When your Norddeutsche Landesbank statement arrives as a scan or a photo of the page rather than a clean export, the numbers on it are just pixels until they are extracted. Statements that arrive as a scan or a photograph introduce a different class of risk: a smudge on the glass, a fold across a digit, or uneven lighting that turns a 3 into an 8. Convert·Into reads these images with OCR, then applies the same running-balance verification it uses for every file. Each line's calculated balance is compared against the statement's own printed closing figure. When a row does not add up because the OCR misread a character, that row is flagged for review rather than letting a corrupted number travel unnoticed into the ledger. The rest of the file proceeds normally.
If a transaction description on your Norddeutsche Landesbank statement is long enough to wrap across two or three lines, it still needs to land in a single row once converted. When a transaction description wraps across two or three lines inside a statement cell, it can fool a generic parser into treating the continuation as a phantom entry: a zero-amount ghost row, a duplicated date, or a fragment of text that lands in the amount column. Convert·Into detects multi-line descriptions and re-joins the pieces into a single transaction, so one long narrative line stays one row in the output. No phantom rows appear in the final file.