Thats more of a problem with the older image generation models, the current models have largely resolved the hand problem