Image Compressor

PRACTICAL GUIDE · UPDATED OCTOBER 29, 2026

Choose compression settings by the result you need

Use the compressor for supported JPG and PNG images up to the displayed upload limit. Keep an original and create a separate candidate. The useful outcome is an image that meets your display requirements at an acceptable size—not a guaranteed savings percentage for every file.

Try a reproducible sample

Use Try sample PNG to load an original synthetic 320 × 200 image with a transparent left edge and seeded color noise. It contains no personal data. Leave resize bounds blank and compare 256 colors, 128 colors, and Lossless using the same source. Inspect the transparent edge and changed colors as well as the byte count. This intentionally noisy test image is not representative of every photograph or logo. You can also download the source PNG to repeat the comparison elsewhere.

Use a preset as a starting point

Website fits the image within 1600 × 1600 at JPG quality 80; Email uses 1200 × 1200 at JPG quality 75. These are editable bounds, not a guaranteed target file size. The original aspect ratio is retained and smaller images are not enlarged. Compare the original and result previews, dimensions and exact bytes saved. When dimensions are unchanged and the candidate is larger, the original is retained. Resized outputs still report their actual size, including increases.

Compress a batch

Select or drop up to 20 images, up to 5 MB each and 30 MB total. Each image must stay within 12 million pixels. Images are processed one at a time locally; PNG encoding runs in a background worker. Use Compare on a result to inspect the before/after previews, Download for one file, or Download all ZIP for successful results. A failed file is reported individually while the batch continues. Settings changes clear prior results; compress again to use the new settings. Batch and palette compression require JavaScript; the existing server fallback handles one file with its original optimizer.

Start with a controlled comparison

1. Choose an image with no sensitive content.
2. Record its dimensions and file size.
3. If appropriate, enter a maximum width or height for your display.
4. For JPG, begin with a moderate quality value such as 80.
5. Compress, download, and open the result at the intended display size.

For a fictional 800-pixel-wide product panel, try an 800-pixel maximum width while checking labels and fine details. That number is an example, not a universal recommendation. If the product needs a larger zoom view, preserve enough detail for that requirement.

The quality slider behaves differently for PNG

PNG uses its own color setting: 256, 128, 64 or 32 colors, or Lossless. Color reduction can shrink a PNG without changing dimensions, but it can change gradients and semi-transparent edges. Lossless preserves the browser-decoded canvas pixels, not original file bytes, metadata or high-bit-depth source values. JPG uses the quality slider. Animated PNG is rejected to avoid flattening its animation. If transparency is unnecessary, compare a JPG with the Image Converter and check the background explicitly.

If the result is larger

Re-encoding does not always beat an already optimized source. Compare actual bytes and keep the original if the candidate fails your size or quality requirement. Try one adjustment at a time, using the same original each time. Repeatedly compressing the latest JPG copy can make it harder to distinguish which step lost important detail.

Inspect the downloaded image

Look at lettering, textures, gradients, and sharp edges. Open the downloaded file rather than relying only on a preview. Check it in the real page if it will be published. The normal browser path processes locally; a fallback request can use the server, with temporary-download expiry and cleanup described in the privacy policy.

Record original and exported dimensions and byte sizes, then compare visible detail before choosing the final file.