Never said to change backend library, but in this case how about implementing unicode support in fpdf and sending a patch to them?

I'll do it eventually (the modifications are not that huge, however there's a serious need to improve the way ufm is accessed), I was just hoping to hear about something already planned.