Deliberate Underexposure for Color Saturation

[nicshow]

Lately I have had good luck with underexposing my 1/3 or 2/3 of a stop. I find that the color saturation is better and there is less clipping. Do I sacrifice anything in this process? Does underexposing add color noise?

[Mark_Cohran]

Underexposure will cause noise in the shadow area, but you won't see it much unless you adjust to bring out detail in the shadow. I've have good success with removing such noise with Noise Ninja.

Underexposing for saturation is common when shooting slides, but I've never really heard of doing this for digital (it's a totally different mechanism for recording color) and I've never really noticed more saturation from underexposure.

Mark

[evandavies]

If you are shooting RAW internal camera parameters are ignored but if you are using JPG then I use these in camera parameters:

saturation to +1 or sometimes +2
Its easier to reduce colour in post than add it.

contrast to -2 (lowest).
This way you keep as much range as possible for altering levels.

What Mark says it true. Why alter the exposure of a shot (which is permanent) when its easy to pp levels and get the same result with more control.
Thats the main advantage of digital, whatever you do can be undone and tried again or you can save different versions....

[Wilt]

Underexposure squeezes the shadow detail into fewer digital 'bins' so that you lose some of the dyanamic range within the shadow detail...that is why the recommendation is 'expose to the right (of the histogram)' so that the shadow details are captured into more digital bins.

Underexposure to increase saturation applies to SLIDE FILM; on the other hand underexposing color neg film simply results in muddy color! If you want color saturation in your photo, expose normally and use postprocessing software!

[JMHPhotography]

I agree with Wilt. I get far better results in terms of color saturation when exposing the photo properly, shooting in RAW, and using the advanced saturation controls in ACR. Using the proper white balance helps too. Even if it's just using a reference card to achieve proper white balance in other frames.

[AJSJones]

I used to use that maxim with slides. With digital, when shooting raw, as you should if you want the bestest possible PP options, it is possible for one color channel to be clipped but not see blinkies. I don't know if it's possible for no blinkies and to actually have a blown pixel or two, given the raw highlight recovery capabilities of the lates converters. However, it is wise to turn the contrast all the way down on the in-camera jpeg settings so that the histogram is least misleading - the histogram comes from the processed jpeg and if the in camera settings are applying contrast adjustment, the resulting histogram may look OK but the raw data may not be.

[PacAce]

I used to use that maxim with slides. With digital, when shooting raw, as you should if you want the bestest possible PP options, it is possible for one color channel to be clipped but not see blinkies. I don't know if it's possible for no blinkies and to actually have a blown pixel or two, given the raw highlight recovery capabilities of the lates converters. However, it is wise to turn the contrast all the way down on the in-camera jpeg settings so that the histogram is least misleading - the histogram comes from the processed jpeg and [b]if the in camera settings are applying contrast adjustment, the resulting histogram may look OK but the raw data may not be.
I'm confused by the above statement. If the histogram looks OK with constrast applied to the image, how can the raw data NOT be OK? Can you clarify that for me, please. :)

[AJSJones]

Because one of the channels may be blown but not enough to cause blinkies - the histogram (unless it's an RGB triple histogram) is based on the luminance after in-camera processing. Had a picture a while ago with bright yellow in in a bird's feather - no blinkies, but I had a really hard time getting it right in PP, R was completely blown and G was very close. In any case, my point was to help people understand that they need to understand the histogram and where it comes from if they rely on it completely, and it doesn't come (directly) from the raw.

[Titus213]

I'm confused by the above statement. If the histogram looks OK with constrast applied to the image, how can the raw data NOT be OK? Can you clarify that for me, please. :)

I'm with you. While the histogram may show an issue the RAW data is still not affected. You may look at the histogram and re-shoot because of it but that's about all.

I've even read (and I can't find it again) that it is easier to extract detail from slightly over exposed than from under exposed. I guess it depends on what you are looking for.

[grego]

Slightly underexposing, slightly overexposing. Generally you'll do pretty well, as long as its slight. Especially if its show in RAW. with Photoshop, kicking up saturday is a breeze!

[nation]

Like the others said you will increase the likelihood of loosing the shadow detail applying -ve EC. The blinkeys often make us think that underexposure is a better option but it's really dependant on whether the highlight or shadow details are more important. A great example where 1/3 or 2/3 -ve EC works well is here in HK is taking pics on bright sunny day in the business district. With the glass buildings and sky in the composition -ve EC works well to reduce blown highlights provide more detailed saturation.

[Lightstream]

When photographing sunsets, or at dusk, I find that -2/3 exposure compensation brings out the best in the image. Colors are indeed richer. I do lose a bit of detail.. but it just adds to the image, rather than taking away from it. Can't say I'm unhappy. :)

[Wilt]

When photographing sunsets, or at dusk, I find that -2/3 exposure compensation brings out the best in the image. Colors are indeed richer. I do lose a bit of detail.. but it just adds to the image, rather than taking away from it. Can't say I'm unhappy. :)

Sunsets are a different beast than underexposing a sign in order to make its color 'deeper' in saturation!

You could meter (with a spotmeter, for the purposes of this discussion) the ambient area in the foreground and end up with an exposure than provides some detail in that region...

http://i69.photobucket.com/albums/i63/wiltonw/IMG_1579.jpg

or you could meter the sky close to the sun and render it with a brightness that causes the colors to saturate more

http://i69.photobucket.com/albums/i63/wiltonw/IMG_1578.jpg

Transmitted light behaves a bit differently than reflected light...try the same EV spread with a sign.

[Lightstream]

Hmmm, cool :) I never thought of it that way.

[Lester Wareham]

Lately I have had good luck with underexposing my 1/3 or 2/3 of a stop. I find that the color saturation is better and there is less clipping. Do I sacrifice anything in this process? Does underexposing add color noise?

I tend to expose as normal and nock the exposure back a little in RAW processing by eye. This gives better saturation but keeps the noise in the blacks down.

[CyberDyneSystems]

Underexposing many types of color film by a few thirds or more was a very valid and effective way to get better saturation..

But for Digital I do not find the benifits to out weight the many cons.. not by a long shot. Especially given that we have so much lattitiude with processing to boost that saturation without the inherent downsides of underexposure.

[wilflee]

With color negs, I usually over expose to get denser negatives. The denser negatives takes a bit longer to expose on the enlarger but it avoids "clear" areas on the negs which are basically useless.

The reverse is true for slides. Slight underexposure on slides gives denser images which appear as more color saturation.

Neither theory works with digital since the darkroom work is on your computer instead of on film/chemistry. But I sometimes do purposely underexpose shots not for the purpose of color saturation (which you can adjust with many popular software) but to prevent color clipping. The sensor in your DSLR sees grey. Period. If you have a bright red flower, it sees it as a grey flower (all 3 color channels, red/green/blue, combined).

The problem with seeing grey is that even though the shade of grey looks fine to the exposure meter, the amount of a dominant color in the picture may blow out the sensor. In other words, when you're shooting a bright red flower, the meter doesn't expose for red. It exposes for grey. So the meter may tell you the exposure ok OK (since the green & blue channels appear to be OK), the red will blow out the red sensors in your camera. What you'll get are white patches on the resultant image when it should have been bright red.

One solution is to get a color meter. But that's a $$$ investment with very little usage. Another solution is to get a camera that shows the levels curve of each color channel separately to check for color clipping. Or, simply underexpose and double check the picture immediately after shooting.

[Lester Wareham]

I agree with wilflee, except that most digital conversions, in camera or via RAW convertor use a tone curve to convert from linear luminance, these seem to be based on slide film type tone S-curves which is part of the story.

As he also points out the main problem with digital is clipping of one channel causing desaturation of one colour - this is what I tune back for in RAW conversion.

[Lesmac]

I disagree with wilflee to an extent, in running water shots for example, I always underexpose from average (although correct exposure for highlights), by up to 2 stops, knowing I can always pull back shadow /midtone detail in PS, you can't replace blown whites.
The trade-off is sometimes increased noise, but with the amount of detail in the 1DS MKII raw files, it's not usually a problem.

And being an previous velvia user, most of my landscapes are shot slightly underexposed for the richer tones.

Les

[Wilt]

OK gang, the principle of underexposure of color slide is due to the fact that the light response of film is NOT a linear one (like digital is), but is an S curve. It has a toe and a shoulder, both which are lower-sloped areas compared to the central area which has a more linear area. What that translates to is the high exposure areas are all rendered near white to white. So to get color back into things, you REDUCE exposure so more things fall into the steeper central area where color lies, rather than in the flatter shoulder area where things are near white to white.
In color neg, since its response is the inverse (higher exposure = denser) you do the opposite of color slide, and overexpose.
The idea of underexposing digital has NOTHING to do with any dose response curve...it ISN'T a curve at all! So there is not the case of having more items fall into the central area rather than the flat shoulder! Make sense?

[Lester Wareham]

OK gang, the principle of underexposure of color slide is due to the fact that the light response of film is NOT a linear one (like digital is), but is an S curve. It has a toe and a shoulder, both which are lower-sloped areas compared to the central area which has a more linear area.


Hi Wilt, as I say above most RAW converters and in camera convertors also have an S-Curve for conversion from linear.

ACR is one of the more linear looking default curves however with just a small curve at the blacks but straight at the highlights and apears to be post conversion from linear.

But theory aside there does seem to be an advantage in tuning back exposure even if this is just to stop a channel clipping as suggested by Wilfee. In the end we tune for the visual effect we want.