I tried to shoot a resolution test chart using 1080p, 2.7K and 4K at 50 fps and found that there is a small resolution improvements between 1080 and 2.7K but 2.7K and 4K look pretty identical.
Is it because 4K is simply 2.7 upscaled? Did anybody else do similar tests? Can anybody confirm or disprove?

My confidence in DJI can't fall because it is already at floor level but at least we won't waste SD card space and batery by freeing it from unnecessary upscale task

I don't have an action 2, but I know the difference between 2.7K and 4K usually is not worth the effort.
At least not with the DJI products I had.
The lens is not able to resolve that amount of resolution and will only add blurred details in unnecessary pixels.  

Agree, no matter whether it is binning or bad compression it appears that they should not claim their at least consumer products support 4K@50/60 as it seems to be deceiving customers.

No, Action2 4K is true 4K.
Action2 sensor output is 4.6K , after rocksteady it is 4K.

would u post the original video so we can check

DJI-ytao Posted at 11-25 05:23
No, Action2 4K is true 4K.
Action2 sensor output is 4.6K , after rocksteady it is 4K.

I don't doubt sensor resolution but another important parameter is the speed meaning how fast data can be read from the sensor and for cheap sensors this speed is low so that manufacturers do the binning trick i.e. reading every second row thus effectively reducing resolution at higher frame rates and do upscaling.
If that is not the case then probably the CPU unit is not able to encode using lower quality parameters that again may lower effective resolution.
Perhaps you know that better than I do since I see you are from DJI. I have already sold my Action 2 and deleted the test video but others can try luck shooting resolution charts

DJI-ytao Posted at 11-25 05:23
No, Action2 4K is true 4K.
Action2 sensor output is 4.6K , after rocksteady it is 4K.

Thanks a lot for the info, this is big. I am surprised something this important is not in the specs.

I would be surprised if it is hitting the diffraction limit past 2.7K as it has even a bigger sensor than GoPros, early Mavics etc..  There must be something else going on.

What bothers me the most is what seems to be a pincushion distortion (judging from the samples I have seen online) which I cannot make much sense of. Can you please post your chart captures here in different modes? I am sure people will appreciate.

At 24fps I can see slightly better details (sharper edges) in captured 4k footage compared to upscaled from 2.7k in Davinci Resolve 17. The OP was asking for 50fps, and there might be a processing/bandwidth bottleneck. So, the loss of detail at 4k50 might be due to higher compression?

After doing a couple of comparisons, I can confirm that the OP is wrong. There is a slight difference between 2.7k and 4k footage at both 24 and 48 fps.