Wrong coordinates in hitbox checks

#1

All functions with hitbox checks (in Bitmap.hx, Tilemap.hx, Video.hx) currently use this logic:

if (px > 0 && py > 0 && px <= __bitmapData.width && py <= __bitmapData.height)

It is 1 pixel wrong for both axes. For example, Mouse.ROLL_OVER event (which is based on hitbox checks) will not trigger when the cursor is 1px over an object from top / left, but will trigger when it is 1px away from bottom / right.

hitbox

To fix this the code should be changed to:

if (px >= 0 && py >= 0 && px < __bitmapData.width && py < __bitmapData.height)
1 Like
#2

Have you had an opportunity to compare this to the Flash target? If this is how Flash handles it then I would suggest offsetting your cursor graphic by (-1, -1)

Let me know if you have a chance to check, thanks :smiley:

#3

Yes, in Flash it’s fairly odd:

if (px > 0 && py >= 0 && px <= __bitmapData.width && py < __bitmapData.height)

I don’t know why they treat X and Y differently, but this has always looked like a bug to me.

#4

Interesting – is there a way to verify this, like a .hitTest() call rather than needing to use a mouse or touch?

#5

I don’t know, I was testing with a mouse. I also don’t know if all hit tests in Flash share one logic (and Mouse.ROLL_OVER is not based on it’s own rules, for example).

#6

Strange so if you have an object at (0, 0) and you move your mouse to (0, 0) in Flash it does not register as a hit? I think that is why we use the logic we have now

#7

Atm OpenFL logic is wrong and not Flash-like.

#8

Does hitTestPoint also work like if (px > 0 && py >= 0 && px <= __bitmapData.width && py < __bitmapData.height)?