Should You Over or Under Expose Your Photos? An Experiment in Understanding Exposure

If you are well steeped in photographical technical terms, then the TLDR is I bracketed 9 photos at 0.5 stops separating each photo to test the recovery of the camera data. I set up the shot with a balanced exposure to have both under and over exposed elements, then shot my bracketed images and brought it into Lightroom to see how much pushing and pulling I could do with the RAW data.

I wanted to come up with a test, a way of visualizing with my eyes, what looks under or over exposed. If the limits of my camera’s sensor are dependent on the software I use to edit, then it seems like a worthwhile endeavor to try to learn what those limits are and if I can internalize them before shooting. 

I also want to learn about software’s exposure recovery in general AND to create a test that you can try at home to better understand the power and limitations of your camera.

If that doesn't make sense, or if you really love a good in-depth technical rambling, let’s break it down into actionable steps so that you can follow along and test your own camera at home.

Step 1 - Learn what bracketing is and if your camera can do it (spoiler, it probably can)

Bracketing in photography is a technique where multiple shots of the same scene are taken using different camera settings. The goal is to increase the likelihood of capturing a well-exposed image, particularly in challenging lighting conditions.

For this test we’re going to be using specifically Exposure Bracketing, so let’s peel apart that little nugget and make sure we’re all on the same page when it comes to what that means.

Exposure bracketing is a technique where photographers capture multiple shots of the same scene with slightly different exposure values (or more commonly known as stops though your camera may show them as EV). When using exposure bracketing, photographers typically take three shots (though some can do less/more). 

If we’re just talking about those three shots the first will be at the camera's recommended exposure settings (measured by your camera’s metering mode), one slightly underexposed (usually by -1 EV) and one slightly overexposed (usually by +1 EV).

Some cameras allow for a wider range of bracketing (5,7,9, or more) with varying amounts of EV between each. As a quick example, I set up my Sony A7R III to take 9 shots (self portraits) each spaced by 0.5 EV. So one photo each at  -2.0 EV, -1.5 EV, -1.0 EV, -0.5 EV, 0 EV, +0.5 EV, +1.0 EV, +1.5 EV, and +2.0 EV.

In case I didn’t achieve the fully blowout of dynamic range, I also did another round of bracketed shots at a full stop, making that -4 EV, -3 EV, -2 EV, -1 EV, 0 EV, +1 EV, +2 EV, +3 EV, +4 EV).

OKAY, WOW SO MUCH JARGON, but we’re through the most burly part of the tech talk (I think) (maybe not)

Step 2 - Take the photos! 

Like I mentioned, I wanted to check the recovery of the camera so I picked a shot that would have both deep shadow and bright highlights. My goal? To see just how much recovery is possible with over and under exposed images. My subject? A tree in the sunlight and a self portrait (because I don’t want to subject anyone else to my extreme version of learning) so that I can also monitor what is happening and changing to my skin tone.

Step 3 - Analyze the data and begin testing!

Let's start with my face (please excuse my missed focus). On the left we have the normal (+0 EV) exposure and on the right the overexposed image (+2 EV). I’ve brought down the exposure in LR -2 stops to match the normal exposure to get a match aaaaaand… wtf is going on with that skin tone?! Not only are the highlights getting lost by dropping the exposure two stops but they are drastically moving away from the pinks in the skin and into an uncomfortable jaundiced yellow. Yikes.

Let's do the same but now the other way around. On the left we have the normal (+0 EV) exposure and on the right the underexposed image (-2 EV) on the left that I’ve brought up +2 stops in LR to match. I’m sure you’re with me on this, but the skin tone color is retained WAY better, and even though I’m a pretty pale dude, its at least “the same”

For the sake of testing, let’s go ahead and use the photos of the tree and see what happens. We’ll start again with the normal (+0 EV) exposure on the left and the overexposed image (+4 EV in this case) that I’ve corrected on the right. Even though I am still seeing some color shifts still, the WORST part is that I’m seeing massive clipping and data loss as well as a gross fading of the entire image.

Let’s move back to the underexposed image where on the left is the normal (+0 EV) exposure and the right underexposed image (-4 EV) on the left that I’ve brought up +4 in LR. Though there are shadow differences, the overall contrast of the image is retained. The data is still there and I feel pretty confident in what’s coming out of the recovery here bringing it up in exposure.

Here's -4 stops corrected:

Here’s -3 stops corrected:

Here’s -2 stops corrected:

 And lastly -1 stop corrected:

Findings from this make me think that underexposing, while it does introduce some grain, is pretty acceptable and surprisingly so at even 4 stops of recovery. 

Let’s move onto the overexposed side and see if there are acceptable margins of recovery there. To test this, I’m going to take our normal exposure and drop it down by -1 stop so that we have a basic understanding of what kind of recovery is even possible and what kind of data retention exists. 

Now, I was going to go through the whole rigamarole and do a similar comparison, between all 4 of the exposures, but after just testing recovery of the +1 exposure and moving it down 2 stops, the answer is pretty simply “don’t do this”.

Now stick with me, because this is a little jumbled. On the left, we have our normal exposure +0 EV, that in Lightroom I’ve adjusted to a -1 EV in Lightroom. On the right, we have the over exposed +1 EV that I’ve dropped down -2 EV in LR to match the normal photo’s appearance after it was adjusted. Aaaaand clipping and fading are both visible. Wow. 

So let’s pause for a moment and do a mid-way recap. What have we learned so far?

1. Underexposing and then recovering introduces grain but seems to keep color consistency for the most part. 
2. Shadow exposure values and some shadow color in the underexposed image AFTER adjustment of exposure in Lightroom differs very slightly from the normally exposed image.
3. Overexposed images when brought down to the normal exposure experience big color shifts that can drastically augment and change the perceived color from initial capture. 
4. Overexposed images also experience clipping and data loss when corrected in the same way.

What’s better? Normally this is where someone will tell you to look at the results and make up your own mind, but uhhhh no. Underexposed is the clear winner by a mile.

How well does an underexposed image work for highlight retention?

If the goal of over exposing an image and recovering is to have control and maintain shadow detail, then shooting underexposed is there to maintain the highlights right? 

Let’s start by looking at the blown out highlights from the normal exposure (+0 EV) on the left, next to the underexposed exposure (-4) on the right that we corrected in Lightroom. 

Both seem pretty similar, the bloom on the highlights is being treated very similarly with of course the underexposed side having more grain, but what happens when we drop both by a -1 EV in LR? Still looking pretty good though if you look closely you can already see a little more detail in the highlights of the underexposed image. At -1 EV, the normal exposure isn’t showing big signs of clipping yet either so I think as far as this goes -1 EV falls into an acceptable category. 

Moving down another stop (-2 EV from our starting point) I’m starting to see that clipping occur in the base exposure, and in the game of highlight recovery, the underexposed image is pulling ahead.

At -3 EV from the starting point It’s not really a contest now. The base exposure is starting to see color degradation and shifts in the greens and clipping is very visible. On the left, because it was underexposed and is still a full 1 EV away from its natural exposure it has plenty of image data to go.

After all those adjustments that base exposure is at -4 EV and the originally underexposed image is at its natural +0 EV level. The base exposure is having major contrast loss, color augmentation and is bad and sad and if images could talk, it would probably feel depressed.

Conclusion

Alright, so that was either a slog of testing and pixel peeping (or lots of fun for you fellow nerds out there), but what are my main overall takeaways?

1. Underexposing your images and then bringing them back up in post-processing is definitely the way to go. Sure, it might introduce a bit of grain, but the color consistency is pretty good and with noise reduction tech available in Capture One as well as the magic of Lightroom’s Denoise, it’s barely a consideration.
2. Overexposing and then trying to recover those highlights later is a nightmare. Funky color shifts coupled with clipping and data loss, you will want to cry (or make me cry for you). It's not worth it and not a pretty sight.
3. All of this, while worth having and thinking about, should come second when it comes to the photo you’re trying to make. If you’re shooting in bright high noon sun and you need that extra detail in the highlights, you know how to achieve that and how to fix issues that might come from it. Are you looking for a high key blown out look with minimal details then? Overexpose and push it to the limits.

If you too want to test your own camera's dynamic range limits, here's the quick and dirty checklist:

1. Set up a scene or find an environment with both dark shadows and bright highlights.
2. Use your camera’s exposure bracketing or manually take a series of shots at different exposures (-2 EV to +2 EV in 0.5 increments)
3. Import the images into your preferred editing software (such as Capture One or Lightroom)
4. Compare the underexposed, normal, and overexposed images and pay attention to what's happening with color, highlight/shadow detail and grain/noise.
5. Play with the underexposed and overexposed images to see what their retention is for that information before they start jumping the shark. 
6. Make a mental note of your camera’s sweet spot for exposure and recovery and then use that knowledge to get the best edit possible.

What works for me and my camera won’t be the same for you and its worth it to learn where that sweet spot for your gear where clipping, grain and color look right to you and your creative eye.

As for me, I’m going to start underexposing everything by a stop and see what happens. Maybe I’ll like it, maybe I won’t, but either way I’ll be sure to share my findings!