Metadata formatting is currently applied during serialization (for example, normalizing fields to Proper Case). The problem is that this ends up overriding manual edits performed in the metadata editor, which a user would expect to be preserved.
A solution might be to apply all metadata transformations at the time the metadata is extracted, but this means changes to metadata formatting settings will only be applied to new downloads.
Another solution might be flagging fields that have been edited manually, but this requires nontrivial bookkeeping.
Metadata formatting is currently applied during serialization (for example, normalizing fields to Proper Case). The problem is that this ends up overriding manual edits performed in the metadata editor, which a user would expect to be preserved.
A solution might be to apply all metadata transformations at the time the metadata is extracted, but this means changes to metadata formatting settings will only be applied to new downloads.
Another solution might be flagging fields that have been edited manually, but this requires nontrivial bookkeeping.