Earlier quoted context omitted.
Another department I doubt these traditional Japanese companies can catch up on is computational photography. I can take 5s handheld night time photos with adequate quality on a bunch of phones these days. There's still no way of getting anything useful out of DSLRs longer than 1/15s without tripods.
I'm not smart enough to do exact calculations with aperture, etc, but I think the amount of light taken in with a 1/15s exposure with a DSLR would be more than a 5s cameraphone image. Just taking the area of the lens, say a 5mm diameter cameraphone lens is pi x 5^2=79mm^2 vs a DSLR, conservatively 60mm diameter is pi x 60^2=11309mm^2. So 143 times the lens area, while your example exposure time is only 75 times more.…
Taking for instance a full-frame Canon 5D II and something like a Huawei P30 Pro on the back of a napkin. A full-frame sensor is 864mm2 and a 1/1.7in sensor is 43mm2. Huawei with a RYYB sensor has an ISO of 409,600 vs 6,400. The phone aperture is /1.6. Realistically, /1.6 glass would be too heavy, so let's call it /2.8.
The ratio would then be 20 (sensor size) * 0.0156 (ISO) * 0.57 (aperture) = 0.178x before AI kicks in for, say, another 75x more exposure time ~= 421x more usable light on the phone.
Based on super unscientific anecdotal experience between the 2 (since the human eye is logarithmic), it seems about right.