rpi: bad livetv with deinterlacing

I had same issue with my RPi3 and tvheadend 4.3 setup (Astrometa tuners, 2x DVB-C + 1x DVB-T)... Same "green artifacts, glitches and distortion", only with 1080i, didn't tried to play with deinterlacing, I resolved after activated the "Force old status" in the tvheadend server tuner setup (web interface).

Can you share a short sample / recording that triggers the issue?

so long,

Hias

Thanks a lot for the sample!

I'm a bit busy with my daytime job ATM but I'll have a look the next couple of days, not sure if I manage to check before next weekend though.

so long,

Hias

Hmm, just checked the sample on my Linux laptop with VLC and mpv and there are blocking artifacts at around 15-17 seconds - so there seems at least to be an issue with the recording. Do you see other artifacts at different timestamps with LE?

so long,

Hias

ISTR there were issues when tvheadend is running on the same device where video is being played back.

It could be that the memory bandwidth required for deinterlacing interferes with operation of the USB DVB hardware and/or drivers and/or tvheadend - might be too short buffers (in HW or driver) or something similar in tvheadend like increased latency resulting in dropped buffers/packets.

I don't think a definitive solution for that was found, other than moving tvheadend (with the DVB receiver) to a separate device.

so long,

Hias

Possible to be something from tvheadend server 4.3 and new deinterlacing method?

Maybe try to install the tvheadend server 4.2. No need to unistall the 4.3, just disable. Search only one transponder for channels where the issues appear and check.

I have 3 RPi3B in network, one I upgraded to LE 10.0.2 few months ago, without to experience any unusual issues with LiveTV.

Recently I upgraded the last two. Upgraded the tvheadend server to 4.3 (with 4.2 had some EPG updating issues, what is present in 4.3 too) and very seldom, was issues to start the LiveTV (only 1080i, stream blocked at first picture, in the media info "no deinterlace", at all clients, not only where the server running), two days ago I returned to 4.2 version, no issues yet. The "memory bandwidth required for deinterlacing" could be an explanation, in my case when the stream starting use too much memory/processor and lead (?) to deinterlacing issue... maybe... (RPi3B - smaller uP).