how to read a histogram and set exposure

most grading advice starts with looks. it should start here. exposure is the layer everything else sits on, and the histogram is the only part of your screen that tells you the truth about it.

this post covers what the histogram is actually showing you, how to set your black and white points, and what you lose when you clip.

what a histogram actually is

a histogram is a bar chart of your image's brightness. the horizontal axis runs from pure black on the left to pure white on the right. the height at any point is simply how many pixels in the frame sit at that brightness.

that is the whole idea. it is a count, not a quality score.

the important consequence: a histogram throws away position. it knows you have two thousand very bright pixels. it has no idea whether they are a window in the corner or your subject's face. that is what a waveform is for, and it is why the two scopes answer different questions.

why you cannot trust your eyes here

your eye adapts continuously. spend ten minutes in a dim room and a correctly exposed image starts looking bright. spend the afternoon grading and your sense of neutral drifts without you noticing.

your monitor lies too. a screen running at a high brightness makes properly exposed footage look blown out, so you correct downward and hand over something dark. the histogram counts real pixel values, which is why it is worth learning to read even when the image already looks fine.

reading the shape

a few patterns worth recognising:

bunched on the left means a dark image. correct for a night exterior, a problem for a daylight portrait.

bunched on the right means a bright image. correct for a high-key product shot on white, a problem for most other things.

a tall spike at either extreme is the one that matters. a spike hard against the left wall means pixels are sitting at pure black. against the right wall, pure white. that is clipping.

a gap at one end means you are not using the range available. an image whose data stops well short of the right wall has no true white in it, which usually reads as flat or slightly foggy.

a log profile, incidentally, produces exactly that last shape on purpose. s-log3 and c-log3 footage sits bunched in the middle with gaps at both ends, because the whole point is to keep highlight and shadow information away from the walls until you grade it.

what clipping costs you

this is the part worth being precise about, because it is the one mistake you cannot undo later.

a pixel stores a number. in 8-bit, that number runs 0 to 255. when a highlight exceeds what the file can represent, it does not keep going, it stops at 255. a moderately bright cloud and a blindingly bright one both become 255, and the difference between them is not recorded anywhere.

so when you pull that highlight down in post, you do not recover the cloud. you get a flat grey patch, because every pixel in it holds the same value. the detail was never written to the file.

the same applies at the bottom. crushed shadows go to 0 and stay there, and lifting them produces grey mud rather than the texture you were hoping for.

raw files give you more headroom, which is why they matter. a 14-bit raw file stores 16,384 levels per channel rather than 256, and typically holds usable highlight detail one to two stops above where a jpeg would have given up. more room, but not unlimited room. clipped is still clipped.

there is a full post on what is and is not recoverable in overexposed footage if you are trying to rescue something specific.

setting exposure in practice

the order that works, using the controls in luttie's editor:

1. brightness first, for the general level. get the image sitting roughly where it belongs before touching anything else. watch the middle of the histogram move as a block.

2. blacks, for the darkest point. pull down until the left edge of your data just reaches the left wall without piling up against it. most images want a true black somewhere. without one they look washed out.

3. whites, for the lightest point. push up until the right edge approaches the right wall, and stop before a spike forms. specular highlights, a glint on metal or the sun on water, are allowed to clip. faces and clouds are not.

4. contrast last. now that both ends are set, contrast shapes what happens between them.

the distinction between step 1 and steps 2 and 3 is worth internalising. brightness moves everything together. blacks and whites move only the ends. reaching for brightness when you needed blacks is the most common exposure mistake there is.

scopes beyond the histogram

once the histogram makes sense, two more are worth knowing:

the waveform keeps horizontal position. a bright patch on the left of your frame appears on the left of the scope, so you can find which part of the image is causing the problem rather than only knowing that something is.

the rgb parade splits the trace into red, green and blue side by side. if the bottom of the blue trace sits higher than red and green, your shadows carry a blue cast. this is the fastest objective way to spot a colour cast, because the scope does not adapt the way your eye does.

you can read all three for any image in the free histogram and waveform tool, which measures clipping percentages for you and runs entirely in the browser.

where this sits in the grade

exposure comes early for a reason. white balance decides what counts as neutral, exposure decides what counts as black and white, and every creative decision after that is made relative to those two. a look built on top of a badly exposed image inherits the problem rather than fixing it.

the order of operations post covers why some steps commute and others do not, which explains why moving exposure later in the chain changes your result while moving saturation does not.

open the editor or start at the beginning of the learning path.