Just a thought can someone with experience in creating embeddings or checkpoints try to train a model with high-res pictures 2048px or more of hands, fingers and eyes to solve that problem of bad mutated hands and ugly eyes once and for all? I think we have now enough models of vaginas, tits and dicks. 😂😂😂 LOL
Bad eyes, fingers and poses need a multi-faceted approach!
We can use a good VAE file, like the one for the Anything v3 model, which does a good job of fine-tuning eyes, and hand anatomy.
We can use this Textual Inversion (Huggingface Link Nerfgun3/bad-prompt) which is trained on bad hand anatomy, bad posture, bad poses, etc. You call it in your negative prompt.
And lastly, your choice of model really affects this. Some models just inherently produce better eyes and hands! My StolenDreams mix or Elldreth's Vivid mix produce good hands and eyes - without any VAE, or the bad-prompt embed.
Hey Ally, thanks very much will give it a try next time. For the picture below I had to use inpaint a couple of times until it looked "okay-ish" and in the end I just blurred the hands in Photoshop. LOL