Colour Temperature in Hot Shoe Flashes

[chilli_vision]

Hi All,

I have just been reading up a little on White Balance, and I was wondering what temperature my hot shoe flash kicked out, and whether my WB Flash setting was in line with this temperature.

My Flash (as per the manual) kicks out 5600K, which looking at my camera body, the WB Flash is set for 5400K.

So, I was have been wondering a number of things:

a) does my flash really give out 5600K or as with most spec's is not 100% true.

b) is there a way of testing this out as if it is closer to say 5800K, then maybe a different WB setting would be better suited. And, if my camera's WB is not quite 5400K the gap could be wider.

c) is 200K noticeable in an exposure?

d) do different makes of flash vary or is 5600K a standard? Can't seem to find any details on the Nikon/Canon/Sunpak websites.

Ok, I understand the conversations about shooting in Raw allows you to fine tune WB in Photoshop so the above question may not matter that much, but I was just wondering about how accurate the tech really is.

Comments welcome.

Matt

 

[Garry Edwards]

200K doesn't matter for portrait-style photography, it does matter for some subjects such as commercial still life, fashion and so on. See this article in the Lencarta Learning Centre - the picture at the bottom shows the same shot at increments of 200K.

5600 is what I would call a theoretical standard, i.e. it's the figure that most manufacturers quote, but my cynical mind tells me that most manufacturers know full well that the average punter doesn't understand colour temperature and has no means of measuring it. One seller quotes colour temperature accuracy of 100K with his lights, but his claims are wildly optimistic.

If your flashgun really does output 5600K then that will be at full power only. At full power on a hotshoe flash (and at any power setting on a studio flash) the flash starts off warm, quickly reaches a cold peak and then, as the power dies off over a period of time, it becomes progressively more warm. The theoretical average is say 5600K. On a hotshoe flash, it doesn't have the 'tail off' of warm colour at low power because it's quenched using IGBT technology that brings the flash to an abrupt end. Therefore, the shorter the exposure, the more blue the flash.

A slight caveat here, apparently Canon flash guns have a means of controlling this, but I don't use Canon and don't have details.

Ok, I understand the conversations about shooting in Raw allows you to fine tune WB in Photoshop so the above question may not matter that much, but I was just wondering about how accurate the tech really is.

It may not be important if light from the flash is the only source, but you can't correct in photoshop if different light sources are outputting different colours.

 

[HoppyUK]

Garry, the thing about Canon (and quite possibly other makers) correcting flash colour is done in camera with the white balance, not to the flash at source. And from what I can tell, is extemely subtle. I can honestly detect zero difference between pictures taken at any level from full power or 1/128th, with my 580EX gun and 5D2, either shooting on AWB or pre-set to flash, or daylight etc.

The information came from Chuck Westfall of Canon USA and he is always totally reliable, but even he cannot get further details from the factory. They don't release much, not even the flash durations at anything but full power etc.

To the OP, really, don't worry about it. Compared to the variations you can get from the influence of spill from coloured surroundings, the flash itself is neither here nor there. If you have any problems, then either do a custom white balance, or shoot Raw and correct in post processing. Dead easy either way.

 

[Garry Edwards]

Richard,

Thanks for clearing that up, I didn't know whether it was the Canon cameras or the Canon flashes that apply the correction. As a matter of interest, do you happen to know whether the technology is on all models?

 

[dalex_257]

My Flash (as per the manual) kicks out 5600K, which looking at my camera body, the WB Flash is set for 5400K.
...
d) do different makes of flash vary or is 5600K a standard? Can't seem to find any details on the Nikon/Canon/Sunpak websites.

The WB cannot be accurately described solely as colour temperature alone. This is like many to one mapping - many different white balances will have the same colour temperature but will look different to the eye. In raw development software this is typically compensated by having tint control as well as temperature but even these are not precise. The in camera WB presets usually only set colour temperature with some sort of default tint (but that tint is probably based on the camera manufacturer's own flashes/defaults). This explains why the WB from light sources of supposedly the same colour temperature may and will look different. The only solution to this is to do a custom WB shooting gray card in your specific lighting conditions.

If you want a a bit more technical explanation - here it is. The camera sensor response is balanced to a certain light source (some sort of tungsten I believe). This means that if you capture an image in that precise idealistic light and won't apply WB at all - it will look just right. The other light sources will all cause certain channels (red and blue) to over or underexpose which is why we have WB. The WB is effectively a multiplication coefficient that is applied to red or blue channels in raw data. The most correct way to think of this is as a separate exposure compensation for red and blue channels.

Now the colour temperature of light is sort of an averaging over the spectrum of the light. Naturally, the spectrums of different light sources can be varied in different parts yet could still addup to the same colour temperature. Look at this image for example (the CIE colour space locus) - there the temperatures along side each marked line (say 5000K) are the same yet colour varies (this is what tint usually describes). The camera's sensor has a colour filter array that defines camera's spectral response so if you illuminate sensor with light sources that have the same colour temperature but different spectra, the sensor will capture that differently. In this case the WB for one light may for example require red channel to be boosted slightly more than the other.

Tint and colour temperature settings for WB unfortunately don't make it very clear what is it that they are doing - the more precise control would be to give the direct control over WB coefficients for red and blue channels (or exposure compensation) but so far there is only one active RAW converter I know that does it.

 

[HoppyUK]

Richard,

Thanks for clearing that up, I didn't know whether it was the Canon cameras or the Canon flashes that apply the correction. As a matter of interest, do you happen to know whether the technology is on all models?

I don't know about different Canon models Garry. I would assume it's a function of E-TTL II. What I do know is that when I shot a whole series of images of a grey wall, at power levels from 1/1 to 1/128th, with AWB or pre-set to flash etc, I could detect no difference in colour whatsoever.

On a related note, I've recently been playing with the 'tail-sync' hack that is supposed to give flash sync at highest shutter speeds - triggering via an optical slave off the HSS/FP pulse which puts the firing right at the start of the shutter cycle (if only to prove how useless it is as a technique ). This captures (almost) the whole of the flash pulse and at high shutter speeds like 1/2000sec any colour change over the duration should be clearly visible. I was expecting to see some colour shift here, but in reality, nothing noticeable - not that I was specifically testing for that but if there was anything significant it would have shown up.

I know that you are very particular about colour temperature Garry, but my experience and quite a lot of testing with my Canon flashes and Elinchrom D-Lite studio heads is that it is both accurate and consistent at all times.

To the OP, I would just add that colour balance is often more subjective than technically accurate. The colour of a subject changes a lot according to the light falling on it, and our eyes/brains adjust and compensate for that so that it 'looks' right. We're actually quite tolerant of colour changes and it's certainly not something I worry about. I generally use the camera pre-sets, or sometimes a custom white balance setting if I'm maybe bouncing the flash off a coloured wall.

The times when colour is critical, I think, are when shooting for a catalogue of something where customers are making a specific colour choice, or a wedding where the bride knows exactly what delicate shade of white her dress is!

Generally, colour consistency is more important than strict accuracy. For example, shooting that wedding again, pictures taken outside in sun, shade, and indoors with both flash and ambient need to match up pretty well as even slight differences really jump out when you compare them side by side.

The other time is when shooting with slow-sync flash indoors - balancing the flash with ambient tungsten light. What usually happens there is that either the foregound subject lit by flash looks too blue or the background too orange - or a bit of both! There is nothing you can do about that in post processing but it's easily fixed with a light orange CTO gel* over the flash and reset the white balance to tungsten. That will get the two light sources quite closely aligned, then tweak the overall result to taste in post. I find a little bit of 'incorrect' warmth usually looks more natural in that situation than strict accuracy.

* coloured cellophane wrapper from Quality Street works pretty well!

 

[Garry Edwards]

I'm inclined to agree. With the Fuji S3/S5 Pro cameras people raved about the good skin tones, in fact they were far too warm, but people generally like warm.

And a point that people often miss, with studio flash, is that virtually any modifier will change the colour quite a lot - it may not make much diffference when the umbrella/softbox is brand new but they yellow with age...

Some time ago, Lencarta produced a 'warm white' beauty dish. It could have been pure white (or as close to it as paint technology allows) but the decision was made to make it a sort of ivory colour, and they sell like hot cakes simply because people like the warm results.

And I also agree with you that colour consistency is much more important than the actual colour of the flash, and that's why personally I get pretty ****ed off when some sellers claim that their flashes only vary by (say) 100K when actual tests show that they vary by very much more. If I was cynical I might wonder whether they do this deliberately, knowing that very few people are able to actually measure colour temperature.