Exposing to the Left vs. Exposing to the Right
Sunday, March 2, 2008 at 12:48AM Google this topic and you will discover a war of sorts, a raging disagreement between those who say you should overexpose digital photography as much as possible, referred to as “exposing to the right” since it piles up the histogram toward the right edge, and those who recommend the opposite: underexpose and create a histogram that is left-leaning.
You must expose to the left because clipping is bad, some say. Overexposure is a non-concept in digital photography. Even when shooting RAW, the rightmost edge of your histogram is a sharp cliff, and it must be avoided at all costs. While you can sometimes recover highlight information from a RAW file, you’ll never get much, and if you miscalculate and clip some highlights, the results will be un-film-like and harsh. Expose to the left and always be safe.
Others purport that you must overexpose your digital photography, i.e. expose to the right, because of how digital sensors work. Unlike film, which has a logarithmic response to light, digital sensors have a linear response. So while film grain is evenly distributed across perceptual values, sensor noise lives predominantly in the shadows. It’s easy to see why when you imagine a linear sensor trying to hold four stops of exposure—if 100% sensor charge is four stops up, then one stop down is half that, or 50%, and one more stop down from there is 25% and one more down is 12.5%. As Jason Rodriguez commented on my dynamic range post, fully half your chip’s sensitivity is devoted to the brightest stop you can hold. Each stop you drop from there doubles your noise. So to maximize your signal to noise ratio and get the cleanest image, you must overexpose as much as you can, to distribute your image across the chip’s least-noisy sensitivity range.
Like the raging war between the half-black, half white aliens in Star Trek episode 70 (oh yeah, I went there), this is a non-argument. Both philosophies are 100% correct, and should be in play in the digital photographer’s mind when deciding on an exposure.
It’s so simple to state the combination of these two philosophies that renders both extremes silly: You should expose as bright an image as you can without clipping.
Man, that’s so much easier. I don’t know why people put so much effort into the debate.
Let’s look at some images. Here are some trucks at f/11, 1/500:
(click images for larger size)
Here’s that same view at f/8, 1/250 (for a total of two stops brighter):
The first image is a about a stop underexposed, although it does have small highlights that are just barely being held. The second image is clearly blown-out, and appears clipped in the highlights, but holds nice shadow detail.
But these images are raw, so we have some flexibility. Here they are again, color corrected in Lightroom into a similar look:
They almost match, but if you look closely at the overexposed image you can see that, while Lightroom was able to recover a surprising mount of detail in the highlights, the backs of the white trailers are still a flat, featureless white, with abrupt, cyan-to-white transitions in shading. Almost worse are the highlights in the clouds, which reveal Lightroom’s desperation.
Meanwhile, in the shadows, the underexposed image is a bit noisy, whereas the overexposed image is cleaner.
But the difference is not very noticeable. In this case, it seems the advantage goes to the underexposed image. Had I opened up a stop I could have reduced noise in the grays by half, but I’d be missing some highlight detail on the foreground truck. My fear of blowing out caused me to expose to the left, with happy results.
And now the counterexample: Another pair of images two stops apart:
Another attempt to make them match:
And while in the second image one could say that the clouds are a bit clipped, and the sky a bit low saturation, this is hardly as noticeable as the noise in the shadows of the underexposed image:
So our second image would seem to indicate a victory for exposing to the right.
You’re probably way ahead of me on the conclusion: You cannot apply a blanket philosophy of underexposure nor of overexposure to digital photography. Instead, you must learn your camera’s nuances and seek the correct exposure for the scene—which will almost always be as richly exposed as possible without clipping. Sometimes the resultant histogram will be left-leaning, and sometimes it will be piled up to the right. Make the shot, not the histogram.
Reader Comments (16)
I've always thought the term 'exposing to the right' meant exposing until something hits the 99%. To avoid overexposure because clipping is bad.
Now you say that this is called exposing to the left.
Then what is exposing to the right? Pushing it until all shadows have no visible noise in them?
Clark, I think your definition of exposing-to-the-right is the common one—it's certainly the one I was paraphrasing at the top of the post. So take a look at the histogram for the first image (trucks at f/11). Would you call that exposed to the right? Most people at first glance would call it exposed to the left, but in truth it obeys the rules of exposing to the right—I opened up as much as I could without letting the small highlights clip.
Yet another example of the value of planning and preparation for the DV rebel. Pretty much any exterior shot (on an average day) is going to contain both extremes of the histogram. Proper planning (using Stu's example) tells me to avoid relying on the master and rather using a 2 camera shot, getting the most from the sensors. Solid article Stu.
Stu, I think I misunderstood the 'piling up on the right'. I thought by that you meant the line that appears when you let stuff overexpose and the white pixels 'pile up' on the right of the histogram.
I would probably have that large peak above the word 'f11' in the histogram lean more to the right of the histogram and sacrifice the tiny highlights when exposing to the right. But I guess that is a personal preference.
By the way, I love your blog. It is very inspriring, and so was your book.
I have to add that it is extremely rewarding to have found your blog and reading stuff that kept my head busy for years. Storytelling, color correction, editing, visual effects, dynamic range, the Red One, the list goes on and on.
BTW, a slight clarification in that with digital sensors, noise doesn't live in the shadows, nor is there "more" noise in the shadows. This is a little bit of a misconception. Digital sensors have a linear distribution of noise, meaning that the noise is constant throughout the linear range of the sensor. In other words the digital code values in the highlights have the same amount of noise as the shadows. The issue of course is that in the highlights you can't see the noise as much (or at all) because the noise is disproportionate to the amount of signal (the human eye can't see the noise in the highlights because we can't dicipher minute changes in levels in bright images where-as we can in dark images), but it's there, and in a photometrically linear image it's the same amount in the highlights as in the shadows. The problem though is that after gamma correction, the noise in the shadows has gain applied to it, so it becomes more visible. A noise distribution value of +/- 3 or 4 in the shadows can become +/- 12-16 or more depending on the amount of gain applied (REC709 applied an approximately 4x gain in the shadows, LOG even more). The highlights of course after gamma correction don't have nearly as strong a gain applied . . . so after gamma correction noise is disproportionately visible in the shadows than the highlights (the noise in the highlights is actually compressed).
BTW, I do want to add that this has been a really great series of posts, and Stu, it's great to see you outline these somewhat counter-intuitive concepts of proper digital exposure so clearly. RAW is definitely a different animal than working with compressed video, but these posts have nicely outlined the power that's under the hood :)
Thanks!
Random question/idea. Given this current thread of ideas concerning dynamic range I was wondering what ideas some of you may have had in this arena. What are the different approaches and ideas that could be undertaken to develop HDR based video cameras. I am just talking acquisition here for the moment. I could see two scenarios right off the bat. One would involve putting a beam splitter right in front of the sensors and splitting the light onto 3 or more sensors which each have a variable sensitivity rate. This would solve the problem of shooting moving subjects on video with different shutter speeds as is common with still image HDR acquisition, and would allow for different exposures to be recombined resulting in a video HDR. The other that jumps out is the idea of having the camera shoot say 120 frames per second and in set of 5, change the ISO of the sensor so that each frame would have 5 different exposures with which to derive the HDR data. The obvious downsides being that you would be limited to shoot a shutter of 1/120 and each frame would not be necessarily identical resulting in a certain degree of sharpness reduction per frame. However, perhaps an algorithm built into the camera could utilize some sort of optical flow based pixel tracking to better combine the set of frames and also allow you to simulate the motion blur that would have happened if you wanted to shoot a normal shutter speed; ie 1/48(180 shutter)..ect. I was wondering what other ideas are out there towards this end, as I would see some sort of HDR based image acquisition as the logical future of digital video cameras. I realize some of this would take an incredible amount of CCU power and we might not be there yet, but just talking conceptually. Once the images are acquired then the problem is in reducing flicker via wavelet or gradient based tone mapping. It would appear though that work has already begun in this arena which is pretty cool. Anyway all just a thought.
Chris LeDoux
Typically with digital sensors the S/N improves by "exposing to the right" because the statistics of temporal noise make it converge to a fixed number given a sufficient time.
Hence, noise which appears random in shadows, "smoothes" out with in increase in exposure value (EV), whether, brought by an increased aperture size, or by larger integration time.
-- Joofa --
Jason,
DAC's are less accurate at deciding what is signal versus noise as the signal becomes weaker, or closer to the signal-to-noise "floor." When talking about an image, that floor is the shadow area, so yes, global low light situations, or even the shadow areas of an image have more noise. This is pretty easy to see in the data of a typical image
a brighter image, like a brighter musical note, is more pleasing to the human senses.
Shoot, somebody better tell Caravaggio that!
In both music and painting Baroque is the exception to the "bright is beautiful" prejudice.
So if I understood it right, in a setting where there are more highlights than shadows you should underexpose or expose to the left, while in a setting where there are more shadows than highlights you should expose to the right or overexpose?
That would mean then that the best way to capture a high contrast low key lit scene with lots of shadows would be to expose to the right?
If so, with digitals lower dynamic range, if exposing to the right for such a look (overexposing as much as possible without blowing any highlights) the shadows would sure be much brighter than they should have been for the look to work. That would mean bringing them down in post. My understanding is that you would end up with a lower contrast image rather than the high contrast one you were initially shooting for. Specially when shooting compressed formats like HDCAM, DVPRO HD or even XDCAM I'm not sure that would be a good approach since you're limited in color correction in post by the compression of the format. Am I on the right track here? But noise and compression artifacts are really annoying and they appear most in underexposed shadows. If the look you are going for is a high contrast low key look and you want to keep info on both extremes of the spectrum, meaning no clipping of the shadows or highlights, it becomes quite difficult with digital. In a scene where you are shooting for a 6:1 or even 8:1 contrast ratio with lots of shadows, overexposing the shadows to avoid noise while protecting for highlights will produce a very low contrast image, totally different from the intended image, which was a high contrast low key image. No?
how would bringing down shadows reduce contrast?
if blacks are at 10 IRE and you bring them down to 0 then you increase contrast, yes?
if you use the gamma to bring down the midtones then you increase contrast, it's unavoidable that's what bringing down gamma does.
If on the other hand the picture is too dark and whites are already at 100 then bringing up the gamma will reduce the contrast.
Personally I have never heard a photo guru advocating pushing the histogram to the left - but relying on zebras (in the DV/HDV world) may well have this affect.
In run and gun style shooting (with limited supplementary lighting) completely eradicating clipped highlights will produce an underexposed image more often than not.
Thanks
this is the must logical answer about which side o-)
I ve ever read...