20
\$\begingroup\$

So reading the questions I came across this Why do cameras include very high ISO settings even though they are so noisy? and my comment on this answer led me thinking, how do we qualify high ISO? Is it the top available ISO value? a percentile? Grain visible when viewed at a certian size?

Usually when speaking to someone you say high ISO, they know what you mean, you know what they mean and it doesn't need further explanation (to me high ISO is the highest setting I would be willing to go to and get a usable shot/the grain would be tolerable)

  • On my old Canon 450D (crop, circa 2008, ISO range100–1600) I would consider ISO 1600 high ISO.
  • On my Canon 7D (crop sensor, circa 2010, ISO range 100–6400) I would consider ISO 4000 high ISO.
  • On my Canon 5D2 (FF sensor, circa 2009, ISO range 100-6400, expandable to 12800 and 25600) I would consider 12800 high
  • On my Canon 5D4 (FF sensor, circa 2016, ISO range 100–32000, expandable to 51200 and 102400) I don't think I've got there yet (haven't needed to shoot above 25600, which was perfectly acceptable)...

When is something considered high ISO and more importantly is it quantative? Or is it all subjective?

\$\endgroup\$
8
  • 2
    \$\begingroup\$ Not an answer: I don't there's High ISO as a value for a camera. There may be only too high (or too low) for a particular situation. \$\endgroup\$ Commented Apr 18, 2018 at 9:40
  • 5
    \$\begingroup\$ High ISO starts at 1273.491, plus or minus a few 1000, depending on who you ask and what cameras can do this week. \$\endgroup\$ Commented Apr 18, 2018 at 10:17
  • 11
    \$\begingroup\$ This question is pointless. It is totally opinion based, and a moving target. I remember when Tri-X at ASA 400 was the "high speed" film. And don't forget "high speed" Kodachrome at 64, when the previous version could only do 32. Now I routinely use 1600 ISO as "normal". \$\endgroup\$ Commented Apr 18, 2018 at 10:22
  • 6
    \$\begingroup\$ @OlinLathrop I disagree. My question is can it be quantified or is it subjective, therefore an accepotable answer to this would be 'its totally opinion based and a moving target. So. Don't put answers as comments i suppose! \$\endgroup\$
    – Crazy Dino
    Commented Apr 18, 2018 at 10:26
  • 6
    \$\begingroup\$ See: Is it poor etiquette to downvote without leaving a comment? It's always nice when people give you constructive feedback, but absent that you should just understand a downvote to be the opposite of an upvote: someone felt that the question was poorly researched, or unclear, or not useful. Remember that votes in either direction aren't rewards or punishments for the author; they're an indication to the community of the quality of the post. \$\endgroup\$
    – Caleb
    Commented Apr 18, 2018 at 13:46

11 Answers 11

18
\$\begingroup\$

Being a man of science, I did some scholar-googling and came across some articles.

Too long, didn't read:

None of the scientific articles I found give a clear definition of "high ISO". However, they all link high ISO with higher noise levels. Therefore, I would say that high ISO is completely dependent on subjective criteria and the camera in question.


My sources:

From Tamer F. Rabie, "Adaptive hybrid mean and median filtering of high-ISO long-exposure sensor noise for digital photography" (2004):

Visible noise in a digital image is often affected by temperature (high worse, low better) and ISO sensitivity (high worse, low better). Some cameras exhibit almost no noise and some a lot and all the time. It has certainly been the challenge of digital camera developers to reduce noise and produce a "cleaner" image, and indeed some recent digital cameras are improving this situation greatly, allowing for higher and higher ISOs to be used without too much noise.

[...]

The resulting effect is the visibility of random noise artifacts in the acquired image that differs in severity from acceptable (at low-ISO settings ISO 2000) such that it becomes visually unacceptable.

[...]

One observation from the experimental results is the fact that different sensors exhibit different noise levels at the same ISO setting.

To me, that sounds like: "It depends on the signal-to-noise ratio (SNR), not on the ISO number."

From Y. I. Pyo, R. H. Park and S. Chang, "Noise reduction in high-ISO images using 3-D collaborative filtering and structure extraction from residual blocks" (2011):

Digital cameras are commonly used to acquire images in everyday life. Digital camera users want to get high-quality images at high-International Organization for Standardization (ISO) setting, in which ISO in a digital camera denotes the gain of image sensor. Images that are captured with low-ISO setting have little noise and vivid color with some blurring. On the other hand, images that are captured in low-light condition with high-ISO setting have higher gain of image sensor and are less affected by blurring, but more affected by noise than images with low-ISO setting.

[...]

Chrominance images have coarse-grain noise at high-ISO setting.

To me, that too sounds like: "It depends on the signal-to-noise ratio (SNR), not on the ISO number."

From Youngjin Yoo, HoCheon Wey, SeongDeok Lee, Chang-Yong Kim, "Profile based fast noise estimation and high ISO noise reduction for digital cameras" (2008):

In general, noise level is high in images taken by high ISO setting.

Once again, to me, it sounds like: "It depends on the signal-to-noise ratio (SNR), not on the ISO number."

\$\endgroup\$
5
  • \$\begingroup\$ I was seeking further clarification on how ISO is calculated (w.r.t. SNR) and was surprised to read all the intricacies involved, but at least in some cases it is directly calculated by SNR (and in other cases is more related to the dynamic range of highlights). Anyone reading this answer might want to check out en.wikipedia.org/wiki/Film_speed#Measurements_and_calculations \$\endgroup\$ Commented Apr 18, 2018 at 19:34
  • \$\begingroup\$ See, particularly, en.wikipedia.org/wiki/Film_speed#Discussion \$\endgroup\$ Commented Apr 18, 2018 at 19:54
  • \$\begingroup\$ @DarrenRinger So SNR-based ISO provides a standardized noise value - that would be nice to have, though a chart with proper SNR-values (that are based on psychophysics) and the correlating SOS-ISO would suffice, most of the time. However, it would be very uncomfortable as (sole) basis for exposure - Imagine shooting with both the EOS 5D II and 5D IV and having to remember that the IV's SNR decreases more linearly than the II's, thus it effectively has a steeper sensitivity-curve when changing between higher ISOs. Not that this is up for debate, but just saying. Thanks for bringing this up! \$\endgroup\$
    – flolilo
    Commented Apr 18, 2018 at 20:18
  • 1
    \$\begingroup\$ I like this answer a lot. It's sciency! \$\endgroup\$
    – Crazy Dino
    Commented Apr 19, 2018 at 8:42
  • 1
    \$\begingroup\$ @CrazyDino Hail science! \$\endgroup\$
    – flolilo
    Commented Apr 19, 2018 at 13:53
9
\$\begingroup\$

ASA Film Speed provides a simple enough answer for this. ISO 400 was considered standard high sensitivity film with anything above that being very high sensitivity film.

It would be easy to argue that anything above 400 ISO is High ISO. The real question is whether or not to include 400 in that or not. I would argue 400 should not be included as high ISO as its more of a standard. By many measures 400 ISO is the standard all-purpose speed which makes anything below it a Low ISO and anything above it (generally this would skip to 800 and above) would be High ISO.

Additionally, it is also recommended to begin with a general-purpose 400 speed film, such as Kodak Tri-X 400 or Ilford HP5 Plus, as these are some of the most flexible, user-friendly films available. ISO 400 film is ideal for handheld shooting in most outdoor conditions, many indoor conditions, and is also the most common film speed to push or pull during development.

Source: B&H Buying Guide: Film Photography Student

Further reading:

\$\endgroup\$
6
  • 9
    \$\begingroup\$ I don't think that comparison with film is really useful. It seems like saying "Well, a fast horse can run at 45km/h. Therefore, a 'fast car' is any car that can go at 45km/h or more." \$\endgroup\$ Commented Apr 18, 2018 at 15:38
  • 3
    \$\begingroup\$ @DavidRicherby there's a big difference when these were what the films were intended for. ASA 400 was the standard all purpose film. A horse running is a terrible analogy because there is no standard. Consider it more like lenses. 50mm is normal focal length with anything under it being wide and anything over being telephoto. \$\endgroup\$ Commented Apr 18, 2018 at 15:44
  • 4
    \$\begingroup\$ The question isn't asking about films: it's asking about digital cameras. The fact that ISO-800 is considered a very fast film is irrelevant to whether or not ISO-800 is considered a "high ISO" on a modern digital camera. \$\endgroup\$ Commented Apr 18, 2018 at 15:49
  • 4
    \$\begingroup\$ I was thinking exactly this when I read the question. Back in the day, sometimes ISO 400 was called "indoor" film and ISO 100 was called "daylight" film, and they were both marketed to regular consumers taking family photos, etc. ISOs above 400 were marketed differently as "high speed" or "sports" or "action" film and their use was sometimes seen as challenging enough that the target market was the serious hobbyists and professionals. 95%+ of Fotomat customers brought in film that was 100, 200, or 400. When we saw 1600, it was a red letter day. \$\endgroup\$ Commented Apr 18, 2018 at 16:17
  • 2
    \$\begingroup\$ @David A resounding "Yes" to the horse comparison. And of course it doesn't stop there. When we'll have the new organo-metallic quantum sensors we'll look back and say "Remember when we were young and took pictures on silicone and 20000 was considered high ISO?" \$\endgroup\$ Commented Apr 19, 2018 at 6:54
5
\$\begingroup\$

High ISO is a traditional expression from the film days. High ISO is 800 ASA and up. It's always been that way. When you reached the 800 ASA film, you were buying high ISO along with the attendant grain. You had to be much more careful with your exposure and your process.

In digital cameras, there's a specific phrase called expanded ISO which is accessible but not recommended ISO. After that it's up to the camera manufacturer to decide what ISO they stop recommending.

On full frame expanded ISO usually starts at 6400 ISO and up (some specific models like the Sony A7S family might consider expanded ISO above 51200 ISO but that's very high).

For APS-C, expanded ISO is also in the above 6400 ISO range now but previously began above 3200 ISO.

For Micro Four Thirds, those little sensors are noisy from 1600 ISO with expanded ISO beginning as low as 3200 ISO.

Expanded ISO is finally really moving up though. The Fujifilm APS-C sensors in the X-T2, X-T20 and X-H1 hold together right through 12800 ISO. Expanded ISO as it's camera specific is a moving target.

But as soon as you are at 800 ISO you are into high ISO. Just go and look at the charts for your camera (and other popular cameras) at DXOmark and see what happens when to dynamic range and signal to noise when you reach 800 ISO and above in comparison to 50, 100 or 200 ISO.

Here's a good MFT, APS-C and full frame camera, compared for dynamic range across ISO

400 ISO is the last ISO where most digital camera images still look their best. The drop off in dynamic range is already two full stops by ISO 800 for the Sony A6300. Signal to noise for the A6300 has gone from 44.4 to 36.4, already out of DXOmark's green zone. Color sensitivity holds up better but is below optimal perceptible levels by ISO 3200.

I've been careful to pick good and representative cameras here: the Sony A6300, Panasonic DMC-G80 and Canon 5D Mark III are considered among best of class for their respective categories (APS-C, MFT, full frame) and are all have been extensively used by pros and semi-pros. Outliers like the Sony A7S series don't change the rule. High ISO is 800 ISO and above.

Personally I won't shoot my Sony NEX-5T above 800 ISO, my Canon 5D above 1600 ISO or my Canon 5D Mark III above 3200 ISO. On each camera the deterioration in image quality is too perceptible at that point. Even looking at the Sony A7S which is tuned for ISO, a usable picture continues to exist at much higher ISO (say to 51200) but a lot of life has gone out of its picture by ISO 6400 (at that point, Sony A7S dynamic range has fallen from 13 stops to 10 stops).

High ISO, IBIS (in body image stabilisation) and fast autofocus should not be a replacement for good photographic technique. Tripods, pre-focusing and slower exposures still play a huge role in creating high quality images.

\$\endgroup\$
4
\$\begingroup\$

I don't think you'll ever get a majority to agree on what ISO is the lower margin for "high ISO". Cameras differ too much for that (we all know that photographers are rational beings...)
But as you say, most have a pretty good idea what it is.

But there are many factors that play a role. Of course, the sensor and camera electronics are omportant. But exposition also influences the result.

Then there's the final use of the image: you can get away with a lot of noise when all you have to do is show it on an average computer screen (it'll get somewhat averaged out when downsampling. If you have to show the same image as a double page spread (or an A3 print), you'll want the minimum noise possible.

Even the skill of the one doing the post-processing plays a role. With all the different options available, optimal noise reduction is not simple.

\$\endgroup\$
2
\$\begingroup\$

In the 1950’s I was a boy loading my first roll of color film, its film speed was 10. Since then, film speeds have climbed astronomically. Choosing the right film for a job is was no easy task. Most times your film holders and cameras were pre-loaded so changing film in the field was often prohibited. What I am saying, fellows and girls working this trade in the digital era have it easy. What ISO is high? Yesterday, optimum results dictated 100 ISO. Today it’s likely that 400 ISO will yield the same results. What happens when you elevate? Likely noise and some loss of scale.

The good news: Optimum ISO is a moving target because both the hardware and the software of the digital world evolve. I think the sky is the limit. On an early post I predicted that a noise free camera could emerge. This was criticized because physics tells us noise free is impossible. That may be true but the bottom line, is the level of noise present detectable by the observer? By the way, I label high ISO as being above 1000 (however ISO is a moving target).

\$\endgroup\$
6
  • \$\begingroup\$ Good input Alan, but I don't think high ISO was ever a moving target. High ISO was always 800 ISO and above. You'll see below in my answer including a couple of camera charts that even the best digital cameras still perform at their best at 400 ISO and lower. But you are right - back then 400 ISO was already considered fast film if not high ISO. \$\endgroup\$ Commented Apr 18, 2018 at 23:31
  • 2
    \$\begingroup\$ @Foliovision Your chart simply shows a near-linear decline starting below 100 ISO, with the possible exception of the Canon EOS where the decline shows a marked increase around 1000 ISO, which can be supposed to mark the upper end of the "usual" range. "High ISO" is very obviously a moving target (which consumer film could shoot at 50000 ISO?). I remember when Kodak's Ektar 1000 came out and enabled me for the first time ever to take night shots without tripod. The quality was probably worse than the EOS at 100k today. Not moving? LOL. \$\endgroup\$ Commented Apr 19, 2018 at 9:29
  • \$\begingroup\$ @Foliovision But what is the relevance of "high ISO film" to the question? "High ISO" might not have been a moving target in film days but, with ever increasing ISOs available on today's digital cameras, and ever-decreasing noise performance well above 400 ISO, why do you believe it's currently static? \$\endgroup\$ Commented Apr 19, 2018 at 11:46
  • \$\begingroup\$ As I mentioned in my answer High ISO is static. It's anything over 800. Cameras can perform better or worse in high ISO. I.e. a Sony A7S performs much better in high ISO than a Panasonic GH5 (even though they cost approximately the same). This sentence makes sense: "Sony A7S high ISO performance exceeds that of any other 35mm digital camera." If high ISO is a moving target it's impossible to compare high ISO between cameras. Compare how awkward it is to write: "High ISO on the Sony A7S starts at ISO 6400 while high ISO on the A6300 starts at 1600 ISO." \$\endgroup\$ Commented Apr 20, 2018 at 3:41
  • \$\begingroup\$ Peter, I've thought about this distinction some more. Another way of looking at High ISO is that it's the level at which the human eye can detect a change in ISO. That still happens after ISO 400. I.e. ISO 100 to 400 can be looked at full quality from a sensor but from ISO 800 you can see the difference between ISO 100 and 800 pretty easily. Like you I remember well the grain in high ASA film. Yes, the difference now is more striking and the quality sinks slower but the principle distinction between full quality and High ISO still starts after ISO 400. Interesting, esoteric subject. \$\endgroup\$ Commented Apr 24, 2018 at 22:12
1
\$\begingroup\$

I think you can go quantitative on it. Noise costs dynamic range. With ISO you get less and less dynamic range, and you can possibly specify when you have "enough noise to make it shitty" as a number - which is quantified.

Is it subjective? Well, the number may be, but again you can compare.

https://www.dpreview.com/news/4302149407/sony-a7r-iii-dynamic-range-improved-nearly-matches-chart-topping-nikon-d850

has a dynamic range chage for the A7RIII. Objectively - some ISO are worse than ofthers (like - why use 400 when 640 has more dynamic range?).

Objectively you can also say that if you have a specific quality in mind, and your camea is better, you can waste this dynamic range to get to the defined target quality.

So, while you can quantify it - the goal is subjective.

\$\endgroup\$
3
  • 1
    \$\begingroup\$ What compromises DR at higher ISO is the amplification off the sensor. If a camera's base ISO is 100 only at ISO 100 do values near full well capacity not clip. At ISO 200 anything over 1/2 full well capacity is clipped. At ISO 400, anything over 1/4 full well capacity is clipped, at ISO 800 it's 1/8, at ISO 1600 it's 1/16, and so on. As the maximum reading which doesn't clip is lowered, it becomes a lower and lower multiple of the fairly constant value of read noise, thus SNR is lowered. \$\endgroup\$
    – Michael C
    Commented Apr 18, 2018 at 17:45
  • \$\begingroup\$ Yes, except that in the real world it is not so easy. Did you even bother looking at the chart I provided: Hint: the A7RIII actually GAINS again at ISO 640, going back to the same level as the ISO 200 setting (after having been abou ta stop lower). \$\endgroup\$
    – TomTom
    Commented Apr 18, 2018 at 18:13
  • \$\begingroup\$ TomTom is right - the very latest Sony cameras have much shallower fall off curves at high ISO. Generally SNR falls off pretty sharply after 400 ISO. From 800 ISO one really remains below the peak of performance, hence the traditional answer - 800 ISO probably still applies. More detail in my answer below. \$\endgroup\$ Commented Apr 18, 2018 at 23:35
1
\$\begingroup\$

In the digital space, the term high ISO (the H marking under ISO) has often been used to describe what we more formally call expanded ISO or extended ISO. So under that definition, high ISO is any ISO setting in which the apparent brightness is obtained by multiplying the digital values read from the sensor, rather than by using an analog amplifier to raise the level of the signal.

The reason this matters is that digital values are quantized. Multiplying by two just means that the bottom bit (ignoring any dithering) will always be zero. You don't gain more information at the bottom, but you do through away information at the top (clipping). When you're converting to a lossy format like JPEG, extended ISO may still provide a real benefit, because you are going to be throwing away most of your bit depth anyway. When shooting RAW, however it is, at best, useless (and at worst, lossy).

The other common definition of high ISO, as others have talked about, is the point at which image quality suffers too badly from noise in lesser-quality cameras. This definition is entirely subjective and arbitrary, but I think of high ISO as anything above about 800. Better cameras do just fine way beyond that threshold; older/lesser cameras don't.

\$\endgroup\$
0
\$\begingroup\$

From my film experience, I generally consider high vs low ISO to relate primarily to shooting experience in relation to light and motion. Even though the ISO rating itself is quantitative, its selection is qualitative but carries certain risks.

Low ISO is useful when a narrow depth of field and/or motion blur is desirable. It carries the risk that camera shake could ruin the shot, necessitating the use of a tripod.

High ISO is useful when a larger depth of field and/or stopped motion is desirable. It also is useful for taking photos in low light without a flash. It also tends to not require a tripod except in low light or when using a long telephoto lens.

Traditionally, color quality and grain effects would lead photographers to carefully scrutinize their choice so they can minimize loss of quality. As noted so many times, however, digital sensors have reached the point where there is no practical loss in image quality for choosing a high ISO.

I am being intentionally vague about specific ISO values. Like any artist, the photographer needs to experiment and discover his/her own personal preferences for each scenario.

\$\endgroup\$
0
\$\begingroup\$

The ISO value is high when it allows fast shutter speeds in low light.

\$\endgroup\$
2
  • \$\begingroup\$ Define "fast shutter speed". define "low light". This question is not about "what happens when increasing ISO", but "what is the definition of high ISO". \$\endgroup\$
    – flolilo
    Commented Apr 19, 2018 at 19:17
  • \$\begingroup\$ @flolilolilo hmm, I should create two questions for that. \$\endgroup\$
    – Carsten S
    Commented Apr 19, 2018 at 19:19
0
\$\begingroup\$

ISO is a sensitivity per sensor area, roughly corresponding to the previous ASA film values defined per film area. It combines with aperture numbers in order to make shutter speed depend on scene brightness, aperture number and ISO regardless of what kind of sensor/film crop/area you use.

For pixel-peeping at images, the noise per pixel is interesting. Since much of the noise level is inversely proportional to the arriving light energy per pixel, the area of a pixel is rather important in determining the noise levels. More modern sensors tend to have higher efficiency and lower inherent noise (not dependent on the actual arriving signal). Larger sensors with the same pixel number have larger areas per pixel and thus lower noise per pixel, higher pixel counts again increase the noise per pixel.

So what ISO level corresponds to what photon (signal-dependent) noise level depends on sensor efficiency, sensor size, pixel count, and the signal-independent noise level may or may not be ISO-dependent (that depends on the sensor technology) but will also depend on pixel area, sensor quality, current sensor temperature (which depends on ongoing use for live view and its update frequency, of course the ambient temperature, and total sensor area and heat conductance to the environment, possibly aided by fans).

From film times, ASA800 and above count as "high ISO": the film grain is a material constant that is independent from what light you actually throw at it. In contrast, a digital sensor's "grain" is defined by its pixel pattern (though some sensors "bin" pixels at "extended" ISO ratings, a technique that Fuji's X-Trans sensors formalize in a semi-religious manner). The resulting noisiness is not as much a matter of the underlying ISO than of the actual exposure you meter for.

In addition with the large impact of sensor size and pixel count, a single answer as satisfactory as "ASA800 and above" is illusionary since the ISO values of digital cameras are not tied to similar area constants like film grain. And even ASA800 film material has improved significantly over the decades, making that particular threshold change its impact over time.

\$\endgroup\$
-2
\$\begingroup\$

The answer to this poor question is really simple. "High ISO" is an ISO value higher than what you use normally.1 This has always been true and will always be true.


1 Obviously that's a fuzzy and moving target, something which David Richerby expressed nicely with his "horse and buggy" comparison.

\$\endgroup\$
2
  • 1
    \$\begingroup\$ 'Poor question' which seems to have received nine answers? Some of which actually provide some solid answers without subjectiveness. \$\endgroup\$
    – Crazy Dino
    Commented Apr 19, 2018 at 8:46
  • \$\begingroup\$ @CrazyDino The most objective answer (Foliovision's charts) simply shows a mostly linear decline of dynamic range with ISO over the entire ISO spectrum; I suppose one could have guessed that (same goes for resolution etc.; that's the reason I found the question poor). The most-upvoted (and accepted, by you) answer, by contrast, correctly states explicitly that the term is subjective. \$\endgroup\$ Commented Apr 19, 2018 at 9:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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