Since the start of this project, I have always wanted to attach an image to the generated text. My hesitation was that, at the time, I didn’t have a way to directly couple the text with AI image generation. So, my solution was to let the system pick keywords from the text, then use those as search terms in an online image database. The system also provided a number. For example, keywords might be “spirit,” “night,” “Earth,” “light,” “sleep,” and “dream,” with the target image number being 26. To make this work, I would enter these key terms into the search and select the 26th image on the page. That image would get combined with the text to create a postcard. It’s low-tech and manual, but I liked having that direct connection between text and image.

However, in the past few months, a new generation of online AI-based image generation tools have become available. These tools take text prompts directly and then use them to create an image. This seems like it is a better match for my project. I have experimented with a couple of these and like the results. There’s just something satisfying about taking the output of an AI and using it as the input for a second AI.