Contrast, Noise, and Sharpness in Night Photography
Here is the problem: The sky is a high contrast subject and the foreground is a low contrast subject.
My background is the profession of Radiology. The making of radiographs, Magnetic Resonance scans (MRI), CAT scans, Ultrasounds, etc. share a great deal with the technology of digital photographs. All are digital images, and the issues of getting quality images and good signal to noise is very similar. More signal is good, more noise is bad in both fields. increased sharpness is good in both fields. Apparent sharpness is primarily due to 2 factors, spatial resolution and contrast resolution. This is one area where medical imaging and photography have a different emphasis. In MRI scanning and CT scanning and digital radiography we emphasize the same two things to make good images, spatial resolution and contrast resolution. Both combine to create the perception of sharpness. Spatial resolution is important in Radiology but the size of the pixels is vastly different. For years CT scanning and MRI scanning used a matrix of 512 X 512 pixels, as that was what the technology could manage. It finally advanced to 1024 X 1024 pixels and is slowly moving upwards. This gives a ONE Megapixel image!!! It is less spatial resolution than the earliest Digital Cameras!
So how could you see or diagnose anything at such low resolutions? Well, it turns out that contrast resolution, or the ability to get contrast differences between normal and abnormal tissues was at least as important, if not more important, than spatial resolution. For years most of the imaging research went into getting better soft tissue differentiation (contrast resolution) rather than spatial resolution. Contrast resolution was more important.
In photography we talk a lot about sharpness, and we usually mean spatial resolution. There is much less talk about talk about contrast resolution. Well, contrast resolution does matter. Some lens have significantly more contrast than others, but it is sometimes hard to even find this data. The perception is that these lens are sharper. It is much easier to find data on spatial resolution in photography than contrast resolution.
Most sharpening filters increase border contrast or edge contrast within the image. Actual spatial resolution is unchanged and fixed relative to the equipment.
So how does contrast matter in Nightscapes?
The night sky has tremendous inherent contrast. This inherent contrast adds to the perception of sharpness. You have a dark sky and bright stars and planets, etc. Also, the contrast in the sky is typically increased even further in processing. One of the first things most photographers do in processing is to increase contrast. This adds the perception of great sharpness. So does the sky need additional sharpening in post processing? I would argue that the answer is no. Sharpening often makes the stars look “crispy” and harsh. In addition, using filters to increase sharpness in Photoshop or other programs increases the noise in the image, or the perception of noise. This degrades the image. Noise is problematic without increasing it further. If anything, it is better to do some noise reduction in the sky and perhaps soften the sky. I typically do some noise reduction on the whole image initially in Lightroom, using a Luminance noise setting of around 50, and a color noise setting of around 18-20 (this is for an ISO of 6400 to 12,800). I then rarely do any more noise reduction or sharpening in the sky. One of the most common problems I see in night sky images (in my opinion) is over sharpening of the sky. A very harsh looking sky can be unpleasant, and moves further away from reality.
The darker foregrounds are completely different. The foregrounds are a very low contrast subject compared to the sky. They usually have much more noise and very low contrast (except for the illuminated areas). As I said above, I do a certain amount of noise reduction on the whole image in Lightroom. I then export to Photoshop. I typically select or mask the foreground or sky and place them in different layers and process them separately. After exporting the image I adjust the contrast and color in Photoshop. On the foreground layer I frequently do some additional noise reduction using the Topaz Denoise filter. I then do a type of sharpening that “increases local contrast”. This is a bit different than regular sharpening, and is more related to the Clarity function in Lightroom. If you use a very high radius of 50-60, and a low amount of 15-20 (threshold of 0), you will increase local contrast and increase the perception of sharpness, without increasing the perception of noise. It is remarkable how much this can increase the perception of sharpness with very little increase in noise. Regular sharpening such as Unsharp Mask, Luminance Sharpening, etc., all appear to increase or accentuate the noise more.
Wayne Pinkston, ©2017