Instead of creating per-frame hitboxes for gun games we (uh, you, not me) need to build some code that allows for linear interpolation of targeting data. For example, a guy enters from the left side, moves to the center, and stands there aiming at you. Instead of creating a couple hundred individual hit boxes, you create key frames. One would be when he enters the screen. The next is when he stops moving. The last is when he fires. Let the computer calculate everything in the middle.