After some investigations, it seems adding unicode support to FPDF isn't impossible, but the lib isn't really a model of best practice in PHP.

I started coding a new PDF lib with Unicode support, and added a FPDF-compatible interface, I however noticed that HTML2PDF is directly using private elements from FPDF lib, including direct reference access to variables, which prevents correct overloading of variables for the compatibility class.

I'll have to add some patches to html2pdf for this to work (especially in some of the methods in "mypdf" part) and add handling of multiple toUnicode tables in pdf when unicode range grows bigger than what is initially expected.