Kelvin (20D)

[aam1234]

(I hope I spelled it right). There is a function in 20D where you can set the kelvin temperature. Was wondering if that is a substitute for custom WB or not. Or Am I way off.

Thanks.

Btw, what's the past tense of "spell" thought it was "spellt"

[Wildman]

Kelvin is a measure of temperature... the light emitted by a "standard black body". When a body is heated up, it glows depending on its temperature. The Kelvin reading is an accurate measure of the color of the light of the body at that temperature. It's a very precise way of letting you set white balance. Tungsten light is about 3200 Kelvin and sunny light is about 5500 Kelvin. The 20D lets you compensate in 100 K steps, a pretty fine gradation.

[aam1234]

Thanks Wildman,

So if you can measure the temperature (in Kelvin) you don't need to set the WB, you just use that temp. Is that correct.

Thanks

[Persian-Rice]

Thanks Wildman,

So if you can measure the temperature (in Kelvin) you don't need to set the WB, you just use that temp. Is that correct.

Thanks

mmmhm, you can use a kelvin meter, which tend to be stupid expensive.......

[aam1234]

Thanks Wildman,

So if you can measure the temperature (in Kelvin) you don't need to set the WB, you just use that temp. Is that correct.

Thanks

mmmhm, you can use a kelvin meter, which tend to be stupid expensive.......

Fair enough. Then why there is a kelvin temp in the camera if it's so expensive and nobody will use it. I'm sure there is a use for it, somehow.

[Wildman]

Yup... a Kelvin meter is an option, tho' an expensive one. If you shoot in RAW mode, you can very accurately adjust the white balance using Kelvin temperatures. I guess if you did enough photography, you could guess pretty accurately the Kelvin value at the time you shoot the picture, then tweak it in post processing.

The advantage to Kelvin values over custom white balance is accuracy (probably more accuracy than 99% of users need). Sun, shade, incandescent, florescent, etc. represent bands or regions of Kelvin values. When shooting film, control of lighting requires changing film types and/or using filters to get the correct Kelvin value. Ain't digital photography amazing?

[aam1234]

Sun, shade, incandescent, florescent, etc. represent bands or regions of Kelvin values.

That's one of the reasons of the original question. I thought if incandescent is normally, say 6k, but at your location it looks a bit brighter than normal incandescent, so you would adjust to maybe 6.5K for example.

[aam1234]

Where is Jon when you need him. He can sort it out very easily, he is so knowledgeable in the technical side of things.

[Persian-Rice]

AAM, there is a use for it, it is used for exactly what you are thinking about. A kelvin meter goes for atleast 700-800 USD................if you own one then its great, if not, use the custom white balance.

[aam1234]

Got it PR :wink:

So basically we shouldn't bother with it.

Thanks again.

[Wildman]

Persian Rice didn't say not to use it... it's just more tweak than most of us need. There are artists and photographers who really want this capability, so Canon opted to include it. It's really only some lines of computer code that will make a small minority of users happy, so why not?

[Persian-Rice]

No no no, use it if you want, I would. It's the most accurate and is exteremly usefull, it's just that it's very expensive.

If you can afford it and are willing to buy the colour meter, then go ahead, your WB will be more accurate then ours, thats for sure.

[robertwgross]

In my old D60 camera, I can set custom white balances, but I can't do it by any specific scale or color temperature. On some of the newer cameras, you can set the custom white balance the empirical way, or you can set it by a specific color temperature. Granted, almost nobody outside a scientific lab will have a color temperature meter, but at least you can have an assigned number on your setting. If nothing else, it is one number that you can remember. For example, "That tungsten light in my garage is 3250 degrees K."

If it were not Kelvin scale, then it might be Rankine.

---Bob Gross---

[dhbailey]

If you are shooting in the same place a lot (or with similar lighting conditions) you could create your own Kelvin test, since the camera adjusts the K setting by 100s, you could start at the low end, and shoot the 40 or so pictures, each at the next K setting higher, with the camera on a tripod, and then see which picture is most faithful to what you actually saw at the location. It will take time, but will cost a lot less than a Kelvin meter, however it will only work for that situation.

Here is a link to a chart which might help to give you a good starting point:
http://www.schorsch.com/kbase/glossary/cct.html (there may be other more exact charts or charts with more varied situations) you could print out the chart and then go and try the settings out in the situations described and see how you like the results. Then keep a copy of the chart in your camera bag for those situations when you don't think AWB will work and you don't trust the Custom WB/GrayCard results.

[Jon]

Where is Jon when you need him. He can sort it out very easily, he is so knowledgeable in the technical side of things.

Nice to be appreciated! Visiting the folks for Thanksgiving. That's where. ;{)#

Kelvin's only marginally related to brightness as in EV. It's like wildman said, as you heat up an object (say a bar of iron), it starts to glow. As it gets hotter, the predominant colour moves through the spectrum from infrared, through red, orange, yellow, green, blue, indigo, violet and ultraviolet (assuming it doesn't melt first!). Each colour will occur at a specific temperature, which we report in Kelvins (not, for reasons which elude me, degrees Kelvin - ask a physicist). 0 Kelvin is absolute zero - as cold as it gets. Normal household lights are around 2700-2800 Kelvins, depending on the wattage. Photofloods are 3200 or 3400, again depending on wattage. Daylight's (depending on overcast, shade levels, and so forth) anywhere from 4800 on to maybe 10000 Kelvins. So if you know the light source, you can use a fairly close approximation of its colour temperature. If you were to send any of these lights through a prism, you'd see a continuous spectrum, with intensities falling off toward the ends (although you might not see a full ROYGBIV spectrum with temperatures at the extremes).

Knowing a light's Kelvin temperature is also helpful if you've ever encountered MiReDs (Micro-reciprcal degrees) (1,000,000/(Kelvin temp)). These are used since a 100 Kelvin shift at the low end of the range will cause a much greater shift in colour than a 100 Kelvin shift at a higher temperature. MiReDs help linearize the evaluation. Most standard lighting conversion filters (80a, 85b, etc.) have MiReD shift values associated with them. If you know this, you can pre-visualize the impact they'll have when used outside their normal environment.

When you're dealing with incandescent light sources, voltage fluctuations and aging of the lamps can cause the light to get progressively warmer. If you're using fluorescents or other gas-discharge lamps (sodium or mercury vapour, for instance), a Kelvin temperature is only a closest approximation, since they don't emit a continuous light across the spectrum, but rather radiate discontinuously depending on what elements are being excited (think spectrophotometry).

Colour temperature meters (colorimiters) are expensive, but if you're doing critical work for reproduction they are very important. Consider trying to set a colour profile for your computer if you're taking your standard image under worn, low-wattage lamps and during a brown-out, and then trying to get colours in a picture taken at high noon in the Sahara to match! Actually, the reader you use is a form of colorimeter.