(2019-07-22, 06:43)wrxtasy Wrote: (2019-07-22, 05:08)Mount81 Wrote: (2019-07-18, 13:29)wrxtasy Wrote: Not sure about the FireTV Stick 4K..
Any of the HDR capable AML S905X, S905D, S922, A311D chipset devices running any sort of OS can do HDR -> SDR mapping at the Linux Kernel level.
But color and brightness outputs will not be 100% correct, you really need specialist software, like MadVR and tailored display mapping as described in this https://forum.kodi.tv/showthread.php?tid...pid2860458
Back to this: I didn't seen actually the other options (like Zidoo), but the Q10 Pro can do a really good and almost seamless HDR->SDR conversion, it works fine and with decent PQ on my FHD TV. I turning back to my previous question, that will (can) it be any better with those new Amlogic chipsets with Core Elec anytime? Or is this such a hardware or Linux kerner limitation, that couldn't be acceptably resolved?
It might look seamless, but I suspect you have just struck it lucky with a device, that can map and that closely matches your distinct display's picture output capabilities.
Anything else, like cheap AMLogic will very likely have strange colors and brightness outputs due to using generic tone mapping.
You really need to use tailored tone maps that matches each distinct SDR display capabilities. Anything else is just guesswork.
What do you base that assessment on @
wrxtasy ?
We have broadcast standards - BT.1886 or Rec.709 with 2.2-2.4 power law gamma for SDR (both have the same gamut, but BT.1886 specs a more precise EOTF, whereas Rec.709 doesn't, so 2.2-2.4 power law gamma is assumes) - and TVs display these reasonably consistently if correctly calibrated.
You will need to convert a Rec 2020 HDR to one of these (they are similar but not the same). As you are trying to squeeze a quart into a pint pot, you have to trade off various aspects of a wider gamut, wider dynamic range source when you tone map it back to a narrower gamut, narrower dynamic range.
Broadcasters are already beginning to produce live multicamera content in UHD HDR Rec 2020 and tone map this to Rec 709 for broadcast in HD - and they don't take into account viewers' individual display characteristics? They convert using an approach that delivers a decent SDR result, but there isn't only one way of approaching this challenge, and which approach is used will dictate what the end result looks like.
There are currently at least two distinct approaches to HDR->SDR conversion being used :-
'Scene Light' where you try and create an SDR signal from an HDR signal, that comes as close as possible to matching what the scene would have looked like if shot on an SDR Rec 709 camera. You effectively map the Rec 2020 HDR signal back to light levels, and then simulate a Rec 709 SDR camera shooting that scene. This will make the scene look like it was shot on an SDR Rec 709 camera.
'Display Light' where you try and create something that looks as close as possible to watching the HDR picture on an HDR display, but on an SDR display.
When tone mapping gamuts you also have decisions to make as to whether you go for a constant hue conversion or not.
So different tone mapping approaches will have different results, but that's because they are aiming to deliver different results.
However if your SDR display is correctly calibrated, I see no reason why you would use a significantly different set of tone mapping variations display-to-display, you just need to be aware that different tone mapping approaches will deliver very different results.
Of course if your display is Rec 709 power law and not BT.1886 - then that will expose a difference.
