Thursday, 6 March 2014

Reducing Noise by Image Averaging

Article by: Jijo John

Image noise has always been a problem in digital photography. Most noise reduction techniques currently popular among photographers end up softening the final image as well. Image averaging is one technique that could successfully reduce noise and greatly enhance your photographs without softening it by compromising detail.

How does Image Averaging Work?

Image averaging is a technique commonly used in high end Astro photography and works on the logic that noise in digital images is truly random, thus as one averages more and more images these random fluctuations will gradually even out producing a neat final image. What is actually happening is that by averaging a number of images we are increasing the signal to noise ratio of our image. The image also benefits by the increased bit depth which is otherwise not possible with a single image.

Put simply take multiple shots at high ISO and combine them in Photoshop. We will discuss the concept and also the method in detail; first let’s go through an example.


noise reduction by image averaging
noise reduction by image averaging

Assume we have our camera on a solid tripod and it is not moving in between shots. We take two images namely A and B at high ISO so each have random noise in locations  1, 2 and 3 for image A and 4, 5 and 6 for image B. Now when we average the two images, pixel value at each of the locations 1,2,3, 4,5 and 6 are averaged and we now get a final image C which only has half as much noise as each of our  original images namely A and B.

Image Averaging Step 1 - Taking the Pictures

First step is to determine what you requirement is, that dictates the number of shots you need to take, and the formula to remember here is that two averaged images usually produce an image with noise levels comparable to an ISO setting which is half as sensitive.

If you are shooting at 1600 ISO and would like your results to be that of 800 ISO you only need to two pictures and if you wish the shot to be like 100 ISO then you will need to take 4 pictures.

The camera should be set up firmly on a solid tripod so as to prevent camera movement between exposures.

Step 2 – Image Averaging in Photoshop

Load all the images as layers in a single Photoshop document.  The key then is to set right amount of opacity to each layer. Remember each layers opacity determines how much the layer beneath is seen. In order for the blending to be effective it is essential to make sure that each layer contributes equally to the final image. This means that to properly average four images,  one should not set each layers opacity to 25%, instead set the bottom (background) layers opacity to 100%, the layer on top of its to 50%, the one on top of it to 33% and the top layer to 25%.

image averaging technique in photoshop
image averaging technique in photoshop

The following formula could be used to calculate each layer’s opacity.

Layer opacity = 100 * 1/# of layers below +1

Here's a practical example of the technique at work. The picture given below is taken at 100 ISO

image averaging
Image taken at 100 ISO
And here's a 100% crop of the bottom left corner of the above image.

image averaging technique in photoshop
100% crop of the bottom left corner
Here is the result of a single image taken at ISO 12800

single image taken at iso 12800
single image taken at iso 12800

And here's the 100% crop of the bottom left corner of the high ISO image.

single high iso image
single high iso image

Now see the result we got by averaging 4 images taken at ISO 12800

image created by averaging 4 high iso images
image created by averaging 4 high iso images

And here's 100% crop of the bottom left corner of the averaged image.

100% crop of the bottom left corner

Here's a comparison of the three 100% crops.

reducing noise by image averaging
reducing noise by image averaging

Click on the image to see larger view.

Recommended Uses

When should image averaging be preferred as opposed to just taking a longer exposure at a lower ISO?

  1. To reduce noise in the shadows when you wish to bring out details while post processing.
  2. To avoid fixed pattern noise from long exposures
  3. When you work on a camera which does not have a bulb mode and you need to go beyond its longest shutter duration. (for e.g. Most cameras have 30 seconds as its slowest speed, if you don’t have a bulb mode then you cannot go slower than that. But remember 2 30 second shots at ISO 200 is equal to one 60 second shot at 100 ISO.)
  4. For situations where you cannot guarantee interruption free exposures beyond a given time.  E.g. Taking a photo in a public place and wanting low noise, but cannot take long enough exposure because people often pass through the shot, you could take several shots in between passersby.
  5. To selectively freeze motion in low detail, faster moving areas while still retaining low noise in high detail, slower moving areas.