I've a white balance problem - any ideas?

Just to throw a spanner in the works regarding Pentaxian's theory, I believe the way in which the light produced by tungsten and fluorescent lamps is fundamentally different.

Tungsten lamps emit light because the filament is heated up by a current passing through it. The amount of current relates to the temperature of the filament, and the colour of the light produced. Low current, yellow/orange colour (think of a torch with a low battery), high current, whiter light (though still yellowish). As stated, as the current cycles, the filament does not have time to cool down, so the light stays constant.

I believe that fluorescent light tubes work like this:

A cathode in the tube emits electrons when heated, which then strike atoms in the gas, which emits ultra-violet light as a result. This UV light then hits a fluorescent coating on the inner surface of the tube, which then emits visible light. Far more detail (than you could ever wish for) is here: http://en.wikipedia.org/wiki/Fluorescent_lamp

What this means, is that either the tube is emitting light (current flowing), or it's not. The flicker is as a result of this. You would not get the changes in light temperature as the both the UV and visible light is produced due to excitation of electrons, which will result in a given colour of light for a particular material. The combination of phosphors (the fluorescent material) dictate the colour of the light.

Basically, the light temperature is not going to fluctuate.

I am more than happy to be proved wrong.

Dan

...and John beat me to it.

And Peter made another good suggestion.

Dan

But the fluorescent tube is never at 0 volts. It's in a plasma state that constantly emits electrons. The phosphors don't switch in and out, they glow constantly.

Therefore, the colour temperature, which is what we are talking about, will be constant.

I have just captured a clear example of the effect happening - under a mixture of fluorescent tube and daylight. With the same camera settings (including white balance), about half the shots have one tint, the other half another.

Pics to follow later.

The complex bit will be to find a solution along the lines of what I suggested. I could probably come up with some instructions for Silkypix. The daylight is fading now so I might have to save it for another day.

The phosphors of some tubes dim as the voltage lowers. There is certainly a flickering effect with some tubes. When there is a mixture of light sources, then the colour temperature will change. Examples to follow later (in addition to the fencing ones already posted).

I didn't say it was.

No it's not, it emits elctrons when the current flows to heat the cathode. Hence the flicker.

The colour temp will be constant, but not because of that. It's because a particular element, or compound, will emit light at a particular frequency when an electron jumps from one 'shell' to another ('probability cloud' if we want to get a bit quantum ). It won't emit a different colour if the energy trying to cause that electron to jump shells is less, because the electron simply won't jump. (Before someone comments, I know it's when the electron returns to its normal posittion that the light's emitted. )

Who'd have thought that the original post would have got us here?

Dan

Took a load of shots with continous shooting mode - under a mixture of fluorescent tube and daylight, with all the same settings, including the same manual white balance.

Each image fits one of two categories: bright and slighty green, or dim and purple. Four typical examples:













K10D 1/500s f/1.4 at 50.0mm iso500

I would think it's fair to say that the FL tube will change in brightness as the current goes through its cycle (the number of electrons emitted by the cathode will go through a similar cycle, therefore the amount of fluorescence will cycle too). So the mix of daylight to artificial light will also go through this cycle.

How this actually relates to the half and half effect Iceblinker has found, I'm not sure, but I would imagine that with very careful study, it could be shown that it wasn't just a choice of two options, but a whole range, but to the eye it would look like one tint or the other.

Be interesting to see the results.

Dan

I can see some slight variations in hue, but they seem to fit into two clear categories: distinct enough for practical purposes.

UPDATE: 2 more pics added to my above post.

Doing an animated GIF shows us a cycle through the 4 images, notice how the colour does change slightly between the two showing similar tints, a bit more magenta seems to creep in. Also, the more orangy ones are darker - is this when the FL tube is 'off'?




Dan

If you use a K20D in burst mode, you will be able to see a lot more of the flickering effect. This is one of the reason they never use fluorescent as main lighting on a movie stage. It makes editing a nightmare. As to where I got the info, well just remember I'm 50 and my old school physics books are long gone. The best thing you can do with fluorescent is to shoot RAW and correct in post.

You're quite right. Good GIF! But still I think it might be useful to divide a large set of images into two lots for some initial bulk processing. Manual fine tuning could be done to the best ones afterwards.

With just a small set of pictures though, I wouldn't bother attempting any automatic process at all. I would simply adjust each one manually untill it looked OK.

That would make sense to me.

I was going to quote one post but there are nuggets of truth in many of them.

here's my take:

Fl lights cycle off and on but it is now very unusual for them all to be doing it together, if the pictures were taken in an old building with old lights then maybe this would be the cause. Flourescent lighting is murder, the spectrum is incomplete, to white balance, as Don says manual WB is the way to go but you are never going to get the range of colours you get in daylight.

My pics above were with manual WB. It doesn't stop the variation if the FL tubes are flickering and there is some other light source in addition.