PGHphoto wrote:
I use compensation so that I always slightly underexpose the shot. It comes from my film days of expose for the highlights - develop for the shadows. I feel that when shooting raw it almost extends the exposure latitude since blown highlights are impossible to recover but shadows almost always contain detail that can be brought out in post. When I change the exposure comp and am using in-camera metering, the raw or jpg file definitely reflects the change i dialed in.....
This is
probably exactly the opposite of what you should be doing with digital!
Back in the days of film, with slides (transparencies) we had to be careful not to blow out highlights. A highlight in a slide is a complete lack of any image info... completely clear film that allows all light to pass through.
With negative films (B&W or color) it's just the opposite. Highlights on film are a great deal of info, completely blocking any light from passing through the film base.
Now with digital, it's more like negative film... there's lots of data and highlights are far more recoverable than most people realize. In other words, digital highlights are not very likely to be "blown out". I see and hear people trying to prevent their histograms from touching the righthand side of the display.... which is a mistake. In fact, any pure white within an image should push the histrogram strongly to the right.
With digital, underexposure causes loss of detail and additional noise in shadows.
The problem is that many people judge their images based upon uncalibrated computer monitors. There are two problems with this. The first is that most computer monitors straight from the box are way, way too bright. They cause users to make their images way too dark (hence too dark prints, or images online that appear too dark to other users). Another problem is that all computer monitors clip both highlights and shadows... none can render "true black" or "pure white". If you make a print from an image with a high quality printer and smooth matte paper you will always find a lot more detail in both the highlights and shadows, than you realized was there looking at the image on your computer monitor. In other words, the dynamic range of the camera is actually greater than what your computer monitor can display. If the monitor isn't calibrated properly, this is further exacerbated... the uncalibrated and overly bright monitor is clipping highlights even more than a calibrated one would.
All this leads one to do just the opposite of what one really should be doing... setting exposure to protect shadows as much or more than highlights. "Expose to the Right" is a broad generalization and not as good as precise exposure control... but it works, on the whole, because it skews slight over-exposure and helps prevent underexposure.
If you have a properly calibrated computer monitor, you can pretty easily tell for yourself what's best. Go shoot 1000 RAW images with Exposure Compensation zeroed out or, if shooting purely manually, always "centering the needle". Now post-process those images in whatever software you use and make note of what exposure adjustments you do to each of them. I bet you'll find the vast majority you are having to "push up" exposure (which will also amplify noise). On average over a large number of images, it's almost always better to have some slight over-exposure. With very high ISO images, in particular, under-exposure needs to be avoided at all costs! In fact, I was advised to try slight over-exposure and "pulling" images back down 1/3 to 2/3 stop in post-processing, to get the "cleanest" high ISO images. That worked well, without any noticeable loss of highlight detail.
But it can vary depending upon brand and even camera model within any given brand. I use different settings (less ETTR) with my current Canon 7D Mark II cameras, than I did with the original 7Ds that I used for about five years. Today I set no ETTR for normal lighting... I only dial in E.C. for unusual subject tonality or strong back lighting, for example. With my original 7D I used +1/3 stop ETTR most of the time.... +2/3 stop for high ISO and in strong, midday sunlight. But I might change that... because I'm finding after about a year and some 50,000 or 60,000 images with them that the Mark II camera images also tend to need a bit of push in post-processing... I just don't think it's as much as was needed with the earlier model. This is probably due to different metering systems used in the models.
There also are differences depending upon the metering method used... Evaluative/Matrix metering is a little different from Center-Weighted. Partial or Spot metering can be more accurate, but requires greater and more careful user tweaking via the manual settings or Exposure Compensation. Actually, a separate, handheld incident meter can be the most accurate means of determining exposure settings. This type of meter measures light falling onto a subject, instead of what's being reflected off it. For this reason, the incident meter isn't influence by subject tonalities (so no E.C. is needed).
Because of the brand to brand and model to model differences you really should observe your own images and judge for yourself... And you really need a calibrated monitor to do so. But, if you take the trouble to test it yourself, I bet you find that the last thing you want to do is deliberately and consistently under-expose... the way you're doing now!