2018-08-09, 02:11
(2018-08-09, 00:41)wesk05 Wrote:Thanks. We're getting claims that Vero 4K / 4K + can't do 'proper' 10-bit quite a bit now.(2018-08-08, 14:00)Mount81 Wrote: It would be informal and clarifying to read more specific and concrete regarding the Amlogic vs proper 10bit output issue. The info fragment I'we found so far were very contradictious and mind-scramblingly evading or mystifying oppositely any assurance.
And what is the source of your presumption, that ATV 4K (or just the MrMC) has some limitations with HDR output?
I know this whole thing about S905X not having true 10-bit processing started with a comment from BayLibre developer. I have checked the output on Vero 4K and it does seem to have a true 10-bit output (or is employing a magical dithering algorithm).
If you are sure that your TV can do proper 10-bit rendering, then you can quite easily check whether the output of a device is true 10-bit or not with the Quants2D test patterns - http://test.full.band/index.html. You really don't need to have any expensive equipment to do this.
If it's true 10-bit, you would see 450 distinct shades of gray squares (this would suggest 1-bit LSB accuracy; you will need an analyzer to confirm it). Something like shown below. In this picture, I have changed the brightness to visualize the distinct squares.
If the output is truncated/dithered 8-bit, then you would see something like shown below:
As for the question regarding ATV 4K, it has been known for quite some time now that tvOS replaces the original HDR10 metadata with its own generic metadata. This could potentially alter the tonemapping on at least some displays that actually considers HDR10 metadata.
I sent two units out to two BayLibre developers that are involved in mainlining efforts, so they should have the capacity to test.
(Just to clarify): we're not of the opinion that there is an issue after extensive testing.
Sam