[Leica] It ain't alright.... "Fix it in SW"

[hoppyman at bigpond.net.au (G Hopkinson)]

Howard, you supplied what is probably a well reasoned scientific post.
Most of which I can follow!
I'm not at all a technophobe but my brain starts to rebel when I read
"There are inherent inequalities in the detective quantum efficiency of the 
sensor ..." Yikes. I'm shooting with the M3 today to get
my head straight.

Cheers
Hoppy

-----Original Message-----
From: lug-bounces+hoppyman=bigpond.net.au@leica-users.org 
[mailto:lug-bounces+hoppyman=bigpond.net.au@leica-users.org] On Behalf Of
HOWARD L RITTER
Sent: Sunday, 19 November 2006 07:31
To: Leica Users Group
Subject: Re: [Leica] It ain't alright.... "Fix it in SW"

Frank--

The fix is not as simple as a magenta-subtracting coating. If the  
problem were excessive sensitivity to a combination of visible  
wavelengths that causes a magenta cast in the final interpretation of  
the color balance, then that's basically irrelevant in an engineering  
sense. This is a fixed proportioning, in visible colors, of the  
differential responses of the individual pixels. It's just a matter  
of how the numbers of electrons in the wells corresponding to the R,  
G, and B pixels are interpreted in software. There are inherent  
inequalities in the detective quantum efficiency of the sensor at  
different wavelengths in the first place, and this differential  
efficiency has to be accounted for in the camera design, either in  
the particulars of the color filters over each pixel or in firmware  
processing of the signal.

However, if I understand correctly, the problem is that infrared  
wavelengths, which are not perceived visually and are disregarded in  
any process of presentation of the final image (monitor, printer,  
etc.), are inappropriately detected by the photon-detecting elements  
in such proportions as to be interpreted as magenta. Thus the  
recommendation of an IR filter, because this problem is, in any  
rigorous or non-arbitrary way, impossible to fix in software. There  
is obviously no way for any algorithm to distinguish between those  
electrons in R, G, and B wells generated by photons of (visible)  
wavelengths that constitute the color perceived as magenta on the one  
hand and, on the other hand, those generated by IR wavelengths. And  
the admixture of IR to visible differs arbitrarily from point to  
point in any picture, just as the colors do, so it's not a matter of  
subtracting some constant in SW.

In the first case, a separate "magenta-minus" filter or coating would  
be clumsy and completely unnecessary, and a software "fix" (not a  
fix, just a proper setting of engineering variables) no issue at all,  
either unnecessary because of filter selection, or trivial and  
inconsequential. In the second case, the addition of an IR-minus  
filter is the only rigorous solution, short of technology that  
results in sensors that are simply insensitive to invisible  
wavelengths like IR. Apparently it is not feasible to build IR  
rejection into the R/G/B pixel filters, as it is usual to build in  
this function as a plate in front of the sensor, but apparently this  
was optically not feasible in the case of the M8. Any software "fix"  
would be a compromise that risks having arbitrary elements, and this  
is why Leica recommends the inelegant filter solution.

--howard