Quality means being clear.
Our tools do real document work. They are not a promise of perfect fidelity for every PDF. Always retain your original and inspect the output before printing, submitting or sharing it.
Browser safety limits, not a subscription gate
| Each input file | 50 MB |
|---|---|
| Combined input batch | 100 MB, up to 30 files |
| PDF page operations | Up to 500 pages (including merged output) |
| PDF image rendering | 100 pages; 8 million pixels per page; 100 MB encoded-image budget |
| JPG/PNG input | Up to 25 million decoded pixels per image |
| Daily task quota | None |
A browser can still run out of memory below these limits, especially on phones. Try fewer files, fewer pages or lower rendering resolution if the operation fails. Cancel takes effect at safe processing boundaries; a browser engine operation already in progress may need to finish first.
Preserving content versus preserving everything
Merge, split, extract, delete and reorder copy PDF page content rather than rasterizing it. Document-level bookmarks and attachments are not transferred. Internal links and unusual annotations may need review. Interactive forms are rejected for page-copy operations to avoid silently losing the canonical form structure.
PDFs with detected signature fields or certification permissions are rejected for editing. Password-protected and XFA files are not supported. Removing a password or bypassing document restrictions is not a feature.
Compression without magic percentages
Lossless repack rewrites supported PDF structures without downsampling their images. Many already-optimized PDFs will not get smaller. Image-based compression explicitly replaces pages with JPG pictures, losing searchable text, links, annotations and accessibility structure. It requires your acknowledgment. If a compression mode does not reduce file size, the original is returned unchanged.
Output checks
Generated PDFs are reopened and their page counts checked before download. These checks do not establish complete visual fidelity, accessibility, PDF/A compliance or suitability for a legal submission. First-page previews help spot obvious problems but are not a substitute for reviewing the entire file.
Specific limitations
- PDF to text extracts embedded text; it does not OCR scans or reconstruct tables.
- PDF to images renders whole pages. It does not extract original embedded pictures.
- Resize produces static pages. Forms, links, annotations, bookmarks and attachments are not carried over. Pages with rotation metadata are currently rejected by this tool.
- Watermarks use a built-in Latin font. They are not secure redaction, encryption or copy protection.
- Page numbers can overlap an existing footer and follow the page coordinate system, including existing rotation.
- Metadata editing changes standard properties only, not all XMP or identifying information.
- JPG/PNG conversion embeds supported images. Check EXIF orientation and color appearance before printing.
Photo and signature tools
Photo/signature resizing accepts JPG and PNG up to 12 MB and 24 million encoded pixels, with no side above 16,384 pixels. Encoded dimensions are checked before decoding. Output is capped at 4,096 pixels per side and 12 million pixels. Current browsers with Canvas and ImageBitmap are required. Output dimensions and byte size are checked after encoding. JPEG compression is lossy; PNG may not meet a small file-size limit. Automatic dimension reduction happens only with your permission. Both formats use a white background in the resizer. The standalone signature maker exports transparent PNG. Identity-photo compliance and acceptance by an application portal are not guaranteed.
Sign PDF places a signature image on one selected page. It is not identity verification, a cryptographic signature, DSC or Aadhaar eSign. Interactive forms, existing digital signatures, rotation metadata and nonstandard crop boxes are rejected by this signing workflow. Only use signatures you are authorized to apply.
Not advertised as working
OCR, PDF-to-Word/Excel/PowerPoint, editing existing paragraph text, secure redaction, cryptographic signing, password protection and unlocking are not implemented in this release. They require separate engines, licensing and quality evaluation. There are no pretend buttons for them.
Open-source foundations
Document manipulation uses pdf-lib (MIT); rendering and text extraction use Mozilla PDF.js (Apache-2.0); ZIP packaging uses fflate (MIT). Full upstream texts: pdf-lib, PDF.js, fflate, Lucide, React, React DOM and Next.js. See the repository notices and supplied asset licenses for additional details.