In the context of cameras, dynamic range is the difference between the brightest and darkest parts of an image that a camera can capture. The greater the dynamic range, the more detail a camera will be able to capture in scenes with a wide range of light intensities. Dynamic range is typically measured in stops of light. The human eye can perceive about 14 stops of dynamic range in a single image. SDR displays, meanwhile, are limited to about 6 stops, while HDR displays can range from around 14 to over 17 stops. Because the output can vary so drastically, it's important to maximize the amount of information you capture in-camera to give you the most flexibility with your footage.
We added this test as part of our 0.13 test bench update, along with the related Luminosity Patch Detection test. For more information, see the test bench 0.13 changelog and the related R&D article, Maximizing A Camera's Video Dynamic Range: Sensitivity Vs. Specificity, A Choice With No Right Answers.
We had no equivalent test prior to test bench 0.13, so any cameras tested on 0.12 or earlier versions won't have these results.
0.8-0.12 | 0.13 | |
---|---|---|
Strict SNR Max Dynamic Range | ❌ | ✅ |
Lenient SNR Max Dynamic Range | ❌ | ✅ |
Background Floor Max Dynamic Range | ❌ | ✅ |
Learn more about how our test benches and scoring system works.
The goal of the video dynamic range test is to measure the number of stops a camera can capture between the brightest and darkest parts of a video at different noise thresholds.
Dynamic range is most important when filming high-contrast scenes, that is, scenes with great differences in light intensities, such as an indoor scene with bright windows in the background or an outdoor scene in harsh midday light with dark shadows. Cameras that can capture more dynamic range will preserve more details in the shadows and highlights of the scene. Details lost in the shadows are typically lost to noise, while details lost in the highlights are clipped into pure white. Shooting high-contrast scenes can be especially tricky in video, as opposed to photography, as the light in a scene isn't always consistent when recording moving images.
We test dynamic range using an illuminated Imatest 36-Patch Dynamic Range Test Chart with a DarkWorld Mask to block additional light. We perform the test in a light-controlled environment, with the doors closed and lights turned off.
Once the camera is set up and the test chart is framed correctly, we record 3-second clips at each of the camera's ISO settings, adjusting the shutter speed and/or aperture as needed to maintain the same exposure as determined at the camera's base ISO. We also include extended minimum and maximum ISO settings.
After recording our video clips, we run a Python script that executes a number of processes, which include isolating each patch on the chart as a region of interest (ROI) and using the selected ROIs to extract frames of all 36 patches for each ISO setting to line them up like so:
The Python script then analyzes each patch through Fiji and Octave to get their light intensity and signal-to-noise ratio (SNR), and then organizes that data into a graph, with a waveform representing the number of stops captured by the camera.
All of the lined-up patches and corresponding graphs are combined for each tested ISO setting and put into a slideshow for comparison.
We measure all of our data automatically through our Python script, which uses Fiji and Octave to analyze and organize the dynamic range stops. The lines on the resulting graph represent different noise thresholds. We define a Strict SNR of 16.67:1. In other words, for the strict threshold, the value of the noise in the patch must be less than 6% of the signal intensity. The strict threshold indicates the point at which noise just starts to become visible. Put another way, this is the number of completely clean stops a particular camera can capture.
For the Lenient SNR, we define the threshold at an SNR of 7.14:1, or a noise value of less than 14% of the signal value. The lenient threshold indicates how much usable dynamic range a camera can capture. If some noise is tolerable, you'll have this many stops to work with.
Finally, the background floor is determined by the mean intensity of the darkest patches on our chart. This is effectively the camera's base background noise floor. Any stops visible above the noise floor are included here, but that doesn't necessarily mean all of those stops will be usable, as some will be so noisy that most information is lost anyway.
Aside from shutter speed, aperture, and ISO, which are adjusted throughout the test, we use the following settings, or the next best setting available, when setting up the camera:
Sensor technology has continued to improve, with most modern cameras able to capture a wider amount of dynamic range than cameras released even five years ago. That doesn't mean dynamic range is a non-issue since capturing as much information as possible is still critical for professional videographers and cinematographers. It does, however, mean that for most video shooters, the camera itself won't be the limiting factor.
Still, there are certain things you can do to ensure you're getting the most dynamic range your camera is capable of capturing. If your camera supports Log recording, shooting in a Log profile will capture more dynamic range than using a standard picture profile. While some noise is inevitable, understanding exposure and properly exposing your videos will also ensure you aren't losing information to noise or inadvertently clipping important highlights.
Do you have any questions or suggestions concerning our methodology? Are you interested in specific details about our setup, process, or the philosophy behind our tests? We'd love to hear from you in the comments below. You can also email us at feedback@rtings.com.