One-loop scattering on a stack of D3 branes was considered in arXiv:0801.0218 [hep-th]. Divergence was found and its cancelation mechanism was proposed, wherein it was conjectured that the D-brane geometry be introduced in the form of counter vertex operators. Here we verify the conjecture at the first few leading orders in an expansion method that we call large-r₀ expansion. We comment on the relation with the Fischler-Susskind mechanism and discuss the implications of our result for AdS/CFT.