

[viral ginkgo]

you got any transform position sets going on other than this?

on rigidbodies i mean

[limpid quartz]

Nope, I only use rigidbody.velocity = velocity, with velocity being set before this based on some raycasts, states, button presses, etc

I assumed it was because of the high speed and was a bullet-through-paper problem, but enabling speculative continuous doesn't seem to help at all 😦

The actual position is never manually set, just the velocity - and I would expect the actual rigidbody to handle collisions detection and resolution itself in the background

[viral ginkgo]

before:

rigidbody.velocity = velocity

after:

Vector3 deltaV = velocity - rigidbody.velocity;
deltaV = Vector3.ClampMagnitude(deltaV, 10);
rigidbody.velocity = rigidbody.velocity + deltaV;

You can limit how much the velocity can change in frame with this

If you that change

This is pretty much moving with force but your control is as good as velocity control

Best of both worlds

[limpid quartz]

Thanks dude! I get what you're saying

But I already clamp my velocity 😦 The player wants to be able to move quite quickly, as you use ramps and drops to gain momentum for jumps. You can also ground pound downwards onto a slope to gain movement speed

So I don't think it's even happening because velocity is too high

[viral ginkgo]

this is not clamping max velocity,
this is clamping max acceleration

i think you got the wrong idea, this is not what you think

[limpid quartz]

I must have

See, I was under the impression that as long as I never manually touch transform.position, the physics system will handle collision resolution in the background

[viral ginkgo]

yea but you are manually touching the terrain technically

[limpid quartz]

Are you saying that you think I'm accelerating into the geometry despite my velocity being capped?

[viral ginkgo]

yes

maybe for one frame

[limpid quartz]

So in the next cycle I'm adding a load to my velocity again when already touching the geometry?

[viral ginkgo]

yes

pushing against the collider nonstop

after penetrating it

[limpid quartz]

Could that override the physics system in the background?

Ahhhhhh

It's hard to debug or use example code, as the movement system I have is quite complex. I used to just use a charactercontroller, and it worked fine and never clipped. After switching to a rb, this has been a problem

[viral ginkgo]

i just think you should try to limit the change in velocity as i said,
not too sure this will fix %100 of the issues you have

[limpid quartz]

But I'd like to use a rb, as it seems to make more sense in the game's context, as it's heavily based on physics and movement

Thank you so much for your time dude!

I'll see if that's causing the problem. I'm at my wit's end trying to think of solutions at this point

[viral ginkgo]

you could also find ways to control the terrain change

so that terrain doesnt change in way that clips the character

or move away the character whenever terrain is transformed

[limpid quartz]

The terrain doesn't change in real time though

It's just static geometry

and pretty blocky geometry, too

[viral ginkgo]

looks like terrain changes when in contact with ball

[limpid quartz]

Mostly cubes, with the odd 45 degree inclines

Sorry! That was probably just my diagram not being very good 😂

That's just the player falling downwards onto a low poly slope

I'm gonna try and screen cap what's happening and get in on youtube. Some video footage might help

[viral ginkgo]

"when geometry surface changes"
then you mean the slope in there

hmm

thats a different case then

and my explainations are irrevelant

but limiting velocity change could still help

[limpid quartz]

That's also probably much more complex haha

[viral ginkgo]

is the ball too fast?

but you have dynamic collision and everything

[limpid quartz]

Yeah, this is just static geometry, but I'll obviously be changing direction as the surface normal changes along the gemotry

[viral ginkgo]

are you changing the collision mesh or not

[limpid quartz]

Exactly. And I only ever change velocity - never direction touch transform.position

Nope

[viral ginkgo]

is the Mesh.vertices changing

or Mesh is moving

[limpid quartz]

The mesh isn't changing, I just meant that the player is touching different parts of it was he falls down

[viral ginkgo]

okay got it

[limpid quartz]

The low poly slope has a few surfaces on it, and on the player touching the next section of the ramp, with a different surface normal, he can clip through

[viral ginkgo]

it should be fine, you shouldn't be having these issues

velocity control should've been fine as well

maybe velocity is too high

[limpid quartz]

It's melting my brain

[viral ginkgo]

maybe you are setting velocity too high

i think you should limit the velocity change like in that example code i told you

other than that, i dont know what might be the issue

[limpid quartz]

The player needs to move at extreme speeds though. I also limit velocity once it hits a certain threshold. It doesn't feel like I'm moving so fast that I should clip though

I'll screen record and send a video through soon my dude

[viral ginkgo]

@limpid quartz i am not talking about limiting velocity

i mean limiting acceleration

you should try the acceleration limit

i have spoken

[limpid quartz]

I SHALL TRY!

Thanks dude

I'll let you know xxx

[pastel scaffold]

how would i go about making a 2d ragdoll move with a moving platform? the platform is using a kinematic rigidbody2d and its being moved with .MovePosition. it's using a physicsmaterial2d with max friction and 0 bounciness. the ragdoll seems to stick for a bit if it was going in the same direction as the platform when it lands on it, but then slides off when the platform changes direction. i've also tried parenting the ragdoll to the platform, which didn't seem to do anything.

[viral ginkgo]

you probably want to have velocity on the platform, move it with forces as well just like you do with ragdoll
@pastel scaffold

Thats number 1

number 2 is

You have a piece of code that slows the character right?

velocity *= 0.9f?

what do you have

rigidbody friction?

what is it?

you need relative friction to the platform which the character is standing
@pastel scaffold

[slender wagon]

Heyy, so i have a little thing, i've been trying for 2 days to get the orbit for a planet around a sun. I've watch a dozen tutorials on youtube but i cant figure it out. Could someone help me ?

[viral ginkgo]

new Vector3(sin(time * angularSpeed), cos(time * angularSpeed), 0) * radius
@slender wagon

seen this before?

[slender wagon]

No i havent wow

[viral ginkgo]

you can see what it does?

circular motion

[slender wagon]

Do i have to search that on yt?

[viral ginkgo]

this is just a piece of psuedo code for a rotating* vector

[slender wagon]

So a rotation vector is a direction that needs to be have AddForce ?

[viral ginkgo]

oh

you want to do it with forces

simulate it

thats a different way

[slender wagon]

Well i'm trying to do it with gravity

[viral ginkgo]

you wanna apply gravity forces then

[slender wagon]

Like a real solar system

Yeah and im trying

[viral ginkgo]

@slender wagon but you know, the real solar system is not stable

your system will most likely eject planets

[slender wagon]

I know, but at least all planets have an orbit

[viral ginkgo]

until there are only two planets left

[slender wagon]

Mine just crash into the sun

[viral ginkgo]

@slender wagon you should give some inital velocity

so they make an orbit

[slender wagon]

I dont want to sound stupid

But what does inital mean?

[viral ginkgo]

i mean at start

[slender wagon]

Ohh yeah

I get it

in a void Start() {

[viral ginkgo]

give some velocity to planets in the begining

yea

so they have some velocity to maintain the orbit

[slender wagon]

But how do i figure out how much velocity they need?

[viral ginkgo]

you can give a random force

it will get in an orbit

just not the orbit you want

[slender wagon]

Haha i just want it to rotate the sun

[viral ginkgo]

or you use that equation
gravity = centripetal force

you know distance, you know mass1 and mass2
you know know gravity constant
you need to figure out velocity in centripetal force equation

[slender wagon]

Im looking at it on wikipedia

Uhm well i think i'll need to figure that out

But what you mean with gravity constant?

[viral ginkgo]

@slender wagon this is very similar to a common highschool physics question type

in that equation you have a constant

its that constant

[pastel scaffold]

@viral ginkgo sorry for late response. ill try using velocity, but i dont understand what you mean by code that slows the character

[viral ginkgo]

what prevents the ragdoll character from endlessly sliding after you stop pressing buttons?

@pastel scaffold

[pastel scaffold]

friction

[viral ginkgo]

of feet?

air friction?

[pastel scaffold]

physicsmaterial2d on feet

[viral ginkgo]

arent the feet dragging behind

when you walk?

[pastel scaffold]

uhhh not really ill double check

[viral ginkgo]

if the feet arent dragging behind when you walk, then you probably arent animating limbs with force and thats not a %100 active ragdoll

[pastel scaffold]

no, they dont drag, i have sort of an animation thing going on that rotates the rigidbody of legs using rigidbody2d.MoveRotation

[viral ginkgo]

@pastel scaffold then its not an active ragdoll you are doing

and animations wont react to external forces

[pastel scaffold]

an active ragdoll would be one that only uses velocity?

[viral ginkgo]

https://www.youtube.com/watch?v=phrudWsx9Gg
This is what my active ragdoll looks like
@pastel scaffold

limbs are moved with force or torque

[pastel scaffold]

interesting, ill try doing that

thanks

[viral ginkgo]

np

[slender wagon]

How do i calculate the range between two masses?

[viral ginkgo]

you can access gameobject position via transform.position

@slender wagon if i were you, i'd make a new gameobject that has references of all planets in the scene

and i'd do a double for loop on that list if you know what i mean

[slender wagon]

Ahh yeah, but how could i get the distance between the two>

[viral ginkgo]

(planets[i].transform.position - planets[j].transform.position).magnitude
@slender wagon

given that i,j are iterators
and planets[i] is a list that contains all planets

[slender wagon]

Ah okay thanks

[slender wagon]

@viral ginkgo I've kinda fixed it, the planets go around the center. But is it possible to make a projectile tractory ?

[viral ginkgo]

@slender wagon projectile?

what projectile tragectory

[slender wagon]

Well of a planet

To see what orbit a planet would make when it goes at a velocity

[viral ginkgo]

you'd need to define these in a function:
velocity, position
whatever your do with planets, you'd do the same with velocity and position

iterate many times, record position each time or every 10 times or so
you'll have your trajectory

[slender wagon]

Alright i'll first try to get the velocity and position

[viral ginkgo]

these are not the velocity, position of your object

[slender wagon]

Im just not sure about the velocity

Oh damn

[viral ginkgo]

you make your own simple physics

[slender wagon]

I feel kinda dumb haha

[viral ginkgo]

just to simulate one tragectory

or you could make a seperate physics scene and simulate the same object multiple times per frame, to render that trajectory
you'll be able to do PhysicsScene.Simulate
and advance time in that scene without breaking your main scene
thats the idea

[slender wagon]

Aha so if i get this right

Im making a new scene

Where everything is the same as the main scene

[viral ginkgo]

not a new scene, it wont do it

you need to have a seperate physics scene

if you make a new scene, it wont make a new physics scene and physics simulation will be shared

and you want be able advance time on a particular object without breaking your main scene

@slender wagon This is overkill anyway

I just explained that so you know your options

I'd just make new vectors position,velocity

do the same logic you do on your planets on them

but 100 times

[slender wagon]

This sounds really complicated haha

[viral ginkgo]

You could define your own orbit physics

And not use rigidbodies

For your current orbit system as well

[slender wagon]

Im using rigidbodies to orbit the sun

[viral ginkgo]

Its not much different then what your are doing now anyways

addforce would be velocity += myForce/mass (given that F = ma and "a" is velocity change)

[slender wagon]

F = ma ?

Whats ma

[viral ginkgo]

force = mass * acceleration

highschool topics

[slender wagon]

Not really

Well it sounds all complicated so i'll try to find something on google

[viral ginkgo]

well you need a little physics/math in gamedev
better understand what you are doing

[slender wagon]

so if i use rb.velocity += myFoce/mass then it will be the same as rb.AddForce(x, y, z)

Cause tbh i kinda need to make the tractory, cause it wont really work without

[viral ginkgo]

using physics:

rb.AddForce(myForce0);
rb.AddForce(myForce1);
rb.AddForce(myForce2);

same thing but no physics:

Vector3 position;
Vector3 velocity;
float mass;
Vector3 myForceSum;

myForceSum += myForce0;
myForceSum += myForce1;
myForceSum += myForce2;

velocity += myForceSum/mass;
position += velocity;
myForceSum = Vector3.zero;

@slender wagon

[slender wagon]

rb.AddForce(myForce0);
rb.AddForce(myForce1);
rb.AddForce(myForce2);

Why do you have this 3 times?

[viral ginkgo]

this is an example to show how to add multiple forces

just wanna show how you'd add some force

[slender wagon]

Ah okay, imma try to make a PsyicsScene

[viral ginkgo]

physics scene is more difficult as i said, but you'll be able to predict collisions

[wide root]

Anyone know how I can truly lock a config joints axis?

Setting locked does the trick except the object still moves

like it'll spring back into place eventhough I have it on locked?

I'm trying to achieve stabbing so I need the Z axis limited while X/Y are locked and unmovable in any direction other than Z

Expected behaviour

It's configured in local space, so the Z axis is in the direction of the stabbing, all other axis are locked.

I've shown only what is relivent to the stabbing functionality, angular related things aren't required atm.

[wide root]

Figured it out

Turns out the connected mass scale was at 1, allowing basically anything to move it, adjusting the values makes it immovable in any direction other than the desired one.

[upbeat wasp]

Anyone here have experience with dynamic bone and colliding with rigidbodies and what not?

[last sundial]

Hey 👋

How do I change the distance between my camera and a target without modifying the camera angle?

I could get the camera to zoom in/out but when I do the LookAt it changes the camera angle when reaching certain distances

[tacit dawn]

Hello, is it allowed to ask questions here?
ah idk i'll just ask. I have a Problem with collision well i think i have a problem with my brain but right now i can't figure out why my collisions don't want to work.
So i got colliders and all set up and they work but it looks kinda scuffed and when i set the player speed to fast it just doesn't give a shit about my "collision". Maybe someone can help? I got a Video that shows what i mean.

this here.

[tacit dawn]

maybe it's my movement script after all?

[stuck bay]

@tacit dawn did you try setting the rigidbody's collision to continuous? Also, what are you using to move the player?

[stuck bay]

Are you talking about field of view when you say camera angle?

[stuck bay]

Hey all, I'm having a problem related to switching from Animator -> Ragdoll. Here's a link to my forum post: https://forum.unity.com/threads/animator-rigidbody-pulse-on-change.1013335/ I appreciate the support.

[last sundial]

Hey! No... By angle I mean the pitch. I have some references for doing this but I don't know how to translate the code to Unity. But to me it seems they take the player position and subtract by the camera position and move the camera in a straight line without using the LookAt

Here's the reference https://www.github.com/open-ragnarok/open-ragnarok/tree/master/roengine%2Fcamera.cc

[last sundial]

Could replicate the code by taking a closer look at the reference.. Thanks anyways!

[tacit dawn]

@stuck bay Thats what i made for PlayerControls. i'll post a picture about my settings for the Player collider and stuff

{
    public float playerBaseSpeed = 5.0f;
    private float playerSpeed;

    // Start is called before the first frame update
    void Start()
    {
        
    }

    // Update is called once per frame
    void Update()
    {
        // playerSpeed manipulation starts here
        playerSpeed = playerBaseSpeed;
        // TODO: make this settable in a menu
        bool inputSprint = Input.GetKey(KeyCode.LeftShift);
        bool inputSneak = Input.GetKey(KeyCode.LeftControl);

        playerSpeed = inputSprint ? playerSpeed + 3.0f : playerSpeed;
        playerSpeed = inputSneak ? playerSpeed - 3.0f : playerSpeed;

        // movement starts here
        float inputX = Input.GetAxis("Horizontal");
        float inputY = Input.GetAxis("Vertical");

        Vector3 move = new Vector3(inputX, inputY, 0);
        transform.position += move * playerSpeed * Time.deltaTime;
    }
}```

@stuck bay this are my settings, so i think the Problem here is that it still can move into that direction after hitting a wall it just hits the wall again until it breaks through the wall.

[vocal phoenix]

I have currently a problem, that my golf ball is sometimes bouncing of flat platforms (paths). Rigidbody settings are also be seen in the video
(Fixed by changing "default contact offset" to 0.00001)

[silver swan]

Here again with a wheel collider problem. When I'm adding torque to the vehicle, it starts moving sideways for some reason and I can't find a fix for it. Can anyone help?

[gilded sparrow]

I have a super weird problem..

My homing projectile uses rb.MovePosition (also tried rigidbody.position)
It "homes" fine except when the target moves, in which it just stops

Doing Debug.Log(rigidbody.IsSleeping() + " : " + tr.position + " - " + newPos + " mag: " + dir.magnitude);

Ohmygod figured it out
Was doing this:

transform.LookAt(lastDestination);

Apparently rb.position or Moveposition doesnt like it if transform is modified right after. Flipping this around fixes it
Wow ok

Actually there's another problem
The above case is fixed, but if the target's moving when the projectile is spawned, then it'll also never start moving
Maybe im not getting something here..

Ok i gotta also do rigidbody.rotation..

[onyx cypress]

Hey everyone :)
Working on a VR game and was wondering if anyone knew how to make the player rig be physics based, similar to Boneworks. I currently have it setup for physics based hands, but I am able to walk through things, and things like climbing wouldn't work at all. I would assume some form of inverse IK would be required for things like climbing, to make sure to pull the main body along instead of the hands going to never land alone. For the body, and mainly the camera, I have no idea how it would work.

So how would one go about rigging up a player in order to allow physics based interaction with the world vs simply phasing through everything?

[quartz imp]

I've been using Unity for education and something is messing with the physics. In the program when the player hits an object the object is supposed to go flying, however it is not registering the collision on the first hit and thus requires multiple before it goes flying. I am unsure if this is better here or in #archived-code-general

[tender gulch]

Share your code.

[sweet flicker]

I am updating my rigidbody velocity in every FixedUpdate like this:

if (rigidbody.velocity == Vector3.zero && newVelocity == Vector3.zero) { return; } rigidbody.velocity = newVelocity;

The if statement is to prevent unnecessary 0 assignments when it's already on 0.

This still causes the rigidbody to have increasing velocity which is causing the transform to move.

There is no other code that does something with the rigidbody. When I disable this code, the rigidbody is not moving at all.
Anybody know why and how I can solve this?

[supple sparrow]

I usually don't mess with directly with velocity to avoid strange results. I prefer to apply forces.

Does newVelocity sometimes get reset ?

otherwise you're always moving yeah

Also you cant' be sure about zero equality because floating point precision, but that's another topic

[sweet flicker]

newVelocity is by default Vector3.Zero untill I receive a MQTT message. Im not receiving any messages at the moment, which means that newVelocity is still on zero and not being reset.

I thought it had to do something with the floating point precision

That's why I changed to my own compare function which works like this:

But it's still moving. Really strange

[supple sparrow]

ok your compare function is ok
Back to the topic so you expect no velocity at all, though you see your rb moving

It's a on a flat ground ? which shape ?

[sweet flicker]

it's on a plane currently

rotations are 0

I think i've just found the problem

Thanks to your question 🙂

The mesh collider convex didnt work as planned

[supple sparrow]

isn't it just falling because gravity maybe ?

oh ok good

so not a flat ground, eh ? 🙂

[sweet flicker]

It's making it unbalanced, which causes the physics engine to add velocity

It is flat ground

Just the object moving on it is not flat haha

[supple sparrow]

oh ok, great you figured it out

[sweet flicker]

Thank you for your help!

Just replaced the mesh collider by a box collider. Problem seems to exist. It's standing still for a few seconds and after that it starts adding a velocity of 2.

Seems like the world center of mass drops down from 6.1 to 5.4 and then it al goes wrong.

The velocities seem to change from -6 to +6 every time. Like it's teleporting?

[supple sparrow]

can you show screenshot of the scene ?

[sweet flicker]

Physically, he's not moving anymore since the mesh collider got replaced by a box collider. The rigidbody velocity is just acting really strange.

[supple sparrow]

Honestly I dont even know if that's to be expected or not, I never look at that. I guess you'll see if that's getting in the way when you try moving ?

[sweet flicker]

Okay this time I really solved it

Yeah it was getting in the way. Every velocity update of X and bigger im sending a message back

The problem was the FixedUpdate.

Changing it to Update made it behave correctly

[supple sparrow]

Uhm... I'd rather mess with Physics in fixedUpdate

I dont see how you compute newVelocity, but let's say you replace it with rb.velocity = new Vector3(0, 0, 1); does the movement still have jittering effects ?

[sweet flicker]

newVelocity is a global variable in the script. With a method like setVelocity (Vector3 velocity) I set this variable. This function is only called through my MQTT Adapter script. So simply: Receiving a message over MQTT containing new velocity value. call the setVelocity method with the new velocity value parameter. Set rigidbody's velocity to this new value

Just did what you said @supple sparrow and it results in the same jitter effect, just a little bit less because it's also moving now

[supple sparrow]

Man, I dont know. Is there a Physics Material on the gameobject ? Is the gameobject slightly above ground when spawned ?

[sweet flicker]

I have it slightly above ground. There is no physics material on the gameobject only on the ground

with bounce 0 and both frictions 0.6 (default)

[north isle]

anyonne know recommended way to make sphere planet world

instead of flat terrain

[supple sparrow]

@sweet flicker if you ignore teh values going all over the place, does it behaves like expected or movement is messy ?

[supple sparrow]

Lots and lots of tutorials on the net, youtube, ...

pick one that looks cool to you

[sweet flicker]

@supple sparrow It is moving like expected but with a little jittering. It moves slightly up in decimals on the Z Axis. And even less down on the X axis. Thats my only worry atm. Even when it's idle it has this jitter.

[supple sparrow]

can you easily make a gif or video about the jittering when idle ?

[sweet flicker]

Im uploading it right now

[supple sparrow]

alright, and does it help if you set collision detection to continuous ?

[sweet flicker]

As you can see, it's falling down really slowly. After hitting the ground. It starts adding velocity and changing the z axis

[supple sparrow]

yeah I dont see a problem here

order of magnitude is -e5, it's tiny tiny

[sweet flicker]

Changing to continious does not change it

Because it's so tiny it moves or?

[supple sparrow]

actually in the scene your object isn't really moving.

The simulation goes crazy but it's not troublesome ?

[sweet flicker]

It's moving the Z axis a little bit up in decimals, but you can't really see it with the human eye. It's a simulation of a real robot, overtime this minimal change in the Z axis can cause a difference that's why I wanted to get rid of it. But it seems that we cant?

[supple sparrow]

I guess velocity applied is not enough to make it move but enough to update values

I'd say you'll see if problem persist when you begin moving

[sweet flicker]

Yeah again just minimal, I will leave this unsolved and will just accept this

Thank you for your time!

[supple sparrow]

No problem, happy coding

[fleet iron]

is there a way to make fixed joint actually fixed? i tried searching everywhere and none of them works for me

[viral ginkgo]

@fleet iron what are you making with fixed joint?

If you have a rigidbody in parent
and have multiple childs with colliders attached(no rb in childs), that will be a single rigidbody, including all those colliders

[fleet iron]

but then if i delete a object the other objects in it will get deleted

and its hard to make that placeable

[soft sparrow]

How do you apply a torque to an object so it aligns with a vector?

I'm currently doing ship.GetComponent<Rigidbody>().AddTorque(Quaternion.FromToRotation(ship.transform.forward, dir).eulerAngles);

where dir is the given vector

[viral ginkgo]

@fleet iron

.
hierarchy:

g.o with rigidbody
g.o with collider
g.o with collider
g.o with collider
g.o with collider
g.o with collider

.
you can delete any child

.
@soft sparrow

Vector3 target = my target normalized direction vector
Vector3 current = ship.transform.forward;

Vector3 torque = Vector3.Cross(target, current);

// i may have mistaken the sign, one of these two should do it
rb.AddTorque(torque)
rb.AddTorque(-torque)

rb.angularVelocity *= 0.9f;

[soft sparrow]

I'll try it out thanks :)

[viral ginkgo]

How does a physics engine know if two bodies overlap?
I assume bounding box check is done first, in the octree grids that contain the two objects.
After engine knows these two objects bounding boxes overlap, what happens next?

I wanna know how a simple cube collides with a complex, concave terrain mesh

[foggy rapids]

it does "solver iterations"
it goes over every potential collision and if it finds one, it calculates the new transform.
It does this for as many times as you specify in the physics settings or until no collision is detected

depending on the type of objects the detection can be really easy or really tricky and there are loads of different ways to do it

circle is easiest, you only need a radius.
box is second easiest
polygon/mesh is a lot more difficult

if at all possible you should reduce your collision geometry to circles and boxes 👍

https://en.wikipedia.org/wiki/Gilbert–Johnson–Keerthi_distance_algorithm
here's one way

[viral ginkgo]

Hmm i see, i plan on using smooth voxels, i wonder if there are volumetric collision checking methods, or people just convert to convex mesh pieces and do these types of things

[drifting quartz]

Hopefully I'm in the right channel for this, sorry if it's incorrect! I'm trying to find a reliable way to have my character slide down a slope if it collides with it mid air. So far, I have it working properly but only if I'm facing the object/terrain nearly perfectly with transform.forward. If I jump onto the slope so I'm perpendicular to it, I just continue sliding forward. How can I get the the vector that I'm using for my controller.Move to be the in the opposite direction of the angled surface that I'm colliding with?

[toxic kite]

is there any benchmarked or anecdotal best practices regarding to which degree a mesh collider should be split up into sub sections? In my own experience on a slightly older project (also mobile), I noticed some benefits to splitting up some big ones, but I never really exactly tried to find the line.

[soft sparrow]

@viral ginkgo About the earlier code you posted, it seems to oscillate a lot, how would I go about counteracting the oscillation?

Nevermind, I'm stupid and forgot to add the last line

[supple sparrow]

Not sure I figure your setup. Can't you just minus the vector ?

Or go down - Vector.up and let your collision glide you along the slope ?

Or are you looking for the tangent ?

I would compute the forward vector with the normal of the collision point (using the dot product) to find the facing direction

and projection on the surface

but depends on your implementation and if you're using custom or Unity components

@drifting quartz It's poorly explained, tell me if you need further help

[stuck bay]

Whats the difference between lerping and simply moving based on Time.deltaTime?

[foggy rapids]

lerp = linear interpolation = inferring data between two known points.
i have no idea what you mean by simply moving based on Time.deltaTime

[stuck bay]

I feel like if you move an object at a certain speed and multiply it by Time.deltaTime you would get a linear movement as well

[foggy rapids]

it would only be as linear as your framerate

[stuck bay]

Wouldnt linear interpolation be the same way?

Because it only calculates a new value every frame?

[foggy rapids]

for lerping you need two points. Moving by lerp is choosing a new point at a fixed rate and using Update to move from the previous point to this point.

you can achieve the same effect with both, but you start to lean more towards lerp the more physics the rest of your game uses (because physics runs at a fixed rate)

[stuck bay]

So in general, the fixedupdate loop should be used for physics CALCULATIONS but not the actual movements? FixedUpdate should set the target and update should lerp towards the target?

[foggy rapids]

yeah. fixed update = make the decisions
update = the stuff in between

also collect input in update

[stuck bay]

Ah - my confusion was I saw a lot of people putting the .Move(delta) in the FixedUpdate

which I guess is inproper?

[foggy rapids]

the whole charactercontroller makes things confusing

[stuck bay]

A lot less confusing than making a character controller out of a rigidbody though

[foggy rapids]

not if you understand the update loops. imo charactercontroller is just a way of getting it done without needing to understand

[stuck bay]

Well my understanding is the character controller's move function basically handles movement, collisions, and slidealongwall

[foggy rapids]

character control is not a trivial thing and deserves a lot of attention in games.

[stuck bay]

I know - I'm coming from Unreal Engine where I was studying the character movement component that is like 30k lines of code

But I had a lot of issues with UE4's debugging process

[foggy rapids]

using delta time in your calculation is a way to decouple it from the framerate but the easy mistake to make there is not multiplying it by another constant as well.

[stuck bay]

My movement code works pretty well - I'm mostly having trouble getting a smooth camera. Wasn't using lerping but it still seemed fairly smooth. Will have trouble when I add impulses in the future though
https://textuploader.com/1enak

[foggy rapids]

if your camera needs to follow something, make it a child of that object. if it moves smoothly so does the cam.
if you also need to reposition the camera then move it's local position at the same time as it's parent's world position
looks like you covered all the bases in that code 👍

[stuck bay]

It's a first person camera but for whatever reason when I use the mouse as the input it is really stuttery but if I use up/down/left/right on the keyboard or a controller it's really smooth. Something to do with the mouse taking in deltapixels vs the keyboard being 0 or 1 i guess

[supple sparrow]

Also check cinemachine cameras, they can do a lot for you

[stuck bay]

This is the stuttering I'm getting
https://streamable.com/21s6pl

Starting to wonder if maybe that's just how mouse movements are (not smooth) and games just smooth it out artificially for you?

[supple sparrow]

Lerp translations and Slerp rotations for smoothy-smoothy

[stuck bay]

One other thing I don't understand about the FixedUpdate - if I multiply my acceleration by deltaTime it shouldn't matter whether deltaTime is fixed or not. I should be moving at a constant speed regardless of my framerate. Are we sure fixedupdate isn't more for performance reasons than this whole "fixed rate" thing?

I know UE4 does not have a FixedUpdate loop - just the regular update loop which may be because it is designed to be run on PC/Console more than Unity

also from my understanding, fixedupdate does not actually have a fixed real deltatime - that's impossible

[supple sparrow]

fixedDeltaTime will return the frame rate of the physics simulation

[stuck bay]

Right, which from what I understand will not always be .02, it will be as close to .02 as it can get

(assuming you're using .02 as default)

[supple sparrow]

yeah I guess the engine will at least try its best

Update() depends on your hardware max capabilities

[stuck bay]

So, other than performance reasons, what exactly is the benefit of using FixedUpdate(). If your code is written correctly, you'll adjust for time.deltaTime regardless of which code block its in

Or is it just that Unity doens't run its physics calculations on Update() strictly for performance reasons which would make sense

but why then does the documentation talk about how its frame rate independent which doesn't seem to be true

[supple sparrow]

If you work with a rigidbody, you should update relevant code in FixedUpdate

[stuck bay]

I'm using charactercontroller

[supple sparrow]

If you move yourself the transform, go ahead in Update()

but still do it as a factor of Time.deltaTime

oh you mean character controller from Unity ?

Don't even remember if it uses rigidbody or not :p

[stuck bay]

Right - it's just contrary to what I'm seeing on forums everywhere which might be because everyone seems to use rigidbody's for their character controller which si weird

From what I understand the character controller has its own implementation of collision detection - it doesn't really simulate physics it just has a function that checks for collision when you try to move and responds accordingly

Gravity and momentum for instance are not implemented which makes sense for a character controller since you dont want your player to bounce off objects (unless you do, but then you probably want to code it yourself so you have more control)

[supple sparrow]

note that you can have both, let's say compute movement vector from input/gamepad in Update(), but applying the vector on rigidbody only when FixedUpdate()

Every controller ends up different

[stuck bay]

Yea - I'm just of the mind that you should not use rigidbody at all for your player character. Maybe in some games but if you were making like a 3D FPS game or something it doesn't seem to be the way to go

[supple sparrow]

yeah sure in this case a simple capsule is enough 🙂

[stuck bay]

As I said, I do come from UE4 and the whole rigidbody based character controller is really strange to me

[supple sparrow]

Maybe people just keep the first think they found and tried and that worked

How do Agents react to physics in UE4 ?

[stuck bay]

Well for non-character objects its mostly the same. But almost no one enables physics on their character controllers, they use capsule based movement controls and code the physics

Its not that Unity and UE4 are capable of doing different things, it's that the community in UE4 would consider rigidbody controllers to be really weird and people here at Unity seem to think capsule controllers are weird

[supple sparrow]

The whole rigidbody thing was weird for me too at the beginning

[stuck bay]

It is definitely not the most common among AAA titles from what I see.

Things actually got a lot smoother after I moved everything to Update() and removed all my fixedupdate code. I'm thinking you just shouldn't use fixedupdate if you're making character controllers for pc/console - if you really need to slow it down it's probably better to write your own code to make your movements trigger every fraction of a second rather than change the default .02 value for physics updates.

I'm still having a problem with the mouse giving me clunky movement but this only occurs with a mouse and not the controller which makes me think there's something weird going on with the inputs

[supple sparrow]

Well FixedUpdate is still useful. Gives you constant interval between simulation steps. So for everything related to the physics it's where you want to be.

for your mouse movement feel free to share a glimpse of code, people here will help if htey can

[stuck bay]

I think the reason we use FixedUpdate when we're calculating movements for rigidbodies is because the rigidbody physics code all executes inside fixedupdate. In the context of character controller (no rigidbody), there is no physics calculations being performed in either the fixedupdate or update. So you don't need to worry about your code being in sync with the physics calculation code

CameraPhysics.cs

    {
        if (m_FirstPerson)
            {
                Vector3 euler = m_Camera.transform.localRotation.eulerAngles;
                euler.x -= m_InputTheta * m_1stCameraSpeed * Time.deltaTime;
                // If our rotation is above '0'/'360' (same thing), decrease by 360f to make negative version
                if (euler.x > 180f)
                    euler.x = Mathf.Clamp(euler.x -= 360f, m_1stMinThetaAngle, m_1stMaxThetaAngle);
                else
                    euler.x = Mathf.Clamp(euler.x, m_1stMinThetaAngle, m_1stMaxThetaAngle);
                euler.y += m_InputPhi * m_1stCameraSpeed * Time.deltaTime;

                m_Camera.transform.localRotation = Quaternion.Euler(euler);
            }
            else // if third person...
            {

            }
        ClearInputs();```

InputFunctions.cs

    {
        m_CameraPhysics.RotateCamera(m_CameraInput.y, m_CameraInput.x);```

```    void OnLook(InputValue value)
    {
        float phi = value.Get<Vector2>().x;
        float theta = value.Get<Vector2>().y;

        m_CameraInput = new Vector2(phi, theta);
    }```

[supple sparrow]

Alright another Unity trick

If your camera follows your player, you can put it in LateUpdate(), so you can be sure it's executed after player movement

[stuck bay]

That makes sense

[supple sparrow]

This can be the source of jittering, or maybe not 🙂

[stuck bay]

Didnt fix the issue on my cameraphysics, thinking about whether it would actually make any difference if I did that in my inputfunctions as well (I wouldnt think so)

What I find really bizzare is this is only happening with my mouse. It does not happen with a controller of if I bind it to up/down/left/right on the keyboard

[supple sparrow]

how do you read your input theta and phi

[stuck bay]

void OnLook(InputValue value)
{
float phi = value.Get<Vector2>().x;
float theta = value.Get<Vector2>().y;

    m_CameraInput = new Vector2(phi, theta);
}

this should be broadcast when it detects new values for Mouse<Delta>. I am unsure of whether it only checks for new inputs once a frame but I would think so

It uses the input system package

forgot this code block
Cameraphysics.cs

    {
        m_InputTheta += theta;
        m_InputPhi += phi;
    }

[supple sparrow]

Oh it's the new InputSystem package ? Didnt use it yet. I'm still on the Input.GetAxis way

Maybe someone else here will be able to answer, or try #🖱️┃input-system

[stuck bay]

Yea I was there first but then I thought I may have had a physics problem so I came here but now I'm thinking it is actually an input-system problem

[supple sparrow]

Yup

[dim kiln]

Hey Guys so I have a ragdoll script which turns my player into a ragdoll when I press U, but when the player is moving and I press U he just stops and falls down instead of tripping is there, does anyone know how I can fix this?

https://hatebin.com/dvdnsgvdrg

[viral ginkgo]

@dim kiln you probably wanna set the rigidbodies of ragdoll limbs

this code has nothing on ragdoll limbs

[drifting quartz]

@supple sparrow Thank you! Your reply made sense, I ended up doing what you had suggested with calculating the angel of the impacting surface and then creating a new heading based off of that and I have it working now. 😄

[supple sparrow]

Great :)

[jaunty barn]

i dont know if its right doing right here

[lime galleon]

Hello guys, i have some problem with trajectory prediction.
Well, i need to predict/calculate perfect velocity for ball to hit exact on target.
So math is pretty basic, trivial physics equations. So It works perfect with Fixed timestep tending to zero, but with 0.02 there is very noticeable imprecision.

Well, although equations are good, and trajectory renders excellent, when i play simulations the actual trajectory differs a little, but with longshot, this imprecisions caused by fixed timestep stacks giving in result projectile missing target.

My question is, do you have any idea, how to take the physics timestep into account in my equations?

I found similar post in google with screenshot that show my problem so i am going to copy paste it here.

[ruby frigate]

Small Unity physics riddle...at least for me:

Why does this do what I expect (rotate the player around the Y-axis):
player.Rigidbody.AddTorque(player.transform.up * speed)

But this does not rotate the player around his forward axis, but instead just
moves him to the left or right:
player.Rigidbody.AddTorque(player.transform.forward * speed)

[ruby frigate]

Also tried it with AddRelativeToque but same effect.

[sterile lion]

Hey guys, I have some issues with some box rigid bodies stacked like bricks when the scene starts. They push eachother at the start and become very unstable. I already tried tweaking the physics material which didn't help. Now I'm trying to get them to sleep when they instantiate but it seems like forces are being applied anyway. Does anyone know what I'm doing wrong?

I'm assuming it should not be done in Start?

[supple sparrow]

Did you try increasing simulation steps in the settings ?

Also, do you compute the equation on Update() or FixedUpdate() ?

[lime galleon]

@supple sparrow Hey, thanks for interest. Do u mean Maximum Allowed Timestep?
Its physics so FixedUpdate

[supple sparrow]

Ok, I would try to do the calculations on Update() though, and use FixedUpdate() only to apply the results

Edit > Project Settings to mess with interval

[lime galleon]

The thing is, I dont kinda understand what do u mean by equations compute.
I use some basic mah to calculate initial velocity, and then i just set rigidbody.velocity and it differs from expectations

[supple sparrow]

how to take the physics timestep into account in my equations?
You mean Time.fixedDeltaTime ?

[lime galleon]

and when fixedTimestep is like 0.0001

It matches expected trajectory, but when its standard 0.02 it fidders like on picture

differs*

[supple sparrow]

I mean I would do the math in Update() and apply velocity to rigidbody in FixedUpdate()

[lime galleon]

thats what i am doing

[supple sparrow]

Alright then go to Edit > Project Settings, you can reduce time steps interval

you might get a performance hit though

[lime galleon]

Yea. thats the thing. and this is mobile game + trajectory simulation iterations would fly to the sky

so i must stick to 0.02 but take it to account somewhere, but dunno how

[supple sparrow]

Time.fixedDeltaTime ?

[lime galleon]

yup. this must stay on 0.02

[supple sparrow]

can you show equation for the parabola ?

[lime galleon]

sure

this calculates launch velocity

and parabola points are gatheres from PhysicsScene simulation

here

[supple sparrow]

So when you hardcode physics.Simulate( withAVeryLittleNumber ) it works as expected ?

[lime galleon]

Yea Exactly

[supple sparrow]

But gives poor performance ?

[lime galleon]

Also true

[supple sparrow]

Maybe Physics.Simulate is not suited for mobile ? Real question here, I don't know the answer

[lime galleon]

hmm, suited or not (well it works fine on Redmi 8) in editor problem is the same

So i maybe better question would be :
How to bypass imprecision of FixedTimeStep grater than InfiniteSmallNumber

Becouse in real life it would probably be infinite small number or 0 and equations relates to reality 🙂

[supple sparrow]

Do you sync the simulation with the framerate like the official example ?
https://docs.unity3d.com/ScriptReference/Physics.Simulate.html

[lime galleon]

No but thats not the point, as whole simulation gets done within one framerate

(or on call, yea there is coroutine to optimize it a bit and get drawing effect, but still its not about replacing physics, but simulating it on call)

[supple sparrow]

Then I have no idea. Would have to spend the time to try it myself.

[stuck bay]

hello

if y= velocity and x= time
i want to do it y = 2x
but also want to change the curve abit,how do i do that ?