AI Text-to-Image Experiments

Poetry to Image
The emergent text-to-image artificial intelligence apps offer visual artists new tools to play with. The first image below was generated from the simple prompt of the song lyric, “Blackbird singing in the dead of night.” The image generators produce a selection of images each a different interpretation of the lyric, poem, or tagline that was entered. The other three images in the first gallery below were generated from poems. I added the words to the final images using Photoshop.

Digital Collages mostly using AI generated images
I have also found them useful for making my digital collages. These have been composed of found or original images using traditional methods. As each collage progresses I look for increasingly specific images to make it work. Online searches can be time-consuming and often the images that best meet my goals have been under copyright. The generators offer an alternative source, although that is not to say that the process is always successful. There are also times when I fuss with the prompt for hours without getting any thing useful. The four images in the second gallery are collages made almost entirely with generated images from prompts employed using a wide range of modifiers to specify style, coloring, mood, perspective, etc.

Various AI generated images
AI generators offer the options of prescribing a particular era or an artist’s or an art movement’s style. The first two in the gallery below are “in the style of Rene Magritte. The third image is an ad agency set in the 1960s. Others are less specific in terms of style. These renderings reflect what Midjourney AI has learned from prompts it is learning to interpret. This process is ongoing, becoming more refined by user choices.

Scroll to Top