Today I am going to use the Surface blur filter to reduce noise. Have you not used this filter before well here is your chance.
This time I am going to start with my Sukiennice Image. I also showcased this image yesterday on my Vlog. It was shot at 200 ISO, so it is not too noisy but the noise is visible when zoomed in at 100%.
I opened the Image in Adobe Photoshop and clicked on the channels palette. When you do this and the image is in RGB mode you will have 4 channels; the combined information channel – RGB, then one for each colour, Red, Green and Blue. You can edit each channel individually and this is the next step. If you want to find out more information about channels click here
Zoom in at 100% and view the channels. Select the first channel and go Filter-Blur-Surface Blur. Surface Blur is a blur filter that blurs an image while preserving edges.
The filter has two options Radius and Threshold.
Radius controls the size of the area that will be sampled and blurred ( the size of the blur)
Threshold defines the number of pixels neighboring each other with the same value that are needed to be classed as an edge. If the threshold is set to 5, groupings with 5 or less pixels will not be blurred.
Once you have applied the blur to one channel move onto the others. Each channel will have a different amount of noise usually with the blue channel containing the most. So you will need to make different adjustments for each channel.
As you can see when using a smart object, the Surface Blur adjustment is added to the smart object. This means that you can make adjustments later to the image.
This workflow is a nod to the Reduce Noise filter and in my opinion more accurate. although it is a three step process. You could create an action and automate this process to make it more stream lined.
Although at full size there is no a visible difference between the images. You can see that the green in the window is smoother on the right (edited image) when compared to the left.
The best example of this technique is with a high ISO image.
With the image of the Krakow Wool Market you can see how strong this technique is. The edited image has considerably less noise than the original. There is still quite a bit of colour noise so i will need to go back and make a few more adjustments. The sky is considerably smoother and the building looks well defined.
Tomorrow I will look at using stacked images in a smart object to reduce noise.
What do you think of this technique, have you tried it before? And any other comments please use the comment box below.
If there is anything you would like me to write a post about or any question please head to “What is Aperture64″ and leave a comment there.