[timb64]

Camera manufacturers are all obsessed with extending the upper level of ISO,although in practical terms it is of limited use.My own K5ii has a limit of 12800,but whilst I have the odd "acceptable" image at 5000, from day to day I try to keep it at not more than 800.

It got me thinking about the other end of the range which is usually 100 (extendable in some models to 80).What are the technical/practical reasons for this lower level?Would it be possible to lower the ISO further to ,for instance,negate the need for ND filters?

Any thoughts/comments?

It's Saturday morning and I've obviously got to much time on my hands/mind!

[Alex645]

It got me thinking about the other end of the range which is usually 100 (extendable in some models to 80).What are the technical/practical reasons for this lower level?Would it be possible to lower the ISO further to ,for instance,negate the need for ND filters?

Digital sensors, not unlike film emulsion, have what is called a ʻnativeʻ ISO. That is the sensorʻs sensitivity without the need to raise voltage, or what videophiles call ʻgainʻ which increases sensitivity, but also increases noise. When the ISO is increased, the base line voltage is amplified. At the highest end of this range (often called Hi-1, Hi-2, etc) the amped effect is simulated.

To a lesser degree, with lowered voltage from the base line or native ISO, the sensor requires more light, and when beyond itʻs physical limitations thru software, can also be simulated (i.e. Lo-1, Lo-2, etc).

There are many articles on the internet that go into a more technical engineering detail about this, but in essence, itʻs similar to an electronic version of push or pull processing. The emulsion was designed and manufactured to work at an optimum exposure and process, but the exposure can be increased or decreased with the appropriate amount of under or over development or chemical processing (dilution or time).

Practical reasons would include slower shutter speeds to increase blur, larger apertures to decrease depth of field or increase bokeh, reduce contrast, reduce noise, increase tonality, and probably more esoteric benefits.

Lowering the ISO significantly to negate the need for ND filters would not really be a practical solution. Many of us that use ND filters need and want anywhere from 3-10EV reduction in light, and that would equate a 100 ISO native sensor to be set between ISO 12 to less than ISO 1.

There is a greater interest and demand for higher native ISO sensors than lower native ISO sensors.

No doubt the hardware and software one day will give us more uber low ISO options one day, but for now an analog filter is a cheaper solution with the help of long exposure noise reduction.

[Bob 256]

It has to do with the sensor's capacity for handling light. You can think of each pixel as being a small bucket which gets filled up with water when it's exposed to light. Only so much water can be held in each bucket. The size of each of these little buckets is built into the sensor through its design and won't be changed thereafter unless a better design comes along. Ideally, you want to set an exposure so none of the little buckets gets overfilled by the brightest area in the scene. All the other buckets will contain less water which is completely ok. For good sensors, the amount of water in each bucket can be read out to extremely small values (let's say 1/64,000,000 of a full bucket), and any where in-between that and a full bucket. This determines how many shades of gray you can have (a lot, for a good sensor).

Now to increase ISO, you give the sensor less light, that means all the little buckets have less water in them in proportion to how much the exposure is decreased (let's say for this example that is 1/100 of a regular exposure). It's a simple matter when the water in each bucket is measured, its value is just multiplied by 100 to get a regular looking exposure. If you increased the sensor base ISO by 1000, just multiply by 1000 and so on........

Now if you have your head wrapped around this so far, we'll continue. First, there's a trade off in increasing ISO. Since there is less water in each bucket and we compensated by multiplying, the number of levels that can be measured decreases by that multiplication factor. In other words, the smallest value was 1/64,000,000 of a barrel originally, but if we gave the sensor 1/100 the amount of light, filling each bucket 1/100th of it original value, and multiplied by 100, the smallest value now becomes 1/640,000 of a barrel and we have fewer levels (fewer shades of gray) that can be captured. This contributes to a decreased dynamic range for the sensor, more noise (for reasons I won't get into) and at some point, really bad photos.

But to get back to your original question about lowering ISO - You can't go the other direction because of the sensor. If the sensor is exposed to more light (meaning the barrel content values would be divided), some barrels are going to overflow and the brighter portions of the image would be lost. The sensor just can't handle more light, period.

A camera manufacture could cheat and set the point where this overflow happens at a lower ISO value (say 10) but then by the time you get to usable ISO values, image degradation would have set in. It becomes a matter of putting the best sensor performance (where only the brightest image areas are near filling the pixel buckets) at a reasonable ISO value, and in fact, this is constrained since it has to correspond to the standard ISO definition. So if a standard exposure (at a given lens opening & shutter speed) produces a good exposure, a certain ISO must result.

Actually you can get your camera to a lower ISO than its lower ISO limit. Take the K-1 for example which has 100 as its lowest value. Set the camera to 100 ISO and put a 1 stop ND filter on the lens - the camera now has an ISO of 50. Forget about all else and for all practical purposes, you're shooting at an ISO of 50. The lens has to open 1 stop more or the shutter has to slow to half its former speed value (or you have to move the camera's ISO to 200 - which is actually only 100 now).

Increasing the capability of sensors probably won't do much for the lowest ISO value being lowered, but it will allow better performance at higher ISO values and allow those to be extended (or at least result in better images at those higher values).

As you can tell by this lengthy diatribe, I too have too much time on my hands, but I hope this helps your understanding of what's involved

[photoptimist]

The ISO value of a sensor is a measure of the total amount of light (the number of photons in a pixel) it takes to create a certain image brightness value. Low ISO sensor means it takes a lot of light to get a bright image. The lower the base ISO, the greater the amount of light hitting the sensor pixels during the exposure.

Each pixel is like a tiny solar cell -- photons of light go in and electrons come out. The efficiency of the pixel to make electricity form light - called the quantum efficiency -- is a major indicator of performance. Better sensors convert a greater fraction of the light into an electrical signal. Yet that implies that better sensors actually have a HIGHER base ISO because they can get a good signal in dimmer light than previous generations of sensors.

Next, the camera must measure the electricity generated by each pixel. For a 24 megapixel camera that means taking 24 million electricity measurements. But, it's currently impossible to have 24 million wires continuously tapping 24 million pixels at the same time. Today's sensors read each pixel in each row, one pixel at a time and one row at a time. But that design implies that each pixel must somehow store the electricity generated during the exposure of the scene at the pixel site. The greater the capacity of pixel to store electricity, the lower the base ISO. Unfortunately, the pixel site is very tiny and can't store much electricity for various technical reasons. Worse, a meaningful reduction in base ISO requires staggering increases in pixel storage capacity -- getting from 100 ISO to 12 ISO implies 8 times the capacity. That is what is limiting the base ISO of modern sensors.

Maybe future innovations can boost the capacity of pixels although the limited need for lower base ISO and the fact that it is so much easier to add an inexpensive ND filter probably reduces the chance that sensor makers will aggressively pursue low ISO sensors.

[normhead]

Maybe future innovations can boost the capacity of pixels although the limited need for lower base ISO and the fact that it is so much easier to add an inexpensive ND filter probably reduces the chance that sensor makers will aggressively pursue low ISO sensors.

IN film days I probably shot 1000 rolls of film and not more than 20 were low ISO. And the ISO I have with a 100 ISO digital sensor is pretty good compared tot he results. Low ISO was used in film almost exclusively because the grain was finer in applications like micro-fiche where the goal was efficient storage of large amounts of information.

Large size hard drives and SSDs have pretty much solved that problem without going the low ISO route. In digital, it's a solution without a problem.

[timb64]

Thanks for all the replies,some of which leave my head hurting!
From memory back in the days of film wasn't there a 64 ISO film?Is the choice of 100/80 for digital just an arbitrary one,equivalent to the old Spinal Tap joke about making the top setting on your amp 11 because it's more than 10?

[AstroDave]

Here's a more numerical explanation of Bob256 and photoptimist's comments:

Clarkvision: Digital Camera Review and Sensor Performance Summary (scroll up and down for lots more info)(fairly heavy on canikon!)

The "full well" value is what really sets the low ISO behavior. Bob256 was a bit optimistic with his 64 million levels! It's more like 50000 - 100000 electrons. The quantum efficiency of a good photodetector these days is better than 50%, so almost every photon that shows up gets converted to an electron. The problem is storing all those electrons without their leaking into adjacent pixels.

[normhead]

Here's a more numerical explanation of Bob256 and photoptimist's comments:

Clarkvision: Digital Camera Review and Sensor Performance Summary (scroll up and down for lots more info)(fairly heavy on canikon!)

The "full well" value is what really sets the low ISO behavior. Bob256 was a bit optimistic with his 64 million levels! It's more like 50000 - 100000 electrons. The quantum efficiency of a good photodetector these days is better than 50%, so almost every photon that shows up gets converted to an electron. The problem is storing all those electrons without their leaking into adjacent pixels.

The problem of electrons leaking into adjacent pixel site was what delayed the K-3 when they went to the 24 MP sensor. If memory serves me well that was about a 6 month delay, and I'm guessing led in part to the creating of Pixel Shift.

[Tan68]

Thanks for all the replies,some of which leave my head hurting!
From memory back in the days of film wasn't there a 64 ISO film?Is the choice of 100/80 for digital just an arbitrary one,equivalent to the old Spinal Tap joke about making the top setting on your amp 11 because it's more than 10?

Some cameras with ISO ranges lower than normal have less dynamic range when those lower ISO ranges are used. I am talking about "expanded" ranges like LO-1 or whatever usually enabled in the camera menu.

Some cameras are turned up to 11 when low ISO range is used.
The K-5 has slightly greater DR at ISO 80 than at 100
I think the Nikon 8xx series has greater DR at its low ISO setting

[RGlasel]

leave my head hurting!

A little more pain, just to clear up a couple of misconceptions and because I'm naturally pedantic. My K-30 has a metering range of 0EV to 22EV, in terms of light intensity, that goes from slightly less than a quarter of a foot-candle to almost a million foot-candles. The photosites simply can't produce a range of voltages to match that range of light intensity and the circuits can't handle that kind of current variation either, so the camera's processor compensates by reading the sensor output several times while the shutter is open and a bias, or base, voltage is applied to the sensor's circuitry and fluctuations from the bias voltage is the signal used by the processor to produce a digital image.

ISO is simply a number provided by the user interface so the processor can calculate how long to keep reading the sensor. If the user interface provides an ISO value that requires exposure settings that result in a meaningless signal from the sensor, you get blown out highlights or noise. If the ISO value is changed after reading the sensor, that is just multiplying the sensor signal after it has been converted to binary data, The binary data is restricted to a fixed number of bits per pixel, so a limited range of colour and intensity levels are represented by that data and that limited range of information is then converted into an analog image by the display medium and the display medium will also be limited in the range of colour and intensity levels that it can accurately represent. DSLRs could allow you to select as low an ISO value as you want, it is just a number. Users want the best images to come from the lowest possible ISO value, so the camera designers have to trade off a lower "base" ISO specification for lower maximum ISO values that still produce usable images. The camera's lowest ISO setting is a marketing decision constrained by the physical limitations of the sensor.

wasn't there a 64 ISO film?

Even lower, ASA 25 film wasn't abnormally uncommon and I recollect reading about ASA 1 monochrome slide film. Film cameras can't adjust the sensitivity of the photosensitive elements in the film emulsion, so ISO/ASA values represent the actual sensitivity of the film to a given intensity of light.

[photoptimist]

Yes, there were a number of low-ISO films such as Kodachrome 25 slide file, Ektachrome 64 slide film, and Kodak Panatomic-X 32. It's worth noting that the quantum efficiency of film really sucks -- silicon is about 25 times more light sensitive than silver halide emulsion.

The choice of ISO for digital sensors is not arbitrary. It's determined by physics, the properties of silicon and other materials used to make sensor chips, and the chip design. It would be easy to make a sensor with much lower base ISO just by adding a thicker color filter array. But that would make the sensor worse for high-ISO and there's a lot more demand for high ISO than low ISO.

[timb64]

Yes, there were a number of low-ISO films such as Kodachrome 25 slide file, Ektachrome 64 slide film, and Kodak Panatomic-X 32. It's worth noting that the quantum efficiency of film really sucks -- silicon is about 25 times more light sensitive than silver halide emulsion.

The choice of ISO for digital sensors is not arbitrary. It's determined by physics, the properties of silicon and other materials used to make sensor chips, and the chip design. It would be easy to make a sensor with much lower base ISO just by adding a thicker color filter array. But that would make the sensor worse for high-ISO and there's a lot more demand for high ISO than low ISO.

I totally get there is more demand for high ISO than low,it's the extent to which the higher limits are useable/give acceptable images as referred to in my original post.

---------- Post added 05-26-18 at 06:22 PM ----------

Some cameras with ISO ranges lower than normal have less dynamic range when those lower ISO ranges are used. I am talking about "expanded" ranges like LO-1 or whatever usually enabled in the camera menu.

Some cameras are turned up to 11 when low ISO range is used.
The K-5 has slightly greater DR at ISO 80 than at 100
I think the Nikon 8xx series has greater DR at its low ISO setting

Part of the reason I have shied away from replacing my K5ii with a K3 or KP is the perceived DR of the 5 (that and lack of funds,perhaps I'm just clutching at straws to console myself!)

[Adam]

I totally get there is more demand for high ISO than low,it's the extent to which the higher limits are useable/give acceptable images as referred to in my original post.

The usefulness of higher ISOs is increasing over time. For example, with the KP you might find yourself getting quite decent results at 12,800 or 25,600, while the K-1 is good at 25,600 or 51,200. Pick up the 645Z and 102,400 isn't too shabby either!

Part of the reason I have shied away from replacing my K5ii with a K3 or KP is the perceived DR of the 5 (that and lack of funds,perhaps I'm just clutching at straws to console myself!)

There could be a valid argument comparing the K-3 and K-5, but the KP trumps both IMO. Here we see that the KP at ISO 102,400 is similar to the K-3 at 25,600:

Pentax KP Review - High ISO Low Light Performance | PentaxForums.com Reviews

Now I just wish they'd release the KP's image quality (or better) in a K-3 style body.

[normhead]

Now I just wish they'd release the KP's image quality (or better) in a K-3 style body.

I bet that doesn't happen until they've perfected a noticeably better AF system. I know they say the system in the K-1 MKii is better, but, no seems mohave actually noticed.

[Adam]

I bet that doesn't happen until they've perfected a noticeably better AF system. I know they say the system in the K-1 MKii is better, but, no seems mohave actually noticed.

Well, on paper the K-1 II's AF system is the same as the K-1's. I see it as more of a bugfix, since now the K-1 II essentially has the tracking performance of the K-3 II / KP.

I agree, what they need is a redesigned system rather than a small update to it. I'm hoping the K-3 III (or whatever it ends up being called) brings some exciting new tech to the table.