Make some fine-ass moms. Well, try to - this is an early model.This is an early LoRA that adds saggy breasts, some fupa/tummy depth, and c-section scars. Still working on training and variety, results may vary based on the core model you use this with!Use in a 0.5 - 2.0 scale - I tend to find 1.2 to 1.5 gets the best results. V1 is narrow, V2 has more variety, but possible for more concept slip.Pretty Good / OK (Needs Work) / doesn't always work well<h2 id="version-2-details-(flux-only-for-now)">Version 2 Details (Flux Only for now)</h2>Trained on these concepts, pair them like "small saggy breasts"<ul><li>breasts: tiny, small, saggy, empty, perky,</li><li>nipples: small, medium, oval, round, dark, light</li><li>pubic hair: shaved, groomed, stubble, strip, triangle</li><li>fupa: small, medium, (pair this with "soft build" or "fatty" desciptors - doesn't work well on skinny)</li><li>stretch marks: on stomach, on thighs</li><li>hips: wide</li><li>build: skinny, soft</li><li>cellulite: on legs, on butt</li><li>venus belly - some base models treat this as pregnant?</li><li>dimples of venus</li><li>c-section scar</li><li>soft pubic area</li></ul><h2 id="version-1-details">Version 1 Details</h2>Some things to try:<ul><li>mombod - will help influence the woman to have some/all of these things below:</li><li>tiny/small/medium/large saggy breasts (use a size, or not!)</li><li>empty breasts</li><li>small/medium/large nipples</li><li>round butt / saggy butt</li><li>wide hips</li><li>cellulite</li><li>small/medium/large fupa (or just fupa)</li><li>soft pubic area (for a fatty/soft looking vulva)</li><li>c-section scar</li></ul>Small dataset, so it doesn't always play nice to specify these all individually. Breasts/fupa tend to follow the weight of the woman (larger women will have more pronounce fupa and slightly heavier breasts) - again more to train!Used the Civ Trainer, so YMMV with this LoRA. Couldn't figure out how to put it in Early Access / request buzz - so enjoy. Tips are nice.<h2 id="flux-2-klein">Flux 2 Klein</h2>Use this Lora at 0.5 up to 2.0, outside of that, results can go wild.<h2 id="sdxl">SDXL</h2>I don't run these models, you just got a LoRA because I accidentally generated that one...

Files with SHA256: f5f1b4588ba48f753dce22d56e37af87a67979b5040b822bf5090138160fe97a