directed by shuffur
Viewing sample resized to 41% of original (view original) Loading...
Description

Generated/Assisted with Stable Diffusion using a mix of models. Made through iterations and hand edits. Upscaled with Gigapixel AI.

  • Comments
  • These are so great, I respect you so much for not taking commissions with AI and for having the prompts needed to generate these on your telegram. Thanks again! If it isn't too much to ask, can you tell me what loras you use? I can find the models just fine, it's the loras that are harder to come across.

    Updated

  • Reply
  • |
  • 1
  • pinasheep69 said:
    These are so great, I respect you so much for not taking commissions with AI and for having the prompts needed to generate these on your telegram. Thanks again! If it isn't too much to ask, can you tell me what loras you use? I can find the models just fine, it's the loras that are harder to come across.

    Thanks!

    For Loras, they are custom and I prefer to keep those private, like the custom merges I use. I use different Loras in the different stages in the image for either concepts or a starting point, then I go in with painting by hand and iterate over the images, so the prompts that are available aren't 100% representative of how the image was made!

    My recommendation is don't be afraid to experiment with styles that were made for one kind of model with another (like anime-focused styles used in realistic models or vice-versa.) They may produce eldritch horrors but some may turn out to work surprisingly well together!

  • Reply
  • |
  • 3