Asking Google I found three possible things, that could be the problem.

1. When creating my mesh I set mainTextureOffset and mainTextureScale.
2. My sprites use the build in transparent shader, which apperently could also kill the dynamic batching.
3. I set the material for all meshes to my spriteatlas texture. Not using shared materials.

The Code is as follows.

public static GameObject CreateSprite(float x, float y, float width, float height, float scale, Material material)
    {
        y = materialHeight - y;
        float uvWidthUnit = 1f / materialWidth;
        float uvHeightUnit = 1f / materialHeight;

        float vertHeightScale = scale * (height / width);

        Mesh mesh = new Mesh();

        Vector3[] verts = new Vector3[4];
        Vector2[] uvs = new Vector2[4];
        int[] tris = new int[6] { 0, 1, 2, 2, 1, 3 };

        verts[0] = -Vector3.right * scale + Vector3.up * vertHeightScale;
        verts[1] = Vector3.right * scale + Vector3.up * vertHeightScale;
        verts[2] = -Vector3.right * scale - Vector3.up * vertHeightScale;
        verts[3] = Vector3.right * scale - Vector3.up * vertHeightScale;

        uvs[0] = new Vector2(0.0f, 1.0f);
        uvs[1] = new Vector2(1.0f, 1.0f);
        uvs[2] = new Vector2(0.0f, 0.0f);
        uvs[3] = new Vector2(1.0f, 0.0f);

        mesh.vertices = verts;
        mesh.triangles = tris;
        mesh.uv = uvs;

        mesh.RecalculateNormals();

        GameObject newMeshObject = new GameObject();
        newMeshObject.AddComponent<MeshFilter>().mesh = mesh;
        newMeshObject.AddComponent<MeshRenderer>();
        newMeshObject.renderer.material = material;
        newMeshObject.renderer.material.mainTextureOffset = new Vector2(uvWidthUnit * x, uvHeightUnit * y);
        newMeshObject.renderer.material.mainTextureScale = new Vector2(uvWidthUnit * width, uvWidthUnit * height);
        newMeshObject.AddComponent<BoxCollider>();
        Vector3 boxColliderSize = newMeshObject.GetComponent<BoxCollider>().size;
        boxColliderSize.z = 0.1f;
        newMeshObject.GetComponent<BoxCollider>().size = boxColliderSize;
        newMeshObject.AddComponent<Rigidbody>();
        Rigidbody rigidbody = newMeshObject.GetComponent<Rigidbody>();
        rigidbody.constraints = RigidbodyConstraints.FreezeAll;
        newMeshObject.isStatic = true;
        return newMeshObject;
    }

So first I tried setting UV for the vertices instead of changing mainTextureOffset and mainTextureScale. For the first test I just removed the two lines, to see if the number of draw calls decreased.

The number of draw calls, on my main game screen, went down from 42 to 9 (I have some GUI taking up some calls).. And when I started shooting, a lot, the number of draw calls stayed solid at 9. So a real improvement. Of cause now I had to test if changing the UV coordinates for each frame changes anything…. It didn’t effect the number of draw calls This post from the Unity Forum says it all. So modifying mainTextureOffset and mainTextureScale is not the way to do it!  So no more testing, the UV way of doing things must be the way to go.

[ExecuteInEditMode]
public class UVSetter : MonoBehaviour
{

    public float pixX;
    public float pixY;
    public float pixW;
    public float pixH;
    public float matrialSizeX;
    public float matrialSizeY;
    private Mesh theMesh;
    public bool updatable;
    // Use this for initialization
    void Start()
    {
        theMesh = transform.GetComponent<MeshFilter>().mesh;

        float uvx = pixX / matrialSizeX;
        float uvy = 1f - (pixY / matrialSizeY);
        float uvw = pixW / matrialSizeX;
        float uvh = pixH / matrialSizeY;

        Vector2[] newUVs = new Vector2[4];
        newUVs[0] = new Vector2(uvx, uvy);
        newUVs[3] = new Vector2(uvx, uvy + uvh);
        newUVs[2] = new Vector2(uvx + uvw, uvy);
        newUVs[1] = new Vector2(uvx + uvw, uvy + uvh);
        theMesh.uv = newUVs;
    }

    // Update is called once per frame
    void Update()
    {
        if (!updatable)
        {
            return; 
        }
        float uvx = pixX / matrialSizeX;
        float uvy = 1f - (pixY / matrialSizeY);
        float uvw = pixW / matrialSizeX;
        float uvh = pixH / matrialSizeY;

        Vector2[] newUVs = new Vector2[4];
        newUVs[0] = new Vector2(uvx, uvy);
        newUVs[3] = new Vector2(uvx, uvy + uvh);
        newUVs[2] = new Vector2(uvx + uvw, uvy);
        newUVs[1] = new Vector2(uvx + uvw, uvy + uvh);
        theMesh.uv = newUVs;
    }
}

On a final note I just want to mention that, to get the editor to show the correct UV in the editor, I just added [ExecuteInEditMode] decoration to my UVSetter test class..

Now, go make games!

– Henning

private shieldFingerId = -1;

...
// If nothing is touched, the shield controller has been released
// Know I should test for the shieldFingerId to be released, but this works for my purpose.
if (Input.touchCount == 0)
{
	shieldFingerId = -1;
}

foreach (Touch touch in Input.touches)
{
	// Fire
	if(touch.phase == TouchPhase.Began && !withinShieldController)
	{
		Fire();
	}

	// "Grap" shield
	if (touch.phase == TouchPhase.Began && withinShieldController)
	{
		shieldFingerId = touch.fingerId;
		UpdateSheldAngle(touch.position);
	}

	// Move shield
	if (touch.phase == TouchPhase.Moved && touch.fingerId == shieldFingerId)
	{
		UpdateSheldAngle(touch.position);
	}
}

As you can see, when I got hold of the information about touch.phase and .fingerId, everything turned out to be very simple indeed.

I found the Unity Script Reference pages to be very limited on the subject of touch input, e.g. information about touch.fingerId only says “The unique index for touch.” no examples, or explanation. But by asking Google, and searching the Unity forums, great resource I might add, it finally made sense.

Working with touch has forced me to use my LG E975 mobile for testing, which is a slow process. :/ I guess I have to figure out how to make Unity Remote work with my phone. If not, development is going to take forever!

Hope the above code makes sense? Now go make games!

– Henning

For the movement script, I wanted the player to move in a continuous motion and not in discreet jumps. The player should also be able to keep moving in one direction holding down a direction button.

To keep track of what the player is doing, I first made an enum with the possible motion states.
I guess the enums are self explanatory

enum Moving
{
    NO,
    UP,
    DOWN,
    LEFT,
    RIGHT
};

Now the the logic. If the player isn’t moving, check for player input.

        if (direction == Moving.NO)
        {
            if (Input.GetKey(KeyCode.A))
            {
                direction = Moving.LEFT;
            }
            if (Input.GetKey(KeyCode.D))
            {
                direction = Moving.RIGHT;
            }
            if (Input.GetKey(KeyCode.W))
            {
                direction = Moving.UP;
            }
            if (Input.GetKey(KeyCode.S))
            {
                direction = Moving.DOWN;
            }
        }

Now if the player is moving, the enum will be holding the direction of motion and we can move the player.

            switch (direction)
            {
                case Moving.UP:
                    if (moveLeft > 0f)
                    {
                        Vector3 v3 = transform.position;
                        v3.y += movementSpeed;
                        moveLeft -= +movementSpeed;
                        if (v3.y > 1f)
                        {
                            v3.y = 1f;
                            moveLeft = 1.0f;
                            direction = Moving.NO;
                        }
                        transform.position = v3;
                    }
                    else
                    {
                        moveLeft = 1.0f;
                        direction = Moving.NO;
                    }
                    break;
                case Moving.DOWN:
                    if (moveLeft > 0f)
                    {
                        Vector3 v3 = transform.position;
                        v3.y -= movementSpeed;
                        moveLeft -= +movementSpeed;
                        if (v3.y < -1f)
                        {
                            v3.y = -1f;
                            moveLeft = 1.0f;
                            direction = Moving.NO;
                        }
                        transform.position = v3;
                    }
                    else
                    {
                        moveLeft = 1.0f;
                        direction = Moving.NO;
                    }
                    break;
                case Moving.LEFT:
                    if (moveLeft > 0f)
                    {
                        Vector3 v3 = transform.position;
                        v3.x -= movementSpeed;
                        moveLeft -= +movementSpeed;
                        if (v3.x < -1f)
                        {
                            v3.x = -1f;
                            moveLeft = 1.0f;
                            direction = Moving.NO;
                        } transform.position = v3;
                    }

                    else
                    {
                        moveLeft = 1.0f;
                        direction = Moving.NO;
                    }
                    break;
                case Moving.RIGHT:
                    if (moveLeft > 0f)
                    {
                        Vector3 v3 = transform.position;
                        v3.x += movementSpeed;
                        moveLeft -= +movementSpeed;
                        if (v3.x > 1f)
                        {
                            v3.x = 1f;
                            moveLeft = 1.0f;
                            direction = Moving.NO;
                        }
                        transform.position = v3;
                    }
                    else
                    {
                        moveLeft = 1.0f;
                        direction = Moving.NO;
                    }
                    break;
                default:
                    break;
            }

That’s it, not pretty but it works. If I have the time, I would like to make the motion even more fluent. The state change if you keep a key down makes the player give a little jump/jitter as the state goes through NO state before returning to a moving state. But for now it gets the job done

All feedback is welcome, I can’t change if I don’t know what to change.

– Henning