Images to PDF

Images to PDF assembles your selected images into a single multi-page document entirely in the browser using the pdf-lib library (PDFDocument.create). Each image becomes one page, and pages are emitted in the exact order shown in the thumbnail grid, which you reorder by dragging tiles (the numbered blue badge on each thumbnail is its page number). How an image is embedded depends on its MIME type: image/png files go through embedPng and image/jpeg (or image/jpg) files through embedJpg, so a JPEG photo keeps its original compressed bytes untouched and does not get re-encoded or bloated. Any other format (BMP, WebP, GIF, etc.) is first drawn onto an HTML canvas and re-exported as PNG (canvas.toDataURL('image/png')) before embedding, which is lossless but can make a photographic source noticeably larger than the JPEG path would.

The Settings panel exposes four controls: Page Size, Orientation, Image Fit, and Margin. Page Size offers fixed presets with exact PDF-point dimensions stored in the PAGE_SIZES table - A4 (595.28 x 841.89), Letter (612 x 792), Legal (612 x 1008), A3 (841.89 x 1190.55), and A5 (419.53 x 595.28) - plus a 'Fit to Image' option that sizes each page to the image's own pixel dimensions plus the margin on every side (pageWidth = imgWidth + margin*2). Because pdf-lib treats one pixel as one point (72 per inch), Fit-to-Image turns a 3000x2000px phone photo into roughly a 42x28-inch page, so the fixed presets are the right choice for anything you intend to print or email. Orientation 'auto' picks landscape whenever the image is wider than it is tall (imgWidth > imgHeight) and portrait otherwise, while 'portrait' and 'landscape' force the choice for the whole document. Margin (the field is labeled 'px' but the value is applied directly as PDF points, default 20, capped at 100, min 0) reserves a uniform border, and the image is then centered in the remaining area.

Image Fit controls how each picture is scaled into its page: 'contain' (the default) scales by Math.min of the width/height ratios so the whole image fits with letterboxing on mismatched aspect ratios; 'cover' scales by Math.max of the two ratios to fill the page, cropping the overflow; and 'stretch' forces drawWidth/drawHeight to the available width and height, distorting the aspect ratio. One real caveat is worth knowing about the per-thumbnail rotate button (90-degree steps): for the canvas/re-encode path it bakes the rotation into the bitmap via ctx.rotate (though the canvas keeps the original un-swapped dimensions, so a 90/270-degree rotation gets clipped), whereas for native PNG and JPEG it only swaps imgWidth and imgHeight in the page-sizing and fit math - the image is passed to page.drawImage with no rotation argument, so a rotated JPEG or PNG changes the page's proportions but the embedded picture is drawn unrotated. The tool is a layout-and-assembly utility, not a scanner: there is no deskew, auto-crop, or OCR, and because every image's bytes and decoded bitmap are held in memory while building the document, very high-resolution stacks are bounded by the browser tab's RAM.