Wednesday, 2015-04-22

*** tpb has joined #timvideos00:00
*** CarlFK has quit IRC02:48
mithroMaZderMind / cfelton: as I mentioned yesterday, the EDID going into the HDMI connectors on the Atlys board are already available via pin headers on the Atlys board.02:50
mithroNo cable dismantling needed - but it does only give you the signal level at the Atlys board end of the system (rather than the other end)02:51
mithrocfelton: you have an Atlys board right?03:19
cfeltonmithro: they are rebuffered, if it is a slew rate issue (not sure why it would be) you could not observe that through the header03:20
cfeltonmithro: yes03:20
mithrocfelton: I'm looking at the Atlys schematics now, it seems that for J2 and J3, you are right - the EDID goes through a buffer on both input and output and JP6/JP7 sit between the two buffers03:29
mithrocfelton: same with J1 but JA is just directly connected03:33
*** CarlFK has joined #timvideos04:32
*** ChanServ sets mode: +v CarlFK04:32
*** hyades has joined #timvideos13:54
MaZderMindI'm currently writing tests for the Caps produced by gst-switch. interestingly for `-f debug` the composite-output and the encoded-output produce 30fps, while the preview-ports only produce 25fps14:31
MaZderMindaccording to gstswitchopts.c `-f debug` should be [email protected]14:33
CarlFKhmmm... not sure that is a bug14:37
CarlFKwhy do the previews need to be the same?14:38
*** Niharika has joined #timvideos14:38
CarlFKI suspect it is better if they are the same, especially if it uses lots of CPU otherwise14:40
MaZderMindhaving to deal with different framerates is a pita, ie when you try to display both ir mix them or what not.14:43
MaZderMindbut i think it's a bug because the formati s explicitly specified as [email protected]14:44
*** Niharika has quit IRC15:04
*** Niharika has joined #timvideos15:10
MaZderMindi opened a ticket - to be discussed: https://github.com/timvideos/gst-switch/issues/20015:22
tpbTitle: Composite-Output is 30fps, independent of cofigured format · Issue #200 · timvideos/gst-switch · GitHub (at github.com)15:22
*** hyades has quit IRC15:59
*** Niharika has quit IRC18:10
*** Niharika has joined #timvideos18:12
*** miselin has quit IRC18:22
*** miselin has joined #timvideos18:25
*** CarlFK has quit IRC18:26
*** CarlFK has joined #timvideos19:14
*** ChanServ sets mode: +v CarlFK19:14
*** Niharika has quit IRC19:27
MaZderMindanother inconsistency in gst-switch: the integration-tests TestMarkTracking, TestMarkFace and TestClickVideo actually don't do anything19:44
MaZderMindthey are called with generate_frames=True so they never assert anyting. Calling them with generate_frames=False fails because there is nothing to compare against, because none of the methods actually change the visual output of the server19:45
CarlFKoh god please don't spend any time on tracking.. it makes me angry.19:54
MaZderMindkk, i don't see it being that useful, too, but well - it's in there somewhere20:33
MaZderMindand another inconsistency i'm not use of it's a bug: it is not possible to attach multiple audio sources.20:35
MaZderMindgst-switch-srv/gstworker.c:283:error: branch_3004: pipeline parsing error: could not link source to gdppay320:35
MaZderMindgst-switch-srv/gstworker.c:808:error: branch_3004: failed to create new pipeline20:35
MaZderMindgst-switch-srv/gstswitchserver.c:750:error: failed serving new client20:35
CarlFKhmm.  I thought I had done that, but maybe not20:45
CarlFKbut done is just testing, nothing has been done in production, so pfft.20:45
MaZderMindwhich makes me think if there are any measures to keep audio and video in sync when switching between them…20:47
MaZderMindwhich is one of the biggest issues when doing a mixing-program20:47
CarlFKwith DVswitch it was "easy" because of everything being live.  there was a very small buffer (2 frames I think) and if new audio wasn't there yet, oh well, it just used the old audio20:49
CarlFKI think that feature was a side effect of everything being dv20:50
MaZderMindit's a question of timestamping the buffers in the sources and syncing the clocks between the sources20:51
MaZderMindand handling restarts in that timestamp-sequence when a source reconnects20:51
MaZderMindthat's no easy thing if it's not designed in from the start20:52
MaZderMindso you never tested a/v sync, did you?20:53
*** CarlFK has quit IRC23:11
*** CarlFK has joined #timvideos23:18
*** ChanServ sets mode: +v CarlFK23:18

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!