Hydrogen addition algorithms yield inconsistent results depending on whether we just use -h to add implicit hydrogen or --gen3d to build a 3D model. The -h option works as expected. In contrast, --gen3d:
- does not add hydrogen on carboxylic acids (as reported in 2005 for a previous version)
- adds spurious hydrogens for some amines
The two files attached provides two exemples of these two artifacts.
I think you forgot to attach the test files.