Trained on fully synthetic images generated by Flux D using ChatGPT prompts, now condensed to the following:

imagine an abstract poster style image of <something>, with <motifs>, in <my color palette>