Recommend to generated with the Lora value at .2 so that the pose is not to far out of the Lora's range then image to image with a lora value of 0.5-0.8 (See example images for optimal config scale and negitives)
FDXLAnusHelper.safetensors