Learning about noise

People talk about noise all of the time. Much of it is about the purging the blight :bow_and_arrow: :joy:. Much of it comes from a lack of understanding of what noise is, or arises from strong opinions on what to do with it. I would like to get a broad overview on the topic. Please share your thoughts or direct me to the appropriate resources. And who knows – this might become the precursor to an article on the main site :wink:.

  1. What types of noise are relevant to post-processing?
  2. Which types are problematic right out of the camera?
  3. Which types are problematic with further processing?
  4. Which types are particularly problematic to get rid of?
  5. Which types are “film-like” or aesthetically pleasing?
  6. What are ways we can examine each type more directly?
  • Such that we can quickly evaluate image quality.
  • Or determine the type of noise we should give the most attention to in a given image.
  1. Which are difficult to isolate due to their relationship with signal or other noise?

The practical questions: noise reduction methods and strategies.

  1. What are they and which type(s) of noise do they typically target?
  2. Where in the processing pipeline do they typically work?
  3. How do we compare methods when each may have different parameters?
2 Likes

Good reading on the subject by Emil Martinec:
http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/

1 Like

Sorry for the late reply, I wanted to be able to sit down and write something more than “yeah!”. :slight_smile:

All of them are usually. Though, different types of photography may address them in different ways (astrophotography vs. photojournalism for instance).

From a processing perspective I tend to break noise down into two main groups, and will approach each of them differently.

  1. Chroma noise
  2. Luminance noise

(This is not counting things like hot pixels or spurious noise that I think an impulse filter usually handles.)

Cambridge in Colour has a good page on this with a nice example image:


Image noise

That they then show as composed of those two primary components, Chroma and Luminance noise:


Chroma noise


Luminance noise

The splotches from the Chroma noise are the type of noise that I personally tend to aggressively attempt to remove.

The luminance noise usually reminds me more of film grain, so I tend to tolerate it better. (I even have a scan of film grain that I very often use over my images to add some ‘structure’ back: relevant pixls article.)

Unless it’s really, really distracting I’ll usually just leave luminance alone.

2 Likes

Thanks @patdavid. I am aware of the two articles.

Could you elaborate on this? What is spurious noise? I found the following on Wikipedia:


I do the same. I also read that our eyes are generally more sensitive to luminance than chroma – probably more so for me since I tend to have trouble with colors, though I am not color blind.

Usually just hot or dead pixels. Single pixels that are significantly different than surrounding pixels (like, really different).