Grayscale, B&W or Color: Picking a Scanned PDF Tone

You need the invoice to look like it went through the office copier, not like it came off a laser printer ten seconds ago. Two controls in Make Scanned PDF stand between you and that: Tone and Scan intensity.

Most people reach straight for Black & White on Heavy, on the theory that more effect equals more convincing. The result looks like a fax that survived a flood, and the blue rubber stamp on the second page has disappeared entirely.

The two settings do genuinely different jobs. Here is what each one changes.

Tone: which machine you are pretending to be

Color

Keeps the original colours and adds grain and a tonal shift on top. Use it when the source has something coloured that matters — a company logo, a blue-ink signature, a red PAID stamp, a coloured letterhead rule.

This is what a modern multifunction machine set to colour scanning produces, and it is also the largest output file, since colour JPEG data does not compress as tightly as grey.

Grayscale (the default)

A proper luminance conversion — the standard weighted mix of red, green and blue, not a naive average — followed by grain, a contrast lift and an edge shadow.

This is the setting that matches what most office scanners actually deliver, because most offices leave their machines on grayscale. If you are not sure which to pick, this is the answer. It is the least likely to look staged.

Black & White

Worth understanding, because it does not do what the name suggests.

It is not a hard one-bit threshold. That approach — everything above a cutoff becomes white, everything below becomes black — is exactly what breaks thin text, hollowing out serifs and making 8-point footnotes vanish. Instead, anything above a fairly high brightness point is snapped to clean white so the paper reads as paper, and everything darker is deepened with extra contrast while keeping its edges. The grain is also deliberately held back in this mode, because speckles on a two-tone page look like dirt rather than like scanning.

The result is a crisp photocopier look with a genuinely white background and the smallest file of the three.

The catch: colour is gone. Your blue stamp becomes a grey-black smudge, and a light-yellow highlight becomes nothing at all. Check page by page in the preview before you commit.

Scan intensity: how tired the paper looks

The tool describes them itself as a subtle clean scan, a typical office scan with light grain and skew, and a heavy grain, stronger skew and contrast look for an old or worn document.

One detail that does more for believability than any of the numbers: the skew is randomised per page within the maximum. Page 3 might lean 0.7° left and page 4 lean 0.3° right. Pages do not all tilt the same way, which is precisely what happens when a stack goes through a sheet feeder. A uniform tilt across every page is one of the clearest signs of a filter applied after the fact.

Choosing by what you are matching

If you are trying to match an existing set of scans in the same folder, open one of them next to your preview and compare the paper, not the text. Real scans differ mostly in background tone — warm grey, cool grey, near-white — and that is what the eye picks up first.

Two things no setting can fix

The output has no selectable text. Every page becomes a flat image at 150 DPI, which is the whole point — a scanned-looking PDF whose text you can highlight gives itself away in one keystroke. It also means search, copy, hyperlinks, bookmarks and form fields are all gone.

Perfect typography still reads as digital. Grain and skew are surface effects. They do not change the fact that the margins are mathematically identical on every page, the baseline grid never drifts, and the ink density is uniform. Against a casual glance, the effect holds up well. Against someone who looks at scans all day, a document that was never on paper usually still looks like one.

Where the file size goes

Pages are rendered at 150 DPI and encoded as JPEG at high-but-not-maximum quality. That is a sensible balance, but the direction of travel is only one way: text that was a few kilobytes of drawing instructions becomes a photograph of text.

A 12-page text PDF that arrived at 240 KB routinely comes out between 3 MB and 5 MB. If it is going into an upload form with a cap, run it through the PDF compressor afterwards — and because there is no text layer left to protect, the image-based Screen level is fair game here and does the most work. To check what you have before uploading, File Info gives you the finished size, page count and format in one screen.

Everything runs in your browser, so nothing about your document is uploaded during the effect itself. On a long document that also means your device is doing the work, and a phone will be noticeably slower than a laptop.

Getting the text back afterwards

If you later need the file searchable, PDF OCR will read the pages and rebuild a text layer.

Note the tension there, though. You deliberately added grain, and grain is exactly what lowers recognition accuracy. Light intensity on Grayscale OCRs cleanly. Heavy on Black & White does not, and you will spend time correcting the output. If you know from the start that both looks are needed, keep the clean original file and produce the scanned version as a copy. That way the searchable version is the untouched one.

The same logic applies if you need the pages as separate images later — PDF to images will export them, but it exports what you made, grain and all.

One line on using this honestly

Matching an archive, building a template, mocking up a demo, or making a digital record sit comfortably alongside twenty years of scanned files — all fine, all ordinary reasons this tool exists.

Passing off a document that was never signed, stamped or issued as a genuine scanned original is fraud, and a grain filter does not change that. It is worth being clear about, because the effect is convincing enough that the line is easy to cross without thinking about it.

The rest of the scanning and recognition tools are grouped under Scanners.