Vector and raster content in a PDF, explained

The difference in one paragraph

A vector PDF stores text and shapes as drawing instructions. A raster PDF stores pictures made of pixels. Vector content is recalculated at every zoom level, so it stays sharp forever. Raster content has a fixed number of pixels, so zooming in eventually just makes each pixel bigger.

Most real PDFs are a mix: vector text and table rules with a raster photograph or logo placed on the page. The question is usually which one dominates, and in particular whether the text itself is vector or a picture of text.

Why a PDF can be either

PDF was designed to describe a printed page exactly, whatever produced it. A word processor exporting a report knows every character and every line, so it writes them as drawing instructions with the fonts embedded. A scanner knows none of that. It sees a sheet of paper as a grid of coloured dots and can only store the grid, so a scanned PDF is a container holding one image per page.

Neither is wrong. They are different kinds of content wearing the same file extension, and almost every surprise people have with a PDF, from blurry printing to search finding nothing, comes down to not knowing which kind they are holding.

How to tell which you have

Three checks, each taking seconds:

• Zoom to 400 percent or more on some small text. Vector text stays crisp, with clean curves. Raster text turns soft or blocky, and often shows a faint grey haze or speckle around letters.
• Try to select a single word. If it highlights as text, the text is vector, or at least has a text layer. If your cursor drags a box over the whole area, you are selecting an image.
• Search for a word you can see. If search finds it, there is text in the file. If it does not, the page is almost certainly a picture.

The awkward middle case

A scanned page that has been through OCR passes the selection and search tests while failing the zoom test. That is because OCR does not convert the image into vector text. It recognises the words and places an invisible text layer over the unchanged picture, so the page looks exactly like the scan and behaves, for search and copy, as if it were text.

World of PDF's OCR tool works that way: recognition runs in the browser, and the output is the original page image with searchable text aligned on top. It is the right fix for a scan you need to search. It does not make the scan any sharper, because the visible page is still the same pixels.

The reverse trip exists too. Exporting a page to PNG renders vector content into pixels at whatever resolution you choose, which is the right move for putting a page into a slide or a document that only accepts images. Pick the resolution for where it is going, since once it is pixels the sharpness is fixed.

Why it decides how compression works

This is where the distinction stops being academic. Vector content is extraordinarily compact: a page of text with its font already embedded costs a few kilobytes. Raster content is what makes PDFs large, because a full-page image at print resolution can run to megabytes.

So there are two very different ways to make a PDF smaller. The first leaves the vector content alone and re-encodes only the embedded images: downsampling anything stored at more pixels than the page needs and re-compressing it more efficiently, keeping the new version only if it is actually smaller. Text stays selectable, lines stay sharp, and the saving comes from the pictures. That is the default in World of PDF's compressor.

The second renders every page to a single image and rebuilds the document from those images. It compresses harder on a scan, and it turns any vector page into a raster one: text stops being selectable or searchable and blurs when zoomed. On a text document it often makes the file bigger, because a picture of a page is larger than the instructions to draw it. The compressor here offers rasterising only as an explicit option, or as a last resort when you set a target size that no text-preserving setting can reach, and it tells you when it has done so.

The practical rule: a mostly vector document will not shrink much and should not be forced to. A mostly raster document compresses well with the first method and rarely needs the second.

Printing and scaling

Vector pages print at the printer's full resolution regardless of how the file was made, which is why a well-exported report prints razor-sharp on any machine. Raster pages print at the resolution they were captured at. A scan made at 150 DPI looks fine on screen and visibly soft on a 600 DPI laser printer, and nothing downstream can add the detail that was never recorded.

The same applies when resizing pages. Scaling vector content up or down loses nothing. Scaling a raster page up only enlarges its pixels.

Frequently asked questions