Back in 2013 yes, that long, I started a research on noise and came with several solutions, each addressing specific noise, none being 'the silver bullet'.
This time my inspiration came from a
thread created by
rmalarz. In his thread
rmalarz posted his own research on predictable noise.
rmalarz posted his test image and I got 'upset' as his image was a JPG and asked for something that not messed up with JPG compression.
rmalarz complied by posting another image, a TIFF. Visually, comparing the two, the change was invisible yet there was something as the histogram had variations in the Blue and Green channels and none in the Red. I decided to compare the images using PS CC. So I took the two images, placed them on top of each other and used 'divide' as blend mode. 'Bingo!' I saw where the modification were made. Great. Problem is something was tugging in my mind.
It took me about two night of insomnia to crack it. If I could isolate difference so quickly, why is it not possible for other purposes?
I took
rmalarz initial JPG as a test file, with his permission. I immediately found that I was able to correct the color artifacts in it. As this could be a fluke I requested folks to send me a few noisy captures in the
post processing section. A few answered and gave me various forms of noise.
It instantly became apparent that I was dealing with a fluke. Then I realized that the solution was specialized in removing JPG color noise.
I tested my 'discovery' with older images I had requested in 2014 and
could not correct. To find my old research I reopened many of my ancient threads.
This led me to a surprise!!!Back in 2013 yes, that long, I started a research ... (