PDF Compressor

{{ engineBadge }} Browser-side analysis {{ profileLabel }} {{ sourceSizeLabel }}

{{ rewriteDownloaded ? 'Downloaded' : 'Download local rewrite' }}

Source PDF:

Drop or browse one unencrypted PDF up to the browser work limit.

Compression profile:

Choose the size-quality target you would use in a production PDF compressor.

Target size:

Enter 0 for no fixed limit, or set an email/upload cap such as 10 MB.

Metadata cleanup:

Keep on for privacy-oriented local rewrites; turn off if document info fields must remain intact.

Output filename:

Example: board-pack-small.pdf. The .pdf extension is added if missing.

Browser work limit:

Raise only for a powerful desktop browser; production compression should use a backend worker.

Metric	Value	Detail	Copy
{{ row.metric }}	{{ row.value }}	{{ row.detail }}

Check	Status	Detail	Copy
{{ row.check }}	{{ row.status }}	{{ row.detail }}

Embed:

Customize

Include current inputs

Size

Advanced

Width

Height

Aspect ratio

Max height

Collapsible embed

Allow fullscreen

Referrer policy

Sandbox tokens

A PDF can fail an email, upload, or archive limit for several different reasons. Some files are mostly scanned page images. Others are office documents with embedded fonts, repeated saves, object overhead, comments, forms, or document information fields. Compression works best when the file's main source of weight is identified before anyone trades away readability.

Portable Document Format is a container, not a single picture. A page can combine text instructions, vector paths, bitmap images, fonts, metadata, annotations, form fields, cross-reference data, and sometimes JavaScript or attachments. That structure is why one PDF can shrink safely after a rewrite while another needs image downsampling or a stronger repair tool.

Common PDF size sources and compression cautions
What makes the PDF large	Typical source	Compression caution
High-resolution images	Scans, photo-heavy reports, slide decks, brochures	Downsampling can blur signatures, stamps, small text, maps, and diagrams.
Structural overhead	Repeated saves, conventional cross-reference tables, loose objects	A rewrite can help, but it may not change visible image bytes.
Metadata and document info	Office exports, reviewed files, authoring applications	Cleanup can improve privacy, but some records must retain title, author, or producer fields.
Encryption or damage	Password-protected files, partial downloads, malformed PDFs	Password-aware repair should happen before size reduction.

Lossless structural cleanup and lossy image recompression are separate jobs. A structural rewrite can compact object storage and neutralize common document information fields while keeping page appearance close to the source. Image recompression and downsampling usually create the largest savings for scanned or photo-heavy PDFs, but they also create the highest risk of blurry or missing detail.

Diagram comparing PDF structural rewrite, image recompression, and final size checking

A target such as 10 MB should be treated as a constraint, not as proof that any document can be reduced safely. Scanned text, signatures, engineering drawings, maps, certificates, and medical or legal attachments need visual review after compression. A smaller PDF is only useful when the file still opens, the pages still render correctly, and the destination accepts the final bytes.

The safest first pass is to separate what can be tested locally from what requires a production optimizer. Local rewrite results can reveal structural savings and metadata cleanup effects. Image-heavy, encrypted, damaged, or very large files need tooling that can downsample, recompress, repair, and validate the PDF more deeply.

How to Use This Tool:

Use this workflow to inspect one PDF, estimate realistic compression paths, and create a local structural rewrite when the browser can handle it.

Choose one file in Source PDF. The file should be an unencrypted PDF and must fit within the current Browser work limit. If the file is over that limit, stop and use a production optimizer instead of forcing the browser tab to process it.
Pick a Compression profile for the production estimate. Balanced sharing is the general starting point, Smallest readable is for strict screen-only limits, and Print-safe quality protects more image detail.
Enter a Target size only when a real upload or message cap exists. Use 0 when there is no fixed cap so the analysis focuses on source size, local rewrite size, and production estimate.
Leave Metadata cleanup on when the local rewrite should replace common title, author, subject, creator, producer, and keyword fields with neutral values. Turn it off when those document information fields must remain part of the record.
Open Advanced only for Output filename or Browser work limit. The limit is clamped between 10 MB and 200 MB, with 80 MB as the default, because rewriting a PDF can duplicate data in memory while parsing and saving.
Use Analyze PDF first. Check Compression Metrics and Optimization Checks for encryption blocks, image-compression review warnings, structure markers, target status, and local rewrite readiness.
Use Create local rewrite only after analysis is ready and no encryption block appears. Download the local rewrite when it is smaller, or when metadata cleanup matters more than a small size increase, then open the output PDF and inspect the pages.

PDF Size Profile is most useful when a target cap is set because it compares source size, local rewrite or local estimate, production estimate, and target limit in one chart.

Interpreting Results:

Compression Metrics separates the downloadable local rewrite from the production estimate. The local rewrite can reduce structure and metadata only. The production estimate is a planning number for a PDF optimizer that can downsample images, recompress streams, repair problematic files, and handle encrypted documents with the proper password-aware path.

PDF compressor result cues
Result cue	How to read it	What to verify
`Local Rewrite Ready`	The browser created a rewritten PDF smaller than the source.	Open the downloaded file and check page appearance, forms, links, signatures, and small text.
`Local Rewrite Larger`	The structural rewrite increased byte count.	Keep the original unless neutral document information fields matter more than file size.
`Image compression: Review`	Image markers were found, so major savings probably need downsampling or recompression outside the local rewrite.	Use the production estimate as a handoff guide and inspect the final optimized PDF visually.
`Encryption: Blocked`	An encryption marker or password-related read failure prevents safe browser-side rewriting.	Use a password-aware optimizer or remove protection through an approved document process.
`Target Limit`	The target compares the requested cap with the production estimate, not with a guaranteed final file.	Leave margin when the estimate is close to the cap, then validate the actual optimized PDF.

A smaller file can create false confidence. If a scan or slide deck drops below the upload cap but text, stamps, charts, or signatures blur, the result is not acceptable. Use byte savings as the first gate and visual inspection as the final gate.

Technical Details:

PDF size reduction is governed by the file's internal structure and by the resources attached to each page. Object streams and cross-reference streams can store indirect objects and lookup data more compactly than older conventional layouts, but they do not change high-resolution images by themselves. A PDF that already uses compact structures may have little structural overhead left to remove.

Image-heavy PDFs behave differently. JPEG, JPEG 2000, CCITT, JBIG2, and image-subtype markers indicate that embedded bitmaps may dominate the file. Downsampling those images can save substantial bytes, but the safe resolution depends on how the document will be read or printed. Screen-only sharing can tolerate more reduction than a print-safe board pack or a drawing set.

Formula Core:

The basic savings calculation compares the source byte count with a candidate output or estimate. Positive savings mean the candidate is smaller.

Savings percent = (1 - \frac{B_{candidate}}{B_{source}}) \times 100

A 20.00 MB source with a 17.40 MB local rewrite saves 13.0%. If the same source has a 9.80 MB production estimate, the estimate represents 51.0% planned savings, but that number remains an estimate until the final optimized PDF is created and checked.

Estimated bytes are computed by applying a bounded savings ratio to the source size.

B_{estimate} = B_{source} \times (1 - r)

Rule Core:

The local estimate, image-share estimate, and production estimate are intentionally bounded. They provide planning numbers rather than a claim that every optimizer will produce the same output.

PDF compression estimate rules
Estimate part	Rule used	Interpretation limit
Local savings ratio	Metadata, object-stream, and Flate signals are combined and capped at 9%.	Models structural rewrite potential only, not embedded image recompression.
Metadata signal	XMP or document-info markers add 1.2%; otherwise 0.2% is assumed.	Signals presence, not whether every metadata stream can be removed locally.
Object-stream signal	Already compact structures add 0.8%; missing object or xref streams add 4.5% potential.	A rewrite can still grow if the source was already efficient or contains unusual structures.
Image share	Starts near 8%, rises with image markers and bytes per page, then clamps from 5% to 82%.	Based on structural markers, not a full visual inventory of each image.
Production savings ratio	Local savings, image-share savings, and profile structural savings are combined and capped at 78%.	Real output varies by content, optimizer settings, repair work, fonts, transparency, and color data.

PDF compressor profile assumptions
Profile	Planned image range	Image savings factor	Extra structural factor	Best fit
`Smallest readable`	96 to 120 DPI with aggressive downsampling	0.48	0.08	Screen-only sharing and strict attachment caps, with visible softening risk.
`Balanced sharing`	144 to 180 DPI with moderate recompression	0.30	0.06	Ordinary sharing where readability matters more than the smallest possible file.
`Print-safe quality`	240 to 300 DPI with light recompression	0.16	0.04	Printed handouts, diagrams, scans, and files that need close inspection.

Analysis checks the PDF header, end-of-file marker, page hints, encryption marker, image filters, image subtype markers, object streams, cross-reference streams, metadata markers, JavaScript markers, and browser work limit. Files above the browser work limit are blocked before rewrite. Files large enough to strain memory can be structurally sampled, so the sampled-bytes warning should stay with the estimate.

When metadata cleanup is enabled, the local rewrite clears common document information fields and saves with object streams. It does not downsample embedded images, remove every XMP metadata stream, repair damaged objects, or prove archival conformance. Those jobs belong to a production PDF workflow that can validate the final file directly.

Privacy Notes:

The selected PDF is read in the browser for analysis and local rewrite attempts. The file is not sent to a compression service by this workflow, but supporting PDF and chart runtimes may load from public script hosts. Use an approved local or managed PDF optimizer for confidential documents that cannot depend on external script loading.

Metadata cleanup replaces common document information fields in the local rewrite when enabled.
Embedded XMP metadata, attachments, comments, form data, or hidden objects may need a deeper document-cleanup workflow.
Password-protected PDFs are blocked because safe compression needs a password-aware path.

Worked Examples:

These examples focus on how to act on the analysis cues rather than treating every estimate as a final compression result.

Board pack under an email cap

A 22 MB board pack must fit a 10 MB attachment limit. Choose Balanced sharing and set Target size to 10. If Production Estimate falls below the target but Image compression: Review appears, hand the file to a production optimizer and verify charts, signatures, and page images before sending.

Text-heavy policy document

A 6 MB policy PDF has no image-heavy markers and shows possible structural savings. Run Analyze PDF, then Create local rewrite. If the result shows Local Rewrite Ready, download it and open the file to check headings, bookmarks, links, and form fields before replacing the source.

Password-protected scan

A protected scan triggers Encryption: Blocked or an encrypted-file error. Do not retry with higher browser limits. Use an approved password-aware workflow, then rerun analysis on the unprotected or properly processed copy if compression is still needed.

FAQ:

Does the local rewrite downsample images?

No. The local rewrite focuses on object-stream saving and common document information fields. When Image compression: Review appears, production image recompression is still required for major size reduction.

Why did the rewritten PDF get larger?

Some PDFs are already compact, and rewriting can add overhead. If Local Rewrite Larger appears, keep the original unless neutral metadata values are more important than byte count.

Can I use a target size as a guarantee?

No. Target size is compared with the production estimate. The final optimized file must still be created, opened, checked visually, and measured against the actual upload or email cap.

What should I do with an encrypted PDF warning?

Use a password-aware PDF optimizer or an approved document process. The browser rewrite blocks encrypted PDFs instead of trying to bypass protection or risking a broken output file.

When should I change the browser work limit?

Raise Browser work limit only on a capable desktop browser and only when the file is trusted. The control is bounded from 10 MB to 200 MB because PDF rewrite work can use much more memory than the source file size.

Glossary:

Object stream: A PDF 1.5 structure that can store multiple indirect objects inside a stream instead of separate body sections.
Cross-reference stream: A compact stream-based form of cross-reference data used by newer PDFs instead of a conventional xref table.
Downsampling: Reducing image resolution, usually measured by pixel dimensions or dots per inch, to reduce embedded image bytes.
Document information fields: PDF metadata fields such as title, author, subject, creator, producer, and keywords.
XMP metadata: XML-based metadata that may live in one or more PDF metadata streams.
Production estimate: A planning value for a full PDF optimizer that can recompress images and streams beyond the local rewrite.

References:

Using PDF Creation Settings, Adobe Acrobat SDK Documentation.
PDF Association Cheat Sheet - Basics, PDF Association, 2024.
Object and Cross-Reference Streams, QPDF 11.9.1 Documentation.
Reduce PDF file size with advanced options, Adobe Help Center, November 5, 2025.