Testing Filters

  • Thread starter Thread starter Dr Jason Polak
  • Start date Start date
D

Dr Jason Polak

Guest
Hey folks, I'm developing a basic testing suite for testing filters and their basic optical qualities including their color fidelity. I wrote an article about it on a sample filter:


If anyone has any suggestions of other things that could be interesting to test I'd be interested in hearing them! Thanks!.
 
Hi Jason, good work.

I am surprised at how neutral the filter is, good job K&F. A difference in illuminant CCT of 24 Kelvin would be unnoticeable* in daylight, even ignoring adaptation (one CIEDE2000 below is supposed to be a just noticeable difference):

From the link at bottom
From the link at bottom

I am wondering whether it is worth calculating a confidence interval, since in theory the only potentially unstable variable is your light source, which you describe as relatively constant, perhaps a simple average would do.

May I ask how you arrive at CCT and tint from the captured raw data?

As for suggestions of other things to try, how about neutral density filters; and comparing different lenses all else equal?

Jack

PS https://www.strollswithmydog.com/white-balance-is-off/
 
Last edited:
Hi Jason, good work.

I am surprised at how neutral the filter is, good job K&F. A difference in illuminant CCT of 24 Kelvin would be unnoticeable* in daylight, even ignoring adaptation (one CIEDE2000 below is supposed to be a just noticeable difference):

From the link at bottom
From the link at bottom

I am wondering whether it is worth calculating a confidence interval, since in theory the only potentially unstable variable is your light source, which you describe as relatively constant, perhaps a simple average would do.

May I ask how you arrive at CCT and tint from the captured raw data?

As for suggestions of other things to try, how about neutral density filters; and comparing different lenses all else equal?
Thanks so much for the reply. The confidence interval is required I think, because between shots there is a very small variance between them. It's not large, and just six shots is more than enough for the T-test to have sufficient power to reject the null hypothesis, but without that one would not be able to really tell whether values are significant or not. I think you are right though; in general a simple average pretty much always tells the story, but I just wanted some statistical validity for rigour.

CCT and tint data, can be determined in darktable because it shows you the color correction matrix and adjusted tint as well, and you can even derive that from the Calibrite color checker!

Yes, neutral density filters are high on the list. In fact, I also tested a K&F concept one (test results unpublished), and there I DID find a statistically signficant change in color temperature around 500K AND a significant color shift as well.

Lenses can be compared too but without a base reference point, some of the tests might be harder to interpret.

--
YouTube: https://www.youtube.com/@JasonPolakPhotography
Website: https://jpolak.org
 

Keyboard shortcuts

Back
Top