editing a PDF in the browser is not the same as turning it into an image

tech · 2026-11-02 · 4 min

editing a pdf in the browser is not the same as placing text over a screenshot. the difference decides whether the document you download is still useful to the next person who opens it.

i built PdfCanvas, a browser-based editor and an open-source TypeScript library. the site says the file stays on the device; there is no account or upload. That is a product boundary, not a claim that every PDF is easy to edit.

a page is more than what it looks like

a PDF can contain text, vector shapes, images, fonts and page-level structure. a renderer turns that into pixels so you can see it. an editor has to do more: identify what is already there, let you change it, and produce a document that still contains usable PDF content.

PdfCanvas uses pdf.js to render pages, PDFium for editing existing text, and pdf-lib to write the result. the project README describes the separation and the export format. existing pages remain vector; new text and shapes are written as PDF objects rather than flattened into a page-sized image.

that matters if someone needs to search, select or print the document later. it also makes the implementation less like drawing on a canvas than the interface suggests.

keeping the file local changes the trade-off

the editor runs in the browser. according to its FAQ, it does not send the PDF to a server, and you can check that in the browser's Network panel. i chose that boundary because contracts, invoices and forms are exactly the documents people hesitate to upload to an unfamiliar service.

the other side of that choice is simple: there is no server copy to recover. download the edited file before leaving the page. local processing is useful for privacy, but it does not replace a backup or make a damaged source document safe.

what it cannot fix

a scanned page is an image, not editable text. PdfCanvas does not perform OCR, so it cannot turn that scan into a text layer by itself. font support also matters: the project documents cases where text can be moved but not rewritten with its original font.

those are not footnotes to hide. they tell you whether the tool fits the document before you start. if the PDF is a scan, use an OCR tool first. if the original text uses an unsupported font, inspect the result before relying on it.

the useful question is not whether a PDF editor can draw on a page. it is whether the downloaded file remains a PDF you can work with. that is the constraint PdfCanvas is built around, and the reason its limitations are visible alongside its features.