Daytime ND filter + longer exposure: Better Color?

[brad_trostad]

<p>When I see all the great pictures of rivers and waterfalls taken during the daytime with strong ND filters and longer exposures it "seems" as if the colors are just more punchy / saturated.</p>

<p>So this has me wondering something in general. If a scene is relatively static and a person has the time, tripod, filters, can fix aperture, etc will a longer exposure with a strong (and high quality) ND filter generally tend to capture more color than its equivalent faster exposure with no filter? </p>

[szrimaging]

<p>Part of what you are seeing may also be the photographer's skills at image manipulation in post. Part may also depend on filter. I use a 7-stop ND+IR from Tiffen, and aside from a bit of a tint, it also seems to give me a more neutral image (akin to lower contrast). However, in post I can usually get a much better/punchier image through curves and levels. That said, there may also be more color information transmitted as compared to brightness, so it could be that there really is more saturation.</p>

<p>That said, I have no definitive proof outside my own experience, others may be able to say they do indeed get more saturated images.</p>

[bob_sunley]

<p>No, but a well positioned pol filter can help a lot. Choice of film, paper and darkroom technique can make huge differences as well. And then came photoshop. :)</p>

<p> </p>

[jim_momary]

<p>I'm getting confused. Please jump in and clarify this thread and my senslessness.<br />Let's say we have a scene with two walls, one lit at a level of ev18 and one at ev12 (to make it easy).<br />The ratio is 18:12 or 3:2, but the difference from one wall to the other is <strong>6 stops</strong>.<br />Let's mount an ND filter that reduces everything by 4 stops.<br />So, the walls become ev14 and ev8.<br />The ratio is 14:8 or 7:4, but the difference is still <strong>6 stops</strong>.<br />Is it not true that the <strong>6 stop</strong> difference is what is important here, so long as the image isn't blown out or total murky dark?<br />The ratio 3:2 or 7:4 is just a matheamtical 'confuser'. (To my tired brain I think Zach is implying that the contrast ratio is lowered and I think that's not by itself the main important factor, right?)</p>

<p>Old Jim in the Back</p>

[tom_mann1]

<p>Hi Jim - The EV scale and the scale of "stops" are both logarithmic, so differences of numbers given on that scale correspond to ratios of absolute light levels. Taking ratios of EV numbers or stops is mathematically defined but has no relevant physical meaning. </p>

<p>To make the numbers a bit easier to work with, I'm going to use your example, except I'm going to reduce the light levels in your hypothetical scene to ev=7 and ev=1. Working in these logarithmic units, the difference is 6 ev. If you insert a 2 stop ND, the values become ev=5 and ev=-1, which also has a difference of 6 ev.</p>

<p>One can do the exact same calculations in absolute, non-logarithmic units. Let's use units of area and time such that ev=0 corresponds to 1 photon per unit area per some unit of time. Then:<br>

ev=-1 is 0.5 photons/area/time,<br>

ev=1 is 2 photons/area/time,<br>

ev=2 is 4 photons/area/time,<br>

ev=3 is 8 photons/area/time,<br>

ev=4 is 16 photons/area/time,<br>

ev=5 is 32 photons/area/time,<br>

ev=6 is 64 photons/area/time<br>

ev=7 is 128 photons/area/time.</p>

<p>So, the lighting ratio of an ev=7 area to an ev=1 area is 128/2, or 64 in absolute units. In stops, this is 6 stops.</p>

<p>If you put a 2 stop ND in front of the lens, the above becomes ev=5 and ev=-1, or 32/0.5, or once again, a ratio of 64 in absolute units or 6 stops. Everything is completely consistent as long as one doesn't try to take ratios of ev units.</p>

<p>HTH,</p>

<p>Tom M<br>

(in absolute units)</p>

[stephen_hipperson]

<p>Do you know, I've never used EV, only stops. Reading the above I'm happy with stops and I'll continue to leave EV alone.</p>

[ariel_s1]

<p>Short answer: No, that doesn't lead to better color. As long as the overall exposure is the same, leading to the same histogram, the colors will be the same.<br>

However, people that are taking the time to use an ND filter are also likely using a tripod, exposing correctly, and leaving their ISO low, which lead to better photos. They're also taking the time to post-process more seriously.</p>

[david_henderson]

<p>Going back to the OP's question for the moment, my experience with 10 stop and 3 stop ND filters is that the bigger the filter factor, the more chance there is of a colour cast. The presence of a colour cast requires to use of post processing tools to remove it. I have not seen any tendency towards greater saturation or higher contrast - and if anything I'd say the reverse, but the fact that you can easily arrive at the degree od contrast and saturation you want makes the "native" effect of the filter itself close to irrelevent.</p>

[jim_momary]

<p>Thanks Tom.<br>

So, ND filters can give me milky water or make moving people disappear from a stationary landscape but provide no color or "DR" advantage. Makes my thoughts more clear.<br>

"Perceived" image quality improvements may be due to the tripod use over hand-holding alone as Ariel states.<br>

Thanks guys.<br>

Old Addled Jim</p>

[brad_trostad]

<p>As I understand it then, all other things being equal (scene, lighting conditions, ISO, aperture, focus point, metering mode, etc) - a RAW shot done without a ND filter should look identical the same shot taken with a ND filter its corresponding (longer) exposure time - with perhaps one exception - color cast.</p>

<p>What lead me to ask the question was examples like this one (which is clearly indicated as a low-light example):</p>

<p><a href="http://www.outdoorexposurephoto.com/photoblog/digital-photography-tips/89/">http://www.outdoorexposurephoto.com/photoblog/digital-photography-tips/89/</a></p>

<p>The text explaining more time to collect the more subtle colors got me to wondering. One thing that makes sense (to me) is that the longer the shutter is open, the more photons the sensor might collect. However, the ND (excluding color cast) must then prevent a certain % of them from arriving right? </p>

<p>I suspected that post processing plays a big role but it sounds like it is probably the only real step in the process that is producing the more "saturated" look.</p>

<p>I also suspect that photographers doing these long exposures (eg. bright daytime waterfalls) are also stopping down the lens significantly to either: render more of the scene in focus or to simply cut down on light if they didn't bring enough ND. Perhaps the generous depth of field in these shots makes the foreground greens and background blues seem even more punchy.</p>

<p>Perhaps another factor is if the long exposure tends to bump the histogram to the right (hope I got that right) then in photoshop, simply running a levels adjustment might really give the scene more contrast and also appear more saturated?<br>

Thanks for your responses so far - this forum is awesome!</p>

[szrimaging]

<blockquote>

<p>The ratio 3:2 or 7:4 is just a matheamtical 'confuser'. (To my tired brain I think Zach is implying that the contrast ratio is lowered and I think that's not by itself the main important factor, right?)</p>

</blockquote>

<p>I am not really saying that it is. But that it seems to be a by product of the setup. Taking two photos (same aperture) with my filter (Tiffen 7-stop ND + IR) always seems to be a much flatter image with less contrast. However, I am not sure that the clipping points are actually altered. I would probably need to setup on a really contrasty day and shoot both with and without the filter and compare the levels. But it could also be a cast/color balance issue that is introduced with the filter and it is deceiving me.</p>

[Jeff Lear]

<blockquote>

<p>There can be a lot of color and light bouncing off clouds and hills that our eyes can’t pick up. Cameras are able to “see” into low light scenes by leaving the shutter open and collecting more light.</p>

 

</blockquote>

<p>This is simply false in the context of this discussion and in the context of his blog post. In essence, what the author is claiming is that by artificially darkening a scene and forcing your camera into longer exposures to compensate, the sensor will capture light that it otherwise wouldn't. Not true. His claim that the only difference between his sample photos is the length of exposure is also ludicrous. A good neutral density filter will be <em>neutral.</em></p>

 

[tom_mann1]

<p>For digital sensors, the only phenomena that I can think of that would cause a long exposure to deviate from the reciprocity rule in either tonality or color are:</p>

<p>a) Dark noise in the sensor will have a larger fractional contribution to the long exposure; and,</p>

<p>b) Small amounts of light could possibly leak onto the sensor if your eye is not on the viewfinder, your large ND filter holder is not perfectly leak-tight (eg, with Cokin square holders), etc.</p>

<p>Both of these phenomena will cause a loss of contrast and saturation, and a possible color cast, but certainly not enhanced saturation. This, aka, lack of reciprocity failure in digital sensors, has been discussed on photo.net before:<br>

http://www.photo.net/nikon-camera-forum/00IdiZ<br>

http://www.photo.net/canon-eos-digital-camera-forum/00U7sH<br>

http://www.photo.net/digital-darkroom-forum/00UNbl</p>

<p>If such a phenomena actually increased the saturation, you can bet that the astro photography community would be all over the (hypothetical) phenomena.</p>

<p>If you want to easily test the hypothesis yourself, at least over a 7 stop range, just dig out your old 50mm f/1.4 lens, and take two exposures of something fairly dim, colorful, and flat: one exposure at f/16 and say 30 seconds, and the second at f/1.4 and one quarter second. I just did this, and other than the expected vignetting and increase in spherical and other aberrations when wide open, the color was, exactly as expected, absolutely identical. To be honest, it isn't even worth my time to post the essentially identical images. If you want to test the hypothesis over a wider range of exposure durations, have at it with very dark ND filters, but be aware of the problems with them mentioned above.</p>

<p>Tom M</p>

 

[bob_sunley]

<p>Brad, look closely at the pic in that link by clicking on it, especially at the clouds above the mountain on the left side. Some seem to be missing in the more intense image. There is an orange ball on top of the mountain in the right image as well. I'm guessing the movement of the unseen clouds to the left changed the lighting conditions from the first to the second image and had a major effect on what you see.</p>

 

[brad_trostad]

<p>I knew the pictures in the post I linked were different and since they were low light I guess I wasn't as much interested in that scenario as I was with all the really rich blurry river shots with super rich greens or blurry ocean water shots with really punchy reds and blues in the skies. It would seem its really just great technique and some extra work in post to get them there.</p>

<p>Maybe another way to say it would be if there were a way to set a camera to mega-low ISO (say 0.05) most photographers would probably use that vs ND filters to blur motion since adding any glass before the lens only seems to complicate things (darker viewfinders, maybe some loss of sharpness, color shifts, etc).</p>

<p>On my wishlist for a future DSLR is a few built in ND filters (assuming they are of high quality) sort of similar to the new C300. Then the viewfinder would be bright and no need to have multiple sizes or step rings or fumbling around.</p>

[dan_south]

For all practical purposes a longer exposure does not change colors over a shorter equivalent exposure. E.g. 1 second at

f/22 will not yield more saturated color than 1/30th at f/4. The image might look different in other ways. More depth of

field might be apparent in the f/22 image, and possibly more motion blur. But the colors should be equivalent.

 

That said, digital sensor don't always have perfect reciprocity, I.e. even though these two exposures are supposed to be

the same, one might end up being a little bit darker, and darker exposures generally look more saturated. And as David

mentioned, filters are not always perfectly uncolored. I stopped using one popular brand because their ND filters added

orange casts to skies. It looked interesting at first, but it became annoying quickly. The filters used might have applied

some color cast.

 

But the the colors that you see in any image are a function of the subject, the quality of light, the exposure value, and

modifications added either at capture time or in post processing such as contrast, saturation, and white balance.

Probably the filter used in these images is a red herring. Other factors created the colors that you saw.

[david_henderson]

<p>On reflection I can see circumstances in which a long exposure could cause the colours to seemingly blend together as the OP's link suggests. But this is pretty much limited to skies and water which can move during a long exposure and what you see in the colours is essentially the by-product of unsharpness. The colours are not more intense, but seem to blend together more smoothly and gently with fewer sharp delineations. It can only happen when there's subject movement though, and its achievement carries a price. Its the sky equivalent of using a long exposure to smooth out water.</p>

[steven_f1]

<p>There can be a lot of color and light bouncing off clouds and hills that our eyes can’t pick up. Cameras are able to “see” into low light scenes by leaving the shutter open and collecting more light.</p>

<p>That statment is false as stated earlier but it may seem true, not because of the camera but because of your eyes. In low light levels the human eye has limited color vision. In very dark conditions you are is simply seeing in black and white only. </p>

<p>this would be obvious if you use a telescope to look at a nebula at night. I recently looked at the orion nebula with a 10" telescope and I didn't see any colors. The view through a 20" telescope was no different. However if you took a picture of it with a camera you would see color in the image. This is because the cameras ability to see color is constant no mater what the light level is. The human eyes ability to see color is not constant and will very with light levels from black and white to full color.</p>

[jian_huang2]

This is a question that have been puzzling me for quite a while too. I am still a little skeptical of the claim that longer exposure would produce "better" color, assuming the overall dialed exposure is correct and unchanged when using ND to achieve longer exposure. However, I have heard quite a few people mentioning this trick and I have a friend who swears by it. He tries to put on a 10 stop ND and dial up the exposure time by 1000x whenever he can, insisting that it produce better color. His pictures(mostly landscapes) do look great though.

 

 

Here is what I can understand so far. From a digital signal processing point of view, longer exposure should not affect the color. It is essentially the same exposure. An 1 stop ND will double the exposure time. It also halves the light through put. The combined effect is that the same amount of light reaches the sensor during the exposure.

 

 

Thus I only have a few hypotheses about what seems to be a visual effect.

 

 

Hypothesis 1, "desirable" color shift, either due to the ND or due to sensor during long exposure.

 

Hypothesis 2, somehow the noise got reduced, causing low light area to maintain a good signal to noise ratio, which in turn cause the color to look a little "purer" or more saturated. Although I don't really understand why the noise would be lower. Maybe the camera is doing some kind of trick in averaging out the noise during the long exposure?

 

Hypothesis 3, reciprocal failure, if there is such a thing for digital sensors. Maybe simply scaling up the exposure time actually reduces the exposure due to reciprocal failure. Essentially, that reduces the exposure. A lot of cameras' grey histogram display is actually just the green channel. The "correct" exposure might be actually clipping the red channel. Thus an under exposed picture would better maintain all the channels, making the colors look better. The picture I've seen with "improved" color are usually dawn/dusk shots, when there is a strong emphasis on red. So this is a possibility.

 

 

These are just some hypotheses I can think of while trying to explain it. I think finding out why would produce an equivalent digital method that could save a lot of time. I'd really like the seasoned photographers here to shine some light on this to see if this phenomenon really exists, and if so, what is the mechanism behind it.

[bob_sunley]

<p>There is also the wall time factor, at dawn and dusk the light is constantly changing, so if you take a sunset photo like the one linked to above, with the short exposure first and then add filters and take the second, the lighting has changed, usually for the better if you are looking for redder oranges.<br>

If you want A/B comparison photos, you would need two cameras, and then there is the dilemma of when to shoot the shorter exposure, at the beginning or end of the longer one or maybe in the middle.</p>

<p> </p>

[tom_mann1]

<p>Jian - see my post of Jan 12th (above) in this thread where I discuss the possibility of reciprocity failure in digital sensors, give a few citations on the subject, etc. I don't know of any credible source that thinks this can happen with digital sensors. I think we can safely exclude that possibility.</p>

<p>Take a scientific approach: Temporarily putting aside what your friend says (because he isn't here to answer questions), have you tried any such experiments yourself? If so, what do you see?</p>

<p>Tom M</p>

[jian_huang2]

Hi, Tom,

 

From the trials I have done, I'd say it is inconclusive, which is why I am still skeptical.

 

I don't have the 10 stop ND my friend has. So my trials have been using 1~2 stop ND's and multiple exposure overlay. The results look different. It is quite subjective. Thus, I cannot say definitively, in terms of color, one is better then the other. As Bob pointed out earlier, it could have been just the effect of longer time duration. Or maybe the improvement is not significant enough for me to tell.

 

I guess to really be scientific, I need to get one of those 10 stop ND's and try it under a controlled light environment. That I have not done.

 

-jian