Generativní umělá inteligence vizuálního obsahu

The aim of this bachelor thesis is to provide a comprehensive overview of existing methods and techniques of artificial intelligence that enable the generation of visual documents, with an emphasis on the DALL·E and Midjourney models. The theoretical part introduces the general structure and principles of operation of generative models, their training process and mechanisms that allow generating new visual content based on user instructions. The practical part of the work focuses on experimental testing and comparison of the performance of selected models in visual content generation. Not only the quality and relevance of the generated images in relation to the prompts will be examined, but also the effectiveness and limitations of individual systems. Prompt engineering, the process of optimizing user instructions to achieve the best results, is introduced as a key skill for effectively utilizing generative AI models. The results and findings can serve as a guide for future research, development, and applications of generative AI in various fields, especially marketing.