>>1477842img2text = you input an image and it tells you what the image contains
text2img = you input a text description and it creates an image to match the text
img2img = you input an image and it creates a new image from it
Except, you don't just input an image, you also input a description.
If you input a shitty mspaint landscape and the description "photograph of a beautiful landscape" it fits a beautiful landscape to your shitty drawing.
What is done here is inputting Pepe but telling the AI: "actually, this is a landscape". That way, you get a landscape that looks like Pepe.
