Why the PDF defaults to A4, and what a wide page looks like on paper

A web page has one dimension paper does not: it is as tall as it wants to be. It is also as wide as your window, which is wider than any sheet you own. Turning that into a PDF means answering two questions, how wide and how tall, and the answers decide whether the file prints, whether the text is readable on paper, and whether it opens at all. Page Scanner's extract sheet shows the answers before you save. This post is the arithmetic behind them.

The two ways to put a page on a page

Single long page makes one PDF page the exact size of your crop, no margins, no scaling. It is the most faithful output there is: a pixel on screen is a point on the page, at 0.75 points per CSS pixel. A 1280 px wide page that scrolls 5,882 px is a page 13.3 inches wide and 61.3 inches tall. That is a banner, not a document. No home printer will take it, and Acrobat clamps any page past 200 inches on a side rather than rendering it, which is why the option says so on its own card.

A4 pages and US Letter pages slice the crop across sheets of paper with a half-inch margin on every edge, because a sheet is a thing that gets printed and no consumer printer reaches its own paper edge. The crop is scaled to fit the width inside the margin, then cut top to bottom. This is what a printer wants, and it is why A4 is the default.

What the scaling does to the text

A4 is 210 × 297 mm, which is 595 × 842 points. Take a half inch, 36 points, off each side and 523 points of width are left. A 1280 px window is 960 points wide. To fit, the page is scaled to 523 ÷ 960, which is 55 %. Body text at 16 px, the browser's default and the commonest size on the web, is 12 points on screen, and at 55 % it lands on paper at 6.5 points. That is legible and small: about the size of the fine print on a form.

US Letter is a little wider, 612 points, so its printable width is 540 points and the same page scales to 56 %, which puts the text at 6.8 points. The difference is not worth choosing a paper over; choose the paper your printer holds.

The extract sheet says all of this on the card for the layout you have chosen: the scale, and what it does to 16 px text, in points. There used to be a warning when the text fell under 8 points. It went, because almost every page is wider than 1,000 px, so it fired on the ordinary case and taught people to ignore the colour. The number is still there without the alarm.

The way round it: capture at the sheet's width

Scaling is only necessary because the page was laid out wider than the paper. Lay it out narrower first and there is nothing to scale.

That is what Narrowed to A4 and Narrowed to US Letter in the popup do. Before the capture, the page is laid out at the width of the sheet's printable area, 698 CSS px for A4 and 720 for Letter, the way it would look in a narrow window, and the site's own responsive layout does the rest: columns stack, the sidebar drops under the text, the type stays 16 px. On paper that capture lands at 100 % scale on A4, and 16 px text is 12 points, the size it is in a printed book. On Letter it is 103 %, and 12.4 points.

The catch is that it is a viewport width, not a print stylesheet. A page that lays itself out differently for print does not get that treatment, because the capture is deliberately of the page on screen. And a page that cannot reflow that narrow is captured at the width it insists on, and the editor tells you what that width is rather than clipping it to make the number come out.

Where the page breaks fall

Cutting a tall capture into sheets means cutting somewhere. A ruler laid every 11.7 inches will land in the middle of a line of text about as often as not. Page Scanner reads the capture's own bitmap instead and moves each break up, by at most 12 % of a sheet, to the middle of the widest quiet run it can find, so no line is sliced in half. A page with nowhere quiet to cut, a full-bleed image, keeps the arithmetic break rather than hunting for one. It costs at most one extra sheet, the sheet count on the card is the count the file will have, and the cuts are drawn on the stage before you extract, so you can move the crop if one falls badly.

When to choose which

Choose A4 or Letter if the PDF will be printed, or read in a viewer that pages, or handed to someone who expects a document. It is the default because that is most PDFs.

Choose Narrowed to A4 or Letter as well if the text has to be readable on paper at its normal size. It is the difference between 6.5 points and 12.

Choose Single long page if the layout is the point and the file is for a screen: a design review, a record of a dashboard, a page whose columns must stay side by side. Check its height on the card first; a page taller than 200 inches is a file that opens wrong.

Frequently asked questions

Why not shrink the text less and let the page overflow the sheet? Because a sheet that is wider than the paper prints with its right edge missing. The width has to fit; the height is what gets cut.

Why A4 and not Letter as the default? A4 is the paper everywhere except North America, and an A4 layout printed on Letter gains margin where a Letter layout printed on A4 loses its edges. The command line and the MCP server default the same way, so the two cannot disagree.

Can I tell the page width from the command line? Yes: --page-width a4 on the CLI and captureWidth on the MCP server's scan_page do what the popup's choice does.

What about a page with a video or a canvas? A video is captured as the still it is paused on, and text drawn into a canvas is a picture, at whatever scale the page gets. Neither changes with the paper.