V1 updated with new data set and improved captioning, may not get accurate finer details but overall quality and stability improved.<s>Test Version: Three girls bodysitting: TEST as I had a 3 picture dataset. I will be updating once I produce a sound dataset and train on site again. It is quite unstable adding extra body parts, if you have any good results, please share and I will add to dataset.</s><s>CFG: 10 recommended</s>

Files with SHA256: c5c1c6ccf29f9ebfd8a1c6500373fcdc87d14fa59916db08458d37f942a5298d