Like my work? Consider <a target="_blank" rel="ugc" href="https://ko-fi.com/avenka">donating</a>.I've used many different breast sizing LoRA's in the past and I really just wanted to give it a try. Mainly because I'm rarely happy with the range of most of them unless they are dedicated to the size I am trying to accomplish. (like flat chest) I've also been trying to refine my LoRA making skills so this was another opportunity for that.I wish you the breast day ever!MBH-TT &amp; MBHU-TT2 'and now Pony' Check "About this version" for small notes on the added triggers for each version.This has become the combination of MyBreastHelper and <a target="_blank" rel="ugc" href="https://civarchive.com/models/104436/asset-manager">ASSet Manager</a>. There is so much going on in this LoRA now that I can't say it all in a short summary.Things to remember.<ol><li>Order matters! you might have the right mix of prompts but in the wrong order to get the result you want.</li><li>Negative prompt what you don't want. Single triggers work pretty well but each one has a tendency to influence other parts of the body.</li><li>Use corresponding "common" body tags when needed in both positive and negative prompts. This helps when you're trying for a contradiction in say breast size vs. body type. Skinny woman with big breasts or a chubby woman with small breasts.</li><li>Use weights to your advantage. Sometimes you almost get what you want but need a small tweak. don't be afraid to use a trigger with a low weight like (saggy breasts:0.2)</li></ol>As always please share those beautiful creations.

MyBreastHelper

Nobody asked for it, and the 36 hours of training was painful.The results however, seem to be worth it. Trained with Pony V6 and primarily tested with <a rel="ugc" href="https://civitai.com/models/341433?modelVersionId=511677">Everclear</a>. The goal was not to recreate the wheel. Pony handles "most" things contained in this version with satisfactory results, but sometimes there's just a little lacking, like the problem we had with 1.5 and 'flat chested' rendering something closer to 'small breasts'. I used (most of) the training data used in MBHU-TT2FRS but removed tattoos and pregnancy data along with a couple other things I'm less worried about. If you want to refer to TT2FRS for tags, the data that was included uses the same exact files, except with Pony, the results are much cleaner and things like 'small areola' and 'inverted nipples' work much easier.

Files with SHA256: 1cf27648e571f7ffc0cf6d7a851973d6c62f1af0f5bcef7b16ffd6b259f89d07