Archived from groups: rec.photo.digital (
More info?)
"Davoud" <see@below.net> wrote in message
news:190220051631034090%see@below.net...
> Winston:
>> On my 20D there's a custom function to reduce noise at high ISO's. I'm
>> curious why this is a custom function & not built into the firmware
>> already? When *wouldn't* you want the noise suppression turned on at
>> ISO's of 1600 & 3200?
>
> You have misunderstood the function. It is noise reduction for long
> exposures. Astrophotographers like me
> <
http://www.davidillig.com/ast-m42m45050127.shtml> use it because we
> make exposures of up to five minutes. With noise reduction on, after
> completing a four-minute (e.g.) exposure the camera makes another
> four-minute exposure with the shutter closed. This is called a "dark
> frame." The camera then subtracts the dark frame -- specifically, the
> thermal and other noise that it contains -- from the four-minute
> exposure. Thus a four-minute exposure with noise reduction turned on
> requires a bit over eight minutes to complete. Some astrophotographers
> prefer to make their own dark frames -- exposures the same length as
> the light exposure, but with the optics completely closed off -- and
> subtract them with the image-processing software, be it Photoshop or
> Images Plus or other.
>
> Davoud
>
> --
> usenet *at* davidillig *dawt* com
I guess the word noise is missleading.
Strictly, noise is a random process and will be different in the main
exposure and the dark frame on a pixel by pixel basis.
What the dark frame subtraction is removing is the pixel to pixel dark
current level, this is an offset or bias in statistics terms, not noise.
In fact as the actual random noise will be different in the two frames, the
noise variance across the pixels will be doubled by this process, so don't
use "noise reduction" unless your exposures are long enough for the pixel
dark current to be visible.
Lester