Long Exposures for Greater Saturation and Luminosity

Note: I originally posted this in the PNWNP forum and it generated some great discussion. I thought it would be helpful to post here for more NPN readers to access.

In situations involving color shifted light (blue shadow light, warm sunset light, etc.) it can often be beneficial to obtain a longer exposure to allow the sensor/film more time to absorb the subtle ambient color and glow (at least I think that is what is happening). There can be a lot of color and light bouncing off clouds and hills that our eyes can't pick up. If I really want to pull as much color from a scene as possible I'll often use an ND filter along with my usual filter stack (GND and polarizer if appropriate)and also set the ISO on my camera down to 50 in order to extend what was already going to be a long shutter speed. The two photos above were taken one right after the other, the first at 6 sec @ f/20, ISO 100 with a 3 stop GND. The second was taken at 30 sec @ f/20, ISO 50 with a 3 stop GND AND a 3 stop ND. The same color temperature was selected for each image in the RAW processor. Quite a difference 24 seconds of shutter time made in this situation.

There has been much speculation about why this occurs. It has been suggested that the ND filter is introducing a color cast. Possible, but Singh-Ray ND filters are supposed to be about as close to color neutral as you can get. Also, the increased saturation can occur with both the red and blue ends of the color spectrum depending on the color temperature of the ambient light, not something you would expect if the filter was causing the effect.

NPN Contributing Editor, Darwin Wiggett had the following to say in an article he published on the Singh-Ray blog: "the 5-stop ND filter creates long exposure times which has two benefits; first it shows painterly blur in any object moving in the frame (swirling tides, running water, blowing clouds) and second it builds up colour intensity especially at sunrise and sunset by allowing colour to wash over the land and sky over time." The article is called the Terrific Triple Threat and you can read it in entirety HERE

I'm interested to learn more about how and why this works, but in the mean time knowing that it does work allows me to utilize it as a creative option in my photography.

It's an interesting question Sean, though I'm not sure I'm totally sold on what you posted. Why is there such a significant contrast and luminosity difference between the two images (note the foreground rocks for example)? Additionally, I see a color balance difference; the first being more cyan/green, the second more red/magenta. I question if there's more going on than simply the exposure time difference.

Interesting. I posed a question about whether long exposure allowed increased color saturation/intensity with digital files in the same way it seemed to with certain films and was told right here on these forums that it did not do so with digital, that such thinking was a fallacy.

I've experimented on a flower in my home studio with a wide range of exposure times (from 1/20 sec to 20 seconds on the same image with the same angle, warm lighting, and other settings) and could see no difference in color saturation/intensity at long exposures. I was not using any filters. Nor have I seen my own unfiltered sunrise/sunset photos more saturated with longer exposures.

I would certainly like to find out the truth here. While I don't really care myself whether it's so (simple matter to handle color saturation/intensity in post processing these days), I'm occasionally asked the question by clients.

Anyone? Class? Class? Anyone? Bueller? Bueller?

I enjoy using longer exposures. They often reveal light that our eyes don't see - the dark side of a crescent moon, for example. I haven't really noticed a color shift due to longer exposures, but I do know that predawn light can change color seemingly gradually and subtly to the eye but the sensor captures a more dramatic change over just moments.

Maybe what you captured could be changing light that wasn't as apparent to your eye as it was to your camera. Just a suggestion, not an argument.

Terrific picture, BTW

Just to clarify the terminology, factoring the filters used in the above example, the image on the right is actually under-exposed by 2 stops compared to the one on the left. It's actually the under-exposure that's responsible for the deeper color. More exposure results in brighter tones that may appear washed out (as seen on the left).

In the past, it was common practice to under-expose film for added saturation. With digital tools it is actually better to expose "to the right" (i.e. brighter tones), then adjust saturation in an imaging program for maximum detail.

If I'm reading Darwin's description correctly, he's actually talking about leaving the shutter open longer (which does not necessarily mean more exposure) to take advantage of the movement of the sun or clouds for a "painterly" effect.

Guy

I think Guy is correct, the saturation comes from slight under exposure. I'll toss in something else here, just for thought. For the past year I've been taking a LOT of photos at night using an extensive lighting set up. I've noticed that with that I am getting some incredibly saturated colors. So far I have been attributing it to a near total lack of glare (from sunlight bouncing around) and also perhaps the strobe light is SO directional it acts as if it was polarized? Whatever it is, it is noticeable.


Kent in SD

Joel - Good eye and thank you for bringing the luminosity and contrast discrepancies to my attention. When I made the two image comparison at the top of the thread I wasn't careful enough in my selection of image files. The first file was not post processed at all, while I had done some basic color, contrast and luminosity work on the second. To make a fair and scientific comparison I should have controlled all the variables more carefully. I guess that since the effect I'm talking about is equally apparent in both my processed and non-processed files I wasn't as careful as I should have been. Allow me to remove all additional variables and consider only identically processed image files.

The four images above were all converted from RAW in Lightroom with WB temp 5000 K and Tint 0, with identical default settings for exposure, recovery, fill light, blacks, brightness, contrast, vibrance and saturation as well as every other available adjustment. Absolutely zero post processing was done on any of the images after RAW conversion other than sizing for the web. The camera settings and filters for the 4 images are as follows:

1 - 2.5 sec @ f/18 ISO 100, 3 stop GND soft only, taken at 6:38 AM
2 - 30 sec @ f/22 ISO 50, 3 stop ND only, taken at 6:39 AM
3 - 30 sec @ f/20 ISO 50, 3 stop ND and 3 stop GND soft, taken at 6:43 AM
4 - 1.3 sec @ f/18 ISO 100, 3 stop GND soft, taken at 6:45 AM

Note that even with all other variables removed, the two 30 second exposures taken with the 3 stop ND still show that increased saturation and luminosity (particularly in the clouds and illuminated ridges) and that the the red/magenta shift is also still there. Further more, it is interesting to me that it appears the color in the scene goes from moderate saturation/magenta in #1 to maximum saturation/magenta in #3 and then to the least saturation/magenta in #4. However, as the scene unfolded to my eye, #1 had the greatest amount visible color at 6:38 AM and the color intensity consistently decreased until it looked like what you see in #4 at 6:45 AM. Yet somehow, setting the ISO to 50, adding the 3 stop ND in #2 and finally the ND with the GND in #3 caused the camera to record increasing saturation and luminosity during that same period of time.

So, I have ruled out post processing and digital trickery as a cause, and I have also ruled out increasing saturation of the visible natural light over time. Guy suggests that the increased saturation is due to underexposing. There are some exposure differences in the foregrounds depending on if I used the GND or not, but looking at the mountains in the middle of the four images I see pretty minor exposure differences. Could such a small amount of underexposure cause such radical changes in saturation and color balance? Perhaps? The two other variables are exposure time and the 3 stop ND filter. Singh-Ray claims that their filters are very color neutral. I'd like to hope that they aren't off this much. Additionally, if the 3 stop ND does cause a serious color boost/shift, then why wouldn't the 3 stop GND also do the same in the top half of images #1 and #4? The filter may certainly be the culprit, but if it is the effect should be repeatable in a variety of situations. I'll have to run some more experiments on this. The final variable is the amount of time the shutter is left open (unless there is a color shift between using ISO 50 and ISO 100 which I highly doubt). I don't have any knowledge of why exposure time would cause this effect, but I also don't have any good knowledge of why it wouldn't. Anecdotally, I feel like I often get increased saturation in longer exposures with and without the use of ND filters and not always in the red/magenta color range. THIS long exposure image posted in ESS by David Fernandez appears to show the saturated blue/cyan effect, without knowing his specific post processing techniques. Darwin Wiggett's article that I reference above also shows similar saturation gains, or "built up color intensity" as he puts it.

So, I'm still where I was before...I don't know precisely what causes this to happen, but knowing that it does happen allows me to use it to my creative advantage. I'm hoping that someone can shed further light. I'd love to hear other ideas.

Interesting question. Isn't this basically a histogram (light or luminosity) question?

I shoot a lot of sunsets. Nearly, if not identical, effect is obtained by various underexposures. The problem of course, this generates noise.

Consequently it is almost imperative to shoot to the right then manipulate (brightness has a great effect as you know) . Has this concept of underexposure (- 0.5 to -3.0 ev, or faster shutter, or smaller aperature?) been tried on your example?

Can you provide the different histograms?

Larry

What Darwin is saying in the article you linked, Sean is that the 5 stops of ND filtration allow for the painterly movement that Guy mentioned but also, particularly at the time of sunrise, in a long exposure the sun will be rising in the sky during the time the shutter is open. This has the effect of changing the colours in the shot over that time. Darwin states this explicitly in his explanation of these 4 photos, http://bp3.blogger.com/_Yubt5qpUPy4/R83DrSin4OI/AAAAAAAAAQI/jXt0xUFgaSQ/s1600-h/Photo3-wide.jpg. It's not a matter that the filter causes increased saturation. Note too, that in the article you linked Darwin also used a polarizer which has a big impact on colour.

With some slide film, slight underexposure would rev up the colours. Velveeta was one film where this was particularly true. Velveeta has an unique response curve unlike any other slide film. But it's the non-linear response curve of film that gives that result. Digital sensors have a linear (or very near linear) response curve and digital sensors record no colour, only light. The colour is effectively added later so the amount of exposure affects colour only to the extent that less light means less colour.

As far as ISO 50, I don't know what camera you're using but I recall seeing a test of the Canon 5D at ISO 50 and there was a difference in the colour balance between ISO 50 and ISO 100 with 50 being warmer. One particular shot I remember was of a room with a candle, and of course I can't find it now. ISO 50 also had/has on the 5D a slightly reduced dynamic range. I really only use it on mine as a last resort when I can't get the shutter speed I want at ISO 100 with the filters I have

One other thought. White balance (kelvin) will shift rapidly during sunset. Adjusting to the same kelvin may not be a fair comparison. Try automatic on camera setting for white adjustment...??? Larry

As I mentioned in the PNWNP thread, I think this is primarily a function of color "movement" through the scene as opposed to a "build-up" of color density or saturation. I think Darwin's choice of words in his explanation ("built up color intensity") is a little misleading, though he later goes on to better describe the effect as one of color that "moves across" the scene.

Another factor in the images above, possibly, is the natural progression of sunrise. It typically starts faint, builds in intensity, and rapidly loses warmth and saturation. I'd expect the earliest and latest images in any series to be the least saturated. It may just be coincidence that you used the ND during the most saturated period of sunrise.

Finally, in the images above, the exposure differences are profound - about 2 stops as Guy indicated - and it extends beyond the foreground into the mountains and sky. The upper part of the sky in #2 is substantially darker than #1, even though you removed the 3-stop GND in that shot. #3 is even more underexposed in the upper portion with the reintroduction of the GND, and only a 1/3 stop increase in aperture to offset.

I think the change from 1 --> 2 can be largely explained by 1) the natural building of saturation as sunrise neared peak intensity, 2) the 2-stop underexposure and 3) enhanced further by the "movement" of color enabled by the ND and the resulting long exposure.

I think the change from 2 --> 3 is strictly the addition of the GND - the top is significantly darker than #2, while the lower third is a little brighter by virtue of the slightly larger aperture (while holding exposure time constant).

I think the change from 3 --> 4 is due to alpenglow leaving the clouds; you still have some on the mountain, but clearly the light quality has changed.

By the way, very nice image. I think the most striking benefit of the ND is the sense of motion it creates in the image. With or without increased saturation/color, the motion is a valuable creative tool.

To expand on what Rick and Guy said with regard to exposure differences, here are some basic calculations. (The softness of the GND means there isn't a quick transition from top to bottom, but I think the numbers are accurate).

Bottom Exposure Difference
1 -- Base Exposure Bottom
2 -- +3.66 shutter, -0.66 aperture, -1 ISO, -3 filter; TOTAL = -1 stops
3 -- +3.66 shutter, -0.33 aperture, -1 ISO, -3 filter; TOTAL = -0.66 stops
4 -- -1 shutter, 0 aperture, 0 ISO, 0 filter; TOTAL = -1 stops

Top Exposure Difference
1 -- Base Exposure Top
2 -- +3.66 shutter, -0.66 aperture, -1 ISO, 0 filter; TOTAL = +2 stops
3 -- +3.66 shutter, -0.33 aperture, -1 ISO, -6 filter; TOTAL = -3.66 stops
4 -- -1 shutter, 0 aperture, 0 ISO, -3 filter; TOTAL = -4 stops

Summary :
1 -> 2 : Top -1, Bottom +2
1 -> 3 : Top -0.66, Bottom -3.66
1 -> 4 : Top -1, Bottom -4

As you can see, the exposure differences are really pretty large, and mostly underexposed. This could certainly create some oddities, and I wonder if there could be something going on in RAW processing, to get them to look similar (even if it's the way the tool handles things under the covers or with defaults or auto correction).

The reason saturation doesn't increase with exposure, as Steve mentioned, is that each pixel on the sensor only records luminosity information at particular wavelengths. And it records it linearly, so double the amount of light that hits and you get double the luminosity information. This data is demosaiced and converted to create the color information we're used to. However, during the development of the RAW file, curves are applied to make the data non-linear at a couple of different steps at least to more closely match how your eye sees. I don't know all the steps, but I've seen files developed without the curves applied, and they look very dark and flat. So these curves could at least be partly responsible. Theoretically, if you do no exposure correction, it should be the same curves from image to image, but I'm not certain of that.

In any case, it could be the developing that's causing the saturation change, it could be changing conditions, if could be some effect of the grads, or it could be all of these. Given the varying exposures and techniques and conditions here, though, I wouldn't read to much into it. If you really want to test it, try something like a color checker indoors in controlled lighting.

Finally - I think #3 is the best one. Really nice FG/BG setup, and I like the bit of motion in the clouds compared to the others.

Some great ideas and serious number crunching coming in here folks. Fun!

Based on the collective wisdom coming through it is sounding to me like underexposure may be accounting for at least some of the effect. However David, I'm not sure that simply comparing the exposures with the first image as the base point of reference is accurate because the sun was rising and the scene was becoming progressively and significantly brighter over the seven minutes between the first and last photo. So, if my thinking is correct, the amount 3 would have been underexposed at 6:38 when 1 was taken is not equal to how much it actually was underexposed five minutes later at 6:43 when the scene was much brighter. Let me know if I'm wrong, but without some light meter values it wouldn't be possible to know the degree of underexposure of 3 compared to 1.

Additionally, the two images that show the greatest red/magenta saturation/shift are 2 and 3. These two images are unique in that they both utilized the 3 stop ND filter and ISO 50, so it seems like one or both of these factors may come into play as well. I AM shooting with the 5D, Robert. I hadn't read about the ISO 50 color shift you are referring to, but it would seem to fit.

Rick, I think your assessment of the course of events is astute and logical, but not accurate with what I actually experienced. The height of saturation to my eye occurred just before image 1. The color was starting to fade so, based on prior experience, I added the ND and changed the ISO to see what effect a longer exposure would have. I immediately noticed on the camera LCD that the image showed more saturated reds, even though the color of light I could see was becoming more white. I added the GND, shot again and immediately noticed even more saturation of red in image 3 despite the visible reduction in color I was experiencing. Finally, I removed the ND, changed the ISO and aperture and shot image 4 just 90 seconds after image 3. To my eye the light changed very little in those 90 seconds, but you can see that this was when the biggest change in what the camera captured occurred. The color moving across the scene over time idea also makes sense, especially in the clouds which were able to move quite a bit in 30 seconds compared to 2 seconds. However, the sun did not move significantly across the bright mountain face between images 1 and 2 or between 3 and 4. As you can see, the face is brightly and evenly lit in both images 3 and 4, but image 3 is considerably more saturated. I'm beginning to think a combination of underexposure and filter/ISO induced color shift is largely responsible here.

So is it possible that in certain situations a perfect storm of compounding effects converge? In this case did the movement of light and clouds in a longer exposure, underexposure, filter cast and ISO color shift all work in conjunction?

Sean - I didn't mean for my calculations to fully explain things, because I don't think they do - just sort of brain storming possibilities. I think you're right that light meter reading comparisons would have been necessary. Mainly what I was trying to get at is that the situation you have here causes exposure to bounce all over the place (including the changing conditions), which might make fully explaining this a bit of a fool's errand.

The height of saturation to my eye occurred just before image 1

I don't know what the "real" light situation was like Sean, but I wouldn't trust my eyes is the light levels were pretty low (i.e. lower than what's shown in the pictures - I don't how documentary they are). And I wouldn't trust the camera to capture exactly what I saw in cases like this either, simply because it may capture saturation your eyes can't see depending on ambient light levels.

Sean,

Timewise it looks as if shot 1 was taken just when color was starting to soak the clouds. In 2 it seems you caught the peak in sunrise/sunset color. #3 has the look of it being captured just before color faded and 4 was taken when the color was obviously gone.

It looks as if the sun is doing this and not the ND filter.

A 30 sec exposure would have it's advantages over a 2.5 or 1.3 second exposure when light is bouncing into different portions of the scene but would also have it's disadvantages if color in the clouds lasted less than 30 seconds or capturing when color is beginning or fading.

Besides all of that beautiful photograph Sean.

There is a very simple test to find out whether the Singh-Ray GND filters create a color cast. I used the following steps and was convinced that they indeed do (you'll have to do this really fast to make sure that time is not an issue).

1) Frame a scene on tripod. Determine appropriate exposure settings, let's assume they are F11 at 1s. Make a photograph with these settings.
2) Get a Singh-Ray 3-stop GND (the graduated one), and insert it all the way down so that the graduated area completely covers the entire scene. Overexpose the scene by 3 additional stops, keeping the aperture at F11.
3) Get a Singh-Ray 3-stop ND (simple ND filter), and mount it on the lens. Again, overexpose the scene by 3 additional stops, keeping the aperture constant at F11.
4) Now take all filters off, and with just the lens and camera, make another photograph at F22 and overexpose the scene by the additional 2-stops required, i.e. at 4s.

In theory, if the length of exposure causes additional saturation and intensifies color, then the shot #4 would appear much more intense than shot #1. If the Singh-Ray GND doesn't introduce any color casts, then all shots should appear identical.

However, after repeated testing, I always find that shot #1 and #4 will be almost completely identical (after accounting for the motion blur and additional brightness caused if you are shooting water bodies). And shot #2 and #3 will be extremely different.

Leaving me to conclude that the Singh-Ray ND filters are really neutral, but the GND aren't.

Any kind of a "test" cannot be done outdoors. If the filters are going to be tested for neutrality, it has to be done under controlled conditions with consistent light and an unchanging subject. All variables except the filters have to be removed from the equation.

Robert, I understand that it is not a scientific test. But it's a result that cannot be ignored easily. You can try that by just looking through the shaded portion with your naked eye. If you try to look at a scene through the fully-shaded portion of a 3G-HS grad ND filter or a plain 3-stop ND filter, the grad filter shows obvious magenta cast.