Introduction


Over the years, selfie photos and videos have become more important in our daily lives, particularly due to social networks. Many different types of photos and videos are captured: single or group videos, group photos with friends at events; selfies with landscapes while traveling or even closeup selfies to document one’s mood, etc.

Depending on the scenes captured, the lighting conditions, and the environment, these can result in challenging use cases that are not necessarily well managed by all smartphones. To bring a comprehensive view on smartphone selfie performances, DXOMARK’s image-quality experts introduced  the DXOMARK Selfie protocol in 2019. Since then, we reproduce real-life usages and scenes in our labs and take pictures and videos in natural conditions.

To keep up with the latest technology trends and usages, DXOMARK regularly updates its protocols and continues to offer exhaustive and meaningful evaluations.  

Selfie Score Structure


50

hours of testing

for each smartphone 

2000

Photos shot and analyzed

Learn more about our Use Cases

Photo 70%

We evaluate the still image quality delivered by the front camera for individual subjects but also with duo and groups of people.

 

Tests are performed:

    • Using the default mode
    • Both in portrait and landscape modes
    • Handheld motion is simulated with a dedicated shaking platform.
    • In an extended range of lighting conditions from 1lux for lowlight to 1000lux for bright light conditions and even higher for actual outdoor scenes

Image quality is measured at various distances, representing various selfie use cases:

    • Close up portraits – at 30 cm, when the photographers’ face takes up the largest part of the frame and the rendering of the background is irrelevant
    • Self-portrait – at 55 cm, at arm’s length. The photographer remains prominent, but the background contains elements that the user wants to capture – a particular monument for example, this distance is also very useful for duofies, when two people are in the photo.
    • Group portrait – at 120 cm (with a selfie stick). Users wish to capture the group, but also as much as the background scenery as possible

To evaluate photo, assessments are made on a total of eight technical quality attributes:

Exposure

How well the camera properly adjusts to and captures the brightness of the subject and the background, as well as if the contrast is unusually high or low. We particularly focus on face exposure and contrast, as well as skin tone rendering, which are tested on a large spectrum from deep to fair to light

Color

For Selfies, we particularly observe the various skin tone renderings but also the overall color rendering and the white balance, which is intended as how accurately the camera can reproduce color in a variety of lighting conditions, as well as how pleasing its color rendering is to its viewers

Focus

How good the camera is at focusing on the subject at different shooting distances. We also evaluate the depth of field to ensure good sharpness even when you’re taking a group selfie photo.

Texture & Noise

How well the front camera renders the details of a captured scene, but also how much noise is visible in different parts of the image.

Artifacts

Any effect appearing in an image that is not present in the original imaged object (distortion for example)

Flash

It evaluates the overall image quality and the undesired effects observed when taking photos with the flash on. In the case of selfie, various technologies can be used for flash such as screen flash or classical LED flash. The protocol evaluates the final image quality without considering the technologies used.

Bokeh

How well the camera separates the subject from the background when taking portraits

Important note:

Our range of acceptable renderings for all attributes in various light conditions are based on the findings gathered from our multi-cultural focus groups. We believe that within this inclusive scope, the risk of bias is eliminated when evaluating a smartphone camera for a particular attribute. In keeping with this approach to testing, we do not evaluate manufacturers’ signature renderings.

Video 30%

We evaluate the video quality of the front camera for individual subjects but also for groups of people

Tests are performed:

    • Using the resolution setting and frame rate that provides the best video quality. If a front camera offers 4K video mode but uses 1080p Full HD by default, we’ll select 4K.
    • In an extended range of lighting conditions from 1lux for lowlight to 1000lux for bright light conditions and even higher for actual outdoor scenes

Image quality is measured at various distances, representing various selfie use cases:

    • Close up portraits – at 30 cm, when the photographers’ face takes up the largest part of the frame and the rendering of the background is irrelevant
    • Self-portrait – at 55 cm, at arm’s length. The photographer remains prominent, but the background contains elements that the user wants to capture – a particular monument for example
    • Group portrait – at 120 cm (with a selfie stick). Users wish to capture the group, but also as much as the background scenery as possible

To evaluate videos, assessments are made on a total of seven technical quality attributes. Even if there are similarities, some image quality aspects are unique to video. For instance, for video, we’ll particularly look at the device’s ability to maintain its quality level over a longer period:

Exposure

How well the camera properly adjusts to and captures the brightness of the subject and the background, as well as if the contrast is unusually high or low. Face exposure and HDR adaptations are evaluated for various lighting conditions.

Color

For Selfies, we particularly observe the various skin tone renderings but also the overall color rendering, which is intended as how accurately the camera can reproduce color in a variety of lighting conditions, as well as how pleasing its color rendering is to its viewers. Adaptation of the white balance and color rendering are systematically evaluated on various selfie/group scenarios.

Focus

How good the camera is at focusing on the subject at different shooting distances. For video, we particularly look at the ability to bring continuous focus.

Texture & Noise

How well the front camera renders the details of a captured scene, but also how much noise is visible in different parts of the image.

Artifacts

Any effect appearing in an image which is not present in the original imaged object (distortion for example)

Stabilization

How well the camera eliminates motions that occur while capturing video walking, running, etc.

Important note:

Our range of acceptable renderings for all attributes in various light conditions are based on the findings gathered from our multi-cultural focus groups. We believe that within this inclusive scope, the risk of bias is eliminated when evaluating a smartphone camera for a particular attribute. In keeping with this approach to testing, we do not evaluate manufacturers’ signature renderings.

Click to get a closer look

What we also Test


Leave a Reply