There must be research in the compression, optical recognition, or error correction fields that could improve on Optar's efficiency and/or reliability.
Optar itself is pretty bare-bones from a feature perspective, unpolished, and hasn't been maintained in ages. It has no official GitHub repo (or any other kind of source code repo) that I could find. The released tarball doesn't even have a version number. It only works with the obscure PGM image format (for which you need ImageMagick or GraphicsMagick to convert to/from) instead of something common like PNG.
It does actually work (or did when I tried it a decade or more ago), so by that measure it could be considered good enough, but it's sad that there's been no more modern open source alternative in the many years since it's been released.