Создание маркетинговых материалов при помощи диффузионных моделей

Abstract

The possibility of using diffusion network models to create images for marketing materials is considered. A sample of test images is made, consisting of the results of generating Kandinsky and Stable Diffusion models as well as a human designer. The sample includes four themes (people, medicines, food products, nature) and built images based on them. The questionnaires developed include a five-point assessment of each stimulus image. A survey of 26 respondents is conducted for each of the images. The instructions to the subjects are neutral in relation to the image source. The survey results are subjected to statistical processing using Student’s t-test, which allows one to conclude that using artificial intelligence diffusion models is a promising technology for creating graphic marketing materials. At this level of diffusion model development, the Kandinsky model has proven itself to be most effective. Creating marketing materials using artificial intelligence diffusion models offers companies new opportunities to improve their advertising strategy by intensifying the production of advertising materials. Diffusion models allow one to automate and intensify the process of creating content, reducing the time for its required quality development, which makes it possible to conclude that it is advisable to use diffusion models of artificial intelligence as a tool for producing marketing materials.