I’ve been running a Spears and Munsil HDR10 colour-space pattern on a few devices and have noticed a difference between the Vero V, and also another Amlogic device, the Ugoos AM6B+ running CE, vs my Panny UB820. The test is design to show up issues in the conversion from 4:2:0 to 4:4:4.
Here is a shot of the test playing from the UB820 - note that the bottom right square has horizontal coloured lines, alternating in colour to match the burst pattern in the top left. The UB820 output is 10bit YCbCr 4:4:4.
With the Vero and the Ugoos, I get the following with their 4:4:4 output. While there are atefacts in these images arising from photographing a display, note that the colour has gone from the bottom right square. This is with “pure direct” enabled on my Panny OLED, which appears to stop the TV from resampling the 4:4:4 input (disabling pure direct degrades the pattern)
Having read this article (see the section High Frequency Detail) Choosing a Color Space, Ultra HD Edition I wonder if this means that the Amlogic devices have an issue stepping from 4:2:2 to 4:4:4 ?
I was also able to run the test on an ATV via Infuse. The output signal via that combo is 4:2:2, so the TV has to do the last step, and the test pattern via ATV/Infuse matches the Panny UB820.
Is there perhaps a command line I can use to force the Vero to max its chroma upscaling to 4:2:2, to see what happens? This area is beyond my technical knowledge, so welcome comments.
1 Like
You could try 422 in Display settings. I’d be interested in the result.
thanks, I should have mentioned that I already tried that (not expecting it to work), and indeed the output is worse. I think I read on this forum at one point that the 422 output option kicks in after the Vero has already upsampled to 444, so if the 444 has issues, then subsampling to 422 will make it worse. Here is a shot with 422 enabled, we now have loss of colour from the vertical stripes in the bottom left panel as well as the horizontal. So I wondered if there is a way to intercept the upsampling chain to get the 422 (or 420?) before it’s converted to 444.
Thanks for that. However 422 is generated, I guess two conversions 420->422->HDMI->444 or, worse, 420->444->422->HDMI->444 you are going to lose something. Whether its due to something in the deinterlacer (yes, that’s in circuit, even for progressive video) or an issue with the chroma subsampling geometry I couldn’t say.
I have a version of that S&M testcard but not sure of its parentage. The clip is 4k but said to be SDR and it displays at 1080p on my 4k Panny with the horizontal lines showing. But at 4k it’s worse.
Not going to be easy to fix.
Is there any update on this? Would be good to have this sorted.
In ten days, no…
What content are you having problems with?
Well Spears and Munsil flags it as an issue so that content for a start. I wasn’t expecting any fix as yet just wondered if it was on the list.
1 Like
