This tutorial series will show you how to make a single player FPS game.
Throughout the course of this tutorial series, we will cover how:
- To make a first person character that can move, sprint, and jump.
- To make a simple animation state machine for handling animation transitions.
- To add three weapons to the first person character, each using a different way to handle bullet collisions:
- A knife (using an Area)
- A pistol (Bullet scenes)
- A rifle (using a Raycast)
- To add two different types of grenades to the first person character:
- A normal grenade
- A sticky grenade
- To add the ability to grab and throw RigidBody nodes
- To add joypad input for the player
- To add ammo and reloading for all weapons that consume ammo.
- To add ammo and health pick ups
- In two sizes: big and small
- To add an automatic turret
- That can fire using bullet objects or a Raycast
- To add targets that break when they’ve taken enough damage
- To add sounds that play when the guns fire.
- To add a simple main menu:
- With an options menu for changing how the game runs
- With a level select screen
- To add a universal pause menu we can access anywhere
While this tutorial can be completed by beginners, it is highly advised to complete Your First Game, if you are new to Godot and/or game development before going through this tutorial series.
Remember: Making 3D games is much harder than making 2D games. If you do not know how to make 2D games you will likely struggle making 3D games.
This tutorial assumes you know have experience working with the Godot editor, have basic programming experience in GDScript, and have basic experience in game development.
You can find the start assets for this tutorial here:
The provided starter assets contain an animated 3D model, a bunch of 3D models for making levels, and a few scenes already configured for this tutorial.
All assets provided (unless otherwise noted) were originally created by TwistedTwigleg, with changes/additions by the Godot community.
All original assets provided for this tutorial are released under the
Feel free to use these assets however you want! All original assets belong to the Godot community, with the other assets belonging to those listed below:
The skybox is created by StumpyStrust on OpenGameArt. The skybox used is
The font used is Titillium-Regular, and is licensed under the
SIL Open Font License, Version 1.1.
You can find the finished project for each part at the bottom of each part’s page
In this part we will be making a first person player that can move around the environment.
By the end of this part you will have a working first person character who can move around the game environment, look around with a mouse based first person camera, that can jump into the air, turn on and off a flash light, and sprint.
Getting everything ready¶
Launch Godot and open up the project included in the starter assets.
While these assets are not necessarily required to use the scripts provided in this tutorial, they will make the tutorial much easier to follow as there are several pre-setup scenes we will be using throughout the tutorial series.
First, go open the project settings and go to the “Input Map” tab. You’ll find several actions have already been defined. We will be using these actions for our player. Feel free to change the keys bound to these actions if you want.
Let’s take a second to see what we have in the starter assets.
Included in the starter assets are several scenes. For example, in res:// we have 14 scenes, most of which we will be visiting as we go through this tutorial series.
For now let’s open up
There are a bunch of scenes and a few textures in the
Assets folder. You can look at these if you want,
but we will not be exploring through
Assets in this tutorial series.
Assets contains all of the models used
for each of the levels, as well as some textures and materials.
Making the FPS movement logic¶
Once you have
Player.tscn open, let’s take a quick look at how it is set up
First, notice how the player’s collision shapes are set up. Using a vertical pointing capsule as the collision shape for the player is fairly common in most first person games.
We are adding a small square to the ‘feet’ of the player so the player does not feel like they are balancing on a single point.
We do want the ‘feet’ slightly higher than the bottom of the capsule so we can roll over slight edges. Where to place the ‘feet’ is dependent on your levels and how you want your player to feel.
Many times player will notice how the collision shape being circular when they walk to an edge and slide off. We are adding the small square at the bottom of the capsule to reduce sliding on, and around, edges.
Another thing to notice is how many nodes are children of
Rotation_Helper. This is because
Rotation_Helper contains all of the nodes we want to rotate on the
X axis (up and down).
The reason behind this is so we can rotate
Player on the
Y axis, and
If we did not use
Rotation_helper then we’d likely have cases where we are rotating
Y axes at the same time. This can lead to undesirable results, as we then
could rotate on all three axes in some cases.
See using transforms for more information
Attach a new script to the
Player node and call it
Let’s program our player by adding the ability to move around, look around with the mouse, and jump.
Add the following code to
extends KinematicBody const GRAVITY = -24.8 var vel = Vector3() const MAX_SPEED = 20 const JUMP_SPEED = 18 const ACCEL= 4.5 var dir = Vector3() const DEACCEL= 16 const MAX_SLOPE_ANGLE = 40 var camera var rotation_helper var MOUSE_SENSITIVITY = 0.05 func _ready(): camera = $Rotation_Helper/Camera rotation_helper = $Rotation_Helper Input.set_mouse_mode(Input.MOUSE_MODE_CAPTURED) func _physics_process(delta): process_input(delta) process_movement(delta) func process_input(delta): # ---------------------------------- # Walking dir = Vector3() var cam_xform = camera.get_global_transform() var input_movement_vector = Vector2() if Input.is_action_pressed("movement_forward"): input_movement_vector.y += 1 if Input.is_action_pressed("movement_backward"): input_movement_vector.y -= 1 if Input.is_action_pressed("movement_left"): input_movement_vector.x -= 1 if Input.is_action_pressed("movement_right"): input_movement_vector.x = 1 input_movement_vector = input_movement_vector.normalized() dir += -cam_xform.basis.z.normalized() * input_movement_vector.y dir += cam_xform.basis.x.normalized() * input_movement_vector.x # ---------------------------------- # ---------------------------------- # Jumping if is_on_floor(): if Input.is_action_just_pressed("movement_jump"): vel.y = JUMP_SPEED # ---------------------------------- # ---------------------------------- # Capturing/Freeing the cursor if Input.is_action_just_pressed("ui_cancel"): if Input.get_mouse_mode() == Input.MOUSE_MODE_VISIBLE: Input.set_mouse_mode(Input.MOUSE_MODE_CAPTURED) else: Input.set_mouse_mode(Input.MOUSE_MODE_VISIBLE) # ---------------------------------- func process_movement(delta): dir.y = 0 dir = dir.normalized() vel.y += delta*GRAVITY var hvel = vel hvel.y = 0 var target = dir target *= MAX_SPEED var accel if dir.dot(hvel) > 0: accel = ACCEL else: accel = DEACCEL hvel = hvel.linear_interpolate(target, accel*delta) vel.x = hvel.x vel.z = hvel.z vel = move_and_slide(vel,Vector3(0,1,0), 0.05, 4, deg2rad(MAX_SLOPE_ANGLE)) func _input(event): if event is InputEventMouseMotion and Input.get_mouse_mode() == Input.MOUSE_MODE_CAPTURED: rotation_helper.rotate_x(deg2rad(event.relative.y * MOUSE_SENSITIVITY)) self.rotate_y(deg2rad(event.relative.x * MOUSE_SENSITIVITY * -1)) var camera_rot = rotation_helper.rotation_degrees camera_rot.x = clamp(camera_rot.x, -70, 70) rotation_helper.rotation_degrees = camera_rot
This is a lot of code, so let’s break it down function by function:
While copy and pasting code is ill advised, as you can learn a lot from manually typing the code in, you can copy and paste the code from this page directly into the script editor.
If you do this, all of the code copied will be using spaces instead of tabs.
To convert the spaces to tabs in the script editor, click the “edit” menu and select “Convert Indent To Tabs”. This will convert all of the spaces into tabs. You can select “Convert Indent To Spaces” to convert t back into spaces.
First, we define some global variables to dictate how our player will move about the world.
Throughout this tutorial, variables defined outside functions will be referred to as “global variables”. This is because we can access any of these variables from any place in the script. We can “globally” access them, hence the name.
Let’s go through each of the global variables:
GRAV: How strong gravity pulls us down.
vel: Our KinematicBody’s velocity.
MAX_SPEED: The fastest speed we can reach. Once we hit this speed, we will not go any faster.
JUMP_SPEED: How high we can jump.
ACCEL: How fast we accelerate. The higher the value, the faster we get to max speed.
DEACCEL: How fast we are going to decelerate. The higher the value, the faster we will come to a complete stop.
MAX_SLOPE_ANGLE: The steepest angle our KinematicBody will consider as a ‘floor’.
camera: The Camera node.
rotation_helper: A Spatial node holding everything we want to rotate on the X axis (up and down).
MOUSE_SENSITIVITY: How sensitive the mouse is. I find a value of
0.05works well for my mouse, but you may need to change it based on how sensitive your mouse is.
You can tweak many of these variables to get different results. For example, by lowering
JUMP_SPEED you can get a more ‘floaty’ feeling character.
Feel free to experiment!
You may have noticed that
MOUSE_SENSITIVITY is written in all caps like the other constants, but is
MOUSE_SENSITIVITY is not a constant.
The reason behind this is we want to treat it like a constant variable (a variable that cannot change) throughout our script, but we want to be able to change the value later when we add customizable settings. So, in an effort to remind ourselves to treat it like a constant, it’s named in all caps.
Now let’s look at the
First we get the
rotation_helper nodes and store them into their variables.
Then we need to set the mouse mode to captured so the mouse cannot leave the game window.
This will hide the mouse and keep it at the center of the screen. We do this for two reasons: The first reason being we do not want to the player to see their mouse cursor as they play.
The second reason is because we do not want the cursor to leave the game window. If the cursor leaves the game window there could be instances where the player clicks outside the window, and then the game would lose focus. To assure neither of these issues happen, we capture the mouse cursor.
see Input documentation for the various mouse modes. We will only be using
MOUSE_MODE_VISIBLE in this tutorial series.
Next let’s take a look at
All we’re doing in
_physics_process is calling two functions:
process_input will be where we store all of the code relating to player input. We want to call it first before
anything else so we have fresh player input to work with.
process_movement is where we’ll send all of the date necessary to the KinematicBody
so it can move through the game world.
Let’s look at
First we set
dir to an empty Vector3.
dir will be used for storing the direction the player intends to move towards. Because we do not
want the player’s previous input to effect the player beyond a single
process_movement call, we reset
Next we get the camera’s global transform and store it as well, into the
The reason we need the camera’s global transform is so we can use it’s directional vectors. Many have found directional vectors confusing, so let’s take a second to explain how they work:
World space can be defined as: The space in which all objects are placed in, relative to a constant origin point. Every object, no matter if it is 2D or 3D, has a position in world space.
To put it another way: world space is the space in a universe where every object’s position, rotation, and scale can be measured by a known, fixed point called the origin.
In Godot, the origin is at position
(0, 0, 0) with a rotation of
(0, 0, 0) and a scale of
(1, 1, 1).
When you open up the Godot editor and select a Spatial based node, a gizmo pops up. Each of the arrows points using world space directions by default.
If you want to move using the world space directional vectors, you’d do something like this:
if Input.is_action_pressed("movement_forward"): node.translate(Vector3(0, 0, 1)) if Input.is_action_pressed("movement_backward"): node.translate(Vector3(0, 0, -1)) if Input.is_action_pressed("movement_left"): node.translate(Vector3(1, 0, 0)) if Input.is_action_pressed("movement_right"): node.translate(Vector3(-1, 0, 0))
Notice how we do not need to do any calculations to get world space directional vectors. We can define a few Vector3 variables and input the values pointing in each direction.
Here is what world space looks like in 2D:
The following images are just examples. Each arrow/rectangle represents a directional vector
And here is what it looks like for 3D:
Notice how in both examples, the rotation of the node does not change the directional arrows. This is because world space is a constant. No matter how you translate, rotate, or scale an object, world space will always point in the same direction.
Local space is different, because it takes the rotation of the object into account.
Local space can be defined as follows:
The space in which a object’s position is the origin of the universe. Because the position
of the origin can be at
N many locations, the values derived from local space change
with the position of the origin.
This stack overflow question has a much better explanation of world space and local space.
https://gamedev.stackexchange.com/questions/65783/what-are-world-space-and-eye-space-in-game-development (Local space and eye space are essentially the same thing in this context)
Each Basis has three vectors:
Each of those vectors point towards each of the local space vectors coming from that object.
To use the a Spatial node’s local directional vectors, we use this code:
if Input.is_action_pressed("movement_forward"): node.translate(node.global_transform.basis.z.normalized()) if Input.is_action_pressed("movement_backward"): node.translate(-node.global_transform.basis.z.normalized()) if Input.is_action_pressed("movement_left"): node.translate(node.global_transform.basis.x.normalized()) if Input.is_action_pressed("movement_right"): node.translate(-node.global_transform.basis.x.normalized())
Here is what local space looks like in 2D:
And here is what it looks like for 3D:
Here is what the Spatial gizmo shows when you are using local space mode. Notice how the arrows follow the rotation of the object on the left, which looks exactly the same as the 3D example for local space.
You can change between local and world space modes by pressing the little cube button when you have a Spatial based node selected.
Local vectors are confusing even for more experienced game developers, so do not worry if this all doesn’t make a lot of sense. The key thing to remember about local vectors is that we are using local coordinates to get direction from the object’s point of view, as opposed to using world vectors which give direction from the world’s point of view.
Okay, back to
Next we make a new variable called
input_movement_vector and assign it to an empty Vector2.
We will use this to make a virtual axis of sorts so map the player’s input to movement.
This may seem overkill for just the keyboard, but this will make sense later when we add joypad input.
Based on which directional movement action is pressed, we add or remove from
After we’ve checked each of the directional movement actions, we normalize
input_movement_vector. This makes it where
are within a
1 radius unit circle.
Next we add the camera’s local
Z vector times
dir. This where when we pressed forward or backwards we add the camera’s
Z axis, so we move forward in relation to the camera.
Because the camera is rotated by
-180 degrees, we have to flip the
Z directional vector.
Normally forward would be the positive Z axis, so using
basis.z.normalized() would work,
but we are using
-basis.z.normalized() because our camera’s Z axis faces backwards in relation
to the rest of the player.
We do the same thing for the camera’s local
X vector, and instead of using
input_movement_vector.y we instead use
This makes it where when we press left or right, we move left/right in relation to the camera.
Next we check if the player is on the floor using KinematicBody’s
is_on_floor function. If it is, then we
check to see if the “movement_jump” action has just been pressed. If it has, then we set our
Y velocity to
Because we’re setting the Y velocity, we will jump into the air.
Then we check for the
ui_cancel action. This is so we can free/capture the mouse cursor when the
is pressed. We do this because otherwise we’d have no way to free the cursor, meaning it would be stuck until you terminate the
To free/capture the cursor, we check to see if the mouse is visible (freed) or not. If it is, then we capture it, and if it’s not we make it visible (free it).
That’s all we’re doing right now for
process_input. We’ll come back several times to this function as we add more complexities to our player.
Now let’s look at
First we assure that
dir does not have any movement on the
Y axis by setting it’s
Y value to zero.
Next we normalize
dir to assure we’re within a
1 radius unit circle. This makes it where we’re moving at a constant speed regardless
of whether we’ve moving straight, or moving diagonal. If we did not normalize, we would move faster on the diagonal than when we’re going straight.
Next we add gravity to our player by adding
GRAVITY * delta to our
After that we assign our velocity to a new variable (called
hvel) and remove any movement on the
Next we set a new variable (
target) to our direction vector.
Then we multiply that by our max speed so we know how far we will can move in the direction provided by
After that we make a new variable for our acceleration, named
We then take the dot product of
hvel to see if we are moving according to
hvel does not have any
Y velocity, meaning we are only checking if we are moving forwards, backwards, left, or right.
If we are moving according to
hvel, then we set
accel to our
ACCEL constant so we accelerate, otherwise we set
our ``DEACCEL constant so we decelerate.
Then we interpolate our horizontal velocity, set our
Z velocity to the interpolated horizontal velocity,
move_and_slide to let the KinematicBody handle moving through the physics world.
All of the code in
process_movement is exactly the same as the movement code from the Kinematic Character demo!
The final function we have is the
_input function, and thankfully it’s fairly short:
First we make sure that the event we are dealing with is a InputEventMouseMotion event. We also want to check if the cursor is captured, as we do not want to rotate if it is not.
See Mouse and input coordinates for a list of possible input events.
If the event is indeed a mouse motion event and the cursor is captured, we rotate based on the relative mouse motion provided by InputEventMouseMotion.
First we rotate the
rotation_helper node on the
X axis, using the relative mouse motion’s
Y value, provided by InputEventMouseMotion.
Then we rotate the entire KinematicBody on the
Y axis by the relative mouse motion’s
Godot converts relative mouse motion into a Vector2 where mouse movement going
up and down is
-1 respectively. Right and Left movement is
Because of how we are rotating the player, we multiply the relative mouse motion’s
X value by
-1 so mouse motion going left and right rotates the player left and right
in the same direction.
Finally, we clamp the
X rotation to be between
degrees so we cannot rotate ourselves upside down.
see using transforms for more information on rotating transforms.
To test the code open up the scene named
Testing_Area.tscn, if it’s not already opened up. We will be using
this scene as we go through the tutorial, so be sure to keep it open in one of your scene tabs.
Go ahead and test your code either by pressing
Testing_Area.tscn as the open tab, by pressing the
play button in the top right corner, or by pressing
You should now be able to walk around, jump in the air, and look around using the mouse.
Giving the player a flash light and the option to sprint¶
Before we get to making the weapons work, there is a couple more things we should add.
Many FPS games have an option to sprint and a flash light. We can easily add these to our player, so let’s do that!
First we need a few more global variables in our player script:
const MAX_SPRINT_SPEED = 30 const SPRINT_ACCEL = 18 var is_sprinting = false var flashlight
All of the sprinting variables work exactly the same as the non sprinting variables with similar names.
is_sprinting is a boolean to track whether the player is currently sprinting, and
flashlight is a variable
we will be using to hold our flash light node.
Now we need to add a few lines of code, starting in
_ready. Add the following to
flashlight = $Rotation_Helper/Flashlight
This gets our flash light node and assigns it to the
Now we need to change some of the code in
process_input. Add the following somewhere in
# ---------------------------------- # Sprinting if Input.is_action_pressed("movement_sprint"): is_sprinting = true else: is_sprinting = false # ---------------------------------- # ---------------------------------- # Turning the flashlight on/off if Input.is_action_just_pressed("flashlight"): if flashlight.is_visible_in_tree(): flashlight.hide() else: flashlight.show() # ----------------------------------
Let’s go over the additions:
is_sprinting to true when we are holding down the
movement_sprint action, and false
movement_sprint action is released. In
process_movement we’ll add the code that makes the player faster when
they sprint. Here in
process_input we’re going to change the
We do something similar freeing/capturing the cursor for handling the flash light. We first check to see if the
was just pressed. If it was, we then check to see if
flashlight is visible in the scene tree. If it is, then we hide it, and if it’s not we show it.
Now we need to change a couple things in
process_movement. First, replace
target *= MAX_SPEED with the following:
if is_sprinting: target *= MAX_SPRINT_SPEED else: target *= MAX_SPEED
Now instead of always multiplying
MAX_SPEED, we first check to see if we are sprinting or not.
If we are sprinting, we instead multiply
Now all that’s left is changing the accleration when sprinting. Change
accel = ACCEL to the following:
if is_sprinting: accel = SPRINT_ACCEL else: accel = ACCEL
Now when we are sprinting we’ll use
SPRINT_ACCEL instead of
ACCEL, which will accelerate us faster.
You should now be able to sprint if you press the
shift button, and can toggle the flash light on and off by pressing the
Go give it a whirl! You can change the sprint related global variables to make the player faster or slower when sprinting!
Phew! That was a lot of work. Now you have a fully working first person character!
In Part 2 we will add some guns to our player character.
At this point we’ve recreated the Kinematic character demo from first person perspective with sprinting and a flash light!
Currently the player script would be at an ideal state for making all sorts of first person games. For example: Horror games, platformer games, adventure games, and more!
If you ever get lost, be sure to read over the code again!
You can download the finished project for this part here: