JPG to PNG: How It Works
Converting JPEG to PNG produces a larger file with no improvement in quality — the lost detail does not return. There are three situations where it is nevertheless the right move, and knowing them prevents a lot of wasted disk space.
What conversion does and does not do
JPEG discards data permanently during compression. PNG stores whatever it is given without loss. Converting therefore preserves the already-degraded pixels exactly — including every compression artefact — in a container typically three to five times larger.
| Property | Before | After |
|---|---|---|
| Visible quality | As compressed | Identical |
| Artefacts | Present | Present, preserved perfectly |
| File size | 400 KB | 1.5–2 MB |
| Further editing | Degrades each save | No further loss |
When it is genuinely worth doing
- You are about to edit repeatedly. Each JPEG save re-compresses and loses more. Convert once to PNG, do all the editing losslessly, then export a final JPEG. This is the strongest reason.
- You need to add transparency. Removing a background requires an alpha channel, which JPEG has none of.
- A system requires PNG. Some print workflows, design tools and upload forms accept only PNG.
Outside these, converting simply consumes space.
Why the file grows so much
PNG uses lossless compression, which works by finding repetition. Photographs have very little exact repetition — adjacent pixels are similar but rarely identical — so there is little for the algorithm to exploit. JPEG's advantage comes precisely from being allowed to discard the differences PNG must preserve.
Ironically, a heavily compressed JPEG converts to a smaller PNG than a lightly compressed one, because aggressive compression flattens detail into repeating blocks that PNG can encode efficiently. The PNG is larger, and it is faithfully preserving worse pixels.
Removing a background
The common workflow is: convert to PNG, remove the background to create transparency, then export. Two points worth knowing. JPEG artefacts cluster around high-contrast edges — exactly where a subject meets its background — which makes clean selection harder than it would be from an original. And once transparency exists, the image can no longer be saved as JPEG without the transparency being filled with a solid colour.
Choosing the PNG variant
| Type | Colours | Suits |
|---|---|---|
| PNG-8 | Up to 256 | Simple graphics, icons |
| PNG-24 | 16.7 million | Photographs |
| PNG-32 | 16.7 million + alpha | Photographs with transparency |
A photograph converted to PNG-8 will show visible banding in gradients, because 256 colours cannot represent smooth continuous tone.
A better option in most cases
If your reason for converting is transparency, WebP supports transparency with lossy compression, giving a file far smaller than PNG. If the reason is lossless editing, that is what PNG is for — but export the final version as JPEG, WebP or AVIF rather than shipping the PNG to a website, where the extra megabytes serve no purpose.
Conversion here happens in your browser; nothing is uploaded.