This paper presents a method that allows for reliable automated acquisition of images in fluorescence microscopy. Specifically, if an image contains more objects that do not lie precisely in one z-plane, traditional methods necessarily fail due to their principle of operation. We take a different approach by reducing the original problem to a set of tasks which can already be easily solved. Namely, we divide the original image into a few subparts and process each of them individually. The results obtained in such a way enable discovering hidden z-planes with rich information content. Our approach therefore outperforms other acquisition methods including manual one. A large part of the paper is devoted to practical examples and suggestions.