> -----Original Message-----
> From: Chuck Vaughn [mailto:aa6g@aa6g.org]
>> Look at a couple examples and maybe you can clarify it for me.
>
...
> My problem with the filters in the example above is that with 50%
> overlapping filters the intensity in the red channel will be
> twice that
> in the green and blue. If you plug that into the PS Color Picker
> you get pink. Is that really correct or should it be gray? If pink
> is correct, then that means it takes twice the level at OIII compared
> to H-a for it to be gray.
H-a at 100%, two OIII lines each at 50% with 50% of it going
to each green and blue channel. So the result is
R = 1.0
G = 2*0.5*0.5 = 0.5
B = 0.5
This will record as pink. I think I see why you wanted full
sensitivity across this region, but you are solving one artifact
of RGB imaging by introducing another.
The ideal filter set can discriminate line sources but it would have a funny
effect with continuous spectra. Consider an "E" source (flat spectrum),
which would appear white to us. Your filters would capture the
energy from the transition regions twice, the result being an
excess of cyan and yellow. This would not balance nicely to white.
If you calibrate the recording of E, you would end up reducing the
overall gain of the green channel, now the balance of the Ha and
OIII lines becomes magenta! As my father has often counseled,
"there are no answers, just different sets of problems". I think
this may be what he was talking about.
The essence of the problem is the reduction of a complex spectrum
into only three numbers. You can find a solution for one spectrum,
but not all.
Thor.