Merge PDF
Combine PDFs and reorder pages by dragging. Nothing is uploaded.
Render every page to an image at the DPI you choose.
DPI determines the pixel dimensions of the output. A US Letter page is 8.5×11 inches, so the arithmetic is straightforward.
| DPI | US Letter pixels | A4 pixels | File size (JPG q0.85) | Use for |
|---|---|---|---|---|
| 72 | 612 × 792 | 595 × 842 | ~90 KB | Thumbnails, quick previews |
| 150 | 1275 × 1650 | 1240 × 1754 | ~350 KB | Screen viewing, web, email |
| 200 | 1700 × 2200 | 1654 × 2339 | ~600 KB | OCR input — the sweet spot |
| 300 | 2550 × 3300 | 2480 × 3508 | ~1.3 MB | Print, archival, detailed OCR |
| 600 | 5100 × 6600 | 4960 × 7016 | ~4.5 MB | Large-format print only |
150 DPI covers most needs. Go to 300 only if the image will be printed or fed to OCR where small type matters. 600 DPI quadruples the file size over 300 for detail almost nobody will see.
It depends on what is on the page, and the difference in file size is large.
Choose JPG for pages that are mostly photographs or scans. JPEG compression is designed for continuous-tone content and produces files five to fifteen times smaller than PNG for the same page.
Choose PNG for pages that are mostly text, charts, tables or line art. PNG is lossless, so text edges stay crisp, and its compression handles large areas of flat color efficiently. A text page as PNG is frequently smaller than the same page as JPG at high quality, and it is unambiguously sharper.
The rule of thumb: a scanned photograph goes to JPG, a page of type goes to PNG.
Four reasons account for nearly all of it.
To use a page somewhere that will not accept a PDF — a slide deck, a social post, a product listing, a forum, a chat message.
To run OCR. Most OCR tools take images rather than PDFs, so rendering pages at 200–300 DPI is the standard preparation step for extracting text from a scan.
To flatten a document so it cannot be edited or have its text extracted. Converting to images and rebuilding as a PDF removes the text layer entirely. Note this is a mild deterrent, not security — anyone can run OCR on the result.
To extract a specific figure or chart for reuse in a report, which is faster than screenshotting at whatever resolution your monitor happens to be.
It renders the page — the full composed view including text, vectors, and images together, exactly as the page appears in a viewer.
That is usually what people want. Extracting the raw embedded image objects instead would give you the source photographs at their native resolution, without any text or annotations that sit on top of them, and often split into tiles.
If you specifically need the original embedded images at native resolution rather than a render of the page, that is a different operation and this tool does not do it.
Yes by default, and you can select a page range instead. Each page becomes its own image file.
Convert to images, then use an image editor to stack them. A single tall image is unusual and awkward to view, so it is not offered directly.
At 150 DPI or above, yes, particularly with PNG output. Below 100 DPI small type becomes hard to read.
No. Long documents render page by page with a progress bar. Several hundred pages at 300 DPI is memory-intensive but works on a desktop.
Yes, with the JPG to PDF tool — but the result contains images rather than text, so the text will no longer be selectable or searchable.
Yes. A scanned PDF is already images; this renders and re-encodes them at your chosen resolution.
No. PDF.js runs in this page and rendering uses your own GPU.