Guide

If sources include scans, run OCR first to give each file a searchable text layer — the merged result will then be fully searchable. For very large merges, compress each source individually first to reduce the final size.

Decide on a naming convention first: range splits produce names like "original-pages-1-3.pdf"; bookmark splits use bookmark titles. To extract a non-contiguous selection, the page-range mode is more flexible than per-N splitting.

High DPI (e.g., 300) consumes more memory — for PDFs over 100 pages, batch the work or lower the DPI. If you only need the original images embedded inside the PDF, the Extract Images tool yields the raw source assets at original resolution, which is more faithful than a re-render.

For contract scanning, use a "document scanner" app (or your phone's built-in scanner) for perspective correction and cropping first — it beats raw photos. For A4-sized submissions, pick fixed A4 + center, not "Fit image".

If the output columns look misordered, copy the same passage in a regular PDF reader and check — the ordering quirk often comes from the PDF itself, not the tool. For strict table preservation, consider a dedicated PDF→Excel tool.

Embedded image dimensions can differ from what's visible — PDFs often display large images at reduced size, so the extracted source is often higher-res than expected. If too many small icons clutter the output, raise the size threshold.

For hundreds of pages of scanned content, "Rotate whole document" is fastest. If only a few pages are off, "Per-page selection" gives precision without touching the rest.

If you want to keep just a few pages, use Extract Pages instead — picking what to keep is faster than excluding the rest. Before destructive deletes, run Split first as a quick backup.

For straightforward sequential extraction, enter ranges like 1-5,10-15. If you want to reorder (e.g., put page 10 first), tick first then drag thumbnails. Batch ranges beats clicking pages one by one for bulk extractions.

If the source came out wholly wrong-order, "Reverse all" or "Sort by filename" beats manual dragging. Always keep an original-file backup elsewhere before bulk reorganization.

The frame set on the first page is used as a template across the selected scope. If page-by-page content varies (charts that stick out on some pages), "Auto-detect content bounds" is more reliable. For pre-print crops, leave a margin for printer-safe area — don't trim all the way to content edge.

For multiple blank pages, set "Pages to insert" instead of clicking + repeatedly. To prepend a cover, set the insertion point to "Before page 1".

For anti-removal robustness: 45° diagonal + tile-fill + semi-transparent (30-50%) is the gold standard. If you'll add a signature image on top, watermark first then run the Signature tool to overlay the signature.

Standard academic recipe: number abstract/TOC with i,ii,iii, then run the tool again for the body with 1,2,3 starting fresh. The tool supports two-pass execution for exactly this.

A transparent PNG at 200×80 pixels strikes the best signature balance — too large stands out, too small loses fidelity. After signing a contract, run Encrypt to set "open but no edit" permissions to deter tampering.

Set both an open password (gates viewing) and an owner password (refines permissions like print/copy). Strong passwords are at least 12 mixed characters — weak passwords negate the encryption.

Before decrypting, confirm what you actually need — often the goal is to lift print/copy restrictions, not the open password. In that case, Remove Password is more precise.

If the PDF has both an open password and permission restrictions, do it in two steps: first Decrypt to lift the view block, then Remove Password to lift permissions. Or just use Decrypt — full decryption removes both at once.

Check the file's makeup first: pure-text PDFs may actually grow when rasterized — convert to Word format instead. High-image PDFs compress most dramatically. To preserve search functionality, pick "Preserve selectable text layer" + "Balanced" mode.

Scan quality is the single biggest factor: 1) Scan at 300 DPI minimum; 2) Keep pages straight (no skew); 3) Avoid shadows and glare; 4) For black-and-white documents, binarize first for better contrast. Use the Chinese+English mixed model for documents that mix languages — single-language models perform poorly on the other script.