Each garment has a list of visual checks (colour, cut, buttons…). After every image generation, a vision model checks them and gives pass / fail / unsure, each with a short reason. Fails get a targeted fix and everything is re-checked; unsure ones need a human decision.