Hi! The black and white or RGB normals is there a difference between them? Thanks!
Bump maps store one intensity value per pixel that is interpreted as height, normal maps store direction of the surface normal (three values in r, g, b) at pixel location.
Blender wiki (general info, specifics apply to Blender Internal)
Cycles: bump node, normal map node
Wikipedia: bump mapping, normal mapping
But they seem to produce the same result, so is there a final visual difference?
Not that I experienced, but I don’t know better than what is in those links I posted. Both perturb the surface normal for an illusion of higher surface details. Depending on the situation one might work better than the other. For example, tangent space normal mapping can’t be done with a bump map, as far as I know.
How about a practical example? 
Thanks for reply. -
Thanks for the example.
I believe that it might help to think that when you give Blender a grayscale ‘bump’ map it generates a normal map internally and then shades the object accordingly, and when you give it a normal map it just uses what you give it.
Inthis fileall the normal maps were generated from the same height map (better term than bump map) using different parameters. I used Njob. (Invert Y matches most closely the height map’s direction, but only in this program, others may already invert the y so you just have to try it out.)
That was good example, I’ll post my own tomorrow (differences).