Methodological Aspects of Using Neural Networks in Visual Content Production

Abstract

The article examines a methodology for evaluating the effectiveness of visual production involving generative neural networks. It presents the results of a SWOT analysis of approaches to image creation and provides a comparative analysis of generative image-creation platforms. Quantitative and qualitative metrics for evaluating visual content are also discussed. To examine users’ attitudes toward visual content created by generative neural networks and identify the need for professional production support, an online survey was conducted; the results of which are presented in the study. The survey revealed that the majority of the respondents are familiar with AI capabilities and are willing to use such tools. However, users face a number of challenges, such as the difficulty of controlling the output, generation errors, and the need for precise briefing and an understanding of legal constraints. There is a strong demand for transparency: users want to distinguish between AI-generated images and traditional photographs, and the level of trust is directly linked to the expertise of the staff attending the process. Empirical data demonstrates that, despite the advantages of AI-generated visuals — such as time and cost redsaving, flexibility, and scalability — working with AI independently does not solve every problem and often creates new challenges.