By: Steve Bilow, Senior Product Marketing Manager
In our last post, “High Dynamic Range (HDR) and Why Light Properties Matter” we spoke about human light perception and color. Without that understanding, High Dynamic Range (HDR) video concepts would have no foundation. Now, having laid that cornerstone, let us move on to a general understanding of Dynamic Range.
Dynamic range is a central principle in cinematography and videography. It is also critical for describing and evaluating display technology. Dynamic range is the variance – typically expressed as either a ratio or a logarithmic measure – between the highest and lowest values that a varying measured value can have. It is not limited to the measure of light. In audio, the log base 10 form of the measure is called decibels. Concerning video and cinematography, Dynamic Range is the ratio between the brightest and darkest values in an image, commonly expressed as log base 2, meaning in bits or “f-stops”; or simply as a ratio such as 12:1 or 14:1.
Let us set aside audio and other contexts where dynamic range may be pertinent and stick with light. In this setting, if the brightest light in a scene or image is 1000 lux, and the dynamic range is 10 stops, then the faintest measured light will be 0.98 lux, which is 1000/210. That is approximately equal to a 1000:1 contrast ratio (assuming you are willing to make the leap of rounding 0.98 up to 1).
Human Dynamic Range Perception
Before we can say whether this is good or bad for a capture or display device, we need a reference point. Since we want to create realistic-looking images, the human visual system is a good place to begin. Compared to cameras and displays, dynamic range is an area where the human eye is often considered to have a substantial advantage. Under the right conditions, the eye can exceed 24 stops. That, however, is in situations where the pupil is continuously adjusting to light levels. Because camera apertures are discreet, it is perhaps more accurate to compare camera behavior with the human eye’s instantaneous dynamic range (where the pupil is momentarily not adjusting). In that case, the best cameras are not particularly different. Instantaneously, human eyes can see somewhere between 10 and 14 stops which is amazingly comparable to digital cinema cameras.
But it is still not that simple because our eyes are more accurate under low luminance; the eye’s dynamic range may still be considered superior with an instantaneous dynamic range in dark situations where it can exceed 14 stops. The point is that today’s cameras may not always exceed the dynamic range of the human eye, but they are remarkably analogous.
Camera Latitude and Dynamic Range
Now, let us consider cameras. When shooting video, dynamic range determines the level of detail that a device can capture within the scene from which it is creating an image. The greater the dynamic range the more detail can be captured in both highlights and shadows. Otherwise, shadows will lack detail, and/or the details in highlights will be “blown out” (beyond the brightness that can be captured). In other words, the broader the dynamic range the more highlight and shadow data that is retained by the image sensor.
An image sensor is an array of small light-sensitive cells, each responsive to a distinct color of light. To ensure purity, that photosensitive panel of cells is preceded by 2 types of filters: an IR filter to eliminate light beyond the visible spectrum and a Beyer filter array that passes only red, green, or blue light. This ensures that the color-sensitive cells only receive the color to which they can respond. After passing through these filters, the amount of light that the cells themselves can react to is called the “latitude” of the imager. It is measured in the base-2 logarithmic measure we referred to earlier as “stops” or “f-stops”. In other words, the dynamic range of a camera corresponds to the latitude of the sensor.
Displays and Dynamic Range
Cameras use “stops” to define dynamic range, so we know how to describe the way that photons become digital data in a manner corresponding to how cinematographers and videographers understand light. But, at the other end of the media flow, we must display something that humans can see. There, recall, dynamic range is described as the “contrast ratio.”
Television contrast ratio is the difference between relative darkness and brightness extremes. In this context, “dynamic range” refers to that ratio.
But, crucial to our perception of realism is the preservation of visible detail throughout that range. The notion of “detail” is paramount here because simply increasing the amount of space between the minimum and maximum intensity tells us nothing about the visible detail at the top and bottom of the image intensity range. Saying that a display functions from 100 NITS to 2,000 NITS does not say anything about what you can see in the shadows or whether specular highlights are, or are not, washed out. Whether “black” is at 0.1 NITS or 0.009 NITS is meaningless because it says nothing about how much information is visible in that “blackish” shadow or that bright streetlight. So, even at the consumption end of the media flow, dynamic range is only a piece of the story.
Here is an example from a TV Technology article by Jim DeFilippis from back in 2016. It illustrates the dramatic differences in shadow detail and also demonstrates that the issue has been discussed for quite some time.
Later, when we discuss “High Dynamic Range,” this preservation of detail will become crucial. For now, simply realize that not all “blacks” are black; nor are all “whites” white.
Next Up… Transfer Functions
It should be clear now that there is a complex path from photons captured by a camera to photons emitted by a display. To get from photons to data, the camera must employ an optoelectrical transfer function (OETF); from digital video to a display there is an electro-optical Transfer Function (EOTF). Since the entire process, no matter how much manipulation occurs along the way, is one from incoming photons to outgoing photons, the full path relies on, not surprisingly, an OOTF (Opto Optical Transfer Function). In our next post, we will explain those transfer functions and, specifically how dynamic range and gamma work into the equation.