Multiplying the velocity by delta in move_and_slide

YoshikaMiyafuji

The documentation says that move_and_slide multiplies the velocity by delta, so you shouldn't do it; however, this tutorial does otherwise https://docs.godotengine.org/en/stable/tutorials/physics/kinematic_character_2d.html?highlight=delta#moving-the-kinematic-character

Check the very last bit of code shown, just under _physics_process it has the gravity as velocity.y = delta * GRAVITY and at the end of the same function it has move_and_slide.

If move_and_slide already multiplies by delta why does that code multiply gravity by delta? Is there a difference because that isn't velocity gained by player input or some other reason? How is it different?

DaveTheCoder

The source code does appear to multiple the input velocity by delta:

Vector2 KinematicBody2D::move_and_slide(const Vector2 &p_linear_velocity, const Vector2 &p_up_direction, bool p_stop_on_slope, int p_max_slides, real_t p_floor_max_angle, bool p_infinite_inertia) {
...
Vector2 motion = (current_floor_velocity + body_velocity) * (Engine::get_singleton()->is_in_physics_frame() ? get_physics_process_delta_time() : get_process_delta_time());

https://github.com/godotengine/godot/blob/master/scene/2d/physics_body_2d.cpp Lines 947-962 Last commit Feb 7, 2021 4:29pm EST

I don't know the explanation for delta * GRAVITY.

DaveTheCoder

On reflection, maybe the reason that you have to multiple gravity by delta is simply because move_and_slide() doesn't deal with gravity.

YoshikaMiyafuji

@DaveTheCoder said: On reflection, maybe the reason that you have to multiple gravity by delta is simply because move_and_slide() doesn't deal with gravity.

Yeah I guess it has to be something like that, perhaps move_and_slide() only deals with movement derived from player input, whether it is from the input singleton or an input event, and as the gravity isn't caused by the player's input it is omitted in move_and_slide() and has to be multiplied by delta

dotted

Check the very last bit of code shown, just under _physics_process it has the gravity as velocity.y = delta * GRAVITY and at the end of the same function it has move_and_slide.

GRAViTY is set to 9.8 (m/s/s). This is is the standard value in physics for Earth's standard gravity.

The units here are meters per second per second.

delta is the time_step set as 1/60th of a second normally. To simulate the acceleration, velocity is updated by G * delta per _physics_process() invocation.

The calculation is correct to update the position of the kinematic WRT to 1 (Earth) G. Where the velocity of an object increases by 9.8 meters per second, thus 0, 9.8, 2 * 9.8.....each second until it approaches the speed of light (at least in our little real universe), but I digress into geeking physics.

dotted

Note it helps to take a primer for the concepts of position/velocity/acceleration (short video)

Learning some fundamental physics (Newton) and the basic concepts of calculus will be of immense value to you.

I recommend 3blue1brown's series on calculus of which the first part is more than enough:

lukepass86

Thanks @dotted for your very detailed answer! So when applying left or right movement I don't need acceleration (so I don't multiply by delta) because that's not an acceleration, right?

Thanks again!