Noise is the speckled grain you see in photos taken in dim light. When there is little light, the camera amplifies a weak signal (high ISO), and random variations between neighbouring pixels become visible as coloured or grey specks. Small phone sensors show it more than larger camera sensors, and heavy JPG compression can turn it into blotches.
Where photo noise comes from
What this tool does
This is a simple filter, not a learned denoising model. It blurs the image slightly, which averages each pixel with its neighbours and so smooths out random specks, then draws the result again with contrast raised to 110% to win back a little punch.
The intensity slider sets the blur radius: intensity divided by 50, in pixels. The default of 50 gives a 1 px blur; 100 gives 2 px. These are small numbers, but on a full-resolution photo even 2 px noticeably softens edges.
The trade-off
A blur cannot tell noise from texture. Grain disappears, and so do fine hair, fabric weave and distant foliage. Smooth areas such as skies and walls benefit most; detailed areas suffer most. Keep the intensity low, compare with the before and after slider, and consider whether a little grain looks more natural than a smoothed picture.
Afterwards, a light pass with the Sharpen Image tool can bring back some edge definition, but sharpening also brings back some noise, so go gently.
Output
The result is saved as WebP at quality 0.9 at the original size. EXIF data is not kept. The filter uses the canvas filter feature, which older Safari versions do not support; there the result may be unchanged.
Where the processing happens
The filter runs in your browser. Nothing is uploaded.