>>53182852>>53182883>>53182818OR here, thanks a bunch anon. This is really cool.
>in addition to the image itself the ai accepts "prompts". the same way you would create the image from text you can guide the ai when making an image from a referenceOut of curiosity, how far can the prompt deviate from the image itself?
I'm sure it'd be a mess if you told the AI an image was something it wasn't, but can you make slight changes with acceptable results? Like changing the candles to sparklers for example.
