PDF Repair deals with files that have broken structures, that make readers complain, or whose pages will not all open. The flow is not “turn the whole file into images”: qpdf inspects the file and collects warnings, then rewrites it by regenerating the object and compressed streams. Only when qpdf cannot produce a result does the tool try a PDF render engine, rasterising the pages that still display and rebuilding them as a fresh image PDF.

Even when the structure is healthy, a rewritten version is produced. If the tool enters rescue mode, the diagnostic card shows the total page count, the pages rescued and the numbers of the pages that failed to render, so you can judge what is missing.

1. 01Select the original PDFrepair always works on the original bytes of the file in the left-hand list, independent of any rearranged, rotated or deleted pages in the workspace.
2. 02Handle the password and rescue resolutionfor an encrypted file, fill in the “Open password” you know; choose 150, 200 or 300 DPI — the setting only takes effect if rasterised rescue mode is entered.
3. 03Start the diagnosis and repairthe tool runs the qpdf check, then attempts a structural rewrite; on success the result is marked “structure healthy” or “repaired” according to what the check found.
4. 04Read the rescue reportif the structural rewrite fails, page-by-page rendering is attempted automatically; when pages can be recovered you download <original-name>_rescued.pdf, and when nothing can be read the tool reports failure plainly rather than inventing a success file.

Repair order

qpdf inspection and rewrite come first; rasterised rescue starts only if qpdf cannot produce a result, and there is no switch that forces rasterisation directly.

Encrypted files

the correct open password must be known; a wrong password is reported at once, and the repaired output does not keep the original encryption.

What rescue costs

the rasterised mode keeps only the visual content of the pages that render; text layers, links, form fields and other interactive structure are lost, and failed pages do not appear in the result.

Resolution

rescue offers 150 / 200 / 300 DPI; higher is sharper but usually slower, uses more memory and makes a larger file.

What it does not do

it cannot promise to recover every damaged file and will not conjure up data that is already missing; if both qpdf and the render engine cannot read the file, the tool returns a failure.

The qpdf check, the structural rewrite and the rasterised rescue all run locally in the browser; the PDF and its open password are never uploaded to a server.