My first attempt at a Lora. I trained this on around 12 images of mid sections for Z Image Turbo. It's not 100% by any means and I think I'll try for a V2. It works best at .75 but will still produce weird artifacts etc. the trigger word is belly, but you can add flat belly, toned abs, hip ridges, outie, innie, inbetweenie. I think there's a difference when I use it, but I welcome any and all feedback so I can get better! I think in the future I'll use more images and follow better guidance. I think this was around 1,500 steps.