You pushed ISO to get the shot. The image is sharp and the moment is captured, but now you're looking at grain covering your shadows and a texture you didn't intend.
Modern noise reduction can recover a lot of quality from high-ISO images. The skill lies in knowing how much to apply and where to stop.
You're looking for the point where your image is clean enough for its purpose while keeping the detail that makes photographs feel real. It's one of the most common post-processing needs covered in the low light guide.
Understanding What Noise Is
Before you remove noise, it helps to understand what you're looking at. Digital noise comes in two main forms, and each needs different treatment.
Luminance Noise
Luminance noise is random variation in brightness, a grainy texture similar to film grain that shows up as light and dark speckles across the image.
It's most visible in solid-toned areas, smooth gradients, and mid-tones, and less obvious in highly detailed areas where texture masks it.
People usually find it more tolerable than color noise. Some photographers even add film grain on purpose, which looks a lot like luminance noise.
Color (Chrominance) Noise
Color noise shows up as random colored speckles, typically magenta, green, cyan, or yellow, in places they shouldn't exist.
You'll see it most in shadows and underexposed areas, and it's especially obvious in areas that should be neutral gray or black.
It's generally more objectionable than luminance noise. The eye expects some variation in brightness, but unexpected patches of color are distracting.
Why Shadows Are Worse
Shadow areas contain less signal (light information) relative to the sensor's baseline noise. When you brighten shadows in post, you amplify the image detail and the noise equally.
That's why properly exposed high-ISO images often show less noise than underexposed lower-ISO images that were brightened in editing.
The Noise Reduction Tradeoff
Every noise reduction algorithm faces the same basic problem. Noise and detail can look alike, since both are variations in pixel values, so the harder you push noise reduction, the more real detail you risk removing too.
Over-processed images show these signs.
- Waxy, plastic-looking skin
- Lost fine texture (hair, fabric, foliage)
- Smeared or blotchy transitions
- Artificial smoothness that looks unreal
Under-processed images have their own problems.
- Distracting grain in smooth areas
- Color speckles that pull attention
- Shadows that feel dirty or unfinished
You're aiming for the point where noise stops being obtrusive and the image still keeps the detail that makes it feel real.
Noise Reduction in Lightroom/Camera RAW
Adobe's noise reduction tools are easy to use and effective for most purposes.
Accessing the Controls
In Lightroom, go to the Develop module > Detail panel > Noise Reduction section. In Camera RAW, it's the Detail panel > Noise Reduction.
Luminance Noise Reduction
Luminance slider. This is the main control. Higher values mean more noise reduction and more smoothing. Start around 20-40 for mild noise, and go higher for severe cases.
Detail slider. This protects fine detail from being smoothed away. Higher values preserve more texture but make the reduction less effective, and the default (50) works for most images.
Contrast slider. This preserves larger-scale contrast variations that might otherwise get smoothed. Higher values keep local contrast at the cost of leaving some noise behind.
Color Noise Reduction
Color slider. This is the main control, and even modest settings (15-30) remove most color noise. The default values often work well.
Detail slider. This protects color transitions in fine detail. Higher values preserve more color detail but may leave more color noise along edges.
Smoothness slider. This removes larger color blotches. Increase it if you see splotchy color areas after applying color noise reduction.
Practical Workflow
- View at 100% magnification. You can't judge noise from a zoomed-out view.
- Apply color noise reduction first. It often needs only a small adjustment from the default.
- Evaluate the luminance noise. Decide whether you need reduction, and how much.
- Apply luminance reduction gradually. Start conservatively and increase it only as needed.
- Check the detailed areas. Zoom in on fine texture like hair, fabric, and foliage.
- Adjust the detail and contrast sliders. Fine-tune the balance between noise and detail.
- Evaluate at output size. Noise you can see at 100% may be invisible in the final output.
Noise Reduction in Other Software
Capture One
Capture One uses a different algorithm from Adobe's, and some people find it better at preserving detail. You'll find it under the Details tool tab > Noise Reduction, with controls for Luminance, Color, and Single Pixel (for hot pixel removal). Its defaults are generally more conservative, so you may need less adjustment.
DxO PhotoLab / PureRAW
DxO uses Deep Prime and Deep Prime XD, AI-powered algorithms that draw on camera and lens profiles. It often recovers remarkable quality from high-ISO images, and it's some of the most effective noise reduction available right now.
PureRAW can preprocess files for use in other editors, while PhotoLab gives you complete editing.
Topaz DeNoise AI
Topaz DeNoise AI is a dedicated AI-powered denoise application. It's very effective at high ISO levels and offers several models for different scenarios.
It runs on its own or as a plugin for Lightroom/Photoshop, and it can rescue images that seem too noisy for other tools.
Adobe's AI Denoise
Adobe added machine learning-based denoise to Lightroom/Camera RAW in 2023, under Enhance > Denoise in Lightroom. It combines excellent detail preservation with significant noise reduction.
It creates a new DNG file and takes more processing power than the manual sliders. It's currently among the best integrated options in the Adobe ecosystem.
Advanced Noise Reduction Techniques
Selective Noise Reduction
Different parts of an image need different amounts of treatment, and a selective approach gets you the best quality. In Lightroom/Camera RAW, you can work like this.
- Use masking to apply noise reduction only where it's needed
- Create a mask for the shadows or sky (the areas with the worst noise)
- Apply stronger reduction to the masked area
- Leave detailed areas untouched, or give them less reduction
For portraits, scale it by area.
- Heavy reduction on out-of-focus backgrounds
- Light reduction on skin
- Minimal or no reduction on eyes, hair, and fabric texture
Exposure Blending for Noise
If you have several exposures of a static scene (or can make them), blending them can reduce noise while keeping detail.
Noise is random, while image detail is consistent. Averaging several frames reduces the random noise and reinforces the consistent detail. There are a few ways to do it.
- Stack and average in Photoshop
- Use dedicated stacking software (Starry Landscape Stacker, etc.)
- Use in-camera multi-shot noise reduction, which some cameras offer
Luminosity Masking
You can also apply noise reduction selectively by tonal range.
This works because shadows contain more visible noise and highlights need less reduction, so applying it only to the darker tones protects highlight detail. In Photoshop, create luminosity masks that target the shadow tones, then apply noise reduction through the mask.
How Much Reduction Is Too Much?
The right amount of noise reduction depends on several factors.
Consider Output Size
Noise you can see at 100% magnification may be invisible in the final output.
Web and social media. Small display sizes hide a lot of noise, so you may need less reduction than you think.
Large prints. These are more demanding. Noise that's invisible on screen may show up in a print, although natural grain can also look good in prints.
Close viewing distance. Prints viewed up close (typical for art displays) show more noise than prints viewed from a distance.
Consider the Subject
Portraits. Viewers expect smooth skin and clean backgrounds, so noise is more objectionable.
Documentary and photojournalism. Some grain can feel authentic and appropriate, and over-processing can make the work feel sanitized.
Landscapes. Fine texture in foliage and rock matters, and heavy reduction can make natural scenes look artificial.
Night photography. Some noise is expected and accepted, because the shot matters more than pristine smoothness.
Consider the Aesthetic
Noise isn't automatically bad. Film photographs had grain, and some photographers add it on purpose. A few questions help you decide.
- Does the noise distract from the subject?
- Does it suit the mood of the image?
- Am I removing it out of habit, or because it genuinely improves this image?
Preventing Noise in Camera
The best noise reduction happens before you press the shutter.
Expose Properly (or Slightly Over)
Underexposing and then brightening in post creates worse noise than a proper exposure at a higher ISO.
If you need ISO 6400 for a proper exposure, shooting at ISO 6400 gives better results than shooting at ISO 3200, underexposing, and brightening by one stop.
Use Your Camera's Best ISO Range
The native ISO range (typically 100-6400 or 100-12800) generally outperforms the extended or expanded ISO values.
Consider Image Stabilization
Stabilization lets you use slower shutter speeds, which means a lower ISO for the same exposure.
Add Light When Possible
Window light, reflectors, or extra light sources all reduce how much ISO you need.
Know Your Camera's Limits
Test your camera to learn its real high-ISO performance. You may be avoiding ISO values that your camera handles well.
Practical Noise Reduction Workflow
For Standard High-ISO Images
- Import and apply lens corrections.
- Set white balance and basic exposure.
- Apply color noise reduction. The default is often enough, or a slight increase.
- Evaluate luminance noise at 100%, in the shadows and smooth mid-tones.
- Apply luminance reduction conservatively. Start at 20-30.
- Check that detail survives in hair, fabric, and foliage texture.
- Adjust as needed to balance smoothness and detail.
- Consider AI denoise if the standard tools aren't enough.
For Severely Noisy Images
- Export to dedicated denoise software (DxO PureRAW, Topaz)
- Apply AI-powered noise reduction
- Re-import the processed file
- Complete the remaining edits
- Apply selective additional reduction if needed
For Images Needing Aggressive Shadow Recovery
- Apply noise reduction BEFORE heavy shadow lifting
- Or use AI denoise, which handles this scenario better
- Accept that heavily lifted shadows will show noise
- Ask whether full shadow recovery is really necessary
When Noise Reduction Isn't Enough
Sometimes the noise is too severe, or the reduction it needs destroys too much detail. You still have options.
- Accept more noise than you'd ideally want
- Convert to black and white (this eliminates color noise, and luminance grain can work as part of the look)
- Add a film grain overlay, which can make the noise pattern more pleasing
- Resize the image smaller, which reduces apparent noise in the output
- Accept the image's limitations
Not every high-ISO image can be rescued to perfection. Sometimes the shot was worth getting despite the technical compromise.
Key Takeaways
- Apply color noise reduction first (it is usually more objectionable), then luminance noise reduction, checking results at 100% zoom.
- Find the balance point where noise becomes acceptable without over-smoothing detail, since plastic-looking images are worse than slightly grainy ones.
- Lightroom's AI denoise and dedicated tools like DxO PureRAW or Topaz DeNoise can recover remarkable quality from high-ISO images.
- Shoot RAW for maximum noise reduction flexibility, since RAW files contain more tonal data for the software to work with than compressed JPEGs.










