Sometimes (For not saying all the times) text instructions are no enough for the AI to 'understand' what you want to do. I use Krita AI Diffusion to draw by hand a base sketch in Krita and then iterate bit by bit with SD (Stable Diffusion) or SDXL over it like a filter. Perhaps if you already have some experience hand drawing it could help you: https://kritaaidiffusion.com/
Or this one. But is more limited: https://www.artbreeder.com/tools/collage
I am not trying to do advertisement haha, i just though this could help you since you mentioned you already draw.