Bulk Extract Images From PDF Online — Every Photo, One Click

The short answer

To bulk extract images from a PDF without uploading the file, open pdfmavericks.com/extract-images-from-pdf, drop your PDF, and click Extract. The tool walks every page, decodes every embedded image stream, and packages the results as a ZIP file. The original format and resolution are preserved — JPEG stays JPEG, PNG stays PNG, no re-encoding. A typical 100-page document with 50 figures extracts in three to seven seconds.

For a deeper walkthrough of single-image extraction and the general workflow, see the extract images from PDF guide. This post focuses on the bulk case — when you have a thesis with dozens of figures, a product catalog with hundreds of photos, a scanned archive with thousands of pages — and you need every image out in one pass.

Extract vs render: a critical distinction

Two operations sound similar but produce very different output:

• Extract images. Pull the embedded image objects out of the PDF with their original bytes intact. A 4000x3000 photo stored at 90 percent JPEG quality comes out as a 4000x3000 JPEG at 90 percent quality. The text, layout, and page background around the image are not included. The output is one file per embedded image.
• Render pages as JPG. Rasterize each PDF page (text, images, vector graphics, all of it) into a flat JPEG at a specified DPI. The output is one JPEG per page. Embedded images become part of the rendered page; their original resolution is lost during the rasterization.

For a thesis with 80 figures spread across 200 pages, extraction gives you 80 image files at original quality. Page rendering would give you 200 page JPEGs with the figures embedded as part of each page's flat raster. The right tool depends on what you actually want.

The PDF specification (ISO 32000-2, available at iso.org/standard/75839) stores images as XObject streams referenced from page content streams. Extraction walks the XObject table and pulls each stream; rendering executes the content stream and produces a raster. Both operations are documented; pdfmavericks.com offers both as separate tools.

How browser-local bulk extraction works

The bulk path uses three browser primitives plus a packaging library:

1. File API — reads the PDF bytes from local disk via the user-permissioned file picker. No network call. MDN documents the API at developer.mozilla.org/en-US/docs/Web/API/File_API.
2. PDF.js — Mozilla's parser at github.com/mozilla/pdf.js walks the PDF object tree, finds image XObjects on every page, and exposes their decoded byte arrays.
3. Stream filter decode — each image XObject lists a filter chain (DCTDecode for JPEG, FlateDecode for compressed raster, JBIG2Decode for bilevel scans). The matching decoder produces the original raster bytes.
4. JSZip packaging — the open-source ZIP library at github.com/Stuk/jszip packages every decoded image into a ZIP archive in memory, with sequential filenames (page-N-image-M.ext). The ZIP is offered as a single download.

All four steps run inside the browser tab. The PDF stays on local disk; the decoded images stay in tab memory; the resulting ZIP downloads through the Save dialog. No upload, no server, no temporary cache outside your machine.

Real bulk-extraction use cases

Three workflows where bulk image extraction earns its place: thesis prep, scanned archive triage, and old photo recovery.

Thesis and journal figures

A typical PhD thesis runs 200 to 400 pages with 40 to 100 figures embedded across chapters. When the author needs to extract those figures to reformat for a journal submission, build slides for a defense, or share with collaborators, manually screenshot each one is a 90-minute task with quality loss at every step. Bulk extraction takes 10 seconds and preserves original resolution.

The same pattern applies to published journal articles. Reviewing a paper's figures in detail — for replication, teaching, or systematic review — benefits from having the figures as individual files rather than embedded in a long PDF. Most academic publishers allow extraction for personal scholarly use; check the specific journal's license. For Creative Commons-licensed papers (the entire PLOS catalog, most preprints on arXiv, many Elsevier open-access papers), extraction and re-use with attribution is explicitly permitted.

For scanned thesis PDFs where the figures are part of a page raster rather than separate image objects, see the browser-local OCR guide — the OCR step adds a text layer but doesn't separate images. For already-rasterized figures, use the page-to-image rendering tool and manually crop, which is slower but works on scans.

Scanned receipts and bank statements

Tax-time workflow: a year of business expenses lives across 60 PDF receipts. Each receipt PDF was generated by a different scanning app — some have the receipt as a single embedded JPEG, some have it as multiple smaller images, some as a page raster. Bulk extraction across all 60 PDFs (one at a time in browser tabs) produces a folder of every embedded image, which the accountant can then sort by date, vendor, and amount.

For Indian GST workflows specifically, expense receipts need to be archived along with the GSTR-3B filing per the CBIC GST portal. The receipts often arrive as scanned PDFs from vendors; bulk extraction gives you the raw images for the archival format. Combined with the PDF/A for India e-filing guide, this is the complete chain from receipt PDFs to compliant archival.

Old archival PDFs with embedded photos

Family archives, employer document scans from the 1990s, university yearbook PDFs — these are full of embedded photos that were scanned years ago and never made it out of the PDF wrapper. Bulk extraction recovers them as individual image files you can sort, tag, back up to a photo service, or import into a memory book.

One caveat for old PDFs: image quality reflects the era of the scan. A 1995 PDF with 150 DPI scans extracts as 150 DPI images — the extraction can't add resolution that wasn't there. For upscaling, AI-based upscalers (Topaz Photo AI, Adobe's Super Resolution) are the typical next step after extraction; those run as desktop apps, not browser-local, so the workflow leaves the strict no-upload chain. For most archival uses, the original-resolution extract is adequate.

Tips for clean bulk output

Five practical notes for getting useful files out of a bulk extraction:

• Run on the largest file your laptop tolerates. Browser memory tops out around 4 GB per tab on most modern systems. A PDF with 800 images of 2 to 5 MB each is the practical ceiling. For larger archives, split the PDF first using the split tool and extract per chunk.
• Watch for duplicates. Many PDFs embed the same logo or watermark on every page. Bulk extraction will produce one copy per occurrence. A quick dedup pass after extraction (using fdupes, rdfind, or even Finder's sort-by-size) removes the noise.
• Look at the filename pattern. The output uses page-N-image-M naming, where N is the PDF page number and M is the order of appearance on that page. This makes it easy to map extracted images back to where they were in the source.
• Check for vector graphics separately. Charts, line art, and equations are usually stored as vector instructions, not as raster images. Bulk extraction won't pull these by default. The extract-images tool flags vector content with a count; if you need the vectors, the page-to-SVG workflow (page render at vector level) is the right path.
• Verify the count before downloading. The tool shows "extracted N images" before the ZIP download. If the count is way off from your expectation (e.g. you expected 50 figures and got 2 logos repeated 100 times), the PDF probably stores its figures as page-level vector content rather than raster — switch to the page render tool.

For background on why browser-local matters for this kind of workflow, see the browser-only PDF editor guide and the jsonformatter breach lesson. The all-tools catalog lists every browser-local operation on the site.