High ISO night street fashion frame in London by Mariya Harley

AI noise reduction

AI noise reduction

The safest AI job in photography, and how to still get it wrong.

Noise reduction is the AI job with the least risk. It changes grain, not shape.

That is why it is the first tool to add to a workflow. Nobody will ever say your portrait looks wrong because the noise went.

ToolWhat it doesWhere it failsPrice
Lightroom DenoiseRuns on a RAW file and writes a new DNG. One slider.Slow on a large batch. Needs disc space for every new file.Part of the Lightroom plan
DxO PureRAW and PhotoLabDeepPRIME processes noise and lens faults together at the RAW stage.A separate step before your normal edit.One-off licence
Topaz Photo AINoise, sharpening and upscale in one pass, as a plugin or on its own.The default strength smooths skin.From 12 US dollars a month for the web version at the time of checking
Capture OneNoise reduction inside the same editor as your colour work.Less aggressive than the dedicated tools.Subscription or perpetual licence

Prices and features on this page were read from each vendor's own website on 25 September 2026. Vendors change plans often. Check the current page before you buy.

Tool by tool

Lightroom Denoise

One slider, applied to a RAW file, producing a new DNG alongside the original.

The quality is good and the cost is disc space: every denoised file is a second copy.

For most work this is enough, and it needs no extra software.

DxO PureRAW and PhotoLab

DxO processes noise and lens faults together at the RAW stage, using measurements it has taken for specific camera and lens combinations.

That is the difference from a general model: it knows what your lens does at that aperture.

It is a separate step before your normal edit, which is either a small annoyance or a deal breaker depending on your workflow.

Topaz Photo AI

Noise, sharpening and upscaling in one pass, available as a plugin or on its own.

The default strength smooths skin, so portraits need the sliders pulled back.

Listed from 12 US dollars a month for the web version at the time of checking.

ON1 NoNoise

Noise reduction and sharpening in one pass, available as a plugin or on its own.

It keeps more skin texture at moderate settings than most of the alternatives, which matters on portraits.

Capture One

Noise reduction inside the same editor as the colour work, which avoids a round trip.

Less aggressive than the dedicated tools, and for a properly exposed frame that is usually the right amount.

Where it goes wrong anyway

Strong noise reduction eats skin texture along with the grain. The face turns smooth in a way no retouch would do.

It also flattens fabric. A wool coat becomes a grey shape.

Run it at a lower strength than the default and compare at one hundred per cent.

Night street fashion frame in London, shot at high ISO by Mariya Harley

Mariya's verdict. I use noise reduction on every frame shot above ISO 3200. Below that the grain is part of the picture.

Shooting so you need less of it

Noise comes from underexposure more often than from a high ISO setting.

Expose for the face, not for the highlight you will never print. A correctly exposed frame at ISO 6400 is cleaner than a dark frame at ISO 1600 lifted later.

On a dark venue, a faster lens changes more than any software.

Evening hotel balcony frame in low light by Mariya Harley

How a denoising model decides

Noise is random and detail is not. The model separates the two by looking for structure.

Skin texture sits close to the boundary. Pores are small, low-contrast and fairly random, which is exactly what noise looks like.

That is why the first thing to disappear on a strong setting is the skin, and the second is the weave of a fabric.

Modern RAW denoisers work on the sensor data before demosaicing, which is why they beat anything applied to a JPEG afterwards.

Expose correctly and the model has less to separate, so the result keeps more of the picture.

Where noise reduction sits in a working week

It belongs at the start, on the RAW file, before any other decision.

On a dark venue it is the difference between a usable gallery and an apology.

Batch it while you do something else. It is the one step that costs processing time rather than attention.

What to ask before you pay

Does it work on RAW files or only on exports.

Does it support my camera and lens combination, and how quickly after a new release.

How long does a batch of two hundred files take on my own machine.

Can I control the strength separately for skin and for background.

Order of operations

Noise reduction goes first, before sharpening and before any upscale.

Run it on the RAW file rather than an exported JPEG. The tool has more to work with.

Sharpen after, and only where it matters. Global sharpening brings the grain back.

If you also upscale, do that last. Upscaling multiplies whatever noise is left.

A worked example

A dinner in a private room, tungsten light, no flash allowed. The camera sits at ISO 6400 for two hours.

Three hundred frames survive the cull. Every one of them needs noise work before anything else happens.

Batch denoising runs while the culling happens, which costs processing time rather than attention.

Strength goes to about half the default. At full strength the faces lose pores and the tablecloth turns into a grey field.

Sharpening comes last, applied only to eyes and edges, because a global pass brings the grain straight back.

Which one for which job

SituationToolWhy
High ISO event, three hundred framesLightroom DenoiseBatch it, no extra software
A known camera and lensDxO PureRAWCorrections measured for that pair
Noise, sharpening and size in one passTopaz Photo AIOne step, one file
Portrait where skin mattersHalf strength, then checkPores go first at the default
Already exported JPEGAccept lessRAW denoisers work before demosaicing and JPEG has lost that

Where these facts came from

Every price and feature on this page was read from the vendor's own website on 25 September 2026.

Plans change often, and a tool that was a one-off purchase last year may be a subscription now.

Check the current page before you buy, and treat the figures here as a starting point for comparison rather than a quote.

Nothing on this page is sponsored. No vendor paid for a place in the tables, and there are no affiliate links.

Who wrote this and how it is kept current

This page is written by Mariya Harley, a fashion, portrait and hotel photographer working from a studio in west London.

The tools described here are the ones that reach a working photographer's desk, and the failures described are the ones that reach a client's screen.

Prices and features are read from vendor pages rather than repeated from other articles. The date of that check is printed above each table.

Nothing here is sponsored, and there are no affiliate links. A tool appears because it does the job, or because people ask about it.

What is deliberately absent: benchmark scores copied from press releases, rankings that change with the season, and any claim that software can replace direction, light or responsibility for the result.

If a price here is out of date, the vendor page wins. Tell me and I will correct it.

Next step

Shooting somewhere dark?

Tell me the venue and I will bring the right glass.

Get in touch