So do you trust the dims given by the flatten/unbend feature? I sometimes find it can yield dims that are .25 mm/.010" off which can make a world of difference.
Should be unbending them based on the bend parameters applied to that bend which by default is a Neutral Factor of 33%. When I was doing sheetmetal with SE, it was extremely accurate as long as you know what parameters you needed to use which vary based on material and bending method/tooling used.
You must spend the time to get the bend factors right for each material you are bending. This all but requires that you make actual test bends on your machine with the specific materials you are working with, and record the results. Until you take the time to do this, your bend results 'out of the box' with SE sheet metal files will be close, but not perfect.
I have not used the "thin part" version of flattening. But using ordered sheet metal, the flat patterns have been right on.
The real trick of making flat patterns right is using the bend radius that the machine creates. Figure out what R your machine uses and you will be in good shape.
The main part of that learning curve for me was programming a folder. the folding machine includes calibration so that you tell it was bend Radius it will use. In that case, pick one, then use it for both machine and CAD. In my case I needed to input the bend deduction that relates to a bend radius.
Attached is an XLS I created by collecting information about bending sheet metal. The first part of the file is a gineric table for break machine tonage.