For my bachelor’s thesis in MCT, I worked on an individual project where I researched how AI and MCP could be used to automatically create a fully print-ready photozine. Because I am interested in both photography and development, this project was a perfect combination of my technical and creative interests. The goal was to build a proof of concept in which users can upload photos, give instructions in natural language, and let an AI-supported workflow generate a photozine automatically.

My main focus during this project was the technical architecture and the development of the application itself. I worked with multiple MCP servers, each responsible for a specific task in the workflow, such as image analysis, layout support, and PDF generation. I used Python, FastMCP, and the mcp-use framework to connect the different components, and I explored layout technologies such as Typst and Paged.js to make the output suitable for print. I also worked with AI models both in the cloud and locally, depending on the role they had in the system.

A big challenge in this project was that the result not only had to look good, but also had to meet real print requirements. That meant taking into account things like page order, bleed, safe zones, image resolution, and the correct export to a multi-page PDF. Besides the technical implementation, I also reflected on the project with people from the print and AI field, which helped me improve the concept and evaluate its practical value.

The result was a working proof of concept that demonstrates how multiple MCP servers and AI can be combined in one creative workflow. It was a very interesting and valuable project because it allowed me to bring together research, coding, design, and photography into one future-oriented application.

Links:

• GitHub: AI Photozine Generato GitHub