The following file can be successfully decoded by OpenJPEG but fails to be decoded by Luratech JPX decoder library: https://github.com/uclouvain/openjpeg-data/blob/master/input/nonregression/issue228.j2k After looking into the JPEG2000 spec and looking at the hexdump of the file it seems likely that the CME marker for the comment is a culprit since it accidentally omits the Rcme field and instead starts the Ccmd field immediately after the Lcme field. OpenJPEG never even tries to parse any of the fields but the specification allows for a validation of the Rcme-field value to be 0 or 1 (depending on if it is a binary or latin1 text comment). I believe that a decoder library should make the comment type available along with the comment data, but it should assume that the comment is binary unless explicitly indicated to be latin1 text.
We no longer support the Luratech decoder.