Hi all, Nasser pointed out one issue with conversion of Unicode characters inside \Picture* \EndPicture command [1]. The problem is that the Unicode values are stored in \special commands followed with bogus "x" characters. I've provided detailed explanation in my answer [2], so I will not repeat it here. The point is that I've created a Lua module for HTF file parsing (it is included in the answer). It can be used in the way I explained on TeX.sx, but it can be also used in the other way - for conversion of 8-bit font glyphs to Unicode.
This was a limitation I've found in my experiments with producing HTML directly from LuaTeX using Lua callbacks, it was possible to get correct characters only for OpenType fonts. This was a huge drawback especially for math. We should add the functionality provided in the answer to tex4ht by default, but some testing before would be nice. Best regards, Michal [1] https://tex.stackexchange.com/q/448564/2891 [2] https://tex.stackexchange.com/a/449022/2891