Sign PDF (draw or image)

No, and the tool is explicit about it. This stamps a picture of a signature onto a page, exactly the way the watermark tool stamps text. A certificate-based signature is a different thing: a detached CMS signature over the document byte range, with a trust chain and a signature form field. None of that is built here.

Draw it with a mouse, trackpad or finger; photograph or scan a signed sheet, and the paper is lifted off by judging every pixel against the paper beside it; or type your name in one of four faces. A PNG or JPG dropped alongside the PDF is stamped as it is, and wins over the rest. Text, boxes, circles, lines, arrows and pen strokes are placed by dragging on the page.

The signature goes on the last page by default. Type a page number to choose another one; it must be between 1 and the document's page count, and you get "page must be between 1 and N" if it is not. One signature per run, but every annotation remembers the page you drew it on, so one run marks up the whole document. Positions are saved as fractions of the page, so they land correctly whatever the sheet measures.

Because no file ending in .pdf was among the ones you dropped. This tool takes several files at once so it can pair a PDF with a signature image, and it picks the PDF by extension. If you gave a PDF but neither text nor an image, you get a different message asking for signature text or a PNG or JPG.

The width setting decides, in points, defaulting to 160. An image keeps its aspect ratio: the height is worked out from the width you asked for and the image's own proportions. Pull the corner of the box on the preview and the width follows, which is the same setting under another name.

No. Both the PDF and the image are read into a Web Worker in this tab, combined by pdf-lib, and handed back as a file with a -signed suffix. A scanned signature is about as sensitive as a file gets, and this one never goes anywhere near a network.