My thoughts on using more than 2 input images with kontext, given current limitations.

→ View original post on X — @fofrai