Dealing With Digital Noise

Photographing with higher ISO values, under less good lighting conditions or with a compact camera. Means you have to deal with digital noise, deviations in the form of arbitrary and blue or red pixels. Between the pixels with correct colors or additional pixels with different gray values. How does digital noise originate and what can you do?

Noise originates in the conversion of the analog light (photons) to a digital signal (electrons) on the sensor. The light that falls on the sensor is picked up by millions of photosensitive dots, pixels. The more light that falls on an individual pixel, the better the result.

Light Sensor
The sensor will convert light to a digital signal

The photons however, will not fall with regularity on the pixels, but arrive randomly. One pixel receives thus more photons to process than the other pixel. Allowing small deviations to occur during this conversion process. If the shutter speed is reduced or the pixel pitch is limited. Then there are fewer photons to the derogation’s from ‘resources’ and increases the digital noise. Under good lighting conditions with a correct exposure, these small deviations are not visible.

Lighting Conditions

Under less good lighting conditions,  these are visible in the form of arbitrary blue or red pixels between the pixels. With correct colors or additional pixels with different gray values.

If you do not make use of a RAW photo there will be extra deviations. That are caused by the compression of the image to the JPEG format. These various forms of noise can have an impact on the sharpness and the colors of the image. Luminance Noise (grains with gray-scale) comes at the expense of the sharpness of a photo. Chromatic noise comes at the expense of some colors.

Noise is not equal across the entire photo, the dark parts have faster load of noise than the lighter parts.Digital noise can arise in different ways: the sensor size &, the selected sensitivity, the shutter speed and the finishing.

Sensor Size

An important factor is the sensor size. The more megapixels, the less space each pixel has on the sensor. The greater the chance that the image quality is suffering. Smaller pixels capture less photons than larger pixels, the smaller the pixels, the higher the noise sensitivity on the sensor. If two cameras will have the same sensor size. Than a 14 megapixel sensor has more noise than a 10 megapixel sensor.

Light Sensor Sizes
Various sensor sizes sensors give various noise performances

In general, it is the rule that for a camera with a crop-sensor the maximum number of megapixels. For a good picture quality is 10, with a margin for improved sensor technology and new ways of noise reduction.

The step from 10 to 12 megapixels for example, is not nearly as large as you would think. And the negative effects of that step can be largely nullified with new technology.

The Megapixel Myth

The site Petavoxel devotes considerable attention to the megapixel myth (a bit technical). It comes down to the following values whether a sensor will generate a good picture quality. Everything above leads to a reduction in the performance if you will search for the extremes:

  • Medium size (33.1 × 44.2mm sensor, 5517 ×7367 ” Medical pixels) – 40.6 megapixels
  • Full-frame size (24 × 36mm sensor, 4000 ×6000 pixels) – 24 megapixels
  • APS-C 1.5x crop (2633 ×3933 pixels) – 10.4 megapixels
  • Micro FourThirds (2250 ×3000 pixels) – 6.8 megapixels
  • Typical 1/ 2.3″ compact camera (770 ×1028 pixels) – 0.8 megapixels

Since recently, there are medium format cameras that have 40 megapixels (Pentax and Hasselblad). And full-frame  cameras have only recently reached 24-megapixel (Sony A900, Nikon D3x). But  crop-camera manufacturers already have stepped over these values. For example the Canon EOS 550D and 7D (18 megapixels).

Sensor Performance

Nikon has made a different choice, not for megapixels, but for sensor performance under lesser lighting conditions. The most Nikon DSLR have up to 12 megapixels. And the higher megapixel values are to be found in the full-frame section. Although the full-frame D3s for example has only 12 megapixels. This is renowned for its good noise reduction under dark conditions.

Chromatic Noise
A 100% zoomed picture taken under less good lighting conditions on ISO 91 with an iPhone releases much chromatic noise

These lists are being ignored especially in compact cameras. Many users select these cameras on the number of megapixels, convenience and size of the images, not on performance. They make less demands on the image quality than enthusiastic (amateur) photographers.

Compact cameras should contain much less megapixels than SLR cameras, but in practice they have a equal number of megapixels. The pixels are in a quarter (or less) of the space which are crammed in the correct noise values at ISO 200. With a compact camera  comparable to ISO 800 on a digital SLR camera.

Sensitivity (ISO)

Another important role is the sensitivity that is set, the ISO value. This is ISO 100 on most cameras, ISO 200 at Nikon, but can be artificially increased up to ISO 6400 or 12800. With a additional ISO (100, 200, 400, 800, 1600, 3200, 6400, 12800). Will provide a doubling of the amount of light that falls on the sensor.

A higher ISO value strengthens all digital light signals under dark lighting conditions but a sharp photo can still be made. In the strengthening of the signal, however, no distinction is made between the real data of the photo and the noise. A higher ISO value can therefore lead to more visible digital noise.

ISO Value
On ISO 1250 (85mm lens, aperture f/2 1/100s) a very acceptable result, or noise in the dark parts

It is therefore always good to use the lowest possible ISO value, but keep in mind that you will get sharp results in less good lighting conditions. Sometimes you’ll also just opt for a compromise, you can go for a sharp picture with a little more noise or for a noise-free but blurry picture.

Noise Reduction
ISO 640 is not very high, but has much luminance noise to see because it is a dark environment,

Long shutter speeds

Converting light to digital data is an electronic process and not the only electronic process that is underway in the camera. During the taking of a photo there are all kinds of electronic transactions going on.

During normal operations in the camera, this will not have any visible impact on the end result, but at slower shutter speeds (from 0,5-1s)the camera sensor processes more and more of these signals. The camera does not know how to address this and more noise will end up in the final image. This is also depending on the size of the sensor, the more favorable the photosensitive pixels distance to each other the less chance that they will be ‘infected’.

Noise Reduction in the camera

Camera producing brands under value the problem of noise and try to resolve this in the camera by applying a noise reduction algorithm to the photo. This is mainly applied to slow shutter speed and/or high ISO values.

How successful this is depends on the algorithm, but often they will remove all noise and in addition provides slightly less sharp images, particularly visible in the parts with fine details. Another application, which is used primarily at slower shutter speeds, is the possibility of a number of cameras to take a second shot after the photo with the same shutter speed but without image.

By combining these two pictures in the camera, to combine these two the camera sees in the dark photos where there is noise is and in which shutter speed and can correct this in the first photo. This is especially every-time when noise occurs at the same place.

Each correction in the camera has consequences for the fine details that are documented, additional reinforcement is required afterwards.


The correct exposure is important, a lot of noise is visible when making the dark parts lighter when editing. The RAW format is less sensitive than the JPEG, because there are more light tints available, but if you are processing more extreme you’ll quickly see the drawbacks. For Example HDR photos, where three images are merged, give relatively more noise.

During the shooting you can already take account to this when processing by adding more light to the right. Right stores on the histogram of the photo, where the right-hand side is for the light parts of the image and the left hand side for the dark parts.

Keep the graph just to the right edge, you can correct it again in post processing.

The further to the right you can get the histogram without the graph running out on the right side(and light tones lost), the better to prevent noise in the post edit afterwards. You better give a light image more contrast and darker than a too dark image lighter. Note that you do not over-exposed to far, check the exposure with the histogram and check if the overexposed indicator flashes red if there is no detail to see in the light parts.

If you work with the RAW format, then it also plays a role how well the RAW engine can convert the RAW photos. For example Adobe brings every once in a while an adjustment of their Camera Raw engine from which usually also improved noise reduction indicates. How well such a RAW engine works may vary per camera brand and type.

Noise Reduction in software

In Addition noise reduction in the camera, you can also apply post noise reduction  in software, where you will also need to take into account that there is always something lost in the picture after the edit. The removal of luminance noise (grains with gray-scale) is at the expense of the sharpness of a photo (of course also to correct again), the removal of chroma noise is at the expense of some colors.

The reduction of noise in the photo is a balance between how much relief and damage of color data you want to accept in relation to how much noise you will find acceptable. It is also a bit to where you want to use the photo for. If you want to create a large poster this is more critical than if you want to create a 750x500px web image.

If you have a smaller file size, you’ll notice that the noise is less visible (like for example, problems with the sharpness) than in a larger file, sometimes it can help to reduce the photo size.

Software Solutions For The Reduction Of Noise

There are many different software solutions for the reduction of noise. Photo-editing and organization tools such as Adobe Photoshop (Elements/CS), Paint Shop Pro, Lightroom and Aperture have features to reduce noise. The more light info there is available in a photo, the easier the digital noise can be suppressed, so it pays to shoot in RAW format.

Adobe Lightroom
Neat Image noise reduction software

There are also specialized programs that have noise reduction as the only function. Examples include: Neat Image, Noise Ninja and Noiseware, but there are many more. Most of the programs work with a database of noise patterns from cameras to reduce the noise. Because noise often has a random character and the effectiveness will vary per photo. Pay attention when editing, that the reduction does not go too far or you get unnatural plastic-like and/or blurred results.

Apply Noise Reduction

You can also choose not to apply noise reduction, but making the image black and white to make the chromatic noise not obvious anymore and the image contains more grain like in an old picture. Better still, the Lightroom 3 beta has specifically a possibility for adding additional grain to a photo!

Camera sensors are better with the years and it is to be hoped that more manufacturers opt to limit the number of megapixels to the benefit of the performance under dark lighting conditions. Although technological developments, noise reduction software and smarter handling of your camera settings can certainly help to reduce noise to an acceptable level, you cannot always escape from it, and sometimes you have to accept some noise in your photos, just as you see with film grain.

In a future article i would like to go in to the various specialized noise reduction programs. If you use or own such a program, then please let them know in the comment section and i will try to take the program to a test.

Leave a Reply

Your email address will not be published. Required fields are marked *