Rock 64 with LibreELEC-RK3328.arm-8.90.007-rock64

  • Thought I'd give the LibreElec (Leia) v8.90.007 Alpha a go. I'm testing via a Denon UHD (HLG/HDR10/DV friendly) AVR and into a Sony UHD HDR TV. (XF9005 aka X900F)


    Positive aspects in italics, negative aspects in bold

    Frame rate and resolution switching via the the Whitelist works well - 720p output at 720p, 1080p output 1080p, 2160p output at 2160p.

    CEC control worked with remote commands from my TV passed to LibreElec, and volume commands passed from LibreElec to my AVR.

    2160/23.976p 10bit HEVC Rec 2020 HDR10 HDR content was output at 2160/23.976p 12-bit 4:2:2 Rec 2020 correctly flagged as ST.2084 EOTF (i.e.HDR10). However it looks like the HDR10 average/peak light level metadata wasn't passed (my HD Fury Vertex flagged them as 0/0)

    2160/50p 10bit HEVC Rec 2020 HLG HDR content was output at 2160/50p 10-bit 4:2:0 Rec 2020 flagged as SDR, not flagged as HLG. (I've not seen any Kodi device flag HLG as HLG yet)

    Audio passthrough of any bitstreamed audio caused nasty stuttering noise to come via my amp - rather than triggering the amp into passthrough. Disabling passthrough I got 48k/16bit PCM 2.0/5.1 etc. output

    1080i25 H264 BBC One HD recorded native interlaced content from DSat (permanent 1080i25) was deinterlaced with the wrong field dominance so was very juddery, at least at times.

    Watching Live TV SD MPEG2 576i25 native interlaced stuff seemed to be OK for the brief period I watched.  Live TV HD H264 1080i25 native interlaced stuff seemed to drift between correct and incorrect field dominance deinterlacing

  • Thanks for your test, I noticed all of the MBAFF interlaced videos are wrong (25i, 29.97i too).

    Yes - all UK HD TV broadcasting is MBAFF. UK Freeview HD also uses dynamic, on-the-fly, 1080p25/1080i25 encoder switching based on source content.

  • 1080i25 H264 BBC One HD recorded native interlaced content from DSat (permanent 1080i25) was deinterlaced with the wrong field dominance so was very juddery, at least at times.

    My previous post wayback and this post with logs is as far as i've got with only 2 HD stations being the problem 25 fps .

    Last reply

    Nothing has changed in latest .007 image

    Feel free to see if you get the same results as me with test clips i supplied in previous post link.

  • My experience seems to be :

    MPEG2 576i25 content on BBC One SD - also known as 576/50i - with native 50Hz interlaced motion is de-interlaced correctly to 576p50 and scaled - as would be expected.

    H264 1080i25 content - also known as 1080/50i - with native 50Hz interlaced motion (actually the same sports match on BBC One HD) is jumping around all over the place between correctly deinterlaced and incorrectly deinterlaced. I got some fluid motion, followed by jerky reverse field-dominance motion (where the fields look to have been temporally swapped over and thus you get worse judder than just i25->p25 rather than i25->p50 deinterlacing)

    (One thing that MAY be an unusual use case on UK DVB-T2 H264 HD content is that the H264 encoders used dynamically switch between 1080p25 and 1080i25 (MBAFF) on a GOP-by-GOP basis based on the content feeding the encoders (this is done by the encoders dynamically detected intra-frame motion and switching to 1080i encoding if they detect it, but if there is no intra-frame motion and the content is 1080p25 within a 1080i25 wrapped - the encoders switch to p25 encoding for that GOP) This has been the case for many years on UK HD DVB-T2.)

  • Could you grab an stream sample (.TS)?

    Yes - will do that now. BBC One HD and ITV HD are both on PSB3 and both use the i/p switching encoders - recording both now during News shows which include native interlaced content. Also recording an entertainment show this evening on BBC Two HD that is native interlaced.