Refining the sampling geometry of a CT scanner is a standard approach used for reduction of aliasing artifacts in CT images. Although this leads to reduction of the artifacts, the principal problem of aliasing streaks artifacts remains unsolved. A different approach is proposed, which in some special cases can solve the problem very efficiently. It is shown that under certain specific conditions, the sum of images reconstructed from the data collected within different sampling geometries is free of aliasing. These conditions are studied and practical situations where they can be realized are discussed.