JPG to PNG: when it helps and when it is pointless

PNG cannot undo JPEG

The most common reason people convert JPG to PNG is a misunderstanding worth addressing first: PNG is lossless, so surely saving as PNG improves quality?

It does not. Lossless means PNG will not add new damage. The JPEG artefacts the blocking in the sky, the ringing around the text, the smearing in fine texture, are already in the pixels. PNG stores them perfectly, at three to five times the file size.

Converting a photograph from JPEG to PNG reliably produces a larger file that looks identical. If that is what you are doing, there is usually a better answer below.

When it is the right move

You need transparency. JPEG has no alpha channel. Anything that has to sit over a background without a white box around it must be PNG (or WebP). This is the main legitimate reason.

You are about to edit it repeatedly. Every JPEG save re-compresses and loses a little more. If a file is going through several rounds of editing, converting to PNG once and working there stops the accumulation, though it does not repair what the first save already cost.

It is not really a photograph. Screenshots, diagrams, charts, logos, and anything with flat colour and hard edges are the case PNG was designed for. A screenshot of text as PNG is usually smaller than the same screenshot as JPEG, and enormously cleaner: JPEG's compression is built around the assumption of gradual change, and text is the opposite of that.

Something downstream demands it. Some print workflows, some game engines, some upload forms. Not a technical reason, but a real one.

When it is not

To "improve" or "restore" a photo. Nothing is recovered.

To make it print better. Print quality is resolution and colour, not container. A 600-pixel image is too small for print as a PNG exactly as much as it was as a JPEG.

To remove a watermark or compression artefacts. Neither is a property of the format.

For the web, as a default. A photograph as PNG on a web page costs your visitors several times the bytes for no visible gain. Photographs want JPEG or WebP; PNG is for the flat-colour cases above.

What PNG actually gives you

• Lossless. Decode and re-encode as many times as you like; the pixels do

not change.

• Alpha. Full 8-bit transparency, not just a single transparent colour.
• Exact colour. No chroma subsampling. JPEG stores colour at half

resolution, which is invisible in photographs and very visible on a thin red line or coloured text.

• Universal support. Every browser, every editor, every operating system,

for twenty-five years.

What it does not give you is small photographs. It has no lossy mode at all.

Consider WebP instead

If the destination allows it, WebP does both jobs better than either: lossy mode competitive with JPEG at smaller sizes, lossless mode competitive with PNG, transparency in both, and support in every current browser.

The reason to still use PNG is compatibility with things that are not browsers (older software, some print workflows, some corporate systems) and that is a real constraint rather than an imagined one.

What happens to the metadata

Converting through a canvas (which is how a browser does it) produces a file containing pixels and nothing else. EXIF does not survive: camera, lens, timestamps, GPS.

For an image about to be published that is usually a feature. If you need the metadata, keep the original alongside.

Watch for the orientation flag: phone photos often store rotation in EXIF rather than in the pixels, so a photo that looked upright can come out sideways once the flag is gone. Reading the orientation and applying it before conversion is the fix, and a tool that does not do it will surprise you on exactly the photos taken in portrait.