PDF to JPG or PNG

No. Rendering happens in this tab with pdf.js drawing onto an OffscreenCanvas. This is one of the tools marked browser-only for that reason: it needs a canvas, which the command line version has no equivalent for. No page image ever leaves your device.

It follows the scale setting, which defaults to 2. A scale of 1 means 72dpi, so the default gives you 144dpi, and 4 would give 288dpi. Higher numbers mean sharper images, more memory and a longer wait. Anything that is not a number above zero is rejected with "scale must be a number greater than 0".

Because the canvas your browser was asked for exceeded its own size limit, and the tool refuses to carry on with a broken context. The message names the page, the scale and the pixel dimensions it tried, then tells you to try a smaller scale. Halving the scale usually clears it.

Yes. The pages field takes 1-based numbers and ranges, like 1-3,5,8- for pages one to three, page five, and everything from page eight on. Leave it blank for the whole document. A number past the end stops the run with a message naming the range and the document's real page count.

The PDF name without its extension, then a three-digit page number: report.pdf gives report-001.png, report-002.png and so on, using the original page numbers rather than a fresh count. Choosing jpeg names them .jpg. They arrive as separate downloads, one per page, not as a ZIP.

No. This one draws each page and gives you a picture of the whole page, text included. Extract PDF images pulls out the photos already embedded in the file, at their original resolution and without re-encoding. Use this to make a page look, and that to recover a source image.