The topic of colour profiles comes up from time to time over at MeettheGimp.org which inspired me.
[Thanks @nachbar for the prod, I have been meaning to do this exercise for a while].
These are the Canon ICC profiles, but I suspect that Nikon's curves function in much the same way.
But these screen captures are probably of general interest, so I am posting them here at 23HQ.
I took some of the ICC profiles installed by DPP and used the GNOME Color Manager to graph them.
[This is a great linux tool written by Richard Hughes, who is a real hero, it was very much needed]
While I have long suspected these results, Rolf's excellent Checkpoint Charley shots reminded me of the value of pictures.
[Basically this is about 10,000 words worth of pictures]
[You can see Rolf's pictures here: http://www.23hq.com/rstein/album/6933556 ]
Anyhow, these show that the Canon in-camera settings affect the JPEGs produced by the camera.
DPP will normally use these when developing your RAW exposures too, but perhaps you can change that.
Actually, the situation is probably not as simple as these screen captures would imply. In fact, there
seems to be some very dark juju at work, with ICC profiles layered on top of other ICC profiles, and
some of the later profiles (for the 500D for instance) seem to be encrypted and cannot be viewed or
used for much at all. But there is an old Open-Source idea that 'information wants to be free' and so
I have graphed these (which is not a violation of copyright or anything) for anyone who is curious.
For myself, I set my in-camera setting to Neutral and then tweak my colours after the fact. This is
so that I get the most unprocessed version of what the sensor receives that I can. I would estimate
that these profiles are about 75% accurate when used correctly. That's more than good enough for
most uses, but 75% was only a 'B' when I was in university. For my pictures I think I can do better.
[My main reason for wanting RAW captures in the first place is as a backup in case my camera settings are wrong]
While sRGB is intended for a gamma of 2.2, I believe AdobeRGB is meant for a gamma of 1.8 (print).
And gamma is a non-linear thing, which is why the Adobe profiles are curved while sRGB ones are straight.