This LoRA was trained on full-body images, which tend to cause distortion in the eyes and hands. However, including 'upper body' or 'chibi' in the prompt can improve stability.Additionally, since the training was done using images with a white background, the generated images are more likely to have a white background.Please share your creations here!

Files with SHA256: 081b36a1cff83469924c95ef8167cf0d53f6c146fbf847fd7498f42f623e1c09