New to Stable Diffusion
Hello All,
I am new to stable diffusion and i am trying to generate images like the attached images that i got from pinterest.
I used chatgpt to generate a prompt based on these images.
i then input the prompt into the text-to-image field and the result was not at all satisfactory, it is not capturing the image quality at all.
is there anyway i can upload the reference image and get a better result?
or is there a way to improve the image results?
Thankyou in advance for your help
Luckily, I know how you could fix this. Are you using a custom script or not? For this, I'll assume you are. Change the Guidance scale, it should be ~6-8(7.5 works the best). You should inference for at least 48 steps, and make sure to include (masterpiece, best quality, great quality:3) in the prompt (heavily weighted buzz words cause the AI to make better images) I hope this helps.