Soft or collinear photon emission potentially poses numerical problems in t
he phase-space integration of radiative processes. In this paper, a general
subtraction formalism is presented that removes such singularities from th
e integrand of the numerical integration and adds back the analytically int
egrated contributions that have been subtracted. The method is a generaliza
tion of the dipole formalism of Catani and Seymour, which was formulated fo
r NLO QCD processes with massless unpolarized particles. The presented form
alism allows for arbitrary mass and helicity configurations in processes wi
th charged fermions and any other neutral particles. Particular attention i
s paid to the limit of small fermion masses, in which collinear singulariti
es cause potentially large corrections. The actual application and the effi
ciency of the formalism are demonstrated by the discussion of photonic corr
ections to the processes gamma gamma --> <t(t)over bar>(gamma), e(-)y --> e
(-) gamma(gamma), and mu(+) mu(-) --> nu(e)<(nu)over bar>(e)(gamma) (C) 200
0 Elsevier Science B.V. All rights reserved.