ConvertBox Logo
ConvertBox

Images to PDF and PDF to images: page size, DPI and OCR

A PDF page is measured in points and an image in pixels, and DPI is the exchange rate between them. That one relationship explains odd page sizes, soft prints and file sizes, while searchable text is a separate matter that only OCR solves.

ConvertBoxPublished

Points and pixels: two ways of measuring a page

PDF measures pages in points, and a point is 1/72 of an inch. An A4 page is 595 x 842 points and US Letter is 612 x 792, whatever is drawn on them. Text and vector graphics on the page have no pixels at all; they are instructions that a viewer draws at whatever resolution the screen or printer has.

An image is the opposite: a grid of pixels with no inherent physical size. A 2480 x 3508 pixel scan is A4 only if you agree to print it at 300 pixels per inch; at 150 it would cover a sheet twice as wide and twice as tall. DPI is the exchange rate between the two worlds, and every conversion between images and PDF has to pick one, openly or silently.

Images into a PDF: what page size you get

Converters make one of two choices. Some place each image on a fixed paper size with margins. Others make the page exactly as large as the image. ConvertBox's JPG to PDF tool does the second and maps one pixel to one point, with no paper-size, orientation or margin options. A 4000 x 3000 pixel phone photo therefore becomes a page of about 56 x 42 inches (141 x 106 cm). On screen nothing looks unusual, since viewers scale a page to fit the window, but a print dialog set to actual size will try to tile it across many sheets; choose fit to page.

Quality is not affected by any of this. PDF understands JPEG compression natively, so JPEG data is placed in the file as it is, with no second round of loss, and PNG pixels are stored losslessly with their transparency. The PDF ends up about as large as the images that went in. For uniform pages, crop or resize beforehand: A4 has a ratio of 1 to 1.414, and 1240 x 1754 pixels is A4 at 150 DPI.

Preparing photos and scans before conversion

  • Format. The tool embeds JPEG and PNG only; HEIC photos from an iPhone, WebP, GIF, BMP or TIFF stop the conversion with an error, so convert those first.
  • Orientation. Phone cameras record rotation as EXIF metadata, which a PDF page does not interpret. If a page comes out sideways, rotate the image itself and add it again.
  • Weight. Camera photos are several megabytes each and go into the PDF at full size. Resizing them to about 1240 x 1754 pixels turns a 20-page document of tens of megabytes into one of a few.
  • Order. Each image becomes one page in list order, so arrange the list before converting.

DPI for print and for screens

To find how large an image prints sharply, divide its pixels by the target DPI. Commercial printing normally expects around 300 DPI, while screens, slides and chat apps are satisfied with far less. ConvertBox's PDF to JPG tool renders at a fixed scale of 2, which is 144 DPI, so an A4 page comes out at 1190 x 1684 pixels. At 300 DPI that is 1190 / 300, about 4 inches wide, roughly half the width of the sheet: the export looks sharp on a screen and soft when printed at full size. The scale cannot be changed in that tool, so print-resolution output calls for a desktop renderer.

Rendering resolutionA4 page in pixelsSuited to
72 DPI (scale 1)595 x 842Quick previews; small text blurs
144 DPI (scale 2)1190 x 1684Screens, slides, messaging, web pages
216 DPI (scale 3)1786 x 2526Fine print and zooming in
300 DPI2480 x 3508Printing at full size

JPG or PNG for page images

The choice depends on what is on the page, not on habit. PNG is lossless and very efficient on flat areas, so a page of black text on white is both small and perfectly crisp. JPEG is built for photographs: it discards fine detail the eye misses in continuous tones, but around sharp edges such as letters it produces halos and blocks at low quality.

The decision rule: text, line art, diagrams and screenshots go to PNG; scanned magazine pages and photo-heavy layouts go to JPEG, where 80 to 90% quality is normally indistinguishable on screen at a fraction of the size. In ConvertBox's PDF to JPG tool PNG is the default and the JPEG slider runs from 10 to 100%. Every page is saved as a separate file, with no page picker and no ZIP bundle, so for two pages out of forty, split them out first. Either way the result is a picture: text can no longer be selected and links stop working, so keep the PDF as the master copy.

One long image instead of many pages

A long image stacks every page top to bottom in a single picture, which suits messengers that preview images but not documents, and notices meant to be scrolled on a phone. Its width is that of the widest page and its height is the sum of all page heights.

The limit is the browser canvas. Chrome and Firefox stop at 32,767 pixels per side, and mobile Safari is stricter about total area. An A4 page at 144 DPI is 1684 pixels tall, so 32,767 / 1684 allows 19 pages; at 72 DPI the ceiling is about 38 and at 216 DPI about 12. Memory is the second constraint: a rendered A4 page at 144 DPI occupies 1190 x 1684 x 4 bytes, about 8 MB uncompressed, and the stitched canvas needs the same again. ConvertBox's PDF to long image tool offers those three render qualities; for longer documents, lower the quality or split the PDF and make several images. Many platforms also recompress very tall uploads, so send the file as a document if the text turns soft.

When you need OCR, and what it gives you

A PDF built from photos or scans contains pictures of words. You cannot search it, select a sentence or copy a number until optical character recognition has read the pixels and produced text. If the text in a PDF can already be selected, OCR is unnecessary and only introduces mistakes.

Know what kind of output you are getting. ConvertBox's PDF OCR tool renders each page at 144 DPI, reads it with the Tesseract engine in one chosen language (English, Korean, Japanese, Simplified Chinese or Spanish), and returns plain text with a marker line before each page, to copy or download as a .txt file. Its optional PDF output is built from the page images and carries no embedded text layer, so it is not a searchable PDF. Producing one, with invisible text placed behind the scan, is a job for desktop OCR software such as Acrobat or OCRmyPDF. Accuracy follows the scan: body text from a 300 DPI scan reads well, while tilted pages, grey text and multi-column layouts lose characters or reading order, and look-alikes such as 0 and O or 1 and l deserve proofreading.

Tools mentioned in this guide

Related guides

All guides