Oyfo, a creative feedback system — July 2022

Context and intent

This project was my Bachelor Creative Technology graduation project. I started with a personal frustration: I dislike filling in surveys, even though I understand their value.

I partnered with Techniekmuseum Oyfo to explore an alternative way of measuring visitor experience. The goal was to design a feedback method that feels rewarding rather than exhausting, and to test it in a real museum context. The full research process and theoretical background are documented in my thesis.

Concept and outcome

The system invites visitors to give feedback in exchange for a personalized AI generated artwork. Visitors first draw their favorite part of the museum experience on a drawing tablet. After that, they answer five short questions about their visit.

The drawing and questionnaire responses are sent to a Stable Diffusion server (self-hosted server for generative AI). Both inputs influence the final artwork. For example, negative feedback on the atmosphere results in colder color palettes, while positive feedback shifts the artwork toward warmer tones.

All generated artworks are displayed together in a large mosaic, creating a visual overview of collective visitor experiences.

My role

This was an individual graduation project, so I was responsible for the full process: concept development, research, system design, implementation, testing, and evaluation.

I designed the interaction flow, built the technical pipeline connecting the tablet, questionnaire, and AI server, and I evaluated the system for a week inside Oyfo Techniekmuseum.

Reflection and learning

This project was highly motivating for me as I created it from scratch and it felt like my own baby. I often worked late simply because I wanted the system to reach its full potential. I was nominated for best bachelor graduation project and received a final grade of 9.

Testing showed that visitors were genuinely more willing to give feedback through this interaction. At the same time, the project revealed challenges in interpretation. For instance, an artwork dominated by cold colors could be caused by the colors in the original drawing rather than negative feedback. This highlighted the importance of carefully balancing expressive input with data interpretation.