Check this out...
The first image was created using text2img with Stable Diffusion v1.4. Final result was created using a mix of img2img loopback from the initial image and inpainting to fix details and create new backgrounds. That was done on SD v1.5, specifically Inkpunk Diffusion to get the style. All images were made using Automatic1111's Stable Diffusion web UI.