Compress PDF makes a PDF smaller by re-encoding the images stored inside it, leaving the text layer completely untouched. A scanned 20 MB report that bounces off an email attachment limit can drop under 3 MB while every word stays selectable, searchable and sharp at full zoom. DocPivot offers five compression levels, all five free, and handles up to 20 files in a single job. The results panel reports the exact byte savings per file, so you see what you traded before you download anything.
What DocPivot Compress PDF Does
DocPivot Compress PDF shrinks a PDF by downsampling and re-encoding only the image objects it contains, then rebuilding the document around them. Most of the weight in a large PDF is images: scanned pages, product photos, screenshots, embedded charts. Text, fonts, bookmarks, form fields and metadata are a rounding error by comparison, so touching them buys almost nothing and costs a great deal.
The distinction matters more than it sounds. A lot of compressors reach their headline numbers by flattening each page into a single picture. The file gets small, but the text stops being text. You can no longer search it, copy a clause out of it, or run it through a screen reader, and it turns blurry the moment someone zooms past 150%. A compressed contract that cannot be searched is not a compressed contract, it is a photograph of one. DocPivot never takes that shortcut, which is why a compressed invoice still returns hits when you search for the PO number.
Who uses it
Job applicants squeezing a CV under a 2 MB portal limit, students uploading assignments to university systems with hard caps, accountants emailing scanned receipts, real estate agents sending property packs, and support teams attaching manuals to tickets all reach for a PDF compressor. Anyone who has hit "file too large" on a government upload form has needed this tool, usually urgently.
The Five Compression Levels
Five levels give you a real gradient instead of a coarse choice between "barely" and "ruined." Each level pairs a maximum longest-side resolution with a JPEG quality value, and the two move together deliberately. Quality 30 at full resolution looks noticeably worse than half resolution at quality 70 for the same number of bytes, because the eye forgives lost pixels far more readily than it forgives compression artifacts.
| Level | Longest side | Quality | Result on an 825 KB photo PDF |
|---|---|---|---|
| Lossless | Unchanged | Not re-encoded | 844,014 bytes (100%) |
| Light | 2400 px | 85 | 422,163 bytes (50.0%) |
| Balanced (default) | 1600 px | 75 | 297,468 bytes (35.2%) |
| Strong | 1200 px | 60 | 180,606 bytes (21.4%) |
| Extreme | 900 px | 45 | 117,889 bytes (14.0%) |
Balanced is the default because it clears most upload limits without visible degradation on screen. Light suits documents that will be printed. Strong and Extreme are for hard caps, where getting under the limit matters more than pixel fidelity, and Extreme reaches an 86% reduction on image-heavy files. Lossless exists for the case where you want structural cleanup with no image re-encoding at all.
Key Benefits of DocPivot Compress PDF
- Every level is free. All five compression settings work without an account, including the strongest one. Smallpdf's own compression page lists Moderate and Strong as Pro features, leaving Basic as the only free level.
- Text survives intact. Fonts stay embedded and real, so compressed documents remain searchable, copyable and accessible to screen readers.
- Predictable output. Published resolution and quality figures for each level mean you can pick the right setting on the first attempt instead of guessing.
- Batch of 20. Twenty files per job, downloaded as a single ZIP, which turns an afternoon of one-at-a-time uploads into one action.
- It refuses to make things worse. If compression would produce a larger file, the original comes back unchanged.
- Short retention window. Uploads are deleted within 30 minutes, shorter than Smallpdf's stated one hour or iLovePDF's stated two.
- No watermark, no sign-up. Nothing is stamped onto your pages and nothing asks for an email address.
- Available in 18 languages. The interface follows your language rather than assuming English.
Core Features of DocPivot Compress PDF
- Image-only re-encoding. Objects that are not marked as images are passed through untouched, which is the mechanism that preserves the text layer.
- Magic byte validation. Every upload is verified server-side by its actual file signature rather than its filename extension, so a mislabeled file is caught immediately.
- Five labelled level cards. Each card states what the setting costs in resolution and quality before you commit to it.
- Byte-level savings report. The results panel shows original size, compressed size and percentage saved for each file individually.
- Automatic ZIP packaging. A single file downloads as itself; multiple files arrive as one archive.
- Background worker processing. Jobs run in a dedicated queue rather than in the web process, so a large scan cannot stall the page.
- Size regression guard. Two separate checks prevent a compressed file larger than the input from ever being returned.
- 100 MB per file. Large scanned volumes and photo-heavy portfolios are within range.
- No AI, no third-party API. Compression runs entirely on our own infrastructure, with no external service receiving your document.
- Rate limiting per IP. Twenty jobs per IP per hour keeps the queue responsive for everyone.
How DocPivot Compress PDF Works
The pipeline has five stages, and the third one is where this tool diverges from the common approach.
- Choose your files. Add one PDF or up to 20. Each is verified by magic bytes on the server before anything else happens.
- Pick a level. Lossless, Light, Balanced, Strong or Extreme, presented as five cards with the trade-off written on each.
- Take the document apart. A background worker uses qpdf to dump the PDF as JSON, writing every stream out to a separate side file. This is what makes each embedded image individually addressable rather than buried in the document body.
- Re-encode the images only. libvips downsamples and re-encodes each image at the level's target resolution and quality. Anything not carrying the image subtype marker is skipped entirely.
- Rebuild and verify. qpdf reassembles the file. If the output is not genuinely smaller, the untouched original is returned instead.
Stage three is the architectural difference. Ghostscript-based compressors, which is what a large share of free online tools run underneath, rewrite the entire document through their pdfwrite device. Ghostscript's own documentation warns that its compression presets alter the input in a variety of ways and can produce output of poorer quality than simply leaving the settings alone. Extracting the streams first and putting back only the image objects means everything else in the file, including the font programs, comes through the process unchanged.
When to Use DocPivot Compress PDF
Reach for compression when a file has to cross a size boundary you do not control. That covers email attachment ceilings, upload forms, storage quotas and page load budgets.
- Email attachment limits. Gmail and Outlook both cap attachments around 20 to 25 MB, and corporate mail gateways are often stricter.
- Government and university portals. These frequently enforce caps of 2 MB or less with no explanation of how to meet them.
- Job applications. Applicant tracking systems reject oversized CVs and portfolios outright, sometimes without an error message.
- Website downloads. A 40 MB brochure on a product page is a bounce risk on mobile connections.
- Scanned document archives. Scanners default to resolutions far beyond what on-screen reading requires, inflating every page.
- Cloud storage costs. Teams storing thousands of contracts see real savings from a single batch pass.
- Client deliverables. Design proofs and photo-heavy reports need to arrive without a file transfer service in the middle.
Skip compression when the document is going to a commercial printer, where full-resolution images matter, and when the PDF is already text-only. Contracts, code listings and plain reports are already close to their floor.
Real-World Use Cases
Scanned application pack under 2 MB
Context: A graduate scans eight certificates at 600 DPI, combines them into one document, and ends up with a 34 MB PDF for an admissions portal capped at 2 MB.
Process: Upload the merged file, select Strong, review the reported output size, and download. Stepping down to Extreme is available if Strong lands just over the cap.
Result: The pack lands around 1.6 MB with certificate text still legible and the document still searchable by the registrar's system.
Agency photo report for a client
Context: A marketing agency assembles a 60-page campaign report with 120 embedded screenshots, producing a 78 MB file no client inbox will accept.
Process: Run the report at Balanced, check the savings figure, and if the source images were oversized to begin with, resize them before assembly on the next report.
Result: The report drops to roughly 27 MB, screenshots stay readable on screen, and clickable links in the document keep working.
Contract archive migration
Context: An operations team migrates 400 scanned contracts to new storage and needs the archive smaller without losing full-text search.
Process: Process in batches of 20 at Balanced. Contracts that arrive as image-only scans get a searchable text layer added first, since OCR before compression works on the sharper original.
Result: Storage drops by around 65% and every contract remains searchable by counterparty name and clause.
Product manual for a website
Context: A hardware company hosts a 45 MB manual that mobile visitors abandon before it finishes loading.
Process: Compress at Balanced, then restructure the file for progressive web loading so the first page renders before the whole download completes.
Result: The manual serves at about 12 MB and opens in seconds rather than after a visible wait.
Why the Text Layer Survives
A PDF is a container of objects, and each object carries a type marker. Images are tagged with the image subtype; fonts, text drawing instructions, annotations and form fields are not. Because the compression stage filters on that marker, non-image objects are copied back into the rebuilt document exactly as they arrived.
The practical consequences show up immediately. Copy and paste still works. Ctrl+F still finds words. Screen readers still read the document aloud, which keeps it accessible. Hyperlinks, bookmarks and form fields all survive, and the file can still be converted back to an editable Word document afterwards because there is still real text to convert. Run a rasterising compressor over the same file and every one of those properties disappears in a single pass, usually with no warning that it happened.
How DocPivot Compress PDF Compares
| Feature | DocPivot | Smallpdf | iLovePDF |
|---|---|---|---|
| Free compression levels | 5 | 1 (Moderate and Strong are Pro) | 3 |
| Compress to an exact target size | Yes, via a sibling tool | Not offered | Not offered |
| Published resolution and quality per level | Yes | No | No |
| Stated file retention | 30 minutes | 1 hour | 2 hours |
| Batch size | 20 files | Batch listed as a Pro feature | Yes |
Smallpdf's compression page describes Basic, Moderate and Strong levels and marks the latter two as requiring a subscription, so the free path stops at the weakest setting. iLovePDF publishes three free levels named Extreme, Recommended and Less compression, but does not state the resolution or quality each applies. Neither tool lets you name a size and have the compressor aim at it. Both are capable products; the gaps here are specific and narrow.
Honest Limits
A text-only PDF will barely shrink, and that is the correct behaviour. In testing, a 1,493 byte text document came back at 1,094 bytes at every single level, because there were no images to re-encode. Any tool reporting a dramatic saving on a file like that is almost certainly rasterising the pages, trading the text layer for a number.
Extreme is visibly softer on paper. Verified output at Extreme keeps text extractable and fonts real, but an embedded image drops from 274 KB to 37.8 KB at identical pixel dimensions, and that difference is apparent in print even when it is invisible on a phone screen. Choose Light or Lossless for anything heading to a press.
Files are uploaded to our servers rather than processed in the browser, which is what allows qpdf and libvips to run properly, and they are deleted within 30 minutes. Limits are 100 MB per file and 20 jobs per IP per hour. Already-compressed PDFs have little left to give, and a file that has been through Strong once will not halve again on a second pass. If your file is large because it contains too many pages rather than heavy images, splitting it into sections or removing the pages you do not need addresses the actual cause.
Frequently Asked Questions
Yes, all five compression levels are free with no account required. That includes Extreme, the strongest setting. There is no watermark on the output and no sign-up step.
It depends almost entirely on how many images the file contains. An image-heavy PDF reached 14% of its original size at Extreme in testing, an 86% reduction. A text-only document may shrink by only a few hundred bytes.
No, because the text is never converted into an image. Only image objects are re-encoded, so fonts stay embedded and text renders sharply at any zoom level. Photographs inside the document do soften at the stronger levels.
Balanced works for most situations and is the default. Use Light or Lossless for documents going to print, and Strong or Extreme when you need to clear a hard upload limit.
Yes, up to 20 files per job. Multiple files are returned as a single ZIP archive, while a single file downloads on its own. Every file in the batch uses the same compression level.
Each file can be up to 100 MB. The rate limit is 20 jobs per IP address per hour. For anything larger, pulling out only the pages you need brings the file under the limit first.
Uploaded and processed files are deleted within 30 minutes. By comparison, Smallpdf states one hour and iLovePDF states two hours. Nothing is analysed, copied or passed to a third-party service.
Your file most likely contains little or no image data. Text, fonts and vector graphics are already compact, so there is nothing substantial left to re-encode. This is expected behaviour rather than a failure.
No, two separate checks prevent that. If re-encoding would produce a larger file, the original is returned unchanged. This matters because re-compressing already-compressed data plus container overhead genuinely can increase size.
No, only image objects are modified. Fonts, hyperlinks, bookmarks, annotations and form fields all pass through the rebuild untouched. The document stays searchable and accessible after compression.
Yes, through the target-size compression tool, which aims at a number you specify rather than applying a preset. This is possible because images are re-encoded individually rather than handed to a fixed preset.
Often yes, since oversized source images are the usual cause of a bloated PDF. Running photos through an image compressor before assembly, or converting compressed JPGs straight into a PDF, produces a smaller starting file.
Yes, and scans usually benefit most because they are entirely image data. Scanned files often shrink by 70% or more at Balanced. If the scan needs a searchable text layer, add it before compressing rather than after, since recognition is more accurate on the sharper original.
Not directly, since this tool accepts PDF input only. Convert the document to PDF first, then compress the result. Files are validated by magic bytes, so a renamed file will be rejected.
