It’s theoretically possible to reduce the voltage-based sag on DD FET PWM output for levels in the middle of the FET’s range, using hardcoded estimates of what the curve might look like… but it’s not very practical.
The MCU doesn’t have a way to measure the LED voltage or current. It can’t regulate those because it can’t see them. So it would have to use hardcoded estimates.
But it varies per LED type. It might get flat-ish output on one LED, but buy the light with a different LED and now the output has a weird response curve. Making one better makes the other one worse, and the firmware typically doesn’t know what LEDs are going to be used.
With careful calibration though, it could theoretically make the medium-high part of the ramp have more consistent output over time, on FET+1 style lights.
However, the levels which could potentially benefit from this are generally high enough that they’ll be limited by thermal regulation, which would override any attempts at flattening the FET. Instead, it ends up with a different type of flattening… and regulating by temperature tends to keep the brightness reasonably stable.
Recent lights are also shifting more toward regulated drivers, which makes the whole thing moot. Wurkkos already has working prototypes for a boost driver.
Personally though, I barely ever use more than 100 lumens, and when I do, it’s in short bursts. So the levels I use are already regulated (even on FET+1), and regulation isn’t as important for burst use.