In classrooms where computers are used as tools for text-making, images and photographs from e.g. Google, here called "prefabricated images", can be selected and copied into texts and combined with writing. In this article children's use of prefabricated images as resources for personal texts is investigated with specific focus on cohesion between the modes of image and writing. When prefabricated images occur in combination with writing about a personal experience the specific motifs shown in the image are unrelated to the text-maker, but the results of this study show that cohesion may still be obtained, for example via colour, naturalistic modality or decontextualization of the motif in the image via a close-up or a distant perspective. Copying and recontextualization of photographs are common not only in schools but also in professional settings as image banks supply images to, for example, news editors and journalists, and contemporary text creation is often characterized by "representation-as-selection" (Adami and Kress, 2010). The ability to obtain cohesion across modes can be regarded as a defining feature of success in multimodal text-making (Wyatt-Smith and Kimber, 2009), and also for the interpretation of contemporary texts.