Sprite Renderer Performance

Questions & Answers
1

We are developing a 2D Game with many sprites.

Only the background has 729 sprites. These sprites are semi transparent.

All sprites are in one atlas; so that our draw calls are relatively ok; maximum 20 draw calls due to layering system we use.

Each sprite has 2 vertices; so there are 1458 vertices just for background. Although background is created dynamically it is static.

What would be the best approach to take a snapshot of the background image and use it as a single sprite? If we disable those sprites; FPS increases.

I told background is static; but actually it is moving; but it moves as a whole. It is semi-transparent and actually there is a “real static” background image.

2

We figured out hat this can be solved with Unity - Manual: Render Texture.

However; with proper sorting; draw calls dramatically reduced to 2-3 calls; so we did not need id.

Edit: Later on we need it for some other performance enhancements :slight_smile: here how the code looks like:

@SmashingSuccess here is the code:

public class Photographer : MonoBehaviour
    {
        public Camera MasterCamera;
        public Camera SelfCamera;
        public SpriteRenderer BackgroundDisplayer;

        public LevelBuilder CurrentLevelBuilder;
        public LayerMask Layer1;

        public void TakeShot()
        {
            SelfCamera.transform.position = MasterCamera.transform.position;
            SelfCamera.orthographicSize = MasterCamera.orthographicSize;

            var w = SelfCamera.pixelWidth;
            var h = SelfCamera.pixelHeight;


            //Background

            SelfCamera.cullingMask = Layer1;
            var texture = new RenderTexture(w, h, 0, RenderTextureFormat.ARGB32);
            SelfCamera.targetTexture = texture;
            SelfCamera.Render();

            RenderTexture.active = texture;

            var textureFormat = TextureFormat.RGB24;

            //https://issuetracker.unity3d.com/issues/error-regarding-invalid-textureformat-when-calling-supportstextureformat
#if UNITY_ANDROID
// if (SystemInfo.SupportsTextureFormat(TextureFormat.ETC_RGB4))
// {
//    textureFormat = TextureFormat.ETC_RGB4;
// }
#elif UNIT_IOS
// if (SystemInfo.SupportsTextureFormat(TextureFormat.PVRTC_RGB4))
// {
//    textureFormat = TextureFormat.PVRTC_RGB4;
// }
#endif

            var virtualPhoto = new Texture2D(w, h, textureFormat, false);
            virtualPhoto.name = "PhotoBackgroundTexture";
            var r = new Rect(0, 0, w, h);
            virtualPhoto.ReadPixels(r, 0, 0); // you get the center section
            virtualPhoto.Apply();
            var bg = Sprite.Create(virtualPhoto, r, new Vector2(0.5f, 0.5f), 100f);
            bg.name = "PhotoBackgroundSprite";
            RenderTexture.active = null;
            BackgroundDisplayer.sprite = bg;

            SelfCamera.targetTexture = null;
            Destroy(texture);


            var scaleNeeded = 100 * (MasterCamera.orthographicSize * 2) / h;
            BackgroundDisplayer.transform.localScale = new Vector3(scaleNeeded, scaleNeeded, scaleNeeded);


            SelfCamera.enabled = false;
            BackgroundDisplayer.enabled = true;

            MasterCamera.clearFlags = CameraClearFlags.SolidColor;

            var all = GameObject.FindGameObjectsWithTag("ToBePhotographed");
            for (var i = 0; i < all.Length; i++)
            {
                Destroy(all*);*

}
Destroy(gameObject);
}
}
SelfCamera is responsible to take screenshot. It only renders “Layer1”. All other cameras; renders everything except “Layer1”.
Items who are going to be photographed are tagged with “ToBePhotographed” and are configured to be rendered in “Layer1”. After screenshot is taken; they are destroyed…