Compress Images Without Losing Quality: What Works

What happens when you compress images

A digital photo is a grid of pixels, and its file size depends on four things: how many pixels it has, how much fine detail sits in them, the format, and the compression setting. Compression works on the last two.

• Lossy formats such as JPG and WebP split the picture into small blocks and discard color and texture detail the eye barely registers; savings of 80 to 90 percent are common for photos
• Lossless formats such as PNG keep every pixel exactly and only pack repeated data more tightly, so they excel at flat graphics but stay heavy on photos
• Busy scenes like foliage, gravel or a crowd hold more detail and compress less than a clear sky or a studio backdrop
• Pixel count scales everything: halving the width and height leaves a quarter of the pixels, and usually roughly a quarter of the bytes

So compressing without losing quality really means losing the right data. The encoder removes information you cannot see at the size the image is viewed, which is why a photo that looks identical on a laptop can show faint blocks when you zoom to 400 percent. Judge the result at the size your audience will actually see it, not under a magnifying glass.

Quality settings that stay invisible, by kind of picture

There is no single perfect number, because different content breaks down in different ways. These starting points hold up for most people:

• Portraits, pets and travel shots: 70 to 80 percent looks clean on any screen
• Product photos with fabric, wood grain or jewelry: 80 to 88 percent, since buyers zoom in on texture
• Screenshots, charts, receipts and scanned forms: 85 to 92 percent, because smeared letters are the first thing people notice
• Sunsets, skies and soft gradients: watch for visible stripes called banding, and go higher if they appear
• Thumbnails and background images: 60 to 70 percent, where softness is invisible at small sizes

In the tool's Quality mode, start at the default of 75, compare, then step up or down by five. Each result card shows the percent saved, so you can see when an extra five points of quality costs 30 percent more bytes and decide whether that is worth it. For a website, a sensible goal is keeping most images under about 200 KB and large hero images under roughly 400 KB.

Choosing JPG, WebP, PNG or Auto when you compress images

The Format menu in Compress Image has four choices, and the right one depends on where the file is going.

• Auto saves any PNG, WebP or GIF with transparent pixels as WebP, which keeps the see-through areas, and everything else as JPG
• JPG opens everywhere, from government portals to old photo viewers, but it fills transparent areas with white
• WebP is usually the smallest at the same look and suits websites, though some upload forms and older apps still reject it
• PNG keeps transparency: lossless mode re-packs every pixel more tightly and keeps your original if it cannot be made smaller, while Lossy PNG cuts the palette to between 256 and 8 colors for a much smaller file

The tool opens JPG, PNG, WebP, GIF, BMP and AVIF files your browser can read, and saves each result with -min added to the file name so it never overwrites your original. Animated GIFs become a single still frame. If you need an exact number rather than a look, switch to Target file size and type a maximum in KB; the tool then hunts for the highest quality that fits. Lossless PNG cannot aim at a size, so for a PNG under a limit turn on Lossy PNG.

Photos for the web often shrink further as WebP; the Image to WebP tool converts them, and the Metadata Remover strips extra file data.

When compression is the wrong fix

Compression is powerful, but a few situations call for a different tool or no change at all.

• The image is far larger in pixels than its display slot: a 6000-pixel photo shown 800 pixels wide needs resizing first, and no quality setting will match what that alone saves
• The file was already heavily compressed, such as a picture saved from a messaging app: every extra lossy pass stacks new artifacts on the old ones
• The picture is headed to a print shop: printers need pixels and detail, so send the original and compress only the copy you share online
• It is a logo or icon made of flat shapes: an SVG or a well-made PNG is often cleaner and just as small
• It is already small, say an 80 KB thumbnail: the savings are tiny and not worth another generation of loss

A good habit is to compress once, from the best original you have, at the moment you know where the image is going. Keep the untouched file as a master so you can export again when a new use comes along.

▶ Watch: How to compress an image (0:39)

0:39

Video transcript

1. Set a target size. Pick a preset or type the largest file size you want, here 100 KB.
2. Pick a format. WebP is usually the smallest. Choose JPG if the site asks for it.
3. Add your photo. Drop one photo or many. Each one is squeezed under the target size.
4. Compare the sizes. The card shows the size before and after, so you know it will fit.
5. Download it. Save the compressed copy and attach it to your email or upload form.

Step-by-step

Common mistakes to avoid

Pro tips

Frequently asked questions