• @BallShapedManOP
      link
      English
      18 months ago

      Thank you! I’m trying hard to get my own style to show through and not just be a prompt with a standard style.

      I have taken three models and linked them together in Stable Diffusion and have been working on a set of styles and prompts to get an output that fits what I’m looking for. I’ve probably put hundreds of hours into it this year already.

      I can get more detailed and but does that scratch the itch?

      • @[email protected]
        link
        fedilink
        English
        28 months ago

        I’m interested. If I knew more about your process I might be able to suggest some resources I came across recently, if you’re not already making use of them that is.

        • @BallShapedManOP
          link
          English
          18 months ago

          My process: Took an image I liked on DA and put it into Chat GPT and asked it to describe the image in detail as a base prompt. I took that prompt and edited it to take it from a nice painting of a garden scene to this. I then added in some of my style prompts I’ve been working on for this art style and passed it through SD via A1111 a few times until I got what was pretty close to what was in my head.

          I use three models glued together in SD, one for photographs, one for digital art/painting, and one for anime.

          I’m planning on making a lora for the style I am trying to get but I just became a grandfather and have been spending a lot of time with my new granddaughter.

            • @BallShapedManOP
              link
              English
              18 months ago

              I use Loras on occasion but not often. I didn’t here. I find that don’t tend to get me what I’m looking for, but I’m still learning. And I’m not sure what an embedding is.

              • @[email protected]
                link
                fedilink
                English
                28 months ago

                It’s sort of like a crystallized collection of tokens that tell the model something precise you want without having to enter the tokens yourself. I explain it more here.

      • @[email protected]
        link
        fedilink
        English
        28 months ago

        Can you elaborate on how you ‘link’ three models together? Do you import one image into another to act as a starting point or something?

    • @BallShapedManOP
      link
      English
      18 months ago

      That’s probably why I like it!

      And I don’t know why either.