Convert Images to PDF — Free, No Upload
Drop in your images, put them in the order you want, and get one PDF with one image per page. A JPEG is copied into the document without being re-compressed, so it comes out exactly as sharp as it went in. All of it happens on your device — nothing is uploaded, and the page keeps working with the network off.
Drop images here or browse
JPG, PNG, WebP · one image per page · nothing is uploaded
What a page in a PDF actually holds
A PDF page is a dictionary. It carries a MediaBox — four numbers giving the corners of the sheet in points, where a point is 1/72 of an inch — and a content stream, which is a very short program in a drawing language. The image is not part of the page. It is a separate object, an XObject, that the page refers to by name, and the content stream says where to put it using six numbers: a matrix that maps the image's unit square onto the sheet. The page itself costs a few hundred bytes; everything else is the image.
Why a JPEG goes in untouched and a PNG cannot
PDF has a filter called DCTDecode, and DCTDecode is JPEG. The compressed data inside a .jpg file is exactly what an image XObject wants, so the bytes are copied across without ever being decoded. Not one coefficient changes.
Most converters do not do this. They decode the JPEG to pixels, draw it on a canvas and encode a new JPEG, which stacks a second generation of lossy compression on the first. Edges get softer, blocks get blockier, and nothing is gained.
PNG has no equivalent. There is no PNG filter in PDF, and the nearest match — Flate, the same deflate compression PNG is built on — works on raw samples rather than on PNG's filtered scanlines, so a PNG cannot be copied across. Choosing keep every pixel stores the exact RGB values deflated: nothing is lost, but the result is often larger than the PNG was, because PNG's per-row prediction does work that deflate alone cannot. The JPEG option re-encodes instead: smaller, and lossy. WebP sits in the same position. Transparency survives the lossless route as a soft mask, a second grayscale image holding the alpha channel; the JPEG route flattens it onto white, because JPEG has no alpha at all.
Why phone photos come out sideways in other converters
A phone usually stores a photo in the sensor's own orientation and writes a small EXIF tag saying which way to turn it. PDF has no EXIF tag and no concept of one. A converter that copies the JPEG bytes and ignores that tag produces a document full of photos lying on their side. That is exactly why so many tools re-encode every image.
There is a better answer. The rotation is already a matrix and the placement is already a matrix, so this tool multiplies the two together and writes the single result into the content stream. The JPEG stays untouched and the photo stands up. All eight EXIF orientations are handled this way, including the mirrored ones.
The EXIF block itself does not travel. The compressed picture data is copied byte for byte, but the metadata wrapped around it — GPS coordinates, camera serial number, the embedded thumbnail, any comment — is dropped on the way in. A PDF viewer would never show it, and you probably did not mean to hand it to whoever you send the file to.
Page size, and why DPI is not part of it
Fit-to-image makes each page exactly as large as its image, counting one pixel as one point. A 1600-pixel-wide screenshot becomes a page about 22 inches across, which mostly does not matter: a viewer scales the page to the window and a printer scales it to the paper. It starts to matter when you are printing to a real size, which is what A4 and US Letter are for.
Enlarging an image to fill an A4 page does not resample anything. The pixels are stored once and the page merely records how big to draw them, so the only thing that changes is the effective resolution. A 600-pixel-wide image on A4 prints at roughly 72 dots per inch and will look like it — if the file is meant for print, start from a bigger image rather than stretching a small one. One hard limit applies in fit mode: a PDF page cannot be more than 14400 points, 200 inches, on a side, so an enormous image has its page scaled down to that ceiling.
Where this tool stops
- The text in the result is not text. A picture of a page is a picture. Nothing is searchable, selectable or copyable, because there is no OCR here — OCR needs a trained model, and a model needs either a server or a very large download.
- HEIC and camera RAW depend on your browser. The tool can only use formats the browser itself decodes. Safari opens HEIC; Chrome and Firefox generally do not, and nothing here opens a RAW file.
- CMYK JPEGs are re-encoded. A four-channel JPEG needs a colour inversion that is guessed wrong more often than right, so those go through the canvas and come out as RGB.
- A damaged JPEG is rebuilt, not copied. If the markers stop making sense before the picture data starts, the file is decoded and re-encoded instead. Copying it would mean carrying metadata the tool could not read past. The row says when this happened.
- Animated GIFs keep one frame. The first one.
- Colour profiles are dropped. An image in Display P3 is converted to sRGB by the canvas, and a JPEG copied verbatim loses its embedded profile, so viewers read it as sRGB.
- No password, no signature, no shrinking pass. The output is a plain unencrypted PDF, and the image data is stored exactly as described above. If the file is too big, the answer is to compress the images first.
- One image per page. There is no contact sheet and no two-up layout.
Frequently asked questions
Does converting an image to PDF lose quality?
It does not have to. A JPEG is copied into the PDF without being decoded, so the picture is bit-identical to the file you started with. PNG and WebP cannot be copied that way, so they are either stored as exact pixels with deflate compression, which loses nothing but makes a larger file, or re-encoded as JPEG, which is smaller and lossy. You choose which.
Can I select or search the text in the PDF this makes?
No. The PDF contains pictures of your pages, not text, so there is nothing for a search to match. Making the text searchable needs OCR, which needs a trained model that would have to be downloaded or run on a server. If you need searchable output, scan directly to PDF with software that runs OCR.
Why is my PDF bigger than the images that went into it?
Usually because those images were PNG or WebP. Storing them without losing pixels means deflating the raw samples, and deflate alone compresses worse than PNG does, because PNG also predicts each row from the one above it. Switching the setting to re-encode as JPEG will make the file much smaller, at the cost of some quality.
Which page size should I pick, fit-to-image or A4?
Pick fit-to-image when the file is for reading on a screen and you do not want borders; each page then matches its image exactly. Pick A4 or US Letter when the file is going to a printer or into a stack of other documents, so every page is the same physical size. Margins only apply with a fixed page size in mind.
Why do photos from my phone come out rotated in other converters?
Phones store the photo the way the sensor saw it and add an EXIF tag saying how far to turn it. PDF has no such tag, so a converter that copies the file across without reading it leaves the photo on its side. This tool reads the tag and folds the rotation into the matrix that places the image on the page, which keeps the photo upright without re-encoding it.
Are my images uploaded anywhere?
No. The images are read, the PDF is assembled and the file is handed back to you entirely inside your browser. Open your developer tools, watch the Network tab while you build a document, and you will see no request at all. Turning the network off after the page has loaded does not stop it working.
Last updated September 19, 2026