Drop PDF files here to combine
or click to browse · Up to 30 files · Drag rows to reorder
Drop two or more PDFs, drag to set the order, and download a single combined file. Lossless merge — original page quality preserved.
Drop PDF files here to combine
or click to browse · Up to 30 files · Drag rows to reorder
VeloTools Merge PDF never sends document bytes to a server. After you pick files, the browser reads them via the File API into an ArrayBuffer; pdf-lib then copies page trees (/Pages, /Page) into a new document in tab memory. The network stack is not involved: no fetch, WebSocket, or hidden XHR carries PDF body data.
This is an architectural guarantee, not a privacy policy claim. With Wi-Fi off after the page loads, merge still works — proof that computation is local. For NDAs, medical, and financial PDFs, no third party can access file content.
| Max files per operation | 30 PDFs |
|---|---|
| Max file size | No hard cap — limited by device RAM (typically 100–500 MB total) |
| Supported PDF versions | PDF 1.4 – 2.0 (Flate/DCT stream objects) |
| Processing mode | Lossless copyPages — no JPEG rasterization |
| RAM usage | ~1.5–3x total source size during the operation |
| Network | 0 bytes of PDF content leave the browser |
| Encrypted PDFs | Require unlock first |
Merge in VeloTools is not pasting images together — it is importing the PDF object graph. Each source page is a node linking to /Resources: /XObject (raster and vector images), /Font (embedded subsets), /ColorSpace, and optional /Metadata.
Images and color profiles. With lossless page copy, JPEG/JPX/CCITT streams inside the PDF are not decoded or re-encoded. A photo with an ICC profile (sRGB, ISO Coated v2, etc.) keeps that profile on the same XObject. Merge does not unify color spaces: document A (sRGB) plus document B (CMYK) yields a combined PDF with two independent color contexts — normal concatenation behavior, not quality loss.
Fonts and subset merge. Each source PDF embeds a subset of glyphs used on its pages. pdf-lib copies each file's font objects into the output catalog. If Arial appears as separate subsets in two files, the merged PDF contains two font objects, not one merged subset. File size is roughly the sum of sources plus xref/trailer overhead; automatic font deduplication is not performed — a deliberate trade-off for exact page fidelity.
Quality vs compress. Merge does not shrink or degrade images. For a smaller file after merging, use Compress PDF (rasterize + JPEG) — a separate lossy operation.
Bookmarks and metadata. /Outlines (bookmarks) from multiple files are not merged automatically: the output keeps page structure without a combined table of contents. /Info metadata is rewritten with a minimal set when building the new file.