2018-10-02, 01:50
I don't have any issues with delay playback with MPC-HC. It seems fine. But some files played over the network won't play at all. I can't unpause them or play them from MPC-HC. It only seems to be files I opened at some point in the past. I'm not using MPC-HC, but whatever has happened to the player on my end is different.
The report BT.2020 option only impacts SDR playback. This would be relevant if you were sending a DCI-P3 input to a display that only accepts BT.2020. A person using a DCI-P3 3D LUT would be most likely to need this option, as DCI-P3 3D LUTs are superior to BT.2020 3D LUTs for current displays.
I did have a long delay at playback start when using an external player after upgrading, but this hasn't continued to happen. It isn't always that slow on subsequent playback.
I can't disagree that HDR -> SDR isn't valuable to everyone. The mismatch in absolute PQ gamma to relative SDR gamma looks, to me, to impact the accuracy of things such as color saturation. I also think the HDR effect is dulled somewhat. If you owned a projector, you may not want to use a PQ curve to watch HDR content because all of the values would be under 100 nits. In this case, HDR -> SDR can make the image significantly brighter than passthrough. There is also the issue of the brightness discrepancy between HDR and SDR. On my display, SDR content looks best with the backlight set to 11/20, while HDR content looks better at 18-20/20. That is a fairly large mismatch in brightness. I've dealt with this by manually adjusting the backlight before playback, which is not very user-friendly.
Pixel shader with a PQ HDR output could be of value, but I'd like to see some screenshots to see exactly what is happening to the image. If the image is darker, more tone mapping could be happening, which could mean more detail at the expense of some loss in brightness. You could potentially have less accurate color, though, because the image may be tone mapped by madVR and again by the display. Like I said, I'd like to see some screenshots first before concluding pixel shader with HDR passthrough is or isn't useful. madshi appears to be tweaking tone mapping for a while, so things will change further. I was hoping for VapourSynth integration in madVR soon because this would open the possibility of using numerous existing VapourSynth shaders.
The only significant issue with consumer displays when it comes to HDR tone mapping is that many manufacturers seem to have been given poor advice on how to tone map when HDR10 content was first released. When he first started out, madshi said one industry expert gave him terrible advice on how to tone map accurately. This is mostly an issue with titles that are very bright: mastered to 4,000 nits. These movies are sometimes more cool in color than they should be because the hue is being accidently shifted by clipping certain values. Not everyone seems to notice or care, but it is a common flaw of some early HDR displays from many manufacturers.
I assume your comment about reading the logic of using this setting was referencing my long post. I didn't consider that someone would want some explanation as to why to use madVR pixel shader over passthrough. I didn't consider that. With that in mind, I will alter that part the next time I update that post.
The report BT.2020 option only impacts SDR playback. This would be relevant if you were sending a DCI-P3 input to a display that only accepts BT.2020. A person using a DCI-P3 3D LUT would be most likely to need this option, as DCI-P3 3D LUTs are superior to BT.2020 3D LUTs for current displays.
I did have a long delay at playback start when using an external player after upgrading, but this hasn't continued to happen. It isn't always that slow on subsequent playback.
I can't disagree that HDR -> SDR isn't valuable to everyone. The mismatch in absolute PQ gamma to relative SDR gamma looks, to me, to impact the accuracy of things such as color saturation. I also think the HDR effect is dulled somewhat. If you owned a projector, you may not want to use a PQ curve to watch HDR content because all of the values would be under 100 nits. In this case, HDR -> SDR can make the image significantly brighter than passthrough. There is also the issue of the brightness discrepancy between HDR and SDR. On my display, SDR content looks best with the backlight set to 11/20, while HDR content looks better at 18-20/20. That is a fairly large mismatch in brightness. I've dealt with this by manually adjusting the backlight before playback, which is not very user-friendly.
Pixel shader with a PQ HDR output could be of value, but I'd like to see some screenshots to see exactly what is happening to the image. If the image is darker, more tone mapping could be happening, which could mean more detail at the expense of some loss in brightness. You could potentially have less accurate color, though, because the image may be tone mapped by madVR and again by the display. Like I said, I'd like to see some screenshots first before concluding pixel shader with HDR passthrough is or isn't useful. madshi appears to be tweaking tone mapping for a while, so things will change further. I was hoping for VapourSynth integration in madVR soon because this would open the possibility of using numerous existing VapourSynth shaders.
The only significant issue with consumer displays when it comes to HDR tone mapping is that many manufacturers seem to have been given poor advice on how to tone map when HDR10 content was first released. When he first started out, madshi said one industry expert gave him terrible advice on how to tone map accurately. This is mostly an issue with titles that are very bright: mastered to 4,000 nits. These movies are sometimes more cool in color than they should be because the hue is being accidently shifted by clipping certain values. Not everyone seems to notice or care, but it is a common flaw of some early HDR displays from many manufacturers.
I assume your comment about reading the logic of using this setting was referencing my long post. I didn't consider that someone would want some explanation as to why to use madVR pixel shader over passthrough. I didn't consider that. With that in mind, I will alter that part the next time I update that post.
