2016-12-15, 19:50
(2016-12-15, 14:00)flyingernst Wrote: NO x86 device, NO OS (Linux or Windows) has an 10bit oder 12bit video output. Whatever you decode, output ist maximum 4:4:4 8bit.
Are you specifically talking about video decoder output, or about video/gpu output to display in general? A year ago I could output YCbCr 444@10bpp from my nVidia GTX 970 when I was testing certain LG monitor. Another matter was I didn't have any 10bpp-enabled software at that time to test if it really worked. My understanding is that software has to use 10bpp-specific DirectX surfaces to be able to use 10bpp depth, which kinda makes sense. No idea though if the output from HW video decoder (H264, HEVC, VP9...) is capped at 8bpp even for 10bpp source. Maybe it is, but it doesn't make much sense. I'd expect that if you create proper 10bpp target surface for the decoder, then it should render 10bpp source in 10bpp depth.
Edit: just did a quick test with my current PC and it's showing 10bpp whole way up to the display surface, which is only 8bpp, because my current display is only 8bpp.
Screenshot: https://dl.dropboxusercontent.com/u/9162...enshot.png
I don't have time going through the hassle of hooking the PC to TV at the moment to see how it would look like with 10bpp video mode selected at nVidia control panel, but I suppose I'd then see whole path from decoder to display as 10bpp. What exactly do you mean by saying that whatever I decode, output is maximum 8bit? And sorry for highjacking this thread with nVidia-specific data. I'm not sure if the statement "NO x86 device, NO OS (Linux or Windows) has an 10bit oder 12bit video output" was generic, or only related to Apollo Lake.
