DPChallenge: A Digital Photography Contest You are not logged in. (log in or register
 

DPChallenge Forums >> Tips, Tricks, and Q&A >> underexposing raw?
Pages:  
Showing posts 1 - 20 of 20, (reverse)
AuthorThread
01/06/2005 09:38:25 PM · #1
maybe i'm missing something, but something i've been doing lately is underexposing my raw shots. not only because it's easier to correct than an overexposure, but i've been using it for another handy purpose.. in conditions where i may only achieve a 1/15 or 1/20 shutter speed on my 50mm lens, i have been underexposing a stop in order to get a 1/30 or 1/40 out of them when i am not around my tripod. of course, the negative to this is that i may have some completely black areas (which usually are not the focus of my photo and haven't really seemed to harm them so far).. has anyone else tried this? any thoughts?
01/06/2005 09:41:55 PM · #2
We all find a comfortable way to go. I have always under exposed by a stop. Sometimes a half stop. Hardly ever expose on the money as there is always a disturbing burn out somewhere.
01/06/2005 09:44:46 PM · #3
The better way to expose is to get as close to the right of the histogram as possible without blowing the highlights.

And here is the reason why you should be doing that.
01/06/2005 09:50:31 PM · #4
hmm - it provides a very interesting case as to why we should expose to the right of the histogram for better detailed shots (if i understand correctly), i wonder how big the tradeoff is by underexposing a stop? would it be better that i move from iso400 to iso800 (or 1600, since i am essentially underexposing a stop to begin with?) and then expose to the right?

edit: my main concern with my exposures is often camera shake and thus my concern with getting an appropriate shutter speed.

Message edited by author 2005-01-06 22:01:19.
01/06/2005 09:55:22 PM · #5
off topic: what software are you using to develop your raw files?
01/06/2005 09:57:17 PM · #6
i currently use photoshop's raw capabilities (using the 2.4 converter beta right now).. i know there are also captureone and breezebrowser (i think?), but i prefer to keep the number of programs installed to a minimum if i can.. plus, not so sure they provide anything photoshop doesn't.

Message edited by author 2005-01-06 21:57:49.
01/06/2005 10:00:42 PM · #7
As long as you expose as far to the right, but without blowing out ANYTHING, then you will get a better print/webimage.
read Real World: Camera Raw, by Bruce Frazer. An eye opener.
01/06/2005 10:05:17 PM · #8
Hmmm...I've always underexposed 1/2 stop or so. At least ever since I got my Rebel. Before that (digital at least) I always bracketed RAW -1, -.5, 0 off of the meter using my G2. It should be easy to go back and compare some of the -1 versus 0 shots.

I stopped bracketing when I discovered with the Rebel you could exposure compensate during raw processing. But the real question is: what does that really do in comparison to the real exposure. In other words, if I underexpose -1, and then compensate +1, is it the same as a metered exposure (0)? I suspect not, but I've not seen that discussed anywhere.

Message edited by author 2005-01-06 22:05:44.
01/06/2005 10:38:12 PM · #9
I have exposed according to this nice article and the information is mostly accurate. For example, the 10D has less noise than the Minolta.
But there are often very small areas which get blown even under optimal exposure. With the minolta I did a 1/2 stop and the noise ratio was very high. The Canon 1/2 stop works best, but it is all relative to what you are after. I only shoot isa 100.

When you have an ideal subject it is best to shoot on the money: Bang Sore loser, the chess shot was shot on the money. The broken 1/2 underexposed. It all depends on what is on the scene.
01/06/2005 10:47:11 PM · #10
hmm.. i wish i could shoot iso100, but that would've been 2 more stops of light i would've been losing (i took a walk through the botanical gardens on my campus after class, but this was around 4pm and under the shade of trees - certainly wasn't carrying my tripod with me to class all day :D). when i do have enough light, i'll certainly try to begin to adhere by the right-side of the histogram rule. thanks for the links/responses guys.
01/07/2005 10:49:41 AM · #11
The article off Luminous Landscape echoes exactly what I read out of a book (Real World Camera Raw with Adobe Photoshop CS) by Bruce Fraser. Good to read through that article (and book) and see why. Underexposing will only drop all detail from your "shadows." You can always bump up the Shadow slider and Contrast, and down on the Brightness and Exposure if needed.

Message edited by author 2005-01-07 10:50:14.
01/07/2005 11:40:16 AM · #12
I haven't noticed blow-outs when the histogram is exposed to the right on my D70, but maybe it's a difference in how Canon and Nikon handle things. When I can't get the shutter speed I want I always bump the ISO rather than underexposing. I get a touch more noise as it goes up, but I can usually manage that with noise reduction in the converter much easier than managing motion blur.
01/07/2005 11:43:17 AM · #13
And you can always bracket both high and low - just in case. Sometimes the little slightly blown hightlights are critical, sometimes the shadow detail is critical.

But reading the articles, I think I'll start favoring the right side of my histogram a bit more.
01/07/2005 11:47:28 AM · #14
Another interesting twist on the whole expose-to-the-right theory... Bibble 4.1 now has a higlight detail recovery option. The theory is that for minor highlight blowage, there's usually at least one channel that retains the detail. Bibble tries to find that channel, and then applies some kind of voodoo to restore it to the channels where it was lost. They have some good examples on their site.

I don't know if all converters have similar functionality or not, but it's a cool feature. I've used it a few times successfully on cloud detail. Not that I cared too much since it was intentionally blown out, but it does work. I become more impressed with Bibble each release.
01/07/2005 12:24:44 PM · #15
I use Canon's DPP (more than PSE2 actually).

Anyways, if you underexpose intentionally, bringing it back up to the correct level increases noise. I think of RAW adjustment as fine-tuning the ISO--noise increases as you brighten. I think if you expose right on or a little under, you'll have the best results.
01/08/2005 12:42:51 AM · #16
Originally posted by brianlh:

maybe i'm missing something, but something i've been doing lately is underexposing my raw shots. not only because it's easier to correct than an overexposure, but i've been using it for another handy purpose.. in conditions where i may only achieve a 1/15 or 1/20 shutter speed on my 50mm lens, i have been underexposing a stop in order to get a 1/30 or 1/40 out of them when i am not around my tripod. of course, the negative to this is that i may have some completely black areas (which usually are not the focus of my photo and haven't really seemed to harm them so far).. has anyone else tried this? any thoughts?


What you are doing is the same thing in the camera as turning up the ISO from say 100 to 200. I tried this taking moon shots when I first got my 1Ds. To prove this try shooting a picture at ISO 100 while underexposing by 3 stops, then shoot the same image at ISO 800 at normal exposure. the results shoud be identical. Elarge both pictures until you can esily see the pixels to compare. Both should have some degree of noise.
01/08/2005 12:52:26 AM · #17
Originally posted by nsbca7:

What you are doing is the same thing in the camera as turning up the ISO from say 100 to 200. I tried this taking moon shots when I first got my 1Ds. To prove this try shooting a picture at ISO 100 while underexposing by 3 stops, then shoot the same image at ISO 800 at normal exposure. the results shoud be identical. Elarge both pictures until you can esily see the pixels to compare. Both should have some degree of noise.


thanks to everyone for the replies - i guess i'll definitely favor the right side of the histogram now.. and thanks for clearing this confusion up for me nsbca7.. looks like i should maybe just increase the iso on those tough-to-get shots in order to get highlight detail, rather than underexposing at a lower iso.. thanks for all the replies! :D
01/08/2005 04:24:43 AM · #18
Here are some tests I conducted to illustrate differences in the various methods. A visual example sometimes helps more than theory. All these shots were tripod-mounted since very long exposures were used to better highlight differences in a low-light scene. All images are 640x427 crops (100%) of RAW images converted (using ACR) to 8-bit sRGB (fyi, not my normal workflow). All images had the same white balance and calibration settings applied. Sharpness and ColorNoiseReduction were set to 0. All other settings were left at their defaults (except exposure as noted below).

4994: Normal exposure at ISO 100
4995: Normal exposure at ISO 800
4996: 3 stops underexposure at ISO 100 (then 3 Exposure in ACR)
4997: 1 stop overexposure at ISO 100 (then -1 Exposure in ACR)
4998: 2 stop overexposure at ISO 100 (then -2 Exposure in ACR)

499449954996
49974998

Note that ISO 800 is brutally noisy on my camera. It should be clear that ISO 800, while introducing noise, looks considerably better than the -3ev image due to severely clipped shadows. The ev2 image looks cleaner than the ev1 image, but at the cost of (considerably more) highlight clipping (in the upper right).

To illustrate (somewhat) the "expose right" benefits, I've taken 4994 (ev0) and 4997 (ev1) and applied a simple levels adjustment (0,1.00,127):

4994-L4997-L

4997-L has more detail and less noise in the mid-tones and shadows. The upper right is more blown out in 4997, but the original (4994) was already blue channel clipped in that region. The in-camera histogram did not indicate any clipping.

Caveat: These on-screen examples are not necessarily a good indication of similar differences in prints.
01/08/2005 06:16:11 AM · #19
wow, that's an excellent side-by-side test. granted, it's for a different camera - but the general trend/results should be the same across the board. thanks for sharing these tests with me (us)! i was actually thinking of doing something along these lines, but who knows when i would've actually gotten around to it. i guess my main concern with underexposing was in order to reduce some camera shake - but i'll definitely try to avoid that when possible now that i can see the results.

thanks so much!
01/08/2005 12:33:48 PM · #20
Originally posted by brianlh:

granted, it's for a different camera - but the general trend/results should be the same across the board.

Yeah, the D60 is the sibling that didn't so well in school. It uses essentially the same technology (different AF system though), but they did a great job cleaning up the sensor noise for the 10D/300D. Anyway, I'm glad you found the results useful.
Pages:  
Current Server Time: 09/15/2025 01:32:34 PM

Please log in or register to post to the forums.


Home - Challenges - Community - League - Photos - Cameras - Lenses - Learn - Help - Terms of Use - Privacy - Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 09/15/2025 01:32:34 PM EDT.