PDF to Image renders each page of a PDF into a bitmap and exports it in the format you choose (PNG or JPEG) at the sharpness you choose (72, 150 or 300 DPI). Everything runs in the browser and the file is never uploaded.

People reach for it not to “read” a PDF but to put a page somewhere else: a figure from a paper into a slide deck, the first page of a contract into a chat window that only takes pictures, an e-poster exported as an image that previews directly on social platforms, or a set of page images for a device that cannot open PDFs.

What comes out is the rendered whole page: text, vector art, embedded images and annotation appearances are all drawn into one picture. If you want the photo that was originally embedded in the PDF (at its original resolution, without the surrounding text), use “PDF Extract Images” — different job, different tool.

The export can cover every page or only the pages you click in the middle area; multiple pages are packed into a zip automatically and a single page downloads as an image file. File names carry a zero-padded page number (for example report_page03.png) so that sorting by name keeps the right order.

1. 01Add a PDFdrag one in or click to choose. An encrypted file needs its password removed with “PDF Unlock” first — entering the viewing password alone is not enough to export.
2. 02Set the range“All pages” is the default. For a few pages, click them in the middle area first (Shift-click for a run); the panel's “Selected pages (n)” counts them live, and you switch to it.
3. 03Choose a format“PNG (lossless)” suits text, charts and line art; “JPEG (smaller)” suits photos and scans. Choosing JPEG adds a “JPEG quality” slider (30%–100%, 90% by default).
4. 04Choose the sharpness72 DPI is the smallest and fine for on-screen thumbnails; 150 DPI is the everyday balance for viewing and dropping into documents (the default); 300 DPI suits printing or zooming in on fine detail.
5. 05Click “Export images”the progress bar advances page by page, and a multi-page run adds a “pack zip” step at the end. One page downloads directly; several download as filename_images.zip.

An eight-page A4 text report (report.pdf) produced these real outputs under four settings:

Settings                Pixels per page   Single file   8-page zip
PNG  · 72 DPI           595 × 841         ≈ 74 KB       483 KB
PNG  · 150 DPI          1240 × 1753       ≈ 300 KB      1.84 MB
JPEG · 150 DPI · 90%    1240 × 1753       ≈ 338 KB      2.41 MB
JPEG · 300 DPI · 90%    2480 × 3507       ≈ 840 KB      6.05 MB

File naming
  report_page1.png … report_page8.png      (no zero padding below 10 pages)
  report_page01.png … report_page12.png    (from 10 pages up, padded to the widest page number)
  report_images.zip                        (multi-page packaging)

Note the second and third lines: at the same resolution, a JPEG of a text page is larger than the PNG. That is not an anomaly — the “How it works” section explains why.

From points to pixels: DPI is the conversion rule

A PDF page has no pixels; it describes its size in points (pt), and 1 pt = 1/72 inch. A4 is 595.28 × 841.89 pt, which is 8.27 × 11.69 inches. To draw it as a bitmap, you first have to decide how many pixels go into an inch — that is DPI (dots per inch):

So A4 at 72 DPI is 595 × 841 pixels (a scale factor of exactly 1), at 150 DPI it is 1240 × 1753, and at 300 DPI it is 2480 × 3507. The pixel count is proportional to the square of the DPI: a 300 DPI image has 17 times the pixels of a 72 DPI one, and the size grows roughly in that proportion (for PNG) or a little more slowly (for JPEG).

The three steps match three destinations: 72 DPI is close to the physical density of a classic screen and is enough for thumbnails and chat; 150 DPI still looks clean on a high-density display and is the sensible compromise for Word or slides; 300 DPI is the usual requirement in print and the choice when you want to zoom into small type.

What happens during rendering

The tool draws the page onto a canvas with pdf.js at a scale factor of DPI ÷ 72, applying any rotation the page carries in the workspace. Canvas dimensions are rounded down, so 841.89 becomes 841 and not 842. Text is rasterised from font outlines, vector art is filled along its paths, and images embedded in the PDF are resampled to the current scale (an embedded 300 DPI photo loses most of its pixels when you export at 72 DPI).

The canvas then goes to the browser's image encoder:

• PNG: lossless — a per-row prediction filter first, then Deflate compression. Large flat colours, sharp edges and repeating patterns compress very well; photographic gradients compress poorly.
• JPEG: lossy — an 8×8 block discrete cosine transform followed by quantisation according to the quality parameter. Gradients and photos compress extremely well; sharp edges (text!) need a lot of high-frequency coefficients, which keeps the size up and adds “ringing” around strokes. JPEG has no transparency, so a white background is laid down first.

That is why a text page is smaller as PNG than as JPEG (90%): a page of black type on white is “very few colours plus lots of repetition” to PNG, and “high frequencies everywhere” to JPEG. In the other direction, a scan or photo page as JPEG is usually a third to a fifth of the PNG size.

A feel for the JPEG quality slider

Dropping quality from 90% to 75% roughly halves the size; in photos the difference is almost invisible, while text edges start to show light noise. Below 50%, block artefacts become obvious. The default 90% is the setting that does not go wrong when you would rather not think about it; when the goal is a quick look by someone else, around 75% is often enough. The slider stops at 30% on purpose — below that there is no reading value left.

Which page number "selected pages" uses

The page number in the file name is the page's ordinal in the current arrangement: if you deleted page 2 in the middle area first, the old page 3 exports as “page2”. The digits under a thumbnail are always that ordinal.

• Export only the page you need: click one page, switch to “Selected pages (1)”, and the result is a single image file rather than a zip — ready to drag into a chat or a document.
• Straighten before exporting: rotate a landscape-scanned page in the middle area, then export, and the picture comes out upright; the rotation affects only this output.
• Slide material: a 150 DPI PNG dropped into a slide deck still looks clean full-screen on a 1080p projector; at half a slide, 72 DPI is enough.
• Input for OCR: most OCR engines perform best at 300 DPI; use PNG to keep JPEG noise out of recognition. The export can go straight into this site's OCR tool.
• When the PNG is too big: a text page as PNG is already the better choice; switch photo pages to JPEG; if it is still too large, hand the exported images to an image compression tool for a second pass.
• Rebuild a PDF from the images: group the exported images back with “Image to PDF” and you get a PDF of pure pictures with no copyable text — essentially the same product as “PDF Compress”, except that you can pick lossless PNG instead of JPEG.

Input format

PDF; one file at a time, the one selected on the left.

Encrypted files

remove the password with “PDF Unlock” first.

Output formats

PNG (lossless, RGBA bitmap) or JPEG (lossy, white background, quality 30%–100%). WebP, TIFF, SVG and others are not offered; convert afterwards with “Image Converter”.

Sharpness

three fixed steps, 72 / 150 / 300 DPI; no custom value and no pixel-width specification.

Pixel dimensions

set by the page size and the DPI — A4 at 300 DPI is 2480 × 3507. A very large page (an A0 drawing, say) may exceed the browser's canvas limit at 300 DPI and fail; lower the DPI.

Exported content

the rendered whole page, including text, vectors, images and annotation appearances; form field interaction, bookmarks and layer toggles are not exported.

Multi-page output

packed as a zip (DEFLATE); the images are already compressed, so the zip barely shrinks the total.

Size and memory

an uncompressed 300 DPI A4 bitmap is about 35 MB, and a few dozen pages can exhaust a mobile browser; split the run or lower the DPI.

What it does not do

it does not extract the original images embedded in the PDF (use “PDF Extract Images”), does not recognise text (use the OCR tool), and does not turn a PDF into an editable vector format.

• "300 DPI is always better": it only matters when printing or zooming. For chat and web use, 300 DPI just makes the image bigger and slower, and many platforms compress it again anyway.
• "JPEG is always smaller than PNG": false for text, charts and line art, where PNG is often smaller and free of noise. JPEG's advantage is limited to photos and scans.
• "The exported image is the picture embedded in the PDF": no. What you get is the whole rendered page, sized by the DPI; for the original asset use “PDF Extract Images”.
• "The page number in the file name is the original file's": it is the ordinal in the current arrangement, which changes after a deletion or a reorder.
• "The text will still be selectable after export": an image has no text layer. If you need text, use “PDF Extract Text”.

Parsing and rendering the PDF happen locally via pdf.js, PNG/JPEG encoding is done by the browser's built-in encoder, and the zip is packed locally too; no page content leaves your device. It works offline, and closing the page frees the memory.

• ISO 32000-1:2008, PDF 1.7, §8.3.2.3 Default user space (1/72 inch): https://opensource.adobe.com/dcs/pdf/PDF32000_2008.pdf（访问日期：2026-09-08）
• W3C, Portable Network Graphics (PNG) Specification (Third Edition), filtering and Deflate compression: https://www.w3.org/TR/png-3/（访问日期：2026-09-08）
• ITU-T T.81, the JPEG standard, DCT and quantisation: https://www.w3.org/Graphics/JPEG/itu-t81.pdf（访问日期：2026-09-08）
• MDN Web Docs, HTMLCanvasElement.toBlob(), formats and the quality parameter: https://developer.mozilla.org/zh-CN/docs/Web/API/HTMLCanvasElement/toBlob（访问日期：2026-09-08）