Generative image tools have moved fast enough that many designers are using them without a clear workflow — generating images ad hoc, struggling with consistency, and unsure how to handle client ownership questions. This course was built around those specific friction points.
The technical side
Session one covers the mechanics of diffusion models at a conceptual level, then moves into prompt anatomy: how weight, style references, aspect ratios, and negative prompts interact. Participants work with Midjourney and Stable Diffusion side by side to see how the same prompt behaves differently across platforms.
Maintaining visual consistency
One of the harder problems in AI image work is keeping a coherent visual language across a project. Session two addresses this through seed locking, style transfer techniques, and ControlNet for Stable Diffusion. Participants build a small mock brand campaign using only AI-generated assets.
Legal and ethical considerations
Session three steps back from the tools to examine copyright status of AI outputs, disclosure norms with clients, and the ongoing debate around training data sourcing. These are not settled questions, and the session presents the current landscape rather than definitive answers.
Facilitator Renata Voss has worked as an art director for twelve years and began integrating AI tools into client work in early 2023. She brings specific project examples, including cases where AI outputs created unexpected problems.