Compare actual bytes for native-text and scanned PDFs separately

native-text.pdf contains TEXT-1 and TEXT-2; scanned.pdf contains two image forms. Process them independently rather than treating any re-save as satisfying an upload cap.

1. Record each source’s byte count. Run Compress PDF structure separately and retain both outputs.

2. Using one file manager, record before/after bytes and compute (source−output)/source.

3. Reopen both outputs and check every page, marker, form field and orientation. Also check selectable text in the native-text result.

Each output must retain two pages. Size passes only when actual bytes meet the recipient’s cap. Structural re-saving can decrease, barely change or increase size; no fixed saving percentage is promised here.

Repeated structural compression does not resample a large scan. Retain the source and choose a route with explicit image-resolution and quality controls when necessary. Fewer pages or unreadable fields cannot pass merely because the file is smaller.
