Raster vs Vector in PDFs
A PDF is a container, not a single kind of image. Inside one file you can have crisp text, mathematically described line art, and pixel-based photos, all sitting on the same page. Knowing which parts are vector and which are raster explains almost everything about why a PDF looks the way it does, how big it is, and what you can safely do to it.
The two kinds of content
Vector content is described by math: coordinates, curves, font outlines, and fills. The text you select and the shapes in a diagram are vector. Because they're instructions rather than fixed pixels, they're resolution-independent — the renderer redraws them at whatever size and DPI you view or print, so they stay sharp at 100% or 1000% zoom. Vector data is also usually compact, since "draw a line from A to B" takes far fewer bytes than storing every pixel along that line.
Raster content is a grid of pixels at a fixed resolution. Photographs, screenshots, and scanned documents are raster. A raster image has a set number of pixels; zoom past its native resolution and you see softness or blocky artifacts because there's no extra detail to reveal. Raster data tends to dominate file size, especially photos and high-DPI scans.
Most real PDFs are a mix. A report might have vector body text over a raster company logo photo. A scanned contract is the opposite extreme: the entire page is one big raster image, so even the "text" is just pixels — you can't select or search it until you add a text layer.
Why the distinction matters
Quality. Vector text prints razor-sharp at any size. Raster text is only as good as its scan resolution — a 150 DPI scan will never look as clean as native text, no matter how you process it.
File size. If your PDF is unexpectedly large, the culprit is almost always raster images, not text. Vector pages are tiny by comparison.
Searchability and accessibility. Vector text is real text: selectable, searchable, copyable. A raster scan is a picture of text. To make it searchable you need OCR — /ocr-pdf adds an invisible text layer over the image so search and copy work while the original scan stays visible.
What happens when you rasterize a page
Rasterizing means rendering a whole page — vector text, line art, and all — down to a single flat pixel image at a chosen resolution. After that:
- Text is no longer selectable or searchable; it's now pixels.
- The page is locked to one resolution, so zooming or large-format printing can look soft.
- Hidden layers, form fields, and metadata are baked flat and gone.
Rasterizing is sometimes exactly what you want. It guarantees the page renders identically everywhere, it strips interactive layers, and it can shrink a file that's bloated with inefficient vector data. That's why /compress-pdf offers two modes: smart, which recompresses the embedded raster images while leaving vector text selectable, and rasterize, which flattens each page to a JPEG for maximum size reduction at the cost of selectable text. Reach for smart first; use rasterize only when you genuinely need the smaller file more than the text layer.
Converting between the two
You can turn vector PDF pages into raster images with /pdf-to-images, which renders each page to PNG or JPEG at the resolution you pick. The reverse — wrapping raster images back into a PDF — is /images-to-pdf. Note that going PDF to image to PDF discards the vector text permanently; it's a one-way trip unless you OCR afterward.
A practical rule: keep content as vector for as long as you can. Only rasterize when you have a specific reason — a smaller file, a flattened layout, or a fixed-render guarantee — and understand you're trading away sharpness and searchability for it.
Related tools
To work with this in practice, try /pdf-to-images to render pages as pixels, /compress-pdf to shrink raster-heavy files, and /ocr-pdf to add searchable text back to a scan.