This LoRA name is nippleclamp_chain_half to remind users of both the keyword and the recommended LoRA weight. I would set the LoRA weight to 0.50 to start and lower or increase as needed.Note: This LoRA was mostly trained on zoomed in photos of nippleclamps in order to avoid it affecting generated faces. This also means it has a habit of generating faceless zoomed in breast shots.However, tweaking the prompts to include details or aspects of the face will help it include the face. Try prompting for things such as eyes, eyebrows, or hair style. Prompts like "Headshot face portrait" in the positive also seems to help. Putting "cropped, out of frame" in negatives also seems to help. With the right tweaks, it should include the face in the majority of generated photos.Limitations: Doesn't do complex angles well, if at all. (I couldn't find enough photos of complex angles for training).

Files with SHA256: c7ef2938a3a270a433f2b0dbe4a3be49cddf07649681a9fe29fb31667d309403