Reciprocity Failure Is a Film Problem, Not a Digital One

Type “reciprocity failure” into a search bar and you’ll find charts telling you to add anywhere from half a stop to two stops once your shutter speed crosses a few seconds. Almost none of them mention that the chart is for film. If you’re shooting digital, the correction those charts describe doesn’t apply to your camera at all - a 30-second exposure on a modern sensor gathers exactly 30 times the light of a 1-second exposure, no compensation needed.

Why film needed the correction and sensors don’t

Reciprocity is the rule that shutter speed and light intensity trade off evenly: half the light for twice as long gives you the same exposure. Film breaks that rule at the extremes. Each grain of silver halide needs a certain density of photons arriving close enough together to trigger a stable reaction; stretch the exposure out and the photons arrive too sparsely for some grains to ever cross that threshold, so the film records less total density than the metered time predicts. Push past roughly a second on most film stocks and you’re not just losing light linearly - you start losing it faster than the clock would suggest, which is why the correction isn’t a flat add-on but a curve (Kodak and Ilford both publish them: Ilford’s own datasheet gives HP5+ at a metered 10 seconds needing about 20.4 seconds actual, and the multiplier keeps climbing from there).

A digital sensor doesn’t have grains or a chemical threshold. A photosite is a bucket that counts photoelectrons - every photon that lands and gets converted adds one more electron to the count, whether it arrives in the first millisecond of the exposure or the last. There’s no threshold to fail to reach and no chemistry that gets tired partway through. That’s why CMOS and CCD sensors are described as having a linear response: double the exposure time, double the electron count, double the signal, at any shutter speed a camera will let you dial in. The exposure math from the exposure calculator holds at 1/8000s and it holds at 8 minutes - stops are stops, with no long-exposure exception to memorize.

What actually goes wrong on a long digital exposure

The image can still come out worse on a long exposure, which is exactly why the reciprocity myth persists - something is visibly different, so people reach for the explanation that used to be true. What’s actually happening is noise, not underexposure, and it has a different fix:

• Thermal noise (dark current). A sensor generates a small number of stray electrons from heat alone, with no photons involved. On a 1/500s exposure that’s nothing next to the real signal. Stretch the same sensor to 30 seconds or several minutes and the thermal electrons keep accumulating the whole time, showing up as a grainy texture and scattered bright “hot” pixels, worse in warm ambient temperatures and on older or smaller sensors.
• Hot pixels. A handful of photosites leak more than their neighbors regardless of temperature. Invisible on a fast shutter, they turn into bright red, green or blue dots that get more obvious the longer the shutter stays open.
• Amp glow. Some sensors show a faint glow near the readout electronics on very long exposures - a genuine light source inside the camera, not underexposure.

None of these make the frame darker. They add noise on top of a correctly exposed signal, which is why the fix isn’t more exposure, it’s noise reduction: long exposure noise reduction (LENR) shoots a second, equal-length “dark frame” with the shutter closed and subtracts its noise pattern from your image, astrophotographers stack multiple exposures so random noise cancels out while the real signal (the stars) reinforces, and cooled astronomy cameras exist specifically to push dark current down by chilling the sensor. All three are noise-suppression strategies aimed at the same electrons - none of them are compensating for lost light, because none was lost.

The one place “reciprocity” still applies to digital work

There’s a genuinely digital reason a long exposure can come out dimmer than the meter promised, and it’s worth not confusing with film reciprocity: rounding. Most cameras display shutter speed in whole- or third-stop increments, so a metered 47-second exposure might get rounded to a 50-second or 60-second shutter setting, or your intervalometer might only accept whole seconds. That’s a rounding error in the display, not a sensor behaving nonlinearly, and it’s the same order of magnitude whether your exposure is 1/4s or 4 minutes.

Where this actually matters day to day is stacking a strong ND filter for a long daytime exposure - the ND filter exposure time guide walks through converting a metered daylight reading into the correct shutter speed once you’ve cut the light 10 stops, and that math is pure stop-doubling with no reciprocity curve layered on top, because you’re still shooting digital. The same goes for a scene your meter can’t read correctly in the first place - that’s a metering problem, not an exposure one, and worth ruling out before you start chasing noise you don’t actually have.

What to actually do on a long digital exposure

Trust the meter’s math at any shutter speed - a stop is a stop whether it’s 1/1000s or 4 minutes, so there’s no chart to consult and no extra time to add. Then manage noise on its own terms: turn on long exposure noise reduction for single long frames when you can afford the doubled shooting time it costs, shoot in cooler ambient conditions when the shot allows it, and for astro or other noise-critical work, take multiple shorter exposures and stack them in post rather than one very long one. None of that changes what you dial into the exposure triangle - it changes what you do with the frame after the shutter closes.

If you’re the kind of photographer who wants this sort of thing sitting in your inbox instead of re-Googled every time it comes up, it’s the kind of note that shows up in our occasional field-notes emails.