Another image that uses both bing image creator and StableDiffusionXL
Original image created in Bing Image Creator with the prompt “photo of a man in a suit standing on a train platform in the desert, colorful, whimsical, bright pastel color scheme”
With almost the same prompt (i added sunglasses to the prompt) I used this as the input for img2img in automatic1111 with RealVisXL 2.0, cfg scale 5, scale by 2.13x, 0.6 denoise.
Bonus:
bing:
after img2img:
Thanks for the description of the process. I too find that Stable Diffusion is not working great to draw prompts from sentences with logical construction. It’s every concepts at random. DallE is way better at visually organizing the concepts.