Why JPG files are large in the first place

A 12-megapixel phone photo is about 4000 × 3000 pixels. Cameras save it at a very high JPEG quality so no detail is lost while shooting, which is why a single photo is often several megabytes. Most of that data describes fine texture and noise you can't see when the picture is shown on a phone, in an email or on a web page.

The two settings that matter

1. Quality

JPEG quality is a scale from 1 to 100. Very high values (95–100) cost a lot of bytes for almost no visible gain. Between about 75 and 85, photos look the same to most people at normal size; below about 60, edges and smooth areas such as skies start to show blocky artefacts. File Compressor's presets use 85 (High quality), 75 (Balanced) and 60 (Smaller file).

2. Pixel dimensions

Halving the width and height leaves a quarter of the pixels, so it's the biggest single reduction available. Resize when you know where the image will be shown:

Where it's usedLongest side that is usually enough
Full-width image on a website1920–2560 px
Image inside an article or email1200–1600 px
Profile picture or thumbnail400–800 px
PrintingKeep the original size

These are rules of thumb, not requirements; check your platform's own recommendations.

Step by step

  1. Start from the original photo, not a copy that was already compressed. Every JPEG re-save loses a little detail.
  2. Open Compress JPEG and choose the photo.
  3. Pick Balanced for web and email, High quality if you may edit or print it later.
  4. If the image only needs to be shown on screens, set a Maximum dimension under Advanced settings.
  5. Compress, compare the original and new sizes, then download.

Metadata: the part you can't see

Photos carry EXIF metadata: camera model, date and often GPS location. It's small in bytes but can be sensitive. File Compressor applies the photo's orientation to the pixels and then removes this metadata, while keeping the colour profile so colours don't shift.

When a JPG won't get much smaller

If a JPG was already saved at low quality or small dimensions, there's little left to remove without visible damage. In that case the tool tells you it's already well optimized and lets you keep the original rather than producing a worse copy.