We know that the ray misses the box, when t_min and t_max of intersections with both x and y planes are all negative.
@rasterize That's certainly one case where there's a "miss." Can you think of some others?
t_max < t_min
Also if the t-interval calculated here is disjoint from the ray's own t-interval
@Cake Right. Also if t_max is negative, for instance.