Channel Noise on D70

[kipper]

Which channel is more likely to get noise in than others?

Just took a photo at 400ISO, there seems to be a fair bit of noise in the blue colors.

[gstark]

Channel 10. Usually after midnight.

[Nnnnsic]

In my experience it's been red, but that could possibly be because of the D70's preference to the colour red.

[mikephotog]

My experience on my D70 and on a Fuji S2 is that the blue chanel is noisest.
I think I've read that the blue is usually the noiser.

[MattC]

It is interesting to look at the raw output (prior to WB adjustments) from dcraw. Even with a daylight illuminated subject, the blue channel really does seem to suffer from underexposure (and therefore a tendency towards noise). Probably by a couple of stops or more. The situation is far worse with incandescent lighting. The sensor really does like red. Green is somewhere in between running fairly close to the red channel, and rarely seems to be an issue.

On face value there would appear to be a fairly strong case for using a colour temperature meter and colour correction filters for best results. The photogs that have travelled this route all seem (me included) to correct to 5500K which may not be optimal. I suspect that the point at which the three channels are equalised (or close to) is something higher than 5500K. 5500K does however yield good results and is an easy number to correct to.
The downside to using filters is the loss of light through the lens. Bumping ISO to compensate has its limits (assuming that shutter speed and aperture cannot be adjusted further). For indoor work, where the blue in ambient light tends to come up short, I use filters. Outdoors where colour temperatures tend to be higher anyway they mostly stay in my bag - unless I am mixing it with flash. Actually, my use of filters came more from shooting flash in mixed lighting - the mix being ambient + flash (and I use filters on the flash as well). It just happened that there are other benefits to using filters.

One of these days I may do some testing, but I am no hurry. While I am getting acceptable results the rest is just pixel peeping. When print is the final output media, a little noise is not going to be an issue - it may not even be visible. HST, I aim to get the best results possible at the time of shutter release.

All up, I think Nikon has done a decent job given what appears, at a glance, to be a lack of sensitivity in the blue channel. BTW, this is all based on observations. I have done nothing to quantify those observations... Heck, I would not even call it qualatative and I could be waaaay of the mark.

Cheers

[Onyx]

Matt, I was considering colour corrective filters for use under incandescent lights. How many stops of light does the filter lose?

[MattC]

2 - 2 1/2 stops - and there lies the rub.

Cheers