Another big advantage, is that the mouse can be read. This way you can program LV to select object in your picture, and drag them around, delete them etc.
And obviouslly, it is much easier to draw (e.g.) a circle in a picture, then to draw a circle in a graph. (And if you succeed in drawing it, try filling it with a color! It's doable, but no fun at all...)
Disadvantges are: If you draw data points, you'll have to make your own scales, autoscaling, 'scrolling' of the date, etc.
How often you use them depends greatly on the field you're in. If you make games (yes, in LabVIEW), vision, imaq, ..., you'll use pictures. If you're in DAQ, T&M, research, you might never need them...