It is to do with the radial distribution of the orbital - basically, the shape, yes.
The radial distribution functions (look these up) show that for a given principle quantum (n), the penetration of ns>np>nd>nf, and so the magnitude of shielding for ns>np>nd>nf.
ie. the f orbitals are generally speaking poor at shielding, and this is what causes the lanthanide contraction, and this is the reason why the 4d transition metals are very similar in size to the 5d.
The d orbitals are also generally speaking poor at shielding and cause a similar (although not as drastic) contraction accross the d block, which is particularly noticable when you compare Al and Ga's atomic radii (or for that matter, Si-Ge, P-As, S-Se or Cl-Br).
The effect of this is not only apparent in atomic radii. Take a look at, for example, electronegativity - and you will see an uneven or alternating pattern down group 3A. This general phenomenon is often referred to as the "alternation effect" or "alternating effect", thats probably worth looking up too.