colour correction filters

[geesbert]

since i switched to digital i never used any filter other than polarizers.

but i just wonder, sometimes i kie to do a few shots with just my flashes modelling lights, and the are really warm, so when i try to get them to balance the colours are just not like when i'd shoot with flash at 5500k.

i wonder, is anyone of you still using colour correction filters and do they help if your light is at the lower spectrum of temperature?

Stefan

[pixjohn]

I have used CTO and CTB just like the film days with good results. I have had problems with minus green on florescent bulbs  but that could be because of all the different color temps of different florescent bulbs.

since i switched to digital i never used any filter other than polarizers.

but i just wonder, sometimes i kie to do a few shots with just my flashes modelling lights, and the are really warm, so when i try to get them to balance the colours are just not like when i'd shoot with flash at 5500k.

i wonder, is anyone of you still using colour correction filters and do they help if your light is at the lower spectrum of temperature?

Stefan

[Dale Allyn]

Stefan,

I currently use a Phase One back (P25+) and find that setting a custom WB for shots like you describe (using modeling lights or other artificial lighting) is nearly spot-on and avoids the use of filters or significant adjustment in post. In my case, I find that shooting a white card or piece of paper to set the custom WB works great. Some shoot gray cards, but I like the results from the white paper in my setting. Phase made setting custom WB extremely easy, but I can't speak of other backs. I assume they're similar.

Dale

[Doug Peterson]

When you white balance a raw file which was shot without on-lens optical filtration you are giving a defacto push to the underexposed channels. For instance shooting in a heavily orange/yellow light your blue channel will be underexposed and when you WB you will be pushing the blue channel, sometimes by several stops.

When shooting a digital back like Dale's Phase One P25 at low ISOs it is not that detrimental to the final product because digital backs have such great latitude and clean shadow information that a 2 or 3 stop push of one color channel won't be noticed. However with a dSLR (even high end dSLRs) the effect can leave you with noticeably increased noise in the blue channel. This will show up most in lower midtones and shadows and will result in murky, color-inaccurate, noisy shadows.

Any optical color correction that gets you closer to equal exposure on all three channels (closer to "daylight") will decrease noise in the final product, even if it isn't perfect. Naturally using the color correction filter reduces effective ISO, costs extra money, darkens the viewfinder, and requires extra effort, so most photographers don't bother. Even for the quality minded it's likely only worth it if you are especially interested in clean detailed shadows or if the color cast of the light is especially strong.

Doug Peterson,  Head of Technical Services
Capture Integration, Phase One Dealer
Personal Portfolio

[bjanes]

When you white balance a raw file which was shot without on-lens optical filtration you are giving a defacto push to the underexposed channels. For instance shooting in a heavily orange/yellow light your blue channel will be underexposed and when you WB you will be pushing the blue channel, sometimes by several stops.

When shooting a digital back like Dale's Phase One P25 at low ISOs it is not that detrimental to the final product because digital backs have such great latitude and clean shadow information that a 2 or 3 stop push of one color channel won't be noticed. However with a dSLR (even high end dSLRs) the effect can leave you with noticeably increased noise in the blue channel. This will show up most in lower midtones and shadows and will result in murky, color-inaccurate, noisy shadows.

Any optical color correction that gets you closer to equal exposure on all three channels (closer to "daylight") will decrease noise in the final product, even if it isn't perfect. Naturally using the color correction filter reduces effective ISO, costs extra money, darkens the viewfinder, and requires extra effort, so most photographers don't bother. Even for the quality minded it's likely only worth it if you are especially interested in clean detailed shadows or if the color cast of the light is especially strong.

Doug Peterson,  Head of Technical Services
Capture Integration, Phase One Dealer
Personal Portfolio

This is a medium format forum, but for discussion, I will present data from a high end dSLR (the Nikon D3) with exposures in daylight and under tungsten illumination that was quite yellow (white balance in ACR = 2000K). The medium format people like to brag about the dynamic range and low noise of their sensors, but I have seen no data to indicate that they have much advantage over the dSLRs, except for having a higher pixel count. Full well electron count and read noise would be interesting.

Anyway, here is an image exposed under daylight with a raw histogram generated by Photobola Rawnalyze. The white point is indicated by the dotted line in the green channel. The x-axis is linear with minor tic marks in 1/3 stops and the major tics in full stops. Although the image appeared well exposed on the camera histogram, it is not fully exposed to the right. The blues are about 0.3 EV below the greens and the reds are a little over 0.67 EV down. One could use a magenta filter to hold back the green by about 0.5 EV and achieve approximate balance in the channels. The required density would be about 0.15 or CC 15M.

[attachment=9052:BirdsHistogram.jpg]

For the tungsten exposure, one would not use an 80A type filter, since this would bring the light up towards daylight, and daylight is not the natural white balance of the camera. Rather, one would need to hold the red back by 2 EV and the green by about 1.33 EV. One could use a combination of magenta and cyan filters to hold back the green and red by the appropriate amounts or compromise and use a blue filter to hold back the red and green by about 1.5 EV. However, the shot was taken at ISO 3200 and the exposure correction required for this filtration would compromise the shutter speed. One could use this filtration for stills with the camera on a tripod.

[attachment=9053:BasketBHisto.jpg]

[Graham Mitchell]

I still use color correction filters with digital, because they give me better RAW files. This way there is more info to work with in post. Fine CC filtering is not needed, but the rough correction can give dramatically better RAW histograms.

Interesting, I have had similar thoughts lately. The streets here are lit at night by sodium lamps which make colour photography hopeless. The RGB channels are so mismatched that a WB isn't very satisfactory and there are noise problems too.

Has anyone found a good filter for shooting under sodium lights? I heard Singh-Ray makes a filter called the "Lucalox Type B". Anyone tried it?

[Panopeeper]

Color correction filters can be used to enhance the dynamic range of the sensor. For example in daylight the green channel clips much earlier than the others in the typical landscape. The magenta filter reduces the green exposure relatively to the red and blue.

Reducing the greens allows to increase the exposure, thereby bringing some of the blue and red pixles out of the noisy shadow. Thus the dynamic range can be increased by about 1/3 to 1/2 stop; not much, but it may be crutial.

However, this works only if there is plenty of light. The magenta filter reduces the reds and blues as well (but to a lesser degree, than the greens), so if the exposure can not be increased, then the filter causes loss of dynamic range, instead increasing it. Furthermore, the "red" and "blue" pixels capture some of the green as well. Thus the filtering effect on the exposure per channel is not straightforward.

Moreover, even the very highest quality filter causes some loss of quality and a cheapo filter can ruin the shot. Unfortunately, B+W discontinued the magenta filter production.

See following *raw* histograms:

- the first one is w/o filter; the green-red proportion is 13900:6100 ~2.28,

- the second one is with a Tiffany CC30M, the exposure is 1 EV greter; the green-red proportion is 11500:6200 ~1.85.

The difference in the proportions is ~0.43, close to one half of a stop.





Notes:

1. The chosen exposure did not cause any clipping on raw level; I picked this one, for the change can not be measured if clipping occurs (if saturation occured, then it is difficult to impossible to say, how high the exposure exceeded the well capacities),
 
2. the proportions between the channels depend on the actual scenery and illumination,

3. the shot with the CC30M filter on required 2/3 EV higher exposure to reach the same level of red and blue. The point is, that this could have been increased a further 1/3 stop, and that would be the gain in DR.