I was trying to get jaggy lines but instead get paranoidal schizophrenia from every picture i generate with this LORA
Try make this generations as shitty as you can, use worst sampler, make 20 steps, make upscale in extras or choise worst upscaler, can even add worst quality, bad quality to positive promt
important use noise in negative prompt because of my poor choice of scraping of speech clouds and set low cfg scale to get more jaggy and shitty(in good way) results