Looking for a rainbow.

[horvendile]

horvendile:
Yes, that would be apreciated. Either here, or by mail.

I'll start with a small picture here, which just happens to be the view from my kitchen window:

It's taken with a Nikon Coolpix 885, though unfortunately I don't seem to have saved the information on what settings were used - if that is of any interest.

I have that one and several others in full 2048x1536 glory readily accessible if you are interested - might be easier to scrutinize colours in full size.

 

[Basic]

Interesting. The rainbow seem to have a fairly constant intensity over the colors, except maybe for the purple end of the spectrum. I must admit that it was a long time since I saw a rainbow around here, so I don't remember how high intensity that part use to have. I do remember that that end of the spectrum use to be fainter, but not by how much.

Would you say that the image is a good representation of what you saw with your eyes?

I'm still hopfull that there's no "holes" in the spectrum received by a digicam.

 

[Entropy]

I'm still hopfull that there's no "holes" in the spectrum received by a digicam.

I'm afraid that's optimistic.
This pdf is from the site of Sinar, the Swiss manufacturer of large format cameras. It shows both the "ideal" spectral sensitivity curves, and the "real". What is "real" for Sinar is not necessarily the reality of the cheap CCDs in cheap and cheerful consumer cameras.

http://www.sinar.ch/sinar/imaglog/picts/Spectral.pdf

I'd simply dismantle the digital camera in question, and see if I could make out what capture element it uses, then contact the manufacturer for the data of the part.
Sony manufactures a large percentage of the consumer devices.

 

[Basic]

But the curves labeled "real" looks quite OK for this application, except in the violet range. (I assume that they mixed up the wavelength scale.)
It doesn't have to folow human vision. The only requirements are that it should be able to detect all wavelengths with reasonable sensitivity, and that it should show a difference between diferent wavelengths. (You could even relax that to "show a difference between some wavelengths".)

I finally made a rainbow, but it's not practically useful for these photos, since I can't sweep it over the obect in a practical way, and it's low intensity. But it was enough to see what my Canon Digital IXUS400 thought about it. It seems that it show the rainbow with rather constant intensity, but some colours are a bit off in hue.

 

[Zaphod]

What is "real" for Sinar is not necessarily the reality of the cheap CCDs in cheap and cheerful consumer cameras.

If using a camera capable of capturing RAW I guess you could be able to calibrate your way out of this as you then have control over the colour-space conversion. Most (if not all) consumer level digicams does funky undocumented things to the colours when converting into (s)RGB to make the images more pleasing to the human eye.

 

[horvendile]

Would you say that the image is a good representation of what you saw with your eyes?

Maybe this isn't so interesting anymore, but:
Yes, I'd say so. Likely not a perfect representation, and I took the picture almost a year ago and can't remember exactly how it looked, but the picture looks fairly right to me.
If anything, my camera has a tendency to exaggerate red.