I have a Sony a7 (original). And on the camera it has a feature my old Canon DSLRs never had: it allows me to select ISO values LOWER than 100.

I understand the point of these lower ISO values: I take a lot of landscapes and sometimes need longer exposures in bright lighting conditions.

But do these lower ISO values have more noise, like the really high ISO values?

up vote 2 down vote accepted

They have lower level of noise, but also a lower ceiling for highlights. ISO 50 is effectively doing ETTR and then scaling back a stop. Whether or not the ceiling for highlights has been reduced too much is up to the scene and what you're trying to do.

ISOs lower than 100 on the A7 are not "real" in the sense that they don't lower the gain on the sensor, they just instruct the camera to increase exposure time as if the sensitivity was lower. The net result of this is reduced highlight headroom.

If you shoot RAW there is nothing really to be gained from any ISO setting less than 100.

  • Seriously? It doesn't affect the gain on the sensor at all? It just artificially increases the exposure time? So, if I shoot in fully manual mode (which I do a lot), will it make any difference at all? – theJollySin May 13 '15 at 16:13
  • 1
    @theJollySin You can check the histogram to verify this. You should see that it shifts to the right. But this is then still useful to reduce the noise, although you can just as well do that by manually increasing the exposure time. – Count Iblis May 13 '15 at 16:22
  • Footnote: maybe not nothing to be gained if you're in dire need of a slower shutter speed and you forgot to bring your ND filter. – inkista May 14 '15 at 18:30
  • @inkista only if you're shooting JPEG, if you're shooting RAW you can just keep the camera at ISO100 and overexpose if you're in dire need of a slower shutter speed, and get the same result. – Matt Grum May 15 '15 at 10:12

The answer is not that simple, also because of the compression scheme SONY are using. Technically, between ISO 50 and ISO 100 all that is supposed to happen is exposure meter shift, 1 stop. That is, ISO 50 shot is supposed to be ISO 100 shot overexposed by 1 stop, just as Mr. Grum wrote. However if one starts to split hairs and perform noise measurements, ISO 50 results in 0.3 EV wider dynamic range, compared to ISO 100; while at ISO 80 dynamic range is 0.3 EV less, compared to ISO 100. I have no idea why is it, but at least three people using different camera bodies got the same result. You can see it on Bill Claff's page http://home.comcast.net/~NikonD70/Charts/PDR.htm

  • Well, that's extra annoying. What use is the feature on the camera if it's so complicated and hard to predict? – theJollySin May 14 '15 at 1:54
  • @theJollySin it's important to note that you're unlikely to actually be able to see a difference of 0.3EV in DR, so the results are not exactly unpredictable, DR will appear to be pretty much the same at all settings between ISO50 and ISO200. As to why the feature exists, it's to allow for longer shutter speeds for people shooting JPEG, who would otherwise not be able to benefit from the extra headroom available. – Matt Grum May 14 '15 at 9:44
  • Looking at the graph on Bill's page, the only anomaly appears to be the ISO80 result, the others are all very close to 10.5 EV. I don't think this can be the result of compression (I've been studying their algorithm a lot recently) all that happens in the shadows is that values are shifted a bit to the right (i.e. they go from 14 -> 13 bits). My guess is that there's some sort of systematic error in how the RAW values are corrected digitally for for ISO 80. – Matt Grum May 14 '15 at 9:59
  • @Matt Grum : No, this particular thing is not the result of compression. Sorry if it was unclear. – Iliah Borg May 14 '15 at 12:33

A digital sensor only has ONE sensitivity, its one native sensitivity, probably rated ISO 100 for most DSLR today. All a sensor can do is to collect the light photons hitting its cells. It cannot attract more photons. :)

Then basically, all any higher ISO setting can do is to multiply the existing data values, shifting the data up in the histogram, which shifts the noise up with it. Low noise aids this effort, high noise makes it less feasible. ISO values lower than native (but these probably are Not technically actually called ISO) shifts the histogram down, shifting noise off scale to a degree. A darker picture needing more exposure. This shift does lower noise, but also limits dynamic range (since the range of the top part of the histogram is unused - however the camera does normalize it so it looks normal).

You can do these shifts in your photo editor too, however it is best done on Raw data, before gamma is added.

  • Actually, when you adjust your ISO on a modern digital camera from 100 to 800, a gain is applied. That is, the voltage is increased, causing the sensor to be more sensitive to light and detect photos of a lower energy. This gain also increases the noise from the sensor signal, so there is some cost-benefit. But increasing the ISO on a digital sensor actually does increase the number of low-energy photons detected, so the effect can not be exactly reproduced by RAW edit in post. – theJollySin May 14 '15 at 19:47
  • This answer is not correct. If all the ISO setting did was to multiply the values captured at base ISO with a given number the setting would not need to be set at the time of exposure and could just as well be set in post. This is definitely not the case. Look at this answer for example: photo.stackexchange.com/a/35141/21986 – Hugo Sep 7 '15 at 10:30
  • LOL. You are trying to use bad logic instead of the known facts. A sensor photo cell can only collect the photons available from any one lens aperture, which only passes the one value of light. A larger pixel area could collect more photons from it. Otherwise digital ISO does multiply the collection sum to simulate more light, but with greater noise of amplification. This is essentially same effect as lowering White Point in Levels histogram (that's what the raw Exposure slider does). You should read more, maybe from better sources. google.com/search?q=what+is+digital+iso – WayneF Sep 7 '15 at 14:12
  • @WayneF Be nice at photo.SE! Of course a sensor can not collect more photons than physically available during the exposure. The problem is that the ISO setting is not a purely digital setting. It affects what happens before the AD circuitry by applying analog gain. This is not the same as multiplying values that was obtained by the AD in post. Regarding the validity of the source I chose: Matt Grum is a highly trusted user on this site and in the answer he performed the tests himself and provided images showing the difference. – Hugo Sep 9 '15 at 3:43
  • A generic google search on the other hand is not a good source at all. – Hugo Sep 9 '15 at 3:45

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.