I wonder if text written using chinese characters is more compatible with such vision centric compression than latin text.