• swordsmanluke@programming.dev
      link
      fedilink
      arrow-up
      2
      ·
      7 months ago

      That’s pretty cool! I like the Max Headroom variants. Somehow, I think Mr Headroom in particular would approve of generative AI tech.

      I’m just getting into this realm myself. I’m using ComfyUI, with SDXL 1.0 and the new LCM LoRA, but I’m really struggling to get, e.g. consistent framing. (Like, I’ll ask for “full length photo of X” and get nothing but close-up headshots for a dozen images)

      Any advice or good resources you recommend?

      Either way, very cool work, thanks for sharing!

      • IvanOverdrive
        link
        fedilink
        arrow-up
        2
        ·
        7 months ago

        Frankly, I’ve gotten nothing but shyte from LCM on the initial image. BUT it’s fantastic for upscaling img2img with a denoise of 0.1 and and Ultimate SD Upscale. Not sure how ComfyUI would do it though. I find its UX is too slow on my pc, so I stick to A1111.

        But to solve your problem specifically, learn controlNet. By far my most used extension.

        And for photorealistic images, the PDF at the link as been a godsend: https://promptgeek.gumroad.com/l/photoreal

        YouTube channels: Olivio Sarikas: https://www.youtube.com/@OlivioSarikas Sebastian Kamph: https://www.youtube.com/@sebastiankamph

        BTW, you have a civitai profile?

        • swordsmanluke@programming.dev
          link
          fedilink
          arrow-up
          2
          ·
          7 months ago

          Awesome! Thank you!

          I don’t have a civitai profile. I got into this to try to generate profile pics for a home ttrpg session, but I really enjoyed it and I’ve been having a ton of fun learning and trying to create better images.

          I appreciate the resources! I know what my evenings are gonna be for awhile! 😁