How to make rts camera smooth when zooming in Unity

Question

right now I am trying to make rts camera zooming with panning when close to the ground. The problem I have now is that I use mouse scroll wheel for zooming and it makes the zooming feel like it's laggy. It looks like it jumps some Y value and teleports rather than smoothly move to the desired position. Also, I would like to know how to make the camera stop at the minimum Y value because what is happening now is that it stops at around 22 rather than 20 which is my minimum Y value for the camera movement.

I tried zooming with + and - on my numpad and it worked how I wanted (smoothly zooming in and out without skipping) but not all players have numpad and I feel it is more suitable to zoom with mouse wheel.

{

private const int levelArea = 100;

private const int scrollArea = 25;
private const int scrollSpeed = 25;
private const int dragSpeed = 70;

private const int zoomSpeed = 50;

// Maximum/minimum zoom distance from the ground
public int zoomMin = 20;
public int zoomMax = 120;

private const int panSpeed = 40;

// Minimal/maximal angles for camera
private const int panAngleMin = 30;
private const int panAngleMax = 90;


void Update()
{
    // Init camera translation for this frame.
    var translation = Vector3.zero;

    // Zoom in or out
    var zoomDelta = Input.GetAxis("Mouse ScrollWheel") * zoomSpeed * Time.deltaTime;
    if (zoomDelta != 0)
    {
        translation -= Vector3.up * zoomSpeed * zoomDelta;
    }

    // Start panning camera if zooming in close to the ground or if just zooming out.
    var pan = transform.eulerAngles.x - zoomDelta * panSpeed;
    pan = Mathf.Clamp(pan, panAngleMin, panAngleMax);
    // When to start panning up the camera
    if (zoomDelta < 0 || transform.position.y < (zoomMax -20))
    {
        transform.eulerAngles = new Vector3(pan, 0, 0);
    }

    // Move camera with arrow keys
    translation += new Vector3(Input.GetAxis("Horizontal"), 0, Input.GetAxis("Vertical"));

    // Move camera with mouse
    if (Input.GetMouseButton(2)) // MMB
    {
        // Hold button and drag camera around
        translation -= new Vector3(Input.GetAxis("Mouse X") * dragSpeed * Time.deltaTime, 0,
                           Input.GetAxis("Mouse Y") * dragSpeed * Time.deltaTime);
    }
    else
    {
        // Move camera if mouse pointer reaches screen borders
        if (Input.mousePosition.x < scrollArea)
        {
            translation += Vector3.right * -scrollSpeed * Time.deltaTime;
        }

        if (Input.mousePosition.x >= Screen.width - scrollArea)
        {
            translation += Vector3.right * scrollSpeed * Time.deltaTime;
        }

        if (Input.mousePosition.y < scrollArea)
        {
            translation += Vector3.forward * -scrollSpeed * Time.deltaTime;
        }

        if (Input.mousePosition.y > Screen.height - scrollArea)
        {
            translation += Vector3.forward * scrollSpeed * Time.deltaTime;
        }
    }

    // Keep camera within level and zoom area
    var desiredPosition = transform.position + translation;
    if (desiredPosition.x < -levelArea || levelArea < desiredPosition.x)
    {
        translation.x = 0;
    }
    if (desiredPosition.y < zoomMin || zoomMax < desiredPosition.y)
    {
        translation.y = 0;
    }
    if (desiredPosition.z < -levelArea || levelArea < desiredPosition.z)
    {
        translation.z = 0;
    }

    // Move camera parallel to world axis
    transform.position += translation;
}

}

I would like to have a smooth transition from the position where the camera is now and the desired position after scrolling in/out. And also I would like to know how to make the camera to stop at the minimum/maximum zoom distance rather than to stop close to it. Thank you for helping. Video how the camera movement looks like: https://youtu.be/Lt3atJEaOjA

Answer 1

Okay, so I'm going to do three things here. First and foremost, I'm going to recommend that if you're working with advanced camera behavior, you probably want to at least consider using Cinemachine. I'd walk you through it myself, but given my lack of personal experience with it, I'd probably be doing you a disservice by even trying. There are many good tutorials out there. Youtube and Google should provide.

The second thing I'll do is solve your problem in the most direct way I can manage, and after that, we'll see if we can't come up with a better method for solving your problem.

So, the key here is that Unity's scrollwheel input is pretty binary. When you check a scrollwheel axis, the result is directly based on how many "clicks" your wheel has gone through since the last frame update, but what you really want is something with a bit of give. By default, Unity can actually do this with most of its axis inputs: You might notice that if you use WASD in a default Unity project, there's a sort of "lag" to it, where you'll take your fingers off the keys but you'll still keep receiving positive values from Input.GetAxis() for a few frames. This is tied to the Gravity value in your input settings, and Input.GetAxisRaw() is actually used to circumvent this entirely. For whatever reason, scrollwheel axes don't seem to be affected by axis gravity, so we essentially have to implement something similar ourselves.

// Add this property to your class definition (so it persists between updates):
private float wheelAxis = 0;

// Delete this line:
var zoomDelta = Input.GetAxis("Mouse ScrollWheel") * zoomSpeed * Time.deltaTime;

// And put these three new lines in its place:
wheelAxis += Input.GetAxis("Mouse ScrollWheel");
wheelAxis = Mathf.MoveTowards(wheelTotal, 0f, Time.deltaTime);
var zoomDelta = Mathf.Clamp(wheelAxis, -0.05f, 0.05f) * zoomSpeed * Time.deltaTime;

Right, so we do a few things here. Every update, we add the current scrollwheel values to our wheelAxis. Next, we apply the current Time.deltatime as "gravity" via the Mathf.MoveTowards() function. Finally, we call what's mostly just your old zoomDelta code with a simple modification: We constrain the wheelAxis with Mathf.Clamp to try and regulate how fast the zooming can happen.

You can modify this code in a couple of ways. If you multiply the Time.deltaTime parameter you can affect how long your input will "persist" for. If you mess with the Mathf.Clamp() values, you'll get faster or slower zooming. All in all though, if you just want a smooth zoom with minimal changes to your code, that's probably your best bet.

So!

Now that we've done that, let's talk about your code, and how you're approaching the problem, and see if we can't find a somewhat cleaner solution.

Getting a good camera working is surprisingly non-trivial. Your code looks like a lot of code I see that tries to solve a complex problem: It looks like you added some feature, and then tested it, and found some edge cases where it fell apart, and then patched those cases, and then tried to implement a new feature on top of the old code, but it kinda broke in various other ways, etc etc, and what we've got at this point is a bit messy.

The biggest issue with your code is that the camera's position and the camera's rotation are closely tied together. When working with characters, this is usually fine, but when working with a camera, you want to break this up. Think about where the camera is and what the camera is looking at as very separate things to keep track of.

So here's a working camera script that you should be able to plug in and just run with:

using UnityEngine;

public class RTSCamera : MonoBehaviour
{
    public float zoomSpeed = 100f;
    public float zoomTime = 0.1f;

    public float maxHeight = 100f;
    public float minHeight = 20f;

    public float focusHeight = 10f;
    public float focusDistance = 20f;

    public int panBorder = 25;
    public float dragPanSpeed = 25f;
    public float edgePanSpeed = 25f;
    public float keyPanSpeed = 25f;

    private float zoomVelocity = 0f;
    private float targetHeight;

    void Start()
    {
        // Start zoomed out
        targetHeight = maxHeight;
    }

    void Update()
    {
        var newPosition = transform.position;

        // First, calculate the height we want the camera to be at
        targetHeight += Input.GetAxis("Mouse ScrollWheel") * zoomSpeed * -1f;
        targetHeight = Mathf.Clamp(targetHeight, minHeight, maxHeight);

        // Then, interpolate smoothly towards that height
        newPosition.y = Mathf.SmoothDamp(transform.position.y, targetHeight, ref zoomVelocity, zoomTime);

        // Always pan the camera using the keys
        var pan = new Vector2(Input.GetAxis("Horizontal"), Input.GetAxis("Vertical")) * keyPanSpeed * Time.deltaTime;

        // Optionally pan the camera by either dragging with middle mouse or when the cursor touches the screen border
        if (Input.GetMouseButton(2)) {
            pan -= new Vector2(Input.GetAxis("Mouse X"), Input.GetAxis("Mouse Y")) * dragPanSpeed * Time.deltaTime;
        } else {
            var border = Vector2.zero;
            if (Input.mousePosition.x < panBorder) border.x -= 1f;
            if (Input.mousePosition.x >= Screen.width - panBorder) border.x += 1f;
            if (Input.mousePosition.y < panBorder) border.y -= 1f;
            if (Input.mousePosition.y > Screen.height - panBorder) border.y += 1f;
            pan += border * edgePanSpeed * Time.deltaTime;
        }

        newPosition.x += pan.x;
        newPosition.z += pan.y;

        var focusPosition = new Vector3(newPosition.x, focusHeight, newPosition.z + focusDistance);

        transform.position = newPosition;
        transform.LookAt(focusPosition);
    }
}

While I encourage you to go through it in your own time, I'm not going to drag you through every inch of it. Instead, I'll just go over the main crux.

The key idea here is that rather than controlling the camera's height and orientation directly, we just let the scrollwheel dictate where want the camera's height to be, and then we use Mathf.SmoothDamp() to move the camera smoothly into that position over several frames. (Unity has many useful functions like this. Consider Mathf.MoveTowards() for an alternative interpolation method.) At the very end, rather than trying to fiddle with the camera's rotation values directly, we just pick a point in front of us near the ground and point the camera at that spot directly.

By keeping the camera's position and orientation completely independent of each other, as well as separating out the "animation" of the camera's height, we avoid a lot of headaches and eliminate a lot of potential for messy interwoven bugs.

I hope that helps.

Answer 2

Cinemachine is pretty good, but I suggest people learn on your own quite a bit before using Cinemachine so that you better understand what's going on in the background.