Actually, mirorless cameras CAN generate a depth map from their phase detection sensors. They also can take multiple frames and stack them automatically, eithe

Actually, mirorless cameras CAN generate a depth map from their phase detection sensors. They also can take multiple frames and stack them automatically, either for higher resolution or lower noise. But the thing is, there's no reason you would want 100 raw frames in 100ms (which no phone can do, that's 168Gbps of bandwidth, the sensor interface can only do around 8-12 Gbps), instead of a single 20ms exposure from a large, high resolution sensor. A 35mm sensor will, in that 30ms exposure, accumulate around 20 times as much light in the best conditions for the phone (normal focal length) taking 100 pictures in 100ms. If you're using the wide angle or the "telephoto" lens, now the big sensor is now accumulating anywhere from 40 to 200 times more light. The big sensor can have three times the resolution or more, and actually make full use of that resolution. At that quality, no amount of image stacking can improve the resolution unless you have a 2000$ lens and a tripod to use sensor shift mode - which a mirorless camera can also do. In the end, even if your algorithms are perfect, you will always have a much lower resolution and much higher noise on the phone, no matter how much image stacking you do. At that point, you improve by using AI to replace the imperfections with plausible looking generated replacements. You can also do that well on a computer, but in most cases this only makes it worse, because there aren't the imperfections to begin with. If you decide to remove frames without motion blur and imperfect faces, now you go from 100 pictures to stack to maybe 10. At that point, the critical parts of your picture will only have 8ms or so of actual exposure vs 20ms on the MILC. But the 20ms exposure also has 50-300x more light in it, so now your phone camera is way behind. So it ends up being better to just crank up the shutter speed to 1/200 and get a 5ms full sensor exposure, as the critical parts of your picture will have more light, less noise, and more detail than the stacked version you take on your phone.