Why BMP files still exist
BMP is the native bitmap of Windows. It is what Paint saved by default for years, what many older scanner drivers and industrial cameras produce, and what some medical and laboratory software still exports. The format stores every pixel uncompressed, so a scanned A4 page at 300 dpi is around 25 MB regardless of whether it contains a dense contract or a mostly blank sheet. Folders of such files are painful to email or archive, and a PDF solves both problems.
The conversion, step by step
Browsers decode BMP natively, so each file is read on your device, drawn onto a canvas at its full pixel size and encoded as JPEG at quality 92. jsPDF then creates one page per image with the same pixel dimensions as the bitmap, portrait or landscape as appropriate, in the order the files were added. The whole document is assembled in memory and downloaded as images.pdf. No server sees the bitmaps, which matters for scanned contracts and medical images.
File size: where the savings come from
This is the conversion on the site with the most dramatic size drop. A 24 bit BMP of 2480 by 3508 pixels, an A4 page at 300 dpi, is 26 MB on disk. Encoded as JPEG at 92, a typical text scan comes out between 600 KB and 1.2 MB per page; a colour photograph of the same dimensions is more like 2 to 3 MB. Ten scanned pages that occupied 260 MB as BMP usually become a 6 to 12 MB PDF. These are typical values and depend on how much detail the scan contains. If the result is still too big to send, run the BMPs through the compress and resize tool first and reduce the width to around 1700 pixels, which keeps text at roughly 200 dpi legibility.
What is lost and what is not
JPEG is lossy, so the PDF page is not a pixel-perfect copy of the bitmap. At quality 92 the difference on a text scan is invisible without magnification. For material where every pixel matters, such as measurement images or images that will be analysed by software, keep the BMP and produce the PDF only as a viewing copy. BMP files rarely carry metadata, so nothing meaningful is dropped there. Colour depth is reduced to 8 bits per channel, which only matters if the source was a 16 bit greyscale export from specialised hardware, and most browsers cannot decode those exotic variants anyway.
Preparing scans before combining
Scanner output often includes a black border from the lid or a page that is slightly rotated. The crop and rotate tool handles both and its output feeds straight back into this converter. Keep pages in a consistent orientation. If you have a mix of colour and greyscale scans, that is fine; each page is encoded independently. Because BMP stores nothing about page order, the sequence in the PDF is exactly the order you dropped the files, so name and select them carefully.
Limits to know
Large bitmaps are memory hungry twice over: once as the decoded pixels and once as the JPEG data string before it is written into the PDF. A phone may struggle beyond ten or so A4 scans at 300 dpi; a laptop handles far more. The PDF has no text layer. If you need to search the scans, the OCR tool can extract the text page by page. Nothing is uploaded at any stage.