80A Filter in the Digital Age

[JoeKitchen]

So for a little while now I have been wondering if digital had an optimal color temperature range like film, but never thought of a way to test for it.  Then, Chris wrote something about thinking about using an 80A in a earlier topic.  So this morning, I got bored and decided to set up a pretty crappy still life of my small Christmas shrub on a wood table with some beer and wine bottles, cookies, plates, a blue Calculus book, etc.  I lit the table with two Tota lights, one reflecting out of a 50 inch umbrella and the other straight on.  Using a P45+ with C1 Pro v6, I captured an image with and without an 80A, using the exposure calibrator in C1 to get the exposures as close as possible.  

When noise is concerned, the differences are mute in the mid-tones and highlights.  In the shadows, color noise was worse on the non-80A image, especially when pushed hard, although not by much.  I did not have anything very dark and blue with varying tonalities, so I had to look into the shadows of the green tree to test for this.  Maybe with something darker and bluer, the noise would be much more less with the 80A.  

On the topic of color, it seems the 80A may be more useful.  I used the same spot on a white wall behind the scene to color correct.  The 80A image looked cleaner and more true, even though the same exact spot was used for color correction.  This was especially the case for objects with more red in them.  

In conclusion, if you are concerned about noise, using an 80A would only be beneficial for darker images.  However, if you need to reduce the color saturation/strength of only the red channel, using an 80A would help a good deal, it appears.  

See below, non-80A on the left, 80A on the right.  

[Chris Barrett]

So... not a whole lot of benefit?  Another thought I had was to always bracket with a +1 or +2 exposure.  Process that at -1 (or -2) to match Normal.  This should yield cleaner shadows and could be layered in with a luminosity mask.  Actually, I've done this a couple times with good results.

[JoeKitchen]

So... not a whole lot of benefit?  Another thought I had was to always bracket with a +1 or +2 exposure.  Process that at -1 (or -2) to match Normal.  This should yield cleaner shadows and could be layered in with a luminosity mask.  Actually, I've done this a couple times with good results.

I do not see too much of an advantage here, other than reducing the saturation of the red channel, although this could also be done in post.  For very dark images though, I could see using the 80A and getting some benefit out of it.  But with dark images on a wide lens with a CF, the exposure could already be 40 seconds, add an 80A, and now you're at 2 minutes, which leaves more time for a possible camera shake from something (say a very loud speaker sounding a canon blast that vibrates the whole concrete floor). 

[Chris Barrett]

Remember bracketing Ektachrome?  Felt like it took a half hour to bracket one damn shot!  On my interiors nowadays, though, I don't find myself ever exposing for more than 3 seconds.

[JoeKitchen]

Remember bracketing Ektachrome?  Felt like it took a half hour to bracket one damn shot!  On my interiors nowadays, though, I don't find myself ever exposing for more than 3 seconds.

fortunately I am young enough to never had to have felt the pain of shooting interiors at a pro level with chromes.   

Depending on the filters and lighting, my tungsten shots run between 0.5 and 6 seconds.  The 40 second I referenced above was in a very dimly lit museum, and yes, the canon blast really happened and ruined the first real exposure.  After thinking about the colors, the 80A may have benefited me here

[ErikKaffehr]

Hi,

I would expect that setting a white balance would take care of most differences, but I don't know as I have not tried.


Sory, I have mixed up the filter name, now I see what you mean!

Best regards
Erik

So for a little while now I have been wondering if digital had an optimal color temperature range like film, but never thought of a way to test for it.  Then, Chris wrote something about thinking about using an 80A in a earlier topic.  So this morning, I got bored and decided to set up a pretty crappy still life of my small Christmas shrub on a wood table with some beer and wine bottles, cookies, plates, a blue Calculus book, etc.  I lit the table with two Tota lights, one reflecting out of a 50 inch umbrella and the other straight on.  Using a P45+ with C1 Pro v6, I captured an image with and without an 80A, using the exposure calibrator in C1 to get the exposures as close as possible.  

When noise is concerned, the differences are mute in the mid-tones and highlights.  In the shadows, color noise was worse on the non-80A image, especially when pushed hard, although not by much.  I did not have anything very dark and blue with varying tonalities, so I had to look into the shadows of the green tree to test for this.  Maybe with something darker and bluer, the noise would be much more less with the 80A.  

On the topic of color, it seems the 80A may be more useful.  I used the same spot on a white wall behind the scene to color correct.  The 80A image looked cleaner and more true, even though the same exact spot was used for color correction.  This was especially the case for objects with more red in them.  

In conclusion, if you are concerned about noise, using an 80A would only be beneficial for darker images.  However, if you need to reduce the color saturation/strength of only the red channel, using an 80A would help a good deal, it appears.  

See below, non-80A on the left, 80A on the right.  

[eronald]

Joe

VERY Nice simple demo, proving how it is worth using a filter to balance the channels AT CAPTURE.
Visible Quality (= SN ratio) in the tree shadows improves dramatically. BTW, gelling the lights is also a possibility.
Of course a back has good DR, the demo would be much more dramatic with an average dSLR.

Edmund


PS. Erik, the whole point of this demo is that the filter allows you to balance the real physical channel data, if the channels are unbalanced, then wb means  a channel (usually blue) is less exposed and needs lifting by the wb process, thereby introducing noise during raw conversion.

[JoeKitchen]

Hi,

I would expect that setting a white balance would take care of most differences, but I don't know as I have not tried.

Best regards
Erik

At first I thought the color difference was a white balance issue, but then I started playing around with it.  I repeatably adjusted the white balance by sampling different areas of both images, but in the same exact spot every time (hope you followed me there).  The red channel was always more saturated in the non-80A images.  

But, I am sure that this can be solved in post too with the right profile.  

Gelling the lights would work too, but gelling my lights to match ambient is often easier than gelling the ambient, and my lights, to match daylight.  

[eronald]

At first I thought the color difference was a white balance issue, but then I started playing around with it.  I repeatably adjusted the white balance in different areas of both images, but in the same exact spot every time (hope you followed me there).  I red channel was always more saturated in the non-80A images.  

But, I am sure that this can be solved in post too with the right profile.  

Interesting. I guess the filter cuts the spectrum variably moving from red to blue, and in the transition zone objects will change color. I think you can understand what I mean ...

Edmund

[ErikKaffehr]

Hi,

Sloppy reading on my part! Sorry!

Best regards
Erik

At first I thought the color difference was a white balance issue, but then I started playing around with it.  I repeatably adjusted the white balance by sampling different areas of both images, but in the same exact spot every time (hope you followed me there).  The red channel was always more saturated in the non-80A images.  

But, I am sure that this can be solved in post too with the right profile.  

Gelling the lights would work too, but gelling my lights to match ambient is often easier than gelling the ambient, and my lights, to match daylight.  

[eronald]

Gelling the lights would work too, but gelling my lights to match ambient is often easier than gelling the ambient, and my lights, to match daylight.  

Joe,

Yes.

I think an internal filter wheel would make sense for many lenses, video cameras etc, especially now that electronic viewfinders have arrived.

Edmund

[Ken R]

Joe

VERY Nice simple demo, proving how it is worth using a filter to balance the channels AT CAPTURE.
Visible Quality (= SN ratio) in the tree shadows improves dramatically. BTW, gelling the lights is also a possibility.
Of course a back has good DR, the demo would be much more dramatic with an average dSLR.

Edmund


PS. Erik, the whole point of this demo is that the filter allows you to balance the real physical channel data, if the channels are unbalanced, then wb means  a channel (usually blue) is less exposed and needs lifting by the wb process, thereby introducing noise during raw conversion.

Yep, I agree!

One thing of note, the images with the 80A filter have a much longer exposure, that might introduce a bit of noise by itself, specially in low light situations where the high density of the 80A filter will mandate quite long exposures. Would be cool to see what happens with either a lighter cooling filter or in situations with higher light levels. I might have to dig up a few filters from storage. Haven't used colored filters in years.

[bjanes]

VERY Nice simple demo, proving how it is worth using a filter to balance the channels AT CAPTURE.
Visible Quality (= SN ratio) in the tree shadows improves dramatically. BTW, gelling the lights is also a possibility.
Of course a back has good DR, the demo would be much more dramatic with an average dSLR.

For the best SNR, it is desirable to balance the color channels so that each channel is optimally exposed to the right. In the case of tungsten illumination, the 80a will bring the color temperature to daylight. However, with most sensors, daylight is not optimal since the sensor is most sensitive to green. Some exacting photographers place a magenta filter over the lens to hold back green and balance out the channels.

For the D800e the raw histogram for a white target illuminated by tungsten light is shown in the first image below. The tungsten light is quite deficient in blue. An 80a filter will pass blue and hold back red and green to approximate daylight as shown in the second image, where the blue component is better exposed, but now the red is more underexposed. If one placed a CC20M filter over the lens, the green would be moved 2/3 EV to the left and the channels would be approximately balanced. However, the filters could degrade the image and the extra exposure could introduce more camera shake. With low noise sensors, such filtration is not generally worthwhile but might be useful when one has the camera on a tripod.

Bill

[eronald]

I guess you need to choose your filter or sensor with IR absorption cover according to whether you wish to best dscriminate clorophyll a, b, carotenoids or anthocyanins

In other words, one filter may be best for pines, the other for fall colors

It's a bit strange how esoteric stuff immediately cascades into visible differences.

Edmund

[alan_b]

...bracket with a +1 or +2 exposure.  Process that at -1 (or -2) to match Normal.  This should yield cleaner shadows and could be layered in with a luminosity mask.

I do this a lot - even going as far as 3-4 stops.  Works very well, and easier than packing filters.

[Stas Wilf]

At first I thought the color difference was a white balance issue, but then I started playing around with it.  I repeatably adjusted the white balance by sampling different areas of both images, but in the same exact spot every time (hope you followed me there).  The red channel was always more saturated in the non-80A images.

Well, maybe that "saturation of the red channel" doesn't describe the effect well enough. Reds and yellows get lighter, cyans shift towards blue and so on. Color shifts are subtle but complicated (I made an experiment once with a ColorChecker and four different light sources).

But, I am sure that this can be solved in post too with the right profile.  

Indeed. But people tend to skip this part.