[_Shorty]

Well, at least I was wrong about v16 beta 4 with limited being even brighter. It was actually the DXVA test version you just posted that gets even brighter. Here's some brightness measurements I just took now. edit: The black image is a 1920x1080 bmp file with RGB 0-0-0. And all the measurements in Kodi are with the AVS709 black test video.

Black image shown on desktop in Windows Image Viewer
0.022 fL

v15.2 full range with AVS709 black test video
0.022 fL

v15.2 limited range
0.231 fL

v16 beta 4 full range
0.230 fL

v16 beta 4 limited range
0.229 fL

v16 DXVA test version full range
0.229 fL

v16 DXVA test version limited range
0.654 fL

So, given that, I decided to take a guess and measure a couple more images on the desktop. I made a couple of grey images to test. One with RGB of 16-16-16 and one with 32-32-32 to see if I get similar brightness measurements with those.

16-16-16
0.239 fL

32-32-32
0.719 fL

Something's definitely being altered incorrectly in Kodi v16, and especially in that latest DXVA test version. You would think that the test version and limited range would match the 32-32-32 value, since it seems to be converting black to 16 twice, but it did not. So I decided to test with at least one more image on the desktop, but ended up with two more because the first one at 31-31-31 was still brighter than DXVA-test-limited's black.

31-31-31
0.676 fL

30-30-30
0.658 fL

So that last value is close to the DXVA-test-limited value, which would suggest that for some reason that test version's black is being converted to 30 with that limited range setting turned on. In any event, both full range and limited range are incorrect in all v16 betas, including that DXVA test version. In the case of beta 4 both full range and limited range are outputting black as 16, when only limited range should be doing so. But v15.2 behaves as expected, with full range black being 0 and limited range black being 16.