Manual WB Limits on D700 and 80B Filters

<p>I've noticed that when I set a manual white balance on my D700 directly by inputting the color temperature (in Kelvin), I'm limited to a range of 2,500 K to 10,000 K. When I open a RAW file with ACR, I can choose a color temperature as low as 2,000 K. Is there a reason why the camera's internal software limits me to 2,500 K? Is there no way to get around this? In challenging lighting conditions I try to use RAW+JPEG, and work from the RAW file, but sometimes I really want to be able to use a JPEG directly from the camera for a variety of reasons.</p>

<p>Has anyone had luck with using an 80B (or other 80-series cooling filter) to increase the apparent color temperature of incoming light with a modern DSLR? I'd hate to loose an extra one and third stops of light, but an 80B filter should decrement the inverse color temperature by 112 mireds, or enough to turn even 2000K light into just over 2500K. I also wonder if the cooling filter could help the AWB function usefully in quite amber light. But since those sources also tend to be quite dim, I wonder whether the exposure tradeoff would be worthwhile.</p>

<blockquote>

<p>Is there a reason why the camera's internal software limits me to 2,500 K?</p>

</blockquote>

<p>Yes. Every camera maker and raw converter author has a different way of mathematically transforming the RGB values that come from the camera's sensor into RGB values that correspond to a scene lit by different color temperatures of light. All of these different algorithms (bits of math) perform fairly well at color temperatures in the 3000-6000K range, but, for lack of a better term, turn to fertilizer below about 2800K or above 9500K. It's just too much to do with math, the "crossover colors", the yellows and oranges between R and G, and the blue greens between G anr B (which is more like V than B, anyway) get all screwed up. Now, what's in the yellows and oranges between R and G? The "important stuff", like skin tones.</p>

<p>And, as an added bonus, at very low color temperatures, there's so much red light and so little blue light that, if the camera exposes for the mid tones, the red channel gets blown, the blue channel gets underexposed, and that's just one more reason why the image gets "fertilized". Drop the exposure to try and preserve the highlights, and you get even more blue and red speckled noise in the shadows (red, because everyplace where the noise makes "not enough" blue instead of two much blue, you get an "anti-blue" hole).</p>

<blockquote>

<p>Has anyone had luck with using an 80B (or other 80-series cooling filter) to increase the apparent color temperature of incoming light with a modern DSLR?</p>

</blockquote>

<p>Yes. I've had great luck. Spectacular luck. So much luck that I always carry 80A (not 80B) filters in the sizes of my most commonly used lenses. Good ones, B+W, because this technique is important to me. And you have a D700, which definitely qualifies as a "modern DSLR". I made the mistake of trying this with an old D100, once. A typical 80A filter, even a pricy B+W, "leaks" infrared, so on an older DSLR with an infrared sensitivity problem, you get magenta "infrared contamination" all over the image.</p>

<blockquote>

<p>I'd hate to loose an extra one and third stops of light</p>

</blockquote>

<p>You won't. Because of all that math that mangles sensor output into something "sort of" right, at low color temperatures, the sensor's "effective sensitivity" is very low. With candlelight, you'll see as much shadow noise at ISO 800 on the D700 as you would at ISO 3200 with clean, white 5600K strobe or sunlight. So, whether you "lose" the light from the color math of trying to get a 200K white balance, or you lose it from a filter, it's always "lost". There's no "extra" loss from the filter, you're just "shifting" the loss from the sensor to the glass.</p>

<p>Just to add a bit to Joseph's good explanation. What's happening when you change the colour balance downward (toward red) is that the sensitivity of the blue "channel" is increased relative to the red "channel". There's a limit to how much the blue sensitivity can be boosted without excessive noise appearing, so camera maker's choose to put a lower limit on the Kelvin correction, mainly to avoid complaints from users. If you care to override this in ACR or some other RAW converter then that's your own lookout and the camera manufacturer can duck the blame. In any case a picture taken in, say, candlelight with a totally corrected WB will look quite artificial and any candlelit atmosphere will be completely lost.</p>

<p>As an aside, this is a very good reason not to take sunset or sunrise pictures with AWB set. Otherwise you'll get very washed-out colours.</p>

<p>As Joseph said, the use of an A to D filter of some sort will even out the red/blue balance and may result in pictures with a lower noise. I say "may" because by using the filter you'll reduce the overall camera sensitivity, which results in a higher noise floor. Depending on the light level this may or may not give a better result than simply shifting the WB in camera or in post-processing.</p>

<p>It would be instructive if camera makers let us in on the workings of their sensors a bit more fully. Like telling us what the natural WB (equi-sensitivity) of the sensor filter dyes actually was. Sure we can get this from the sensor data sheet if we can track it down, but it still needs a bit of technical know-how to work out the neutral WB.</p>