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: e52c65049b47cc35d60c6aff81b738b19998e218374235a25726f655914c66b1