Image Denoiser
Smooth out grain and noise from low-light photos using a fast Gaussian blur. Tunable radius.
Quick answer: Image Denoiser: smooth out grain and noise from low-light photos using a fast Gaussian blur. Tunable radius. Runs entirely in your browser, free, no signup.
Last updated
Good to know
Digital noise — the speckled grain you see in low-light and high-ISO photos — comes from the sensor amplifying a weak signal, and a Gaussian blur is the classic first line of defense against it. This tool applies a separable Gaussian with a radius you control: a small 0.5-pixel radius wipes out sensor grain while leaving detail essentially untouched, whereas a 2–3 pixel radius flattens texture and softens the whole frame for a dreamy or stylized look.
Denoising is always a trade against sharpness — every noise-reduction method blurs a little real detail along with the noise, so the art is finding the smallest radius that clears the grain. On portraits, 0.5–1.0 is usually invisible to the eye yet removes most high-ISO speckle; go higher only when you actively want the image softened.
This is a classical denoiser, not an AI one. That is a feature for privacy and cost — it runs entirely on a canvas in your browser with nothing uploaded — but AI denoisers like Topaz or DxO PureRAW preserve fine texture better on demanding shots, at the price of a desktop install. For everyday web images, the Gaussian approach is fast and free. The alpha channel survives, so cut-out logos keep their transparency unless you push the radius very high.
The pro workflow is denoise first, then sharpen: clean the grain with a 1.0-pixel radius here, then restore apparent crispness with the image sharpener at strength 0.5. That order — reduce noise, then enhance edges — is exactly what Lightroom applies under the hood.
Frequently asked questions
- Will I lose detail by denoising?
- Some — denoising and sharpness are always a trade-off. A radius of 0.5 to 1.0 is usually invisible on portraits but kills high-ISO sensor grain.
- Is this AI denoising?
- No — it's a classical Gaussian blur, the foundation every more-sophisticated denoiser uses. AI denoisers (Topaz, DxO PureRAW) give better results but cost money and only run on desktop. This tool runs free in your browser.
- Does the alpha channel survive?
- Yes — RGBA images keep their transparency. Hard edges of cut-out logos are preserved unless you crank the radius very high.
- Are my images uploaded?
- No — the convolution runs entirely on a canvas in your browser.
- Should I sharpen after denoising?
- Often yes. The Image Sharpener at strength 0.5 nicely restores apparent crispness after a 1.0-radius denoise — same workflow Adobe Lightroom uses.