Almost all modern day video and electronic stills cameras have the ability to change the brightness of the images they record. The most common way to achieve this is through the addition of gain or through the amplification of the signal that comes from the sensor.
On older video cameras this amplification was expressed as dB (decibels) of gain. A brightness change of 6dB is the same as one stop of exposure or a doubling of the ISO rating. But you must understand that adding gain to raise the ISO rating of a camera is very different to actually changing the sensitivity of a camera.
The problem with increasing the amplification or adding gain to the sensor output is that when you raise the gain you increase the level of the entire signal that comes from the sensor. So, as well as increasing the levels of the desirable parts of the image, making it brighter, the extra gain also increases the amplitude of the noise, making that brighter too.
Imagine you are listening to an FM radio. The signal starts to get a bit scratchy, so in order to hear the music better you turn up the volume (increasing the gain). The music will get louder, but so too will the scratchy noise, so you may still struggle to hear the music. Changing the ISO rating of an electronic camera by adding gain is little different. When you raise the gain the picture does get brighter but the increase in noise means that the darkest things that can be seen by the camera remain hidden in the noise which has also increased in amplitude.
Another issue with adding gain to make the image brighter is that you will also normally reduce the dynamic range that you can record.
This is because amplification makes the entire signal bigger. So bright highlights that may be recordable within the recording range of the camera at 0dB or the native ISO may be exceed the upper range of the recording format when even only a small amount of gain is added, limiting the high end.
At the same time the increased noise floor masks any additional shadow information so there is little if any increase in the shadow range.
Reducing the gain doesn’t really help either as now the brightest parts of the image from the sensor are not amplified sufficiently to reach the cameras full output. Very often the recordings from a camera with -3dB or -6dB of gain will never reach 100%.
A camera with dual base ISO’s works differently.
Instead of adding gain to increase the sensitivity of the camera a camera with a dual base ISO sensor will operate the sensor in two different sensitivity modes. This will allow you to shoot at the low sensitivity mode when you have plenty of light, avoiding the need to add lots of ND filters when you want to obtain a shallow depth of field. Then when you are short of light you can switch the camera to it’s high sensitivity mode.
When done correctly, a dual ISO camera will have the same dynamic range and colour performance in both the high and low ISO modes and only a very small difference in noise between the two.
How dual sensitivity with no loss of dynamic range is achieved is often kept very secret by the camera and sensor manufacturers. Getting good, reliable and solid information is hard. Various patents describe different methods. Based on my own research this is a simplified description of how I believe Sony achieve two completely different sensitivity ranges on both the Venice and FX9 cameras.
The image below represents a single microscopic pixel from a CMOS video sensor. There will be millions of these on a modern sensor. Light from the camera lens passes first through a micro lens and colour filter at the top of the pixel structure. From there the light hits a part of the pixel called a photodiode. The photodiode converts the photons of light into electrons of electricity.
In order to measure the pixel output we have to store the electrons for the duration of the shutter period. The part of the pixel used to store the electrons is called the “image well” (in an electrical circuit diagram the image well would be represented as a capacitor and is often simply the capacitance of the the photodiode itself).
Then as more and more light hits the pixel, the photodiode produces more electrons. These pass into the image well and the signal increases. Once we reach the end of the shutter opening period the signal in the image well is read out, empty representing black and full representing very bright.
Consider what would happen if the image well, instead of being a single charge storage area was actually two charge storage areas and there is a way to select whether we use the combined image well storage areas or just one part of the image well.
When both areas are connected to the pixel the combined capacity is large. So it will take more electrons to fill it up, so more light is needed to produce the increased amount of electrons. This is the low sensitivity mode.
If part of the charge storage area is disconnected and all of the photodiodes output is directed into the remaining, now smaller storage area then it will fill up faster, producing a bigger signal more quickly. This is the high sensitivity mode.
What about noise?
In the low sensitivity mode with the bigger storage area any unwanted noise generated by the photodiode will be more diluted by the greater volume of electrons, so noise will be low. When the size of the storage area or image well is reduced the noise from the photodiode will be less diluted so the noise will be a little bit higher. But overall the noise will be much less that that which would be seen if a large amount of extra gain was added.
Note for the more technical amongst you: Strictly speaking the image well starts full. Electrons have a negative charge so as more electrons are added the signal in the image well is reduced until maximum brightness output is achieved when the image well is empty!!
As well as what I have illustrated above there may be other things going on such as changes to the amplifiers that boost the pixels output before it is passed to the converters that convert the pixel output from an analog signal to a digital one. But hopefully this will help explain why dual base ISO is very different to the conventional gain changes used to give electronic cameras a wide range of different ISO rating.
On the Sony Venice and the PXW-FX9 there is only a very small difference between the noise levels when you switch from the low base ISO to the high one. This means that you can pick and choose between either base sensitivity level depending on the type of scene you are shooting without having to worry about the image becoming unusable due to noise.
NOTE: This article is my own work and was prepared without any input from Sony. I believe that the dual ISO process illustrated above is at the core of how Sony achieve two different base sensitivities on the Venice and FX9 cameras. However I can not categorically guarantee this to be correct.