Do any of these allow you to upload a picture and generate other pictures in the same ballpark as the one you uploaded, or do you have to prompt it? I’m not having much luck with img2img at the moment.
Ryan Rivera
adding several embeddings w style and negative comes up w cool stuff too
>Do any of these allow you to upload a picture and generate other pictures in the same ballpark as the one you uploaded yes, any of the ones with img2img >I’m not having much luck with img2img ???
Justin Brooks
top left looks pretty cool still
Jason Harris
what aspect of your prompt is leading to the huge circular formations in the landscape? looks nice
>generate other pictures in the same ballpark as the one you uploaded are you describing the image and using the denoising level? that should be doing it, especially as you increase the denoising.
it's in the previous thread. unfortunately there are only four or five prompts and a few of them seem to expect a different model. the very first prompt in the thread is bunny knight girl
It's almost overwhelming the amount of freedom to generate completely insane landscapes. Actually it is literally overwhelming so have this nice beach, my most crazy inventive output yet
For the people playing with clothing/design of characters, I stumbled on something. If you already knew then ignore me, but sharing with the thread I was trying to use hex maniac's danbooru tag (using WD+LD merger at 70/30) as part of a prompt to encourage black hairs, but it seems the prompt picks this up also as "hex" as in hexagon or geometry, and often changes up clothing into a strange cyber-esque style with weird geometric shapes. If you're into clothing stylization with the generator, try out some "hex" alike keywords or even hell, try various geometric shapes and see what pops up. It's pretty fun, sometimes it merges weirdly but the end results are almost always interesting with a good prompt.