Large-scale skymapping with COMPTEL using the full survey database presents
challenging problems on account of the complex response and time-variable
background. A new approach which attempts to address some of these problems
is described, in which the information about each observation is preserved
throughout the analysis. In this method, a maximum-entropy algorithm is us
ed to determine image and background simultaneously. Because of the extreme
computing requirements, the method has been implemented on a parallel comp
uter, which brings large gains since the response computation is fully para
llelizable. The zero level is left undetermined in this method. Results usi
ng data from 7 years of COMPTEL data are presented.