"What do I need to know to understand dxomark.com data and its DxOMark Sensor scale?"
As a first step, please see “DxOMark essentials”. We have also provided answers to specific questions about the DxOMark Sensor scale below.
"Does the DxOMark Sensor scale factor in resolution?"
The DxOMark Sensor scale provides a measurement of sensor performance that is independent and equally important than resolution. Resolution is not factored in the DxOMark Sensor scale formula itself. However resolution is considered when normalizing the DxOMark Sensor scale; to that end, all metrics are normalized considering a defined printing scenario. Read more about normalization.
To best use the DxOMark Sensor scale, photographers should first consider the resolution that corresponds to their needs, then DxOMark scale is very efficient in assessing the sensor performance of cameras within the chosen resolution category. Read more about DxOMark Sensor scale
"How should the DxOMark Sensor scale and its metrics be used?"
Here are three steps to follow to make the best use of DxOMark Sensor scale:
- Identify your preferred scenario(s): General, Studio, Landscape, and/or Action.
- Choose the resolution you need for your photography usage.
- If you shoot in RAW, DxOMark Sensor will help you rank the best cameras according to the resolution and usage scenario that you have chosen.
Read more about DxOMark metrics
"How is DxOMark Sensor computed?"
As with any problem that deals with a very large number of parameters, each ranging over extended scales, one has to select fewer metrics that aggregate the various dimensions of the problem to deliver a simpler evaluation system. DxOMark Sensor is a scale that DxO Labs is proposing to the photographic community based on our experience and expertise. Using the same data, other experts could build a different scale and come to other kinds of conclusions.
The DxOMark Sensor scale is constructed with specific photographic scenarios in mind. From the many possibilities—portrait, wedding, landscape, photojournalism, sports, fine art and so forth—we have selected three, each one critically relying on a specific parameter or metric. We then derived a formula that balances the metrics of each representative scenario equally: Portrait and Color Depth, Landscape and Dynamic Range, Action and Low-Light ISO.
To be able to compare cameras having different resolutions, we have chosen to normalize the metrics against resolution by defining an “on print” set of measurements. If the chosen resolution of the print (8Mpix) is clearly arbitrary, DxO Mark scale has nevertheless been constructed so that the chosen resolution has no impact on the camera ranking – in other words, different print resolutions would have resulted in roughly the same ranking.
Finally, we have scaled DxOMark Sensor so that it spans ranks cameras using values ranging from 0 to 100, meaning that a value close to 100 corresponds to the highest quality medium-format cameras.
Since DxOMark Sensor is primarily a logarithmic scale, values should be compared in absolute terms, not in relative terms. For example, the difference between cameras rated 80 and 60 is the same as between cameras rated 50 and 30.
"DxOMark precision seems meaningless and misleading. What is the practical significance of the scale?"
DxO Mark Sensor scale is logarithmic, with a 5-point difference on the scale corresponding to a change in sensitivity of about 1/3 of a stop. We nevertheless report DxOMark Sensor Scale values to one decimal place to avoid the clustering of data that would occur with any arbitrary rounding.
"DxOMark Sensor is composed of objective and repeatable measurements, but the way they are aggregated is arbitrary!"
We have indeed arbitrarily chosen to balance landscape (Dynamic Range), portrait (Color Sensitivity), and action photography (Low-Light ISO) equally as we attempt to answer the needs of typical photographers. Depending on your personal use of your camera, you may want to weight these measures differently. For the sake of simplicity, we chose the simplest possible combination as an average of all three use-cases.
"How can I know if two cameras are noticeably different?"
Consider a digital gain of one f-stop, or similarly underexposing 1 EV at capture and then adding 1 EV at RAW conversion. This operation leads to a loss of 1 in Dynamic Range, 1.5 in Color Sensitivity, and 1 in log2 (Low-Light ISO), corresponding to a loss of 15 in DxOMark Sensor.
Therefore, you can consider that a 15-point difference on the DxOMark Sensor scale is roughly equivalent to a difference of 1 f-stop, and that a difference of 5 is equivalent to 1/3 f-stop. DxO Mark Sensor differences below 5 can be considered as not noticeable.
"Is DxOMark Sensor computed from on-screen measurements (1:1) or from prints (8Mpix)?"
DxOMark Sensor values are computed from the measurements from prints equivalent to 8Mpix. We chose “on print” measurement rather than “on screen” in order to better compare cameras having different numbers of pixels.
For example, consider two cameras: camera A (8Mpix), and camera B (16Mpix). As pixels of camera B will be smaller than those of camera A, it is expected that camera B will have more noise than camera A when images are viewed on screen at 1:1. However, it is always possible to downsample the 16Mpix of camera B to 8Mpix to reduce noise, with the result that the camera B’s images of camera B may be less noisy than camera A’s.
This example shows that comparing the image quality of the two cameras with on-screen 1:1 measurements would have been unfair to the high resolution camera, which is why we decided to compare the images under the same viewing conditions.
8Mpix print images mean that measurements can be normalized against resolution, so that all cameras can be compared fairly.
Even though on screen measurements cannot be used for comparing cameras with different resolutions, these measurements are still useful for assessing a camera’s absolute RAW image quality.
"Why is print size 8Mpix only?"
We have chosen a resolution of 8Mpix as our reference for normalization because most cameras today can achieve this resolution. In any case, choosing a greater or smaller reference resolution would have very little impact on the ranking but would exclude cameras with lower resolution.
"Is it fair that a 12Mpix camera can be rated as highly as a 24Mpix camera?"
It actually depends on the use case. If you make 8Mpix prints (20x30cm), then the comparison is fair. If you make 40x60cm or bigger prints, the 12MPix would be insufficient, so it makes more sense to use 24Mpix when there is a need for high resolution.
"Should I interpret DxOMark Sensor results depending on the way I use a camera?"
Before evaluating a camera, you should indeed define your own photographic use-cases. The print size that you typically prefer will determine the minimal resolution that you need. Once you know your required resolution, then you can use DxOMark Sensor to look at the best digital camera within the resolution category that you have selected.
If your main use case is to view pictures on screen at 1:1 scaling, then you are less concerned with resolution (so long as your screen has fewer pixels than your camera), which means that you can use the “Screen” measurement that dxomark.com provides.
"Some camera bodies have the same sensor, but DxOMark Sensor shows different results."
The sensor might be the same, but infrared or anti-aliasing filters can be different. Also, the sensor electronic controls may be different, leading to sensor output signals of different quality (and thus different measurements) between cameras.
"Is 100 the maximum possible DxOMark Sensor value?"
We have designed DxOMark Sensor as an open scale to cope with technology evolution that will continuously improve quality. It is therefore not capped at 100, which may at some point result in cameras showing DxOMark Sensor scores above 100.
"Cameras with ISO range starting at ISO200 seem under-rated."
To construct DxOMark Sensor, we considered the maximum values for dynamic range and color sensitivity. These values are commonly higher for lower ISO sensitivity, therefore cameras which do not feature low ISO settings (some start at ISO 200, for example) end up with a lower ranking compared to cameras with the same or even slightly lower performance at higher ISO.
"I tried some cameras with very good DxOMark Sensor rating, but I prefer mine with a lower rating."
A good camera should take good quality pictures, but there are many other factors to consider, with respect to camera usage, such as size, weight, autofocus reliability and speed, ergonomics, price, and so on.
DxOMark Sensor measures only the quality of RAW data, which is correlated with the output JPEG quality when everything else is held equal (optics and RAW converter). Therefore DxOMark Sensor does not necessarily reflect JPEG quality in all the different use-cases.
"Camera should be considered as a whole, measuring RAW only is pointless."
Final image quality obviously depends on the optics mounted on the camera body and on the quality of the RAW converter. With D-SLR cameras, optics can be changed at will, and there are plenty of different RAW converters on the market. As these factors are more or less uncorrelated, ultimately any one of these factors can represent the weakest point on which final image quality depends. DxOMark Sensor gives a way to more adequately assess the different parts of the camera relative to one another, which makes measuring in RAW a sensible thing to do.