Exposure and ISO

Colourful Xiaolanzhen alley lights at night
Xiaolanzhen, Guang­dong, 2016. I cap­tured this image using a Fuji­film X‑T2 cam­era with the XF 56 mm ƒ/1.2 lens. Since I was­n’t car­ry­ing my tri­pod at the time, my only option for cap­tur­ing an ade­quate­ly bright and sharp pho­to was to shoot using ISO 3200.

ISO simulates exposure

To demys­ti­fy the con­cept of exposure—to make it approach­able for beginners—many writ­ers present inac­cu­rate expla­na­tions of what ISO is and how it works. Unlike the aper­ture and shut­ter, which are phys­i­cal mech­a­nisms, the ISO describes an elec­tron­ic func­tion that sim­u­lates changes to expo­sure. ISO is not a vari­able of expo­sure because it doesn’t affect the amount of light the image sen­sor receives. Instead, ISO set­tings deter­mine how bright­ly the cam­era ren­ders a pic­ture giv­en the expo­sure you have set using the aper­ture and shut­ter speed. Thus, ISO lets you change pic­ture bright­ness with­out fur­ther adjust­ing your aper­ture or shut­ter set­tings or, if pos­si­ble, chang­ing sub­ject bright­ness by adding or sub­tract­ing light from the scene.

iso and effective exposure animation
In this exam­ple, both the aper­ture and shut­ter speed remain con­stant, leav­ing the ISO to change effec­tive expo­sure.

The stan­dard ISO scale is easy to remem­ber and fol­lows a sim­ple geo­met­ric pro­gres­sion: 50, 100, 200, 400, 800, 1600, 3200, 6400, 12,800, 25,600, etc. The avail­able range will vary depend­ing on the make and mod­el of your cam­era, and, as with aper­ture and shut­ter set­tings, inter­me­di­ate val­ues are typ­i­cal­ly avail­able. The dif­fer­ence in effec­tive expo­sure between adja­cent val­ues is equiv­a­lent to a change of one stop. For exam­ple, adjust­ing ISO from 200 to 800 quadru­ples your effec­tive expo­sure; switch­ing from ISO 3200 to 1600 halves your effec­tive expo­sure.

Many pho­tog­ra­phy instruc­tors describe ISO as an image sensor’s vari­able sen­si­tiv­i­ty to light. It’s a decep­tive­ly intu­itive attempt at under­stand­ing the process but also cat­e­gor­i­cal­ly incor­rect. There’s no com­mer­cial­ly avail­able, mass-pro­duced image sen­sor with vari­able sen­si­tiv­i­ty to light. Unlike the pho­tore­cep­tors in your reti­nas, which under­go chem­i­cal changes to become dark-adapt­ed, image sen­sors’ chem­i­cal and phys­i­cal prop­er­ties remain unchanged when adjust­ing the ISO.

ISO: image sensors and amplification

Under­stand­ing how image sen­sors work will help you appre­ci­ate what hap­pens when you adjust ISO. All image sen­sors are sen­si­tive to a rel­a­tive­ly nar­row tonal range of light, known as the dynam­ic range or expo­sure range. Your expo­sure must fall with­in this range to ensure that visu­al infor­ma­tion is recorded—not only faith­ful­ly and accu­rate­ly, but at all.

An image sen­sor is a dense­ly packed array of pho­to­sen­sors that detect light. Each pho­to­sen­sor, or pix­el, will accu­mu­late an elec­tri­cal charge when exposed to light, and the strength of this charge is pro­por­tion­al to the inten­si­ty of light at each pixel’s loca­tion. Imme­di­ate­ly after expo­sure, the elec­tri­cal cur­rent gen­er­at­ed by each pho­to­sen­sor pass­es through a sig­nal ampli­fi­er. It con­tin­ues towards the ana­logue-to-dig­i­tal con­vert­er, which dig­i­tizes the sig­nal, mak­ing it read­able by the camera’s micro­proces­sor.

It’s com­mon to raise the ISO when achiev­ing an ide­al expo­sure by oth­er means is not fea­si­ble. When you increase the ISO beyond the image sensor’s base sen­si­tiv­i­ty (typ­i­cal­ly1There are excep­tions. For instance, the base ISO on the Nikon D850 is 64, and it’s 160 on the Fuji­film X‑T3. ISO 100 or 200), the cam­era under­ex­pos­es the image sen­sor and then ampli­fies that sig­nal to ren­der an image of cor­rect effec­tive expo­sure. The extent of the image sensor’s under­ex­po­sure is pro­por­tion­al to how many stops your ISO set­ting devi­ates from its base val­ue. For exam­ple, if you take a pic­ture using ISO 3200 and your camera’s base is ISO 200, the cam­era will ren­der the final image from an expo­sure that’s four stops too dark.

ISO and total exposure, Toronto's financial district at twilight, from Riverdale Park East.
Toronto’s finan­cial dis­trict at twi­light, from Riverdale Park East, 2018. This series of images demon­strates the rela­tion­ship between ISO, effec­tive expo­sure, and how much total expo­sure the image sen­sor is receiv­ing rel­a­tive to its base ISO. Pho­to A is cor­rect­ly exposed for and shot at ISO 200, which is my camera’s base ISO. Below that, the pho­tos in the left col­umn (B, D, F, and H) are set at pro­gres­sive­ly high­er ISO val­ues, with the effec­tive expo­sure (bright­ness) main­tained by rec­i­p­ro­cal adjust­ments to the shut­ter speed. The pho­tos in the right col­umn (C, E, G, and I) demon­strate the cor­re­spond­ing lev­els of total expo­sure received by the image sen­sor before sig­nal ampli­fi­ca­tion ren­ders the pho­tos to their left. Thus, B is pro­duced using the expo­sure lev­els in C (1‑stop under­ex­po­sure), D is cre­at­ed from the expo­sure lev­els in E (2‑stop under­ex­po­sure), F is gen­er­at­ed uti­liz­ing the expo­sure lev­els in G (3‑stop under­ex­po­sure), and H is ren­dered with the expo­sure lev­els in I (4‑stop under­ex­po­sure).

Unfor­tu­nate­ly, rais­ing the ISO to increase pic­ture bright­ness increas­es image noise. There are three types of noise in dig­i­tal cam­eras: read noise, shot noise, and ther­mal noise. Read noise occurs when con­vert­ing the ana­logue sig­nal from the image sen­sor to a dig­i­tal sig­nal for pro­cess­ing. Dur­ing this process, the sig­nal is ampli­fied and con­vert­ed using elec­tron­ics, which can intro­duce ran­dom vari­a­tions in the mea­sured sig­nal. 

Shot noise is caused by the ran­dom arrival of pho­tons at the image sen­sor, which results in vari­a­tions in bright­ness at each pix­el dur­ing. Shot noise is typ­i­cal­ly high­est in low-light con­di­tions (dark­er expo­sures).

Last­ly, ther­mal noise is caused by the ran­dom move­ment of elec­trons in the image sen­sor due to tem­per­a­ture fluc­tu­a­tions. It’s main­ly prob­lem­at­ic for pho­tog­ra­phers oper­at­ing their cam­eras in hot envi­ron­ments, cap­tur­ing long expo­sures, or shoot­ing long video sequences. The high­er the tem­per­a­ture, the greater the ther­mal noise, which can result in ran­dom vari­a­tions in the sig­nal.

Image noise is present in every pic­ture you cap­ture regard­less of ISO; how­ev­er, the sig­nal-to-noise ratio (SNR) is high­est (mean­ing less noise) at your camera’s low­est ISO set­ting because they require bright expo­sures. If you raise the ISO to main­tain image bright­ness in dark­er shoot­ing con­di­tions, your cam­era ampli­fies a pro­gres­sive­ly weak­er sig­nal, which low­ers the SNR (mean­ing more noise). Thus, increased image noise is the most appar­ent prac­ti­cal con­se­quence of using high­er ISO val­ues. How­ev­er, it’s essen­tial to under­stand that on mod­ern cam­eras, increased noise is caused by reduced expo­sures and not by increased ampli­fi­ca­tion. 

ISO speed electronic noise examples
This series of images shows the effect of elec­tron­ic noise on image qual­i­ty. Notice how the noise affects the vis­i­ble sep­a­ra­tion between indi­vid­ual office win­dows on the left, and how it even­tu­al­ly over­whelms all hints of detail in black office win­dows near the cen­tre. To main­tain con­stant effec­tive expo­sure across mul­ti­ple pho­tos, every suc­ces­sive increase in ISO speed was rec­i­p­ro­cat­ed by adjust­ing the shut­ter speed to reduce expo­sure.