Reduce image noise with a median filter

Photographers can clean a grainy or dotted capture before further editing without needing to master a complex desktop application.

Smooth isolated speckle noise with a local median filter and recover a cleaner image while preserving edges better than a uniform blur.

How to use

Load a PNG, JPEG or WebP photo that is no heavier than 20 MB.
Pick the intensity option you want: the lighter setting for mild grain or the stronger one for heavier speckle.
Run the process so the local median filter is applied across the picture.
Inspect the preview closely, especially edges, text and fine textures, to see whether noise has dropped without smearing.
If the result looks too smooth, clear the image and repeat with the gentler setting.

Example

A night shot is full of tiny bright dots. The photographer uploads the JPEG, starts with the lighter strength, applies the median filter, and checks that stars remain visible while random speckles become less distracting.

Limits and formats

The median filter targets isolated noise, not motion blur, missed focus or large smudges.
Too much filtering can erase small details such as skin texture, thin lines or distant objects.
Input is restricted to PNG, JPEG and WebP files, and each image must be 20 MB or less.

Where processing happens

The browser reads the selected photo and runs the median filter on your own device. Nothing is sent away for processing, so the cleaned version is generated locally and the original stays on your machine.

Frequently asked questions

What is the difference between the light and strong options?

The light mode removes mild speckle while trying to protect fine detail. The strong mode suppresses noise more aggressively, but it may also soften textures, hair or small objects.

Can this fix a blurry photo?

No, it addresses speckle noise rather than blur. If the image is out of focus or affected by camera shake, reducing noise will not restore sharpness that was never captured.

Should I check the preview before downloading?

Yes, because every image reacts differently. Zoom in on faces, foliage and text, then lower the intensity if the filter has removed details you wanted to keep.