Posted by Brian Monk on December 29, 2006 at 2:59 PM
I recently read an article in a popular digital photography magazine that discussed the routine use of underexposure in order to save highlight data when shootong digital images. The article dismissed this idea in favor of shooting at proper exposure, stating that this would require less time in post-production, and glean more data from an initial exposure. Logically, this makes sense, though I admit to being and "Underexposer." It's just something I've come to do, rather than being advised or taught to do so, and I've learned to manipulate my images in post-production to develop pleasing photographs.
I would like to know if any of you shoot to underexpose intentionally on a routine basis, or under what specific circumstances you do so.
Actually with digital the opposite is true - there is more image information in the highlight side of the scale than in the shadows. Obviously one should be very careful not to over-expose and risk losing it all, but common wisdom is to get as much of the histogram as far to the right as possible without blowing the highlights.
Correct exposure for me ranks all the way up there amongst all the lofty goals like elimination of global warming, hunger and poverty.
My solution is tedious but makes sure I don't burn my fingers. I determine exposure readings for different parts of the scene, neutralize it if necessary with GND filters, and then bracket upto 3 stops on either side of the neutralized exposure. Generates a ton of images (upto 7 if bracketing 1 stop at a time, and upto 21 if bracketing 1/3rd stop each time )
Every time I return from a session in the field, I have only about 4-5 compositions, and upto 100 images total.
Use the histogram as Guy suggests and you'll save yourself a lot of time, both in the field and later during the editing cycle.. You'll also be more productive in the field and in your post processing, too.
You're right Preston I don't actually bracket 21 times in the field all the time - it's just that I never rule it out ever since I drove 500kms+ on one afternoon, took a few photographs on and around Mt. Washington without bracketing, and drove 500kms+ back the next day morning ...
Never gonna make that mistake again
Posted by Richard Wong on December 29, 2006 at 5:57 PM
Are you referring to this month's Outdoor Photographer? There's not really a formula you can follow. If you shoot backlighting such as a sunset at the beach then deliberately overexposing every shot can leave you with no information sometimes.
I bracket two or three times at most usually for landscapes. Generally one where I think it should be, one 2/3'rds up, and 1/3rd down. If I'm confused then I'll look at the histogram and then I'll do some extreme bracketing.
Posted by Ed Cordes on December 30, 2006 at 9:36 PM
I think it depends on the subject/scene. If there are details such as bird feathers or animal fur or flower petals that are brighter than the rest of the scene, I do use EV - 1/3 to -1. These are important parts not to be lost. However, if there are minor specular highlights or minor bright parts whose loss of detail will not detract from the image, I do not use a minus EV. In this case I try to get the total tonal range in the important parts of the image and let the unimportant highlights go where they may. This does produce better shadow detail.
Well if you are comfortable with adjusting exposure in raw/PS then just get it all to fit in. Personally I tend toward images that have most of the information in the shadow half of the histogram. BUT that little information in highlights is crucial to the feel I want to achieve. If they are blown, the image is blown. Pretty regularly I'm doing -1 compensation to fit the highlights in. I have been successful with this and still getting the detail I need out of the shadows.
Very often I rely on the blinkies when the highlights are bright. I don't underexpose until all the blinkies go away -- just enough till I'm sure I don't have large expanses of clipped area in the frame. A little "twinkle" of blinky in a highlight region is ok. (hows that for technical terms!) Sometimes trying to get rid of all the blinkies (like on the white collar of a kingfisher) can compromise the shadow detail.
What Guy said. Expose so that the histogram doesn't fall off the right hand side.
But that's a very, very general statement. The theory is there is better resolving capability in the lights than the darks, and this technique will serve to reduce noise in the darker areas. But there are many exceptions to it. Obviously if you're trying to do something like shoot silhouettes, you need a different strategy. Or if your picture has very little darks, but lots of fine detail in the brights. I've also found that alpenglow is better slightly underexposed.
Lately, I've been using HDR when possible, so I take a range of exposures (usually three shots at +/- 2 stops) to make sure I get everything, and let the computer sort it out later--this only works for static subjects, of course.
-Paul
Posted by Les Barstow on January 3, 2007 at 4:47 PM
If you expose to push everything to the right until just before something falls off the overexposure side of the histogram, you can almost always recover the detail in PhotoShop or in your RAW processing software. If you expose towards the underexposed end and then recover the detail from the shadows, you'll be enhancing the noise that always hides in the shadows of digital images.
To understand why, you have to understand what you're doing to the digital information: when "toning down" highly-exposed images, you are using the most significant data from the pixel to form the modified value; this data is the most stable data in the image, and at least until you reach near-overexposure, it is also the most information-rich. When brightening shadow details, by contrast, you are enhancing the least significant portion of the data - the part that the camera "wasn't as sure about". That's the part of the data where digital noise resides, too. When you enhance from the shadow side of the histogram, you are stretching a very narrow amout of data to fit a much wider range of tones; this can introduce banding and enhances noise.
A well-exposed single digital exposure shouldn't pile too much on either end of the histogram, but should lean towards the overexposure end of the chart. Scenes with high dynamic range really need traditional (grad-ND) or digital (HDR) techniques; no amount of playing with exposure is going to squeeze all that data into a single shot!
Discussion
10 repliesI would like to know if any of you shoot to underexpose intentionally on a routine basis, or under what specific circumstances you do so.
Thanks for the input!
Guy
My solution is tedious but makes sure I don't burn my fingers. I determine exposure readings for different parts of the scene, neutralize it if necessary with GND filters, and then bracket upto 3 stops on either side of the neutralized exposure. Generates a ton of images (upto 7 if bracketing 1 stop at a time, and upto 21 if bracketing 1/3rd stop each time
Every time I return from a session in the field, I have only about 4-5 compositions, and upto 100 images total.
You are working too hard, mate!
Use the histogram as Guy suggests and you'll save yourself a lot of time, both in the field and later during the editing cycle.. You'll also be more productive in the field and in your post processing, too.
-PB
Never gonna make that mistake again
I bracket two or three times at most usually for landscapes. Generally one where I think it should be, one 2/3'rds up, and 1/3rd down. If I'm confused then I'll look at the histogram and then I'll do some extreme bracketing.
Scott
But that's a very, very general statement. The theory is there is better resolving capability in the lights than the darks, and this technique will serve to reduce noise in the darker areas. But there are many exceptions to it. Obviously if you're trying to do something like shoot silhouettes, you need a different strategy. Or if your picture has very little darks, but lots of fine detail in the brights. I've also found that alpenglow is better slightly underexposed.
Lately, I've been using HDR when possible, so I take a range of exposures (usually three shots at +/- 2 stops) to make sure I get everything, and let the computer sort it out later--this only works for static subjects, of course.
-Paul
To understand why, you have to understand what you're doing to the digital information: when "toning down" highly-exposed images, you are using the most significant data from the pixel to form the modified value; this data is the most stable data in the image, and at least until you reach near-overexposure, it is also the most information-rich. When brightening shadow details, by contrast, you are enhancing the least significant portion of the data - the part that the camera "wasn't as sure about". That's the part of the data where digital noise resides, too. When you enhance from the shadow side of the histogram, you are stretching a very narrow amout of data to fit a much wider range of tones; this can introduce banding and enhances noise.
A well-exposed single digital exposure shouldn't pile too much on either end of the histogram, but should lean towards the overexposure end of the chart. Scenes with high dynamic range really need traditional (grad-ND) or digital (HDR) techniques; no amount of playing with exposure is going to squeeze all that data into a single shot!