Image Noise Reduction - Smooth Grain, Keep Detail

Noise is the speckled, grainy texture that appears in photos taken in low light. It comes from the camera sensor amplifying a weak signal: when there are not enough photons hitting each photosite, raising the ISO amplifies whatever the sensor recorded, including its own random electrical fluctuations. Small sensors suffer most, which is why phone photos taken indoors or at night show it far more than daylight shots.

Every noise reduction method faces the same fundamental tension. Noise is fine random variation between neighbouring pixels, and so is genuine fine detail such as hair, fabric weave, foliage and skin texture. Smoothing operates on both, which is why over-aggressive denoising produces that waxy, plastic look where faces lose their pores and grass turns into a green smear.

The approach here is a two-stage one that gives you control over that balance directly. The smoothing pass averages out the fine variation, taking the noise with it. The sharpening pass then restores edge contrast, bringing back the definition on the boundaries that matter, such as outlines, features and text, without reintroducing the speckle in flat areas like sky and shadow.

Working the two together is the whole technique. Apply just enough smoothing to kill the obvious grain, then add sharpening until the subject looks crisp again, checking at full size rather than on a scaled-down preview, since noise and over-smoothing are both invisible at thumbnail scale. Processing runs on a canvas in your browser, so nothing is uploaded and there is no limit on how many photos you clean up.