In-camera exposure compensation (EC) vs. EC during RAW conversion

[atvinnys]

When bracketing my exposures, I definitely understand how exposure compensation works (double the exposure time is 1 stop, quadruple is 2 stops, etc.). However, if you do this on a RAW file, I am not sure I understand the whole software side of things.

(1) What does RAW exposure compensation actually do? Is it a simple shift of the histogram?
(2) Is there trade-off to applying EC in post processing as opposed to in the camera itself, such as using exposure bracketing. For instance: to avoid a long exposure time, you would think you could just underexpose the photo *in the camera*, and then use software to later overexpose *the file* by 5 stops.

Now, besides the pain of correcting the exposure after the fact, I've got to believe that this is not the only drawback...

Thanks
Vincent

[McQ]

Exposure compensation (EC) performed on the RAW file is done in software, whereas EC performed in the camera just changes the original exposure settings. In other words, in-camera EC changes how the image is captured instead of changing the image after it is captured. EC performed within the camera will therefore have no effect on image noise, clipping latitude or apparent dynamic range. By contrast, there are many drawbacks to software EC. EC performed in software on the RAW file can dramatically increase noise if it is positive EC (because the shadows can become the midtones); if it is in the negative direction then you will have a limited dynamic range in the highlights (these will be clipped much sooner than if EC is applied in camera).

Always perform EC within the camera itself if you have a choice. Applying EC to a RAW file is still better than adjusting the exposure of a TIFF or JPEG because it is performed on the untouched (high bit depth) pixel information from the camera's digital sensor.

[SRH]

...EC performed in the camera just changes the original exposure settings. ... EC performed within the camera will therefore have no effect on image noise, ...

McQ, I agree with the principles you outline, but beware--my Oly E3 in "full auto" mode will adjust the ISO setting to achieve EC, and that, as you know, introduces noise. I don't know if other cameras do the same. I normally have my ISO set to a fixed value, so it doesn't really bother me--it's just something to watch for. The E3 has user settings to limit the range of auto ISO, and that is another way to limit the change in Auto ISO.

[McQ]

Thanks for clarifying your experience with EC, SRH. I was not aware of cameras that changed ISO in accordance with changes in EC. Makes sense though that in full auto mode, if the camera cannot achieve the right shutter speed, it might resort to changing ISO if the aperture cannot get any larger... My response should have had the caveat that that analysis primarily applies to SLR cameras.

[TonyBeach]

...there are many drawbacks to software EC. EC performed in software on the RAW file can dramatically increase noise if it is positive EC (because the shadows can become the midtones); if it is in the negative direction then you will have a limited dynamic range in the highlights (these will be clipped much sooner than if EC is applied in camera).

Actually, beginning with an optimally exposed image (ETTR), applying negative EC via the RAW converter will decrease noise and increase DR. For Nikon cameras increasing EC in the RAW converter will have the same effect as raising ISO in the camera and can actually preserve highlights that might otherwise be clipped if ISO is raised in the camera.

[tmyuen]

I understand that when shooting in RAW, I can do exposure adjustments (+/-) during the post processing stage. With this adjustment capability, it would initially seem that I don't need to care about using EC (Exposure Compensation) bracketing in-camera -- to just get the "right" exposure.

However, the previous posts seem to indicate that I should almost always try to get it right *during* the exposure, and not in software later on. What therefore is the real merit of using EC Bracketing when shooting in RAW...?

[Colin Southern]

When you "adjust the exposure" of a RAW file you're not really adjusting "exposure" per sec (that can only be done at the time you capture the image) - what you're really doing is manipulating the data that's already been captured in the RAW file so that the information captured is better aligned with what we need it to be. If the information is in there somewhere then there's a good chance that we can "dig it out" and make use of it with various techniques, but ...

... on the other hand, if the information isn't there in the first place (for example, large areas of the image are brighter than can be captured by the sensor and just go to full white, or areas that are so dark that the values represent them are practically zero), then no amount of manipulation is going to be able to recover them. (ie "you can't get blood out of a stone").

When bracketed shots are taken at a scene that has ranges of brightness larger than the sensor can capture, each shot captures a portion of the range of tones required - and these are later combined into a high dynamic range image.

So - in theory you might be able to get away with not bracketing in some less extreme situations (and tweak the RAW file to display the captured information that's not initially visable), but with a true high dynamic range scene there will only a limited (and insufficient) range of data present - and you won't be able to recover some of the information that you need.

So in summary, RAW is great for coaxing out stuff that already there (but "hiding"), but if it's not there, it's not there - and you have to bracket to capture what you want.

Additionally, the more you dig shadow detail out of a RAW file, the noisier that data gets - and the less detail that there is - so it can result in pretty poor image quality.

It's really not so much a case of RAW files being the photographic equivalent of "the universal answer to life" as it is a case of the alternative (Jpegs) making processing decisions for you, and throwing away a LOT of information in the process (problem is, it doesn't always throw away the RIGHT information).

Sometimes people create a high dynamic range image by "re-developing" the same RAW file - it doesn't give you any information that wasn't already there in the first place, but the technique can make image processing a bit easier in some circumstances.

Hope this helps

Cheers,

Colin - pbase.com/cjsouthern