optchar and Tried to invert singular LMatrix3

I don’t see anything wrong in the egg-optchar output. I assume the scale was correct before you ran egg-optchar? If you post your egg files, I can try it myself and see if I can understand why it’s breaking the scale.

As to the -flag options, you are passing joint names to -flag, but it is expecting geometry names instead. Presumably your pupil geometry has some name in the modeling package; use that name.

David