For now, that's how I move the player:

rb.velocity = Vector2.right * input.x;

And I affect him a force while he is damaged by an entity:

rb.AddForce(Vector2.right * force);

It works fine, until I try to move while I'm damaged by an entity. Since my velocity = input.x while I move, I can't affect any forces to it. I've tried to summarize velocity:

rb.velocity+=Vector2.right * input.x;

But then I need to clamp the speed and if I clamp it, my force which is affected to player is also clamped.

How can I resolve this problem?


3 Answers 3


You're on the right track! This is something that usually the controller is responsible for. When you jump in a platformer you use a "isGrounded" variable to change how the controls behave while in the air right? You need a similar state for isKnockback. In most games when a player is knocked back, they have control striped from them for a certain amount of time. That or you treat it like being airborn where instead of setting velocity it will just apply a force to adjust the velocity. How you want to implement that control is a design choice, but there is no real physics based answer. The true physics based answer is implementing locomotion in the character's legs and that's not really useful for game physics. Another solution could be instead of applying a force, is to work with knockback at a velocity level like your movement. Apply a velocity, and reduce it by some amount each fixed update to simulate a knockback like effect. This is another common solution in the industry.

Edit: Just to credit the other answers here, they mention making your movement acceleration / force based. This is also an option, depending on the kind of motion you want. This approach is far more intuitive but gives you a lesser degree of control. It all depends on what you want, and it's important you make that decision early (or experiment if you're not sure) because it will impact decisions further down the line. If you want to see how crazy some platforming code can get, Matt released the code for Madeline from Celeste. They use a different engine (framework actually) but you can get the jist of how many variables and pseudo physics were used to achieve that feel. https://github.com/NoelFB/Celeste/blob/master/Source/Player/Player.cs

  • \$\begingroup\$ Thank you, it's done by check true isKnockback when force is applied and check false when player will collide with ground next time, it works very fine \$\endgroup\$ Apr 8, 2019 at 15:35
  • \$\begingroup\$ Glad to hear it. In general don't be afraid to fake physics in your games. Lots of engines have these awesome realistic physics, but in many game scenarios people just end up faking most of it :) \$\endgroup\$
    – gjh33
    Apr 8, 2019 at 15:44

I created a little demo game a few days ago which demonstrates different ways to move a player-character. It might help you to better understand which way of moving is the right one for your particular game.

In general, you should use rigidbody.AddForce whenever feasible. It automatically takes care of managing multiple overlapping forces and ensures that the transfer of momentum on collisions is physically correct.

If you don't want your character to be able to accelerate indefinitely, increase the "Linear Drag" value of the rigidbody. The drag force increases quadratically with the velocity, so at some point it will cancel out the acceleration and effectively limit the maximum speed. A larger drag value will result in shorter acceleration and deceleration times, making the controls feel more "tight", but also greatly limit the effect of collisions.

If you want the character to have tight controls but also be affected by collisions, you could handle this the way the answer by gjh33 suggests. Have two different states in your player-controller. A regular state where the rigidbody has high drag and the player has full control force, and a "just got hit" state where you reduce the drag and the control force temporarily in order to make the character fly around in a temporary state of uncontrolled helplessness.

I am looking forward to playing your game.


Myself, I like to solve this by thinking of all player movement as acceleration-based.

I choose a target velocity using whatever complicated control logic I like, then ask the player avatar to accelerate toward that target, while respecting maximum acceleration rates I set.

Then, depending on the avatar's state (on ground, on ice, in the air, in a knockback state), I can change those acceleration rates to make the control input have a sharper or less pronounced impact.

Rigidbody2D body;

void AccelerateTowards(Vector2 targetVelocity, float maxAccel, float maxDecel) {
    // Compute desired velocity change.
    var velocity = body.velocity;
    var deltaV = targetVelocity - velocity;

    // In 3D, you might want to apply movement only horizontally:
    // deltaV.y = 0;

    // Convert our velocity change to a desired acceleration,
    // aiming to complete the change in a single time step.

    // (For best consistency, call this in FixedUpdate,
    //  and deltaTime will automatically give fixedDeltaTime)
    var accel = deltaV / Time.deltaTime;

    // Choose an acceleration limit depending on whether we're
    // accelerating further in a similar direction, or braking.
    var limit = Vector2.Dot(deltaV, velocity) > 0f ? maxAccel : maxDecel;

    // Enforce our acceleration limit, so we never exceed it.    
    var force = body.mass * Vector2.ClampMagnitude(accel, limit);

    // Apply the computed force to our body.
    body.AddForce(force, ForceMode2D.Force);

Separating acceleration & deceleration rates like this lets me give a sharper braking force, which tends to help the controls feel tight & responsive when stopping or changing directions, while keeping a smooth acceleration for getting up to speed. It also lets you penalize braking specifically when on ice or being knocked back.

We can make a control state parameters object to hold the max speed, acceleration, and deceleration rates for our air, ground, ice, knockback states etc. and use this to adjust our control handling very flexibly:

// Choose our current speed / acceleration parameters based on our state.
var controlState = GetCurrentControlState();

// Compute desired velocity.
var velocity = GetDesiredVelocityFromInput(controlState.maxSpeed);

// Try our best to reach that velocity, with our current parameters.
  • \$\begingroup\$ does deltatime automatically give fixedDeltaTime in fixedUpdate loop? That's really cool! I know I used to have to code around that, do you know when they added that? \$\endgroup\$
    – gjh33
    Apr 8, 2019 at 17:42
  • \$\begingroup\$ I looked into it you're right. That's very useful. I don't remember that being a thing when I started working in unity, definitely makes reusing functions in different update loops way easier. \$\endgroup\$
    – gjh33
    Apr 8, 2019 at 17:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .