This model is inspired by <a target="_blank" rel="ugc" href="https://github.com/wangkaihong/hadm">HADM</a> I took 1000 images from their published dataset, ~350 were hand selected to have very high distortion of human limbs, and another 650 were randomly selected from sdxl subset. Then trained with civitai LORA training tool. You should use <code>human-artefact</code> in negative prompt (sorry for the misspell I was confused by the correction mechanism)

Files with SHA256: c1b12fc319bd73dd5e1c1c0a94306d91171dd963ca4e715f3690c0b34fa7914b