Hello and welcome to PixInsight Forum.
Yes, you're right. The problem isn't with PixInsight itself, which fully supports Unicode strings, but with some third-party support libraries on which PixInsight depends, namely libtiff, fitsio, libjpeg and jasper. These libraries don't seem to support the full multi-byte character set for file names on Windows. For example, I can read and write FITS files with accentuated Latin vocals and "ñ" and "Ñ" characters in their names, but not with Cyrillic or Arabic characters for example.
As a test, try with the PNG and BMP formats. You should be able to use Cyrillic with PNG and BMP files without problems.
Note that this is a Windows-only problem. On Linux and Mac OS X, the operating system uses UTF-8 as the default file name encoding system. This has the advantage that even legacy libraries that rely on 8-bit character sets to store file names can work with Unicode transparently.
I'll try to fix this problem in the next version, by hacking the corresponding libraries as necessary. Sorry for the inconvenience.