[noggin]

It doesn't do HDR or 10bit without dither

https://forum.kodi.tv/showthread.php?tid=332180

 wesk05 (who does full analysis of HDMI outputs using high-end HDMI test equipment) has posted that the Vero 4K appears not to have the 8bit+dither issue that other S905X platforms have - so this may be a kernel issue.

It's also important to remember that HDR EOTF (which is just a flag in the HDMI stream with some metadata in some cases), Colour space and Colour gamut are all independent of each other - though can be linked.  You also have bandwidth limitations on some devices that can be an issue.  This is a complex area, made more difficult by poor quality/incomplete documentation by SoC manufacturers it seems.

Bit depth can be : 8 bit, 10 bit and 12 bit (16 bit is also supported in standards)  Not all bit depths are supported at all frame rates / resolutions / colour spaces.

Colour space can be : 4:2:0 YCbCr, 4:2:2YCbCr and 4:4:4 YCbCr using Rec 601, Rec 709 or Rec 2020 YCbCr<->RGB matrices or RGB (which is 4:4:4).  The relationships between YCbCr and RGB are different in all three colour spaces - so you have to be careful not to treat Rec 601, Rec 709 and Rec 2020 YCbCr in the wrong way. Not all colour spaces are supported at all bit depths and frame rates - and supported combinations change with frame rates (<=30 vs >30).   (There are also non YCbCr standards appearing too - ICtCp is one which is nearer constant luminance. Dolby seem to be investigating it... YCbcYCrc is a constant luminance version of YCbCr but doesn't seem to be going anywhere?)

Colour gamut can be : Rec 601, Rec 709 or Rec 2020 - and is independent of the Colour space chosen to carry the signal. These define what actual colours the R, G and B primaries are in each system.  (Rec 601 and Rec 709 are essentially identical but Rec 2020 is very different and much wider- hence Rec 2020 often being called WCG - Wide Colour Gamut) You can map Rec 709 content accurately and essentially losslessly into Rec 2020, but you can't map full Rec 2020 into Rec 709 without losing information, using techniques like tone-mapping to work out how to cope with colours that are valid in Rec 2020 but which are outside the space that Rec 709 can carry.  

EOTF can be : unflagged (so default to SDR gamma), flagged SDR (so force to SDR gamma - BT.1886), flagged HLG (which is backwards compatible with SDR and uses ARIB STD-B67, also present in Rec.2100 EOTF), flagged HDR10 (which isn't backwards compatible with SDR, and uses a PQ ST.2084 EOTF (also part of Rec.2100) with static metadata using ST.2086 for things like max and average light level etc., plus white point mastering colour temperature etc) plus Dolby Vision and HDR10+.

(Some people use Colour Space to include both YCbCr representation and colour gamut - but it is helpful to remember colour gamut is a separate issue, and I know in broadcast circles gamut is used to describe the Rec 709 vs Rec 2020 difference)

Also you will often see people talk about 444 10bit support and HDR content - but 444 10bit doesn't exist at 2160p50 or 2160p60 in the HDMI 2.0/2.0a standards .

444 10 bit only exists at 2160p30 and below (so fine for most UHD HDR movies which are 23.976p- but not for HDR TV or HFR movies like Billy Lynn's Long Halftime Walk which are 50p or 59.94p). There is no official HDMI 422 10 bit standard at 2160p resolutions either. When people talk about 422 10 bit they are usually describing a 422 12 bit signal with no content in the 2 lsbs (so its 10 bit data carried in a 12 bit signal)

For 2160p at <=30Hz you have RGB and 444 support at 8, 10, 12 and 16 bit in the HDMI standard, and 422 support at 12 bit only.
For 2160p at >=50Hz you have RGB and 444 support at 8 bit only (so not really suitable for HDR), and 420 support at 8, 10, 12 and 16 bit (420 8 bit at 2160p50 and 60 squeezes into HDMI 1.4b bandwidth), and 422 at 12 bit only.