If there is plenty of light, raw exposure is usually not intellectually difficult with modern CMOS full-frame or larger cameras. The mantra is a twist on the old film rule: expose for the highlights, develop for the shadows. The idea is to give as generous an exposure as possible, subject to the constraint that important highlights are not saturated in any channel in the raw file. The word important is, ahem, important. In normal photography, specular highlights of limited extent can be allowed to clip, and usually allowing some clipping there will provide adequate signal-to-noise ratio in the shadows. Trying to keep the reflection of the sun from a car bumper from clipping will usually result in poor shadow detail.
A lot of the above is counterintuitive to many photographers. Some corollaries:
- It doesn’t matter what the EVF image looks like to you.
- It doesn’t matter what the JPEG preview image looks like.
- Low contrast scenes will get, on average, more exposure than normal or high contrast scenes.
- Low contrast scenes will look too bright in the JPEG preview image.
- High contrast scenes will often look too dark in the JPEG preview image.
There is an erroneous idea that a “correct” exposure must place midtones in some predetermined place on the histogram. In fact, if you’re shooting raw and post-processing later, the placement of midtones is flexible. What matters is keeping as high a signal to noise ratio as possible, especially in the shadows where the sensor’s SNR ratio drops. If you underexpose unnecessarily, you may be able to recover tonal values, but see more noise.
What does matter when exposing in raw?
What matters is the raw data itself. Unfortunately, almost no camera today shows you a raw histogram, and the conventional method for approximating the raw histogram in the JPEG preview image results in ugly green MILC finder images. Tools like RawDigger or Fast Raw Viewer can help reveal how close you are to saturation in the raw file, but they are impractical to use in the field. What you can do is educate yourself about the difference between the raw histogram and the JPEG preview histogram so that in circumstances that you’ve previously encountered, you’ll know about what the raw histogram will look like by looking at the JPEG one.
This approach is about deliberately using as much of the sensor’s dynamic range as the scene allows. If your scene doesn’t have important highlights near clipping, then expose more generously. If it does, then back off, but only just enough to protect the highlights you care about.
The key to this approach is not to get greedy. If you have half a stop worth of unused dynamic range at the top of your raw histogram, it probably won’t make or break your image, but if you’ve clipped an important highlight by half a stop, that will be likely to have a visual effect on the developed photograph.
Stepan Kana says
“almost no camera today shows you a raw histogram” – is there a camera that does?
One way to get around this is to use a dslr. The finder shows the scene as it is. You use exposure compensation, or spot metering, to get the scene right. Often on my Canon 6D using multi-metering, I find that I need to overexpose by 2/3 of a stop. Bright scenes need at least 1 stop overexposure. The spot-metering technique is tricky as you need to overexpose by about 2 1/3 stops if the subject is “average grayness”. You need to be thinking about that all the time and compensate for the brightness/darkness of the subject. Often you walk away with perfectly exposed, but banal pics 🙁 The multi-meter, if you’re aware of its limitations, allows you to focus on getting the pics you actually want.
JimK says
Some Canons with Magic Lantern. Betterlight backs.
JimK says
How is that an improvement over a live histogram?
Štěpán Kaňa says
I’ve no experience of ‘live histogram’. If you use f/2.8 lenses (or faster) on the canon, the scene as seen in the viewfinder is accurate 🙂 The problem with the screen at the back of the camera is that it crushes blacks. On the computer there is much more shadow detail. Also in bright conditions relying on the screen makes you overexpose. People want to be looking at pictures, not histograms. Oftentimes the histogram suggests one thing but the resulting photo doesn’t feel right. Yes, you can push/pull in the raw development software, but it’s not terribly good. If you push, you’re losing data. If you pull, the highlights look funny. Ultimately there is no correct ‘exposure’. It’s a decision made at the time of shooting, one you’ve to live with. Maybe 44x33mm sensors have more latitude.
JimK says
This is a discussion about how to set exposure. How does looking at the image in the optical finder help with that?
JimK says
Relying on the histogram on the screen makes you overexpose in bright conditions? How does that happen?
JimK says
This is about getting the right exposure. Do you really think I’m suggesting that a histogram be included in the presentation form of the image?
JimK says
Getting the correct raw exposure is not about what feels right.
JimK says
Why do you say that? If there’s no clipping, you can push without losing data.
JimK says
That’s why I’m saying don’t let the raw data clip.
JimK says
I beg to differ. There most certainly is.
Štěpán Kaňa says
Do you never adjust exposure in the post?
The optical viewfinder only helps in that you have a real reference of the scene as it was. Of course, how it was might not be how you want it to be portrayed.
On the canon the brightness of the screen doesn’t change so in bright light if you rely on the screen the pictures tend to be too bright. Maybe you can pull it to make it right without any loss, but the blinking highlights are unreliable and so is the histogram (on the canon!)
If you push, shouldn’t you have exposed more, correspondingly?
The comment about looking at pics vs histograms was facetious, reflecting my frustration with the canon firmware.
Isn’t it also about the target medium? CMYK commercial printing vs. printing at home vs sharing on screens online in srgb jpegs. For newsprint you’d aim for brighter values overall than when using coated paper and so on. how do you determine your ‘correct raw exposure’ that is not about it feeling right, and is presumably independent of the target medium?
JimK says
You can’t change the exposure in post. The controls in Lr and Ps do different things, but neither one of them actually changes the exposure, which is fixed at the time the shutter is released. Changing the actual exposure affects the photon noise SNR.
JimK says
How does that help you pick the exposure?
JimK says
I am not suggesting you rely on screen brightness to detirmine exposure at all.
JimK says
If your important highlights are just short of clipping, more exposure is not what you want. That will drive them into clipping. Instead, lift the shadows in post.
JimK says
Proper raw exposure is independent of the target medium. Post processing is not.