----------- Your trusted source for independent sensor data- Photons to Photos------------------------ Last revised: 2019-04-30

Previous Article----------------------------------- Table of Contents------------------------------------ Next Article

----------------------- Sensor Analysis Primer - WhiteLevel Study

--------------------------------------------------------- By Bill Claff

Introduction

There are a number of reasons why it's not possible to use the entire range of Analog to Digital Converter (ADC) values.
The ADC value that corresponds to saturation is called the White Level.
This value is stored in the Exif data of the raw file in the WhiteLevel tag.

Analysis

I studied WhiteLevel for over 200 cameras at all ISO settings.
When multiple WhiteLevel values were present I used the average across channels.

In each case I computed how many stops the WhiteLevel. value is below ADC clipping.

Here is a histogram of the overall results:

Note that over 60% of the values were less than 1/24EV below clipping.

Here is a full cumulative histogram of the results:

To better understand the left-hand-side let's look more closely at the values that are -1/6 EV or lower:

So, for example, only about 3.5% of the values are 1/3 EV or more below ADC clipping.

Conclusion

As we might expect the designers generally make a reasonable attempt to use the full range of the ADC.
It's likely that the more extreme values are the result of certain implementations of intermediate ISO setting and extended low ISO settings.
This is born out by quick inspection of the underlying data.

White level ought to be used as the high value in dynamic range and Full Well Capacity (FWC) calculations.
This isn't currently the case at PhotonsToPhotos so Photographic Dynamic Range (PDR) and FWC may be overstated.
That said, the percentage of published data that is overstated in any meaningful way appears to be small.