The pixel-peepers of DxOMark shared some of the interesting measurements and techniques that they use to judge the quality of the artificial bokeh of a smartphone, or blur. background in the photos. Not only is it difficult to do in the first place, but they must systematize it! Their guide should provide even experienced shooters with an overview of the many ways in which "bokeh computing", as they say, varies in quality.
Generally, the effect created by a SLR with a good wide open lens, leading to gently blurred backgrounds and light points of predictable shape, is the gold standard.
The pros and cons of the artificial background blur, found in most flagship phones nowadays, come from the base that they all have in two cameras. Using both cameras to capture information on a scene, then using this information to determine a depth map and scramble things beyond a certain distance or object, a fair simulation of the effect of the SLR can be created.
Look at this blurry background blur … photo by me.
But of course, it can be good or bad. There are telltale signs of having taken this shortcut, many of which have been identified by the DxOMark evaluation team and fit into their overhaul scheme. Some are expected, others are a little surprising. But they all appear in the rather crazy test setup concocted to cause these unwanted behaviors.
For example, you probably know that these artificial bokeh systems occasionally scramble pieces of pieces that they are supposed to keep sharp – a loop or hair, a hand or a nearby plant. Of course, true background blur rises gently on either side of the focal point, which means that the parts near the acute part will only be slightly blurred, while distant parts such as distant lights will be reduced to circular spots.
For a phone to simulate this, it must calculate a precise depth map for everything in the scene and make the blur progressively. This type of treatment costs time and battery, so few people do something like this. Still, that's what they should do if they mimic this optical phenomenon – so DxOMark notes them on.
However, this is only one of the many pieces of the puzzle, so read the rest and the next time you read one of the reviews of the site, you'll have a little more of the ###################################################################### 39, information on the origin of all these points.