• Models
  • SD
  • SD 2: requires more negative prompts
  • https://github.com/lllyasviel/ControlNet copys the weights of neural network blocks into a "locked" copy and a "trainable" copy.
  • Deep Floyd (part of stability AI) IF can spell https://wandb.ai/geekyrakshit/deepfloyd/reports/A-Gentle-Introduction-to-DeepFloydAI-s-New-Diffusion-Model-IF--VmlldzozNTY3Nzc4
  • https://huggingface.co/docs/diffusers/api/pipelines/dit SD with transferred
• Example repos
  • Lexica Art: SD search engine
  • Playground AI: better than Lexica?
  • Prompt Hero
• Model and embedding repos
  • Huggingface
  • Civitai - uncensored
• Fine tuned sd Models
  • SD1.5
    • protogen x3.4 = protogen 2.2 + robodiffusion + analog diffusion + openjourney + rpg v2 (source)
    • protogen v2.2 = f22 + eldreth lucid + hassanblend + seek art mega + modelshoot + … (source)
    • anything v3 = novelai + … (source)
    • novelai = proprietary paid subscription model ($25/mo) that was leaked, better than waifu diffusion
    • waifu diffusion
    • midjourney: https://www.reddit.com/r/StableDiffusion/comments/z622mp/trained_midjourney_embedding_on_stable_diffusion/
• Dream booth vs embeddings vs LORA
  • Dreambooth
    • https://tryolabs.com/blog/2022/10/25/the-guide-to-fine-tuning-stable-diffusion-with-your-own-images
  • Embeddings
    • SD1.5
      • Most models are avail also as embeddings
      • https://huggingface.co/sd-concepts-library/midjourney-style
    • SD2
      • Cinema768
      • InkPunk
      • Hyperstylized
      • ScaryMonsters v2
      • VikingPunk
      • vray-render
• UIs
  • webui
    • fast-stable-diffusion-webui for colab
  • invokeai
  • diffusionbee (inactive)