selmslie Loc: Fernandina Beach, FL, USA

Nearly all digital cameras conform to the exposure triangle in that increasing the ISO is offset by an equivalent reduction in exposure. They also conform to Sunny 16 in broad daylight where 1/ISO seconds at f/16 will produce JPEG images that look the same at any ISO setting. Of course, there is also a limited range of shutter speeds and apertures and high ISO leads to less exposure and more noise.

A daylight scene at f/16 - 1/100s @ ISO 100, 1/125s @ ISO 125, 1/160s @ ISO 160, 1/200s @ ISO 200, 1600s @ ISO 1600, etc., will all look the same.

The actual digital values in the raw file and the raw histograms might be the same with a couple of exceptions that depend on how the analog to digital conversion (ADC or A/D) calculates the raw values (this can be different for different camera models).

Within the normal range the camera’s dynamic range (DR) changes with changes in ISO.

But the DR may not change in direct proportion to the change in ISO. This can be seen in the DR plots from DxOMark and Photons to Pixels. Adjustments need to be applied to overcome these anomalies and to get the images to look the same.

At whole stop increments (ISO 100, 200, 400 …) raw values change in [usually direct] proportion to changes in ISO.

For reasons known only to the manufacturers the intermediate ISO steps (125, 160, 250, 320 …) might not remain proportional so an adjustment value, let’s call it “ADJ”, is placed in the raw file to get the JPEG values to behave as expected (to conform to the exposure triangle).

ADJ can also be used to provide extended high or low ISO ranges beyond the normal range.

Some cameras like the iPhone and the Fuji X100T don’t actually have a full normal range in which the DR is directly driven by the ISO. The X100T can be set to any ISO from 200 to 6400 in 1/3 stop increments. However, above ISO 2500 the sensor’s DR no longer changes and all of the exposures rely on ADJ to arrive at the intended image brightness.

Cameras with an extended low range (below base ISO) do not get an increase in DR. They rely on ADJ to get the intended image brightness. Nikon warns us that the extended low range might result in blown highlights because the DR doesn’t change.

The bottom line is that ADJ can overcome many of the anomalies in the relationship between ISO and the camera’s JPEG. But if you are shooting raw your main concern should be to not blow the highlights while getting enough exposure.

selmslie Loc: Fernandina Beach, FL, USA

See also Photons to Photos

selmslie Loc: Fernandina Beach, FL, USA

Here is some more about the X100T and its DR:



You can see that something is different about the DR once the ISO reaches 2500.

Of course, it a lot easier to simply refer to Photons to Photons to find out much of this information. You can see whether the final step after the A/D conversion to create the camera's JPEG needs to do something to overcome the anomalies in the raw data.

But what happens with the JPEG? I took a series of exposures on aperture priority from ISO 200 through 6400 in 1/3 stop increments and examined the raw values and the JPEG values.



The red line plots the raw values for the green channel. From ISO 200 through 2500 they remain fairly consistent. This is the camera's normal ISO range. But above ISO 2500 the raw values start to drop. This indicates that the reduced exposure is no longer getting amplified. In effect, from ISO 3200 through 6400 the raw file is using ISO 2500 in the A/D converter.

The blue line indicates that the electronic shutter is functioning properly in relation to the ISO setting.

The green line shows the relative JPEG values for what the camera is trying to get to middle gray. It's also fairly consistent. The resulting JPEGs all look about the same.



Since the raw values are clearly dropping above ISO 2500 the JPEG is using an ADJ value of +0.3, 0.7, 1.0 and 1.3 to get the JPEGs to the same level. Since that ADJ value is saved in the raw file the raw converter on your computer can use it to mimic the camera's JPEG brightness even if you do not keep the JPEG. But as we know, that only gets you to a starting point. Everything you do after that on your computer starting with moving the Exposure slider simply builds on that.

selmslie Loc: Fernandina Beach, FL, USA

Here is a test run on the Z7. First, the information from Photons to Photos:



There appear to be two anomalies - below base ISO 64 and somewhere around and something else between ISO 320 and 400. Here is what I found:



In the extended low range (ISO 32-50) the Z7 does something similar to what the Z100T does above ISO 2500 only in reverse.

But there is no evidence of the second anomaly. The raw line is flat from ISO 64 trough 500. The JPEG line is flat from ISO 32 to 500. There is no evidence of anything odd happening around ISO 320 and 400.

The DR numbers used in Photons to Photos come from DxOMark but they only show the DR values at one stop intervals. I can only conclude that there is something wrong in the DxOMark data.

The only odd occurrence is the shutter speed reported for ISO 200, 1/80s. I repeated this test a couple of times and got the same result. We can probably assume that it should have been 1/100s. It's probably just reported wrong in the EXIF and on the camera's LCD, maybe a fluke, because there is nothing wrong with the raw or JPEG data at that ISO.

What I am showing are the actual results as recorded in the raw file and the camera's JPEG. I am not measuring the cameras' DR.

selmslie Loc: Fernandina Beach, FL, USA

Finally the A7 II which has been converted to monochrome by removing the Bayer array.

Photons to Phots shows that there are some variations in DR over the ISO range.



But these variations do not affect how ISO relates to exposure over the normal ISO range from base ISO 100 and up.



But the extended low range (ISO 50-80) is handled differently from the way that Nikon does it for the Z7 and other models.

The JPEG result is consistent from ISO 50 and up and the shutter performance is as consistent as the Nikon shutters.

selmslie Loc: Fernandina Beach, FL, USA

The conclusion we can draw from this is that, within these cameras normal ISO ranges, ISO and exposure show a relationship consistent with the exposure triangle.

Raw values are recorded directly from the sensor via the ADC and the camera's JPEG is generated directly from these raw values.

Only in the extended ranges is there is any intervention to maintain consistency between exposure and the brightness of the JPEG. Each manufacturer handles the extended ranges differently.

Each camera model has a different target level for raw and JPEG values but those targets do not change over the normal ISO range. They don't even change for the extended ranges for the JPEG.

The DR plots available from Photons to Photos and DxoMark have nothing to do with this. They are based on a determination of the range over which we can achieve a relatively noiseless image. The fact that many cameras do not show a perfectly straight decline in DR as ISO is increased does not affect the raw or JPEG values being recorded. They are simply the effect of noise reduction techniques being used in the ADC where different methods might be used at different ISO settings.