JPG or PNG for a PDF: pick by what is in the picture

Convert with the format you have

The tool takes both, mixed in 1 batch, plus HEIC, WebP, AVIF, TIFF, GIF and BMP. It runs on your device, so nothing is uploaded.

The rule, and why it holds

The 2 formats compress in opposite ways:

• JPG throws away fine detail the eye does not notice in a photo: tiny colour steps in skin, sky or foliage. On a photo that saves a lot. On a sharp black letter on white, the thrown-away detail is exactly the edge, and what is left shows as a grey halo or speckles round the letter.
• PNG keeps every pixel and saves space by finding repeats. A screenshot is mostly runs of identical pixels (white background, flat bubbles), so it packs small. A photo has almost no exact repeats, so the PNG stays big.

The numbers: 2 test images, both formats

We saved 2 images both ways and made a PDF of each with getPDF on 11 October 2026. The photo is a 4032 by 3024 pixel crop (12 megapixels, a phone camera’s size) of a landscape photo; the screenshot is a 1080 by 2340 pixel test picture of a chat, a phone screen’s size. “Pixels changed” counts how many pixels differ from the original by more than 16 of 255 brightness levels.

What the table says:

• For the photo, JPG wins outright. The PNG is 8 times the size, and the JPG’s changes touch 2 pixels in 10,000. Our test photo is a smooth landscape that packs well; a typical 12 MP phone photo at quality 92 is about 3 MB, but the gap to PNG stays large.
• For the screenshot, PNG costs little. At quality 80 the JPG saves a third (156 KB to 106 KB), and the 0.57 % of changed pixels are the ones round the letters and bubble edges. On a phone screen you will not see it. Zoomed to 400 % or printed, the letters have a soft grey fringe.
• A PDF from a PNG can be smaller than the PNG. getPDF re-packs the pixels with the compression PDF uses, losslessly; for the photo that took 4.8 MB down to 3.4 MB. A JPG goes in as it is, plus a few hundred bytes.

Which one, by what is in the picture

Do not convert between them first

• JPG to PNG makes it bigger, not better. The detail the JPG dropped is gone. Our 619 KB JPG became a 5.1 MB PNG with exactly the same pixels.
• PNG to JPG smudges it for good. Once a screenshot has been through JPG, its letters keep the fringe, whatever you do later.
• JPG to JPG costs a little each time. Every re-save compresses again. Feed the converter the file the camera or the screenshot key made.

This is why getPDF does not re-encode on the way in: a baseline JPG’s picture data is copied unchanged (the camera details, GPS position, device and time, are left out), and a PNG is stored losslessly. The result card says which happened (“1 JPEG embedded as it is, no re-encoding” or “1 picture stored losslessly”).

The other formats

• HEIC (iPhone photos): PDF cannot hold it, so getPDF decodes it on your device and writes it as JPEG at quality 90.
• WebP and AVIF: often small lossy files from the web. Your browser decodes them; a photo is then written as JPEG at quality 90, the way HEIC is, so the page stays near the picture’s size (somewhat larger, because WebP and AVIF pack tighter than JPEG). A graphic with few colours or a transparent picture is stored losslessly, like a PNG.
• TIFF: stored losslessly; a black-and-white scan at 1 bit per pixel, which is very small.

The honest part

• Usually the choice is already made. The camera saves JPG (or HEIC), the screenshot key usually saves PNG, and both are the right format for their content. This guide matters when you export, save from an app, or are tempted to re-save.
• A small PDF is a compression job, not a format job. To shrink a PDF of photos for email, Compress PDF re-encodes the photos at a screen resolution and shows the size before and after; that saves far more than switching formats. Why a photo PDF is too big has the numbers.
• Neither format makes the text searchable. A picture of text is a picture in either format. OCR adds a text layer, and its result is approximate.

For every other picture question, see pictures to PDF.