When Should You Underexpose Your Images?

Underexposing is one of the most common techniques photographers in many different genres use, and it can be quite a beneficial way to shoot, but that does not mean it is always the appropriate way to shoot. This helpful video discusses the benefits of underexposing your images, when it might be right for your workflow, and when you should avoid it.

Coming to you from Gab Loste from gabpolitely, this excellent video discusses the popular technique of underexposing images. The reason this technique has come about is that in general (with digital sensors), it is easier to recover dark shadows than it is to fix overexposed highlights. And in recent years, when sensor technology has enabled greater and greater latitude in post-processing, photographers have been able to embrace the technique with less fear of a major loss in image quality. That being said, if you do decide to use it in your own work, it is really crucial to know the limits of your equipment. For example, when I am working with my Sony a7R III, I know that I can push the files about twice as far as I can when I am shooting with my Canon cameras, and that has a definite impact on how I think about and shoot my images. Consider what you shoot, your own gear, and how you typically edit your photos when you undertake the technique in your own work. Check out the video above for the full rundown. 

Log in or register to post comments


Jerome Brill's picture

As an a7R III user I am not under exposing my images as much as I use to. I was relying too much on it. Trying to push images when I could have just exposed better. Also not practising steadier shots at lower shutter speeds. If you have no moving subject and a tripod, you should be mostly on unless you are trying to bracket.

I ended up setting my zebra to custom Lower Limit 99+. I'm still erroring on the underexposed side for highlights but by not as much. This personally covers most of what I shoot.

Nick Rains's picture

There is no such thing as underexposure really. Under-exposure or over-exposure, is a relative term and so is only relevant when compared to the camera meter's "best guess". Watch your histogram and/or your blinkies, and you'll be fine.

The thing about histograms is they use JPG-data and not RAW-data. It makes it more or less useless if you expose to the limits. Nikon leaves a margin of about 2/3 to 1 EV. So if your JPG-histogram already shows clipping you'll discover there is much room left while post processing the RAW-file.
This video is not worth looking at. Why does FS even promote it? It is obvious that the dude in the video does know nothing about how the sensor data is stored. And stop waving the hands around all the time!

Nick Rains's picture

You are mostly correct, but histograms are far from useless - they just need interpreting. You get a feel for where the exposure needs to be, a few blinkies on parts of white clouds, for example, is probably about right. It's quite camera dependent too, but it all comes with practice. Still far better than using the light meter.

They are (only) useless if you do expose to the limits. You do NOT want to blow out important parts of your photo. Feelings for the histogram I do not need and do not want to rely on, since with feelings I still have to keep a safety margin.

We need RAW-histograms and I wonder why no camera manufacturer does implement them.

Nick Rains's picture

RAW data has no white balance set yet, and so the histogram cannot be measured. In camera jpeg histograms with correct WB (from the camera settings) are more consistent than just raw brightness values with no WB set.

While true, you could use the measured WB. But it does not matter. A raw histogram would show exactly what channel would "hit the wall", has been blown out. Please check this article for further explanations:
(Edited: typo)

Nick Rains's picture

Yes, I'm aware of that. A good reference.

The problem is that you can't really 'measure' a WB in-camera other than using auto-WB which has its limitations. There is no practical thing as correct WB unless you are in a studio shooting specific coloured products for publication. All else is subjective and IMO an aesthetic choice.

So you are stuck with un-demosaiced raw data, uncorrected for Gamma and WB. As rawdigger shows you can certainly graph this. All well and good but it still does not account for the shift in the output-referenced data (especially the blue channel) when using different WBs.

Raw data can in principle be displayed but I hold that it's not as useful as people think. FWIW, if you have an old version of Apple Aperture you can extend the histogram to values beyond 255, to see what raw data is left in the highlights but not shown in the preview jpeg. It's called Extended Histogram and I have never seen it in any other software.

You may well have thought of this already since you seem very well informed on this subject, but a way to fudge a more accurate in-camera histogram is to actually use auto-WB and turn down the jpeg contrast to minimum. The rear camera screen or EVF now shows a 'flat' image but the lower contrast histogram now shows much of the raw highlight information and your blinkies will be a bit more accurate with respect to the raw data.

Thank you for the well written explanation. You certainly have an argument here. Applying Gamma and correcting WB could be done without blowing out brighter (mostly blue) parts by lowering the overall exposure before applying Gamma and WB.
(Sometimes I struggle with intense blue LED light which is more and more used on stage).
If you like, you could read: https://www.darktable.org/2016/10/raw-overexposed/ which I find interesting.
Thanks again for your input! I appreciate it very much.

I went over the subject considering your thoughts. I think we both are right. The margin, e.g. Nikon keeps at the right side and the use of JPG-histograms is reasonable. So I vote for the option to show the flat RAW-histogram (lets call it this way) too. Why? So one can move the exposure close to the upper limit in scenes with significant parts of dark areas, so that one can gather as much information as possible there. (Background: The way the data is distributed over the whole area).

Nick Rains's picture

Indeed. But this does pre-suppose that you wish/need to capture the image data in a single exposure. I frequently auto-bracket, which removes the need for pedantic exposures. Combined with flat-jpeg based blinkies, my exposures are usually pretty close to optimum!

Sorry for the late replying, Nick.There are occasions when you cannot bracket and I do not want to process a multiple number of photos. It is already without bracketing much work to do. I shoot dance performances or classical concerts quite often in very low light or quickly changing light conditions.
Of course if you are shooting landscapes this is a different story. I would bracket as well and I do so if necessary. Cheers!

Nick Rains's picture

Can I sum up by saying "One does one's best in the circumstances"? And practice helps a lot. I've been doing this for almost 40 years so have kinda got the hang of it!

Catherine Bowlene's picture

Underesxposure is relative as you are the one to decide whether the photo needs to be fixed. Still, at the very beginning I didn't thought it was a must, I just believed it was too difficult or something. Now I see that fixing washed out photos is important when the necessity of it is obvious. https://photo-works.net/how-to-fix-overexposed-photos.php - here is a good tutorial from which I started fixing the photos, maybe it will help someone on here as well (which I doubt in fact since the majority of Fstoppers community are experienced photographers but just in case).
And thanks for the article, Alex, it was good!