Methodology of using generative neural networks for developing advertising video design

The aim of this study was to develop and test a methodology for creating an advertising video using generative digital tools. In the course of the work done, a solution was created that allows for an organized and consistent production of video content based on generative technologies, without using a traditional filming base. The study included both theoretical and applied components. A review of existing platforms for generating images and video scenes, both foreign and Russian, was conducted: Runway ML, Pika, Kling, Kaiber, Stable Diffusion, Sora and Kandinsky. International tools can be used when quality and expressiveness are a priority, but they require a higher level of technical training, and are also limited in matters of animation style. Domestic solutions are easy to use and stable in the local context, but require additional adjustments within the framework of available stylization. The result was a set of advertising videos created using various generative platforms. Comparative analysis showed significant differences between the platforms in both visualization accuracy and the ability to follow a given style and scene structure. The methodology developed within the framework of the study covers the entire cycle - from setting the task and forming a scenario description to generating scenes and final editing. The generative approach makes it possible to significantly reduce costs without a sharp loss of quality, and at the same time allows maintaining a high level of stylistic flexibility and scalability. This makes the technology especially promising for use in educational and social advertising, where each production unit must be as rational as possible in terms of costs. The material will be useful for specialists in the field of media production, digital marketing and creative design, offering a scientifically based approach to the integration of neural network tools into the process of creating video advertising.