Possible simple fix for Leica M8 magenta black

DrTebi

Slide Lover
Local time
3:03 AM
Joined
Jan 5, 2008
Messages
306
I have just been playing around with the latest development version of RawTherapee. If you are not familiar with this program, it's a great free raw conversion software that runs on Linux, Mac, and Windoze.

They keep adding new features, and I really like the program's many ways to adjust tonal values and colors. There is now an "HSV Equalizer," and I figured I could try to get rid of some ugly magenta black that I had on many images of one M8 set. It worked like a charm! I just opened the HSV Equalizer, changed the "S" to "Minima/Maxima control points" and dragged down the magenta point, here is an a/b screenshot:

magenta-black-fix.jpg


This may be a great fix for those that still have some images to fix. I don't even own the M8 anymore, but I did forget to use the correct filter at times, and fixing the images afterwards was a nightmare. Now, during the process or cleaning up my image library, I will use RawTherapee.

When I did a few comparisons of Lightroom vs. RawTherapee raw conversions, I liked the one from RT a lot more as well. It's strange since you would think a big player like Adobe should nail this... I don't know, maybe I just didn't figure out the right settings. However, I have said good-bye to the Windows world about a year ago, and thus am not using Lightroom anymore anyway. 'Living in a free world now :)

I hope this post is useful for some of you.
 
It's good to have a reliable method to remove the most obvious problem.

One down, two to go.

Removing the magenta cast repairs one third of the contamination. The foliage and skin tones would be the next challenge for RawTherapee.

The problem is unsolvable. You have no idea what the real, but unknown photon counts should be for each sensor site in the raw data file if there was negligible IR contamination. In other words: there is a non-systematic error in the raw data. That error is the unwanted photon counts from IR wavelengths. You have no means to estimate the error for each sensor site so you can't estimate the real, but unknown, photo counts for the data you want – the photon counts from visible light frequencies.
 
I found using a Brush set to around -50 to -80 Saturation in LR worked pretty well, and of course gives a purely localised adjustment. And you can vary the desat intensity after applying the brush - which does a pretty good job of localising by colour (if you know what I mean)

No help if you're on a Linux machine, of course...
 
There was a Lightroom profile that somebody on LUF made that served as a good quick-and-dirty substitute for the UV/IR filters, but I can't seem to find it in any downloadable location. Been meaning to track it down and try it, though.
 
I did not know about the problems of contamination throughout the entire image.

However, the day that image was taken, I actually shot pictures with the M8 and an analog film camera in parallel. Later I liked the natural look of the analog pictures so much better, that I started shooting a lot more film, a few month later sold the M8, and today shoot mostly film.
 
The corrected shot is slightly better but still a long, long way from the color you would get by simply using a filter. The skintones are downright ugly and the foliage - well.... The only way that can give some relief is to delve deeply into LAB color and blending masks in Photoshop - and even then it is never optimal.
 
Back
Top Bottom