Convert an image

Which format

• JPEG: photographs. Small files, no transparency, and quality degrades a little every time it is re-saved.
• PNG: screenshots, logos, anything with sharp edges, text or transparency. Lossless, so it never degrades.
• WebP: the web. Roughly 25 to 35% smaller than JPEG at matching quality, with transparency support. Every current browser reads it.
• BMP: uncompressed. Only when some old software demands it.
• TIFF: print and archival workflows.

Transparency and JPEG

JPEG cannot store transparency. Converting a transparent PNG to JPEG has to put something behind it, so the tool composites onto white by default. Untick that and transparent areas come out black instead.

The question that decides the format

Ignore the subject of the picture and ask one thing: does it have hard edges? Photographs are continuous tone, with gradients everywhere and no two adjacent pixels quite alike, which is precisely what JPEG was designed for. Screenshots, charts, logos and anything containing text are the opposite, large flat areas divided by sharp boundaries, which is what JPEG handles worst. Its block-based compression smears exactly those edges, and coloured text on a dark background is where it looks worst.

So: photograph, JPEG. Anything with text or a hard line in it, PNG. Either, if it is going on a web page and you control the audience, WebP, which handles both cases well and is smaller than both.

Quality, and where the returns stop

The default of 85 is the standard sweet spot. On a photograph it is visually indistinguishable from 100 and several times smaller. The saving is heavily front-loaded: dropping from 100 to 85 typically halves the file, while dropping from 85 to 70 saves perhaps another fifth for a much more visible cost. Below 70, artefacts around edges become obvious even on photographic content.

The setting has no effect on PNG and BMP, which are lossless and store what they are given exactly.

Every JPEG save costs something

JPEG is lossy each time it is written, not just the first time. Crop, save, annotate, save, resize, save, and each pass adds artefacts on top of the last, concentrated where the previous pass already did damage. After four or five rounds a screenshot with writing on it visibly degrades. If a picture is going through several rounds of editing, convert it to PNG at the start with JPG to PNG, do the work, and export to JPEG once at the end.

Resizing while you convert

Setting a width scales the picture during the same pass, which saves a second encode and is usually the larger win anyway. An image displayed 600 px wide on a page has no business being 3000 px wide in the file, whatever the quality setting is. For more control over shape, letterboxing and exact dimensions, resize image is the fuller tool.

Common questions

Related tools