On the Limitations of Vision-Language Models in Understanding Image Transforms Paper β’ 2503.09837 β’ Published Mar 12 β’ 10 β’ 2