The most common job: paperwork photographed on a phone
Most JPGs that need to become a PDF are not photographs in the artistic sense. They are a contract photographed page by page, a passport scan from a flatbed, a stack of receipts for an expense claim, or a signed form that has to go back as one attachment. A folder of JPGs is awkward to send and easy to lose pages from; a single PDF keeps the order fixed and opens in every mail client and document viewer.
How the PDF is assembled
Each JPG is decoded in the browser, drawn onto a canvas at its full pixel size and re-encoded as JPEG at quality 92. The tool then creates a PDF with jsPDF, sized so that every page matches the pixel dimensions of its image: a landscape photo becomes a landscape page and a portrait photo a portrait page. Images are placed one per page in the order they appear in the list, which is the order you dropped or selected them. The finished file downloads as images.pdf; rename it afterwards.
Pages are therefore not A4 or Letter. A 4032 by 3024 photo produces a page with those proportions, and a PDF viewer or printer scales it to fit the paper. This is fine for reading and printing, but if you need true A4 pages with margins, print the PDF to a new PDF from your viewer with the page size set.
Quality and generation loss
Because the image is decoded and encoded again, there is a second round of JPEG compression. At quality 92 the loss is minor, well below what a camera or scanner already applied, and text remains legible. A 2.8 MB phone photo typically lands between 1.5 and 2.5 MB inside the PDF; three such pages give a PDF around 5 to 7 MB. If that is too large for an email attachment, run the photos through the compress and resize tool first, capping width at around 2000 pixels, and a three page document usually drops to 1 to 2 MB. These are typical numbers, not guarantees.
Orientation and metadata
Phone photos carry an EXIF orientation flag rather than being physically rotated. Modern browsers apply that flag when decoding, so a page photographed in portrait comes out upright in the PDF. What does not come across is the rest of the metadata: GPS position, camera model, capture time and any embedded colour profile are dropped when the image passes through the canvas. For a document you are sending to a stranger that is usually welcome. If you want to see what was in the original first, the EXIF viewer shows it.
Practical tips for document photos
Straighten and crop each page before combining. A photo of a page on a desk includes the desk, and a viewer will show all of it; the crop and rotate tool fixes that per image. Shoot in even light to avoid a dark gradient across the page. Keep all pages the same orientation so the reader does not have to rotate every other page. And drop the files in reading order, since the tool preserves that order and does not sort by name.
Limits
The whole PDF is built in memory on your device, so very large batches of full resolution photos can exhaust the memory of a phone or an older laptop; twenty pages of 12 megapixel images is a realistic ceiling on many devices. The PDF contains images only, no searchable text layer. If you need the text, run the pages through the OCR tool separately. Nothing is uploaded at any point.