ty for your feedback and I fully agree! There are three limitation at the moment: limited corpus of images to choose from (i predefined the set), no way to incorporate a brand assets (jev can't parse images), page text is based on what a user enters. I believe solving solving 1 using an image search api would be most useful.