The Best JPEG Quality Setting for Photos

What the quality number means

A JPEG quality setting is a number, usually between 0 and 100, that tells the encoder how much detail to keep. Higher numbers keep more detail and give larger files, lower numbers discard more and give smaller files. The number is not a percentage of the original quality. It is a scale that each encoder defines for itself, which is why quality 80 from one program does not look exactly like quality 80 from another.

The relationship between quality and file size is not a straight line. Going from 100 to 90 typically removes a large part of the file size with almost no visible change, while going from 50 to 40 saves comparatively little and starts to show artifacts. Most of the benefit comes in the upper part of the scale.

A sensible range

For photographs on websites and in messages, a setting between 70 and 85 is a good starting range. In that range, files are usually far smaller than at 95 or 100, and the difference from the original is hard to see at normal viewing size. This tool starts at 75, which is a common compromise, and the slider lets you move up or down in single steps.

Settings above 90 are worth using only when the image will be edited again, printed large or inspected closely. They produce much larger files with little gain on screen. Settings below 60 are for thumbnails, previews and situations where size matters more than appearance.

What it depends on

The best value depends on the picture. Smooth images such as portraits, skies and product shots on a plain background show artifacts sooner than busy images such as foliage, crowds and textures, because errors are more visible on smooth areas. Pictures with text, thin lines or sharp edges need a higher setting than pictures without them.

It also depends on the size at which the image is shown. A picture that is displayed small can be compressed harder than one that fills a screen, because compression artifacts are tiny details. And it depends on the audience: a photographer's portfolio deserves more quality than a thumbnail in a list of search results.

How to choose in three steps

First, decide the size at which the image will be displayed and resize to it, if possible. Reducing the pixel count is the most effective step, and it makes every quality setting more efficient. Second, compress at 75 and look at the result at the size where it will be used, not zoomed in. Third, lower the setting in steps of five until you can just see a difference, then go back up one or two steps.

This method finds the value for each kind of image instead of relying on a rule. If you have many similar photos, test on two or three representative ones and apply the setting to the whole batch. The compressor processes all images again whenever you move the slider, so you can see the effect immediately.

Artifacts to look for

When you inspect a result, look at areas with smooth gradients, such as a sky or a wall, for banding and blotches. Look at edges and text for halos and ringing, a faint shimmer around contrast edges. Look at skin and faces for a waxy, over-smoothed look, which is a sign that fine texture was discarded. And look at dark areas, where blocky patterns often appear first.

Zooming in to 100 percent or more shows these problems early. That is useful for diagnosis, but remember that most people view the image at a smaller size. A flaw that you can see only when zoomed in at 400 percent may not matter for a picture that appears at the size of a postage stamp.

Chroma subsampling and other settings

Many JPEG encoders store color at half the resolution of brightness, a technique called chroma subsampling. It saves a lot of space and is rarely visible in photos, but it can blur thin colored lines and red text on a plain background. Browsers do not let a canvas change this setting, so it is determined by the browser's encoder; this tool uses the browser's built-in JPEG encoder.

Other options exist in dedicated tools, such as progressive encoding, which shows a rough version of the picture first and sharpens it as data arrives, and optimized Huffman tables, which make files slightly smaller at the same quality. They are refinements; the quality value and the pixel size matter far more.

Keeping originals

Always keep the original file. A JPEG is a final product: you cannot restore quality by saving it again at a higher setting, and compressing an already compressed JPEG adds new errors. If you may need the picture again at a larger size or for print, store the original somewhere safe and compress copies.

When you compress with this tool, the originals on your device are never changed. The compressed files are new downloads with -compressed in the name, so you can compare them with the originals and delete whichever you do not need.