PDFShrink practical guides
How to Reduce Scanned PDF File Size
By PDFShrink Online · Updated
A scan stores a picture of each page, sometimes with a separate OCR text layer. Large page images can make a short document surprisingly heavy. The aim is to remove unnecessary image detail while keeping the information people need to read.
Try Preserve text & vectors in PDFShrink. If the scan is still too large and an image-only copy is acceptable, switch to Smaller image-based PDF. Begin with Balanced and inspect the result before trying a lower setting.
Is the PDF actually a scan?
Zoom into a letter. If its edges become visibly pixelated, that part of the page is likely an image. Try selecting a line of text, too. A successful selection does not prove the page is native text: scanned PDFs can carry an invisible OCR layer over the picture. Mixed documents may contain both scans and pages exported directly from software.
Common causes of large scans include high image dimensions, full-color backgrounds, photographs, paper texture and repeated full-page images. Increasing the JPEG compression or lowering image resolution can help some of these files. A scan already encoded efficiently may gain little, and some rebuilt PDFs become larger.
Page count alone is a poor predictor of size. A three-page document that reuses the same picture can be smaller than three independent scans. Rebuilding each page separately may remove that advantage, which is one reason savings vary.
Choose a setting based on the smallest detail that matters
| Setting | Nominal output resolution | What to check |
|---|---|---|
| Higher quality | 160 dpi | A starting point when fine text or illustrations are important. It is still lossy. |
| Balanced | 120 dpi | Try for ordinary screen reading; inspect small text, stamps and signatures closely. |
| Smallest | 85 dpi | Use only if stronger reduction is needed and the actual result remains legible. |
These are choices for a delivery copy, not universal scanning standards. Very large pages are subject to pixel and dimension caps, so effective output resolution can be lower. Image mode also uses JPEG compression, which can create halos or blocks around letters. More output pixels do not recover detail that was absent from a blurry source.
For the same physical page, moving from 160 dpi to 120 dpi produces about 44% fewer pixels. File size will not necessarily fall by 44%: image content, JPEG settings and PDF overhead also matter.
A repeatable compression workflow
- Save the original separately. If you need the OCR layer, tags, forms or a verified signature, do not replace the original with an image-mode result.
- Open the PDF in PDFShrink Online. Input is limited to one file up to 50 MB. Image mode supports up to 100 pages. This version does not split larger documents.
- Try preserve mode first. This may make little difference to existing image data, but it avoids intentionally flattening the document into pictures.
- If necessary, select “Smaller image-based PDF.” Read and acknowledge the feature-loss notice. Choose Balanced as a trial, or Higher quality if you already know the scan has fine detail.
- Compress, download and open the saved copy. Check the page count and order. Inspect a dense page, the faintest page, and any page with a stamp, diagram, barcode or handwritten note.
- Adjust from the original. If the result is soft, try Higher quality with the original file. If it is clear but too large, try Smallest. Repeatedly compressing the already reduced copy can compound JPEG artifacts.
As an illustration, our one-page scanned test went from about 300.3 KB to 94.9 KB with Balanced, a 68.4% reduction. Our three-page fixture saved only about 6.0% at the same setting. These were test documents, not expected results for every scan.
What happens to OCR, links and forms?
PDFShrink image mode renders each page and rebuilds it from a JPEG image. It does not carry over an OCR layer and does not perform new OCR. Text will not be searchable or selectable; interactive links and forms, bookmarks, accessibility tags and digital signatures are not retained.
If the recipient needs a searchable scan, use a workflow that explicitly preserves its text layer or applies OCR to an appropriate copy afterward. OCR can misread names and numbers, so review its text as well as the visible page. PDFShrink's preserve mode keeps text content, but is not a guarantee of archival, accessibility or signature compliance.
When a fresh scan is the better fix
Compression cannot correct clipped page edges, camera blur or a shadow hiding a number. If you still have the paper, align it, use even lighting, remove empty borders in your scanning workflow and choose a color mode that retains the meaningful content. Grayscale can be suitable for neutral documents; keep color when a colored annotation or seal conveys information.
Those are preparation steps in your scanner or scanning app, not PDFShrink controls. If the receiving organization requires a minimum scan resolution or original color, follow that rule instead of forcing the smallest setting.
Before sharing, read the reduced scan at the size the recipient is likely to use. For barcodes or QR codes, test scanning the result. For important paperwork, legibility and completeness matter more than the compression percentage.