Hmm.
This is such complex stuff, the easiest solution is just to play with exposure time and look for a flicker compensation option somewhere (should have 50hz and 60hz and outdoor options).
To figure out exactly what is going on, lets think about this a little.
From the snapshot, it is clear to me that the camera has a rolling shutter.
We can assume the LED is being driven directly by AC mains power which operates at 60hz in the USA. The light would blink at double that frequency (120hz) because it would light up when the energy sine wave is at the positive peak and again at the negative peak.
We can graph the sine wave like this:
Conveniently, it looks like the "rolling shutter" takes close to 1 wavelength to finish, so it is around 1/60th of a second to fully read the sensor data. That is about 16 milliseconds.
If the exposure time is set to 1/500, then each pixel is exposed for 2 milliseconds. So it makes sense that the black bars would exist. It looks like the light is off about 40% of the time, which is ~6.4 milliseconds. But that is split across two off-periods, so each off-period would be around 3.2 milliseconds long, followed by light for about 4.8 milliseconds, then dark for 3.2 milliseconds, and so on.
So in theory if you got the exposure up to about 3.5 milliseconds (1/286 second) or longer, the fully black areas of the exposure should go away, but the image would still flicker horribly.
Around 7 millisecond exposure (1/142 second), the flicker should be greatly reduced.
I suspect 1/120 exposure would be pretty safe to avoid the flicker.
I'm not 100% certain the flicker is caused by AC power. If the edges between light and dark were a little more fuzzy, I would be more confident. There remains the possibility that the light fixture is converting the AC power to DC and using relatively slow PWM dimming, such that the power graph would look more like this.
Anyway, I don't think that matters much. Based on my crude analysis I think 1/120 shutter speed should be slow enough to mostly eliminate the flicker.
