Image to PDF combines one or more images into a single PDF in the order you arrange, with one image to a page; the page can follow the image itself or be a uniform portrait or landscape A4, with a margin on all four sides. Everything happens in the browser and the images are not uploaded.

It is built for turning a pile of pictures into a file you can send and print: phone photos of receipts, JPGs exported by a scanning app, a few design drafts, a set of screenshots. Compared with dropping images into a chat app, a PDF keeps the order and the page size, and the recipient will not see them squashed into blurry thumbnails.

What it does is wrapping, not recognition: the PDF it produces has no text layer, so it cannot be searched and its text cannot be copied. For searchable text, use an OCR tool; to add pages to an existing PDF, send the result into “PDF Merge”.

1. 01Add imagesdrag them onto the empty PDF workspace page (they are recognised as “images to convert”), or switch to the “Image to PDF” panel and drag them into the image list or click to add; JPG, PNG, WebP, GIF and BMP are supported, and you can pick several at once.
2. 02Sort and trimeach row in the list has up, down and remove buttons, and the order in the list is the page order; “Clear list” starts over.
3. 03Choose a page size“Fit image” gives every page the same proportions as its image; “A4 portrait” and “A4 landscape” scale each image proportionally and centre it on an A4 page.
4. 04Set the page margin(0–72 pt, 0 by default) and the output file name (default “Image Document”).
5. 05Click “Generate PDF”progress advances image by image and the download starts automatically as file name.pdf; the result card offers “Download again” or “Add to workspace” to keep going — to add a watermark or page numbers, for example, or to merge the file with other PDFs.

Four images from different sources, in “Fit image” mode with a 0 pt margin:

Input file                      Format   Original bytes   Image stream inside the PDF                 Page size
photo-3000x4000.jpg             JPEG     1,320,529        DCTDecode 1,320,529 (byte for byte)         2250 × 3000 pt
logo-800x600.png (transparent)  PNG      22,732           Flate 16,907 + SMask 5,626 (alpha split)    600 × 450 pt
scan-1600x1200.webp             WebP     350,694          Flate 3,167,812 (re-encoded as PNG, ×9)     1200 × 900 pt
screenshot-1920x1080.png        PNG      22,208           Flate 23,535                                1440 × 810 pt
Total                                     1.64 MB          Output 4.54 MB

The same four images switched to “A4 portrait” with a 36 pt margin produce 4,536,687 bytes — 98 bytes more than the table above, and all four pages are 595.28 × 841.89 pt. The page rule changes only the drawing instructions, not the embedded images.

The page and object structure of a PDF is covered in the “How it works” section of “PDF Merge” (see here); how pt relates to pixels and DPI is set out in more detail in “PDF to Image” (see here). This section covers only the two things that happen when an image enters a PDF: how the bytes get in, and how the page size is worked out.

JPEG and PNG go in “exactly as they are”

PDF natively supports two image compression filters that happen to match these two formats: a JPEG data stream can be embedded directly as a DCTDecode image object, and PNG pixels after zlib compression are FlateDecode. The tool neither decodes nor re-encodes these two formats — that is why the JPEG in the table above has identical byte counts going in and coming out. It means the quality and the size of a photo are decided by the file you hand over, and converting to PDF costs nothing further.

PNG is slightly different: a PDF image object has no alpha channel, so a PNG with alpha is split in two — the colour data in one part, the alpha stored separately as a soft mask (/SMask). Transparent areas show the white page through them, so a logo does not arrive on a black rectangle.

WebP, GIF and BMP take another road

PDF has no WebP filter. Those formats are decoded to a canvas by the browser first (anything whose longest side is over 4096 px is scaled down proportionally), and the canvas is then encoded as PNG and embedded. PNG is lossless, and for photographic content with noise it is far larger than lossy WebP — the 350 KB WebP scan above became 3.17 MB inside the PDF. GIF contributes its first frame only; animation is not kept.

The practical conclusion: for photographic WebP or HEIC, convert to JPEG with “Image Converter” first, and the size can differ by an order of magnitude; line art and screenshots are smaller as PNG, so there is nothing to worry about there.

How large is a “Fit image” page?

An image has pixels and no physical size, while a PDF page has to be measured in pt (1/72 inch). The tool follows the convention of the web world, 1 px = 1/96 inch = 0.75 pt, and simply multiplies pixels by 0.75: 1600 × 1200 px becomes 1200 × 900 pt, and a 3000 × 4000 px phone photo becomes 2250 × 3000 pt — the equivalent of a 79 × 106 cm sheet of paper. Nothing is wrong with that; readers scale the display, and printer drivers “fit to paper”. But if the result is going to be merged with an A4 document, you would end up with one giant page among normal ones, and A4 mode is the one to pick there.

The EXIF resolution in an image (300 DPI, say) plays no part in the calculation, and when a JPEG is embedded directly its EXIF orientation flag is not applied either: a portrait photo that other viewers “straighten” from the flag may lie on its side in the PDF — rotate it upright with an image tool first.

How the A4 mode lays out

The page is fixed at 595.28 × 841.89 pt (swapped in landscape). The usable area is the page minus twice the margin, the scale factor is min(usable width / image width, usable height / image height), and the image is scaled by that factor and centred — never cropped, never stretched. A landscape image on a portrait A4 page occupies a band in the middle with white above and below, which is a good reason to choose “A4 landscape”; a mixed set of images can only use one orientation, and if pages have to differ, generate the PDF and then rotate individual pages with “PDF Page Organizer”.

• Compress first, then convert: phone originals are 3–5 MB each, and a dozen of them add up. Cap the longest side at 2000 px and quality at 0.8 with “Image Compress” before converting; converting to PDF will not change image size again.
• Make receipts uniform for printing: expense documents come in all sizes; choose “A4 portrait” with a 36 pt margin so each one prints centred with white around it, ready for filing.
• Merging with an existing PDF: after generating, click “Add to workspace” and switch to the “PDF Merge” panel to sort and merge with other files; A4 mode avoids pages of mismatched size.
• When the text has to be searchable: this tool does no OCR. Convert to PDF for archiving first, and recognise the pages that need searching separately.

Input

JPG/JPEG, PNG, WebP, GIF (first frame only), BMP; HEIC/HEIF cannot be decoded directly, so convert to JPEG with “Image Converter” first.

Count and size

no limit on the number of images; JPEG and PNG are embedded with their original bytes, so the output is roughly the sum of the inputs, while WebP, GIF and BMP are re-encoded as PNG and may grow considerably.

Canvas path resolution cap

WebP, GIF and BMP are scaled down when their longest side exceeds 4096 px; JPEG and PNG are not limited in that way.

Pages

“Fit image” = pixels × 0.75 pt plus margins; A4 = 595.28 × 841.89 pt (or landscape), image scaled and centred; margin 0–72 pt.

Orientation

EXIF orientation and DPI are not read; one output has one page orientation throughout.

Output

a new PDF with one page per image, no text layer and no bookmarks; the file name is yours to choose.

What it does not do

no OCR; no cropping or stretching; no putting several images on one page (use “PDF N-up” or the collage tool); no inserting images into an existing PDF (convert first, then “Merge” or use the page organizer).

• “Converting to PDF compresses images”: it does not. JPEG goes in untouched and keeps roughly its size; WebP can even grow. To shrink files, run “Image Compress” first, or put the finished PDF through “PDF Compress”.
• “Text in a PDF can be searched”: text inside an image is still pixels. Convert to PDF for archiving and run OCR separately if you need to search it.
• “Fit image shrinks large images”: it does not shrink anything; it converts pixels to pt at 0.75 and the page comes out large. Choose A4 to pin the paper size.
• “GIF animation is kept”: only the first frame is. PDF has no native animation.

Reading and decoding images and assembling the PDF all happen inside your browser: nothing is uploaded and the tool works without a network connection. See the privacy policy for more.

• ISO 32000-1:2008, PDF 1.7, §7.4.4 FlateDecode Filter, §7.4.8 DCTDecode Filter, §8.9.5 Image Dictionaries (including the /SMask soft mask): https://opensource.adobe.com/dcs/pdf/PDF32000_2008.pdf（访问日期：2026-09-09）
• W3C, CSS Values and Units Module Level 4, §6.2 Absolute Lengths (1 in = 96 px = 72 pt, that is 1 px = 0.75 pt): https://www.w3.org/TR/css-values-4/#absolute-lengths（访问日期：2026-09-09）
• pdf-lib API, PDFDocument.embedJpg() and embedPng() — JPEG embedded directly, PNG parsed with the alpha separated out: https://pdf-lib.js.org/docs/api/classes/pdfdocument#embedjpg（访问日期：2026-09-09）