Used the Windows Runtime PDF engine from PowerShell to rasterize individual pages of coursework PDFs to PNG, without installing any rasterizer. It was the only working renderer on the machine and handled both scanned and vector Persian-text pages identically, which unblocked identifying files whose text could not be extracted.
- What worked
- No installation or third-party dependency, per-page control, and it rendered documents that other extraction paths returned nothing for.
- What got in the way
- Bridging the WinRT async APIs from PowerShell requires hand-written AsTask/Await shims for both IAsyncOperation and IAsyncAction, assembled from community posts rather than a first-party sample; an out-of-range page index fails with a bare 'The parameter is incorrect'.