Spent the past two days optimizing the book system it has been a significant source of performance drops for quite a while due to it relying on widgets. Built tool to merge a shell of the book and then another tool to capture all the pages and package them into a texture array and then inject that into a material instance we then update which texture is being pulled from the array to mimic the functionality of the previous widgets! Works great! of course there is a memory cost, but I think we can handle a couple dozen more megabytes of memory better than a 6ms initialization time.