Posts by chorca

    So, looking further into this, I think an issue I'm running into is that the TV is an older type that can do 4k at 60Hz, but it does it at 4:2:0 chroma while using an older HDMI 1.4b chip. This lets it get 4k60 over the older tech, but at a lower chroma sampling.

    Supposedly the intel chips will support this as well, but via a driver patch that was added back in 2017.. However, it appears that it's not being detected properly and this is causing some issues with (initially) getting 4k60 to show up on my resolution list.

    Spent a few hours digging around xrandr and xorg code, but there's no good way I can tell to force the driver into 4:2:0 chroma mode.

    Dug a little further, completely wiped out and reloaded the system with ArchLinux and installed Kodi via their repo. Messed around, and i'm getting the exact same issue here.

    at 1920x1080, 60hz, i get gui audio

    1920x1080, 30hz, no gui audio

    3840x2160, 30hz, no gui audio

    3840x2160 set via modeline, gui audio, but no video audio.

    It almost seems like the Intel driver is to blame, but i'm not sure it this is just an incompatability with the Sony TV or if it's the driver itself that's having issues here.

    Downloaded the latest 9.0.1 version and put it onto my NUC 7i3BNK hooked via USB-C to HDMI to my TV (Sony XBR-65X850B), and noticed that while I'm able to get GUI sounds moving around, the second I play a video, all audio stops, and there's no sound from the video. If I stop the file, GUI sounds come back and work fine.

    I tried with several different types of files, have my audio set to defaults aside from selecting the HDMI output in the device list (2.0 channel, optimized, medium, no passthrough)

    I noticed that if i turn my TV off and back on while the video is playing, the audio starts to work again, but once I stop the video and start another, the no-audio issue returns.

    This didn't happen before with the 8.x release, seems to be something different here.

    Anyone else having something similar?

    Tried with HSBS content in an .mkv file and getting the same issue. Since running on generic linux, I can only decode SBS or OU files.

    Doesn't seem to be related to the file, both full and half SBS seem to have the same issue.

    Tested with the output set to both 4K and 1080p, getting the same result with both, so doesn't appear to be the output resolution either.

    I grabbed the latest version of libreelec with Leia and am currently testing with the release (9.0.0).

    The big thing I'm running into is that I have a 4K Sony TV with passive 3D, and so have been using interlaced mode in Kodi for awhile now.

    However, in Leia, interlaced mode is only displaying one complete frame of video, it's not interlacing the video as it has in previous versions, so I'm not getting 3D.

    It's correctly detecting that it's a 3D movie and asking for the type of display, and putting it into any other mode (even anaglyph) seems to display the 3D image as two distinct frames, but interlaced mode just operates the same as the '2d' mode in that it shows only one of the two eyes.

    Went back to the older version of Libreelec and it was working fine there, so seems to be something updated with the latest version.

    Have attached a logfile from a boot and attempt to play a 3D movie in interlaced format.

    Hardware is an Intel NUC 7i3BNK running at 3840x2160p, 59.97Hz via USB-C to HDMI cable

    Audio is digital out from TV into small amplifier

    TV is a Sony XBR-65X850B

    Code
    xrandr --newmode "3840x2160p60"  593.41  3840 4016 4104 4400  2160 2168 2178 2250 +hsync +vsync
    xrandr --addmode DP1 3840x2160p60
    xrandr --output DP1 --mode 3840x2160p60

    I'm really curious about this: How did you come up with this modeline?

    Every calculator i've tried says somewhere between 700 and 1000Mhz for the pixel clock, is there something wrong with using these to calculate digital resolutions?

    It actually works, which i'm happy about, but I'm playing around with them and the closest i've come is from eizo's site:

    The frequency with which dots (or pixels) are illuminated on the screen; how quickly a single dot can be produced on the screen. A general formula for calculating the dot frequency is: D x 1.25 x fH (the number of dots in a line multiplied by a constant between 1.25 and 1.3 multiplied by the horizontal scanning frequency).