How to compress an image

Open the Image Converter, load your picture, choose JPEG or WebP, and lower the quality setting until the reported output size is small enough. The default is 82, and most photographs survive a drop to around 75-80 with no visible change.

If the file is still too big, resize it. Halving the width and height quarters the pixel count and usually cuts the file more than any quality change can. Format, quality and dimensions are three separate levers and they are not interchangeable.

Last reviewed: 2026-09-19. Checked against the tool’s own source on that date.

The three levers, in the order you should try them

Format first. A photograph saved as PNG can be five or ten times larger than the same photograph as JPEG, for no visible gain, because PNG stores every pixel exactly and photographs have no exact structure to exploit. Converting a photographic PNG to JPEG or WebP is often the whole fix.

Dimensions second. This is the lever people forget, and it is the strongest one. Pixel count scales with the square of the linear size: an image at 50% width and height has 25% of the pixels. If the picture will be displayed 800 pixels wide, storing it at 4000 pixels wide is paying nine hundred percent extra for detail nobody will see.

Quality last. The quality setting controls how aggressively the JPEG or WebP encoder discards detail. It has a sharp knee: from 100 down to about 80 the file shrinks quickly and the picture barely changes, and below about 60 the artefacts become visible in flat areas such as skies and walls. Past that point you are trading a lot of appearance for very little size.

Which format to compress into

The quality control only affects JPEG and WebP. Dragging it while PNG is selected changes nothing, and the tool says so rather than letting you believe you achieved something.

  • JPEG — photographs, and anything with smooth gradients. No transparency; transparent areas are filled with a background colour you pick.
  • WebP — usually the smallest at a given visual quality, and it keeps transparency. Your browser encodes WebP lossily even at quality 100, so treat it as a JPEG that supports an alpha channel.
  • PNG — screenshots, logos, diagrams, anything with sharp edges or large flat areas. Lossless, so the quality setting does nothing at all for it.

Do not compress the same file twice

Every lossy save re-approximates pixels that were already approximations. Do this repeatedly and the result becomes visibly mottled around text and edges, which is called generation loss and cannot be undone.

Keep the original. Always re-export from it rather than from your last export. If you need an intermediate working copy — cropping in one place and resizing in another — save that copy as PNG, which adds no further loss, and make the single lossy export at the end.

Converting an already-compressed JPEG to PNG does not restore anything. The PNG will faithfully store the degraded pixels in a much larger file.

Doing it

  1. Open the Image Converter and add one image or several.
  2. Pick the output format. The panel states whether that conversion is lossy before you commit to it.
  3. Set the dimensions if the picture is larger than it needs to be — by percentage or by exact pixels, with the aspect ratio locked unless you unlock it.
  4. Adjust quality for JPEG or WebP, then convert.
  5. Read the measured before and after sizes. Adjust and convert again if needed; it costs nothing because no upload is involved.

Worked example: a 6.8 MB phone photo for a web page

A phone photograph is 4032 by 3024 pixels and 6.8 MB as it came off the camera. It will be shown in a content column about 900 pixels wide, on screens that may be twice that in device pixels.

A sensible target is 1800 pixels on the long edge — enough for a high-density display at that column width, and less than half the linear size of the original.

  1. Load the photo into the Image Converter.
  2. Set the output format to JPEG.
  3. Resize by pixels: 1800 on the long edge, aspect ratio locked.
  4. Leave quality at the default, convert, and read the output size.
  5. If it is still larger than you want, try quality 75 and compare — not 40, which will show.

Result: The pixel count falls from about 12.2 megapixels to about 2.4, roughly a fifth. A typical result is a few hundred kilobytes rather than 6.8 MB, with no difference visible at the size it will actually be displayed. The exact number depends on the picture and on your browser’s encoder, which is why the tool reports the measured size rather than predicting one.

Open the tool

Compress an image in the Image Converter

Converts and resizes in your browser and reports the real measured sizes. It cannot read HEIC or camera RAW files.

What this does not cover

  • Input is JPEG, PNG, WebP or GIF. HEIC, AVIF, TIFF and camera RAW are not supported, because browsers cannot reliably decode them without a platform codec.
  • Encoding is done by your browser, so the same settings give slightly different file sizes and results on different browsers and devices.
  • GIF is read as a single frame. Animation is lost, and GIF is not offered as an output.
  • Colour profiles are not preserved through the canvas, so an image with an unusual embedded profile may shift slightly in colour.
  • A large batch of high-resolution photographs is limited by your device memory, and will be slow on a phone.
  • Metadata removal is a side effect of re-encoding rather than a guarantee. Use the Image Metadata Privacy Tool if that is the actual goal.