12-21-2019, 12:27 PM
You can automate this with a "median filter". This filter computes the median value for a given pixel using the values of that pixel on several layers.
In math, the "median" is the value for which there are as many samples below than above. For instance, in 1,1,1,1,6 the average is 2 (10/5), but the median is 1. When your set of samples contain a preponderant value, this value ends up as the median.
So when you apply this to pixels in an image, this select the pixel value that appears the most often in the pictures, which is usually the fixed part. So in practice you take 7 shots or more. There is a median filter in GMIC, and I did one in Python
In math, the "median" is the value for which there are as many samples below than above. For instance, in 1,1,1,1,6 the average is 2 (10/5), but the median is 1. When your set of samples contain a preponderant value, this value ends up as the median.
So when you apply this to pixels in an image, this select the pixel value that appears the most often in the pictures, which is usually the fixed part. So in practice you take 7 shots or more. There is a median filter in GMIC, and I did one in Python