White balance choice for ngiht photography

[flipfrog]

assuming there is not really any artificial light sources in the image, what works best for you guys and gals?

[Oneputt]

Auto

[MCWB]

Indeed. I shoot auto almost 100% of the time, because it just works. The only time it doesn't work is when you have a very cool or warm coloured background, e.g. shooting against a red or blue wall.

[birddog114]

Auto here also.

[Mal]

Shoot RAW Shoot AUTO WB.

[sheepie]

Shoot RAW Shoot AUTO WB.

Likewise - but if you must choose a WB mode, go with Direct Sunlight - the stars are the same sort of stuff as our sun you know

[Zeeke]

Auto here aswell..

Tim

[Gordon]

I do a lot of night time photography... and I'll join the chorus


alltogether now... AUTO


Quite often I still have to give it a slight adjustment in PS though.


Gordon

[marcotrov]

Auto here too
cheers
marco

[Onyx]

Here's another question leading on from this one: what Kelvin temperature does Auto choose?

[Mal]

Here's another question leading on from this one: what Kelvin temperature does Auto choose?

According to the manual
"White balance adjusted automatically based on colour temperature from 1,005 pixel RGB sensor and CCD image sensor"
3,500 - 8000K

[Onyx]

Yes Mal, but in the context of at night with no artificial light sources... ???

[Mal]

Yes Mal, but in the context of at night with no artificial light sources... ???

Yes well here is where the manual lets me down
And where this forum picks up!......

The correct answer will follow in the next couple of posts.

[Gordon]

I'm not sure that it does select a certain K temp. Maybe it looks for the brightest almost white part of the scene and makes it white.
I know the D70 sensor does not record emission line light very accurately. For example, the bright emission nebula in the Large Magellanic Cloud known as the Tarantula nebula should be pink, but here it is in a 30 sec exposure last night with WB set to auto

30mm @ f/1.4

Its looking very blue, so theres something weird about the colour sensitivity of the CCD.

Gordon

[Alex]

I use AUTO -2.

Alex

[digitor]

I'm not sure that it does select a certain K temp. Maybe it looks for the brightest almost white part of the scene and makes it white.
I know the D70 sensor does not record emission line light very accurately. For example, the bright emission nebula in the Large Magellanic Cloud known as the Tarantula nebula should be pink, but here it is in a 30 sec exposure last night with WB set to autoIts looking very blue, so theres something weird about the colour sensitivity of the CCD.

Gordon

Hi Gordon, this is very interesting - I may be able to do some testing on this - can you tell me what the line is? (wavelength or gas - I'm easy!) This is a subject which I have an interest in, and I can do some checks with a monchromator at work.

Cheers

[Gordon]

Its mostly a mix of H alpha and OIII or about 656nm and 500nm.

I suspect its not seeing the H alpha at all, and just recording O III plus the lesser green and blue emissions and continuum there.

Strangely, it does see the IR from the remote trigger, I had it in my hand when setting off the camera for a pic the other day, and I can see a brightish point of light there.

Gordon

[digitor]

Hmm, while I would imagine it would be full on around 500, given the photopic(ish) response of the sensor arrangement, I would have thought it would still be going strong at 650! Maybe they've sneaked a laser pointer filter in there, and not told anyone. I'll see if I can check it out next week, and post my results.

Cheers

[Gordon]

Yes, they have done something weird at the red end, which results in the strange results when photographing anything red, such as cherries. A sort of washed out, almost saturated appearance when there should not be one.

I'm keen to see what you discover!

I'm heading bush for a few days to the rainforest on a mountain biking/photography trip, so there wont be much redness to photograph there

Gordon

[Onyx]

Yes, they have done something weird at the red end, which results in the strange results when photographing anything red, such as cherries. A sort of washed out, almost saturated appearance when there should not be one.

Gordon, I believe this is a separate issue, and that of the red channel reaching saturation before the green and blue. Hence the red channel ends up being clipped, while increasing exposure increases only the green and blue. If you photograph those red cherries with -1EV, then 'fix' in PP, you're likely to get a better result.

I find the other difficult colour for the D70's sensor is violet. Same thing happens, colour shade different from expected, histogram shows clipping (blue channel).

[Gordon]

Thats what I dont understand- the CCD is clearly not sensitive to 656nm wavelength, yet red reaches saturation first! So it must have a severe red cut-off filter in there to cut off the long wavelength end of the red, maybe in order to reduce the overall red sensitivity, but still its not quite enough. Violet is a mix of red and blue, so maybe that is why it is not recorded properly?

Gordon

[Big Red]

Yes, they have done something weird at the red end, which results in the strange results when photographing anything red, such as cherries. A sort of washed out, almost saturated appearance when there should not be one.

Gordon, I believe this is a separate issue, and that of the red channel reaching saturation before the green and blue. Hence the red channel ends up being clipped, while increasing exposure increases only the green and blue. If you photograph those red cherries with -1EV, then 'fix' in PP, you're likely to get a better result.

I find the other difficult colour for the D70's sensor is violet. Same thing happens, colour shade different from expected, histogram shows clipping (blue channel).

my pentax is the same but i guess they are the same sensor as the D70 and get similar results.

[robboh]

AUTO -2 for me usually (use raw).
Only time I dont use AWB is inside at night.

The AWB doesnt seem to reach down to the light temps for incadescent, so I manually set that.

[LIVE4EVA]

Hi all.
Newbie question here
What sort of WB would you recomend for a night shot of a city.
and also if the city happens to have lightning would this effect the choice?
atm i am just on auto but it seems to be rather red.
Thanks
LIVE4EVA

[BullcreekBob]

The correct answer will follow in the next couple of posts.

No it didn't.

Please check your facts before posting such claims

Bob in Bull Creek

[digitor]

Yes, they have done something weird at the red end, which results in the strange results when photographing anything red, such as cherries. A sort of washed out, almost saturated appearance when there should not be one.

I'm keen to see what you discover!

I'm heading bush for a few days to the rainforest on a mountain biking/photography trip, so there wont be much redness to photograph there

Gordon

WARNING: Long and boring post!!

Here is one of a series of 19 shots I took today of the output of a monochromator, from 651nm to 660nm, in 0.5nm steps. The output slit of the instrument was illuminating a standard reflectance tile, which has a substantially flat reflectance from 400 to 700 nm. The images were taken of the tile with a Nikkor 85/1.4, focussed at infinity, from a distance of approximately 300mm, so are significantly out-of-focus.

As can be expected, all of the shots looked exactly the same, here is the one at 656.0nm:



All the wavelengths were captured by the D70 at what appears to be the same brightness. They all appear exactly the same colour not only because of their very similar wavelengths, but also because only one channel is exposed, due to the filters in front of the ccd photosites. If a line source (single wavelength of light) is not in the region where the filters' response overlaps, only one colour, red or blue or green, will result.

What is interesting, though, is what happens when the image is overexposed, as you can see here.



This effect, caused by the blown red channel, could explain some strange colour shifts experienced by Gordon in his astrophotography.

Interestingly, the camera histogram never indicated overexposure. I assume this is because on the D70, the histogram is an addition (or some sort of combination) of the three channels. Assuming each channel is given equal weighting (which I doubt) and the light was exciting only one channel, a single line light source could never indicate more than two stops underexposure. I'm not sure if the histogram axes are linear or logarithmic either, finding this out can be a project for another day!

I would think the colour iconsistencies noted in the astronomy images would be as a result of one or both of two things:

1: Overexposure of a single channel

2: The inability of a three-colour imaging system to reproduce a combination of line sources in the same way as is perceived by human vision

I intend to do some further testing on the second point, by taking some pictures of a wavelength calibrator light source, which has three main emission lines, and looks pinkish to the eye.

Cheers