Comment by Legend2440

Vision models are typically bad at small details. If there’s too much stuff going on at once, they can’t focus on the entire image.