:: #mythtv

Daily chat history

Current users (44):

aloril, biffhero, blahdodo, brfransen, ChanServ, eharris, ElmerFudd, Fearful, ghoti, gigem, gregl, GreyFoxx, hampton_, Hydr0p0nX, ijc, ikevin, jpabq, jya, knowledgejunkie, kwizart, libsci, mad_enz, markspieth, markspieth2, MitchCapper, mkbloke, MythBuild, MythLogBot, nephyrin`, Panic, peper03, poptix, pppingme, rhpot1991, sdrik, sphery, stuarta, taylorr, tgm4883, The_RamShadow, tris, Warped, xris, _charly_
Thursday, September 13th, 2018, 00:17 UTC
[00:17:08] knowledgejunkie (knowledgejunkie! has quit (Ping timeout: 245 seconds)
[00:18:24] peper03 (peper03!~peper03@mythtv/developer/peper03) has quit (Ping timeout: 272 seconds)
[00:24:22] knowledgejunkie (knowledgejunkie! has joined #mythtv
[00:25:45] peper03 (peper03!~peper03@mythtv/developer/peper03) has joined #mythtv
[00:25:45] Mode for #mythtv by ChanServ!ChanServ@services. : +v peper03
[02:53:19] hampton (hampton!~hampton@mythtv/developer/hampton) has quit (Ping timeout: 244 seconds)
[03:07:00] hampton (hampton!~hampton@mythtv/developer/hampton) has joined #mythtv
[03:07:01] Mode for #mythtv by ChanServ!ChanServ@services. : +v hampton
[03:30:31] jheizer_ is now known as jheizer
[03:31:58] jheizer: I was finally getting around to giving the vaapi2 code a try and see it has made its way to master already. FYI it looks like the ubuntu ppa for master last built in April.
[05:12:33] Warped (Warped!~Warped@unaffiliated/warped) has quit (Quit: ChatZilla 0.9.92-rdmsoft [XULRunner 35.0.1/20150122214805])
[05:34:41] Warped (Warped!~Warped@unaffiliated/warped) has joined #mythtv
[06:39:32] SteveGoodey (SteveGoodey! has joined #mythtv
[06:39:32] Mode for #mythtv by ChanServ!ChanServ@services. : +v SteveGoodey
[06:54:36] Steve-Goodey (Steve-Goodey! has joined #mythtv
[06:54:36] Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey
[07:37:02] Steve-Goodey (Steve-Goodey! has quit (Quit: Konversation terminated!)
[07:43:17] willcooke (willcooke!~willcooke@ubuntu/member/willcooke) has joined #mythtv
[07:45:54] markspieth2 (markspieth2!~yaaic@mythtv/developer/markspieth) has quit (Remote host closed the connection)
[09:40:17] markspieth2 (markspieth2!~yaaic@mythtv/developer/markspieth) has joined #mythtv
[09:40:17] Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth2
[09:53:42] markspieth2 (markspieth2!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 256 seconds)
[11:17:27] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:4a51:b7ff:fe21:2979) has joined #mythtv
[11:17:27] Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey
[12:01:49] Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7da3:4501:4a51:b7ff:fe21:2979) has quit (Quit: Konversation terminated!)
[12:09:10] ooshlablu (ooshlablu!~ooshlablu@2601:18d:4600:5f31:a41d:40b0:725b:6115) has quit (Remote host closed the connection)
[13:04:07] jheizer: peterbennett, This means the new deinterlacers are working I assume? Player(0): 2000 codec-deinterlaced frames seen.
[13:07:12] jheizer: Full log
[13:09:58] jheizer: Since comcast has basically gone all 720p and so have mot of my local OTA broadcasts, I don't have a lot of test recordings left except old ones from hd-pvr days and they're aren't the best judge of quality always. I also haven't paid as much attention to it up close since leaving nvidia and getting a 4k tv last year.
[13:10:22] jheizer: But one sample I tried last night had a person in water and it was extremely blocky, but it could just be the source.
[13:10:39] ** jheizer just realized he should try it here on this machine, derp **
[13:10:55] stuarta: how is that newer content encoded?
[13:11:29] Mode for #mythtv by ChanServ!ChanServ@services. : +v jheizer
[13:12:16] jheizer: h264
[13:12:52] stuarta: that should work with vaapi, provided your hardware supports the profile
[13:13:02] stuarta: pastebin vainfo output
[13:13:06] jheizer: OTA is still all mpeg2 here, but our local stations had a merger so they threw them on as sub channels of each other and dropped to 720
[13:13:16] jheizer: oh yeah vaapi is working for sure in general
[13:13:25] jheizer: or least least the playback data says it is
[13:13:48] jheizer: I just know some chips don't do all deinterlacers and what not, and wasn't 100% sure it was doing what I thought it was
[13:14:32] jheizer: vainfo –
[13:15:00] stuarta: you've got very good support in hardware there
[13:15:11] stuarta: pretty much everything current
[13:15:21] jheizer: J4205 all in one board
[13:15:23] jheizer: pretty new
[13:15:38] jheizer: went a bit overkill on it because it did support more
[13:16:23] stuarta: curious i don't see the init messages from libva in there
[13:16:50] stuarta: you should get something which looks a like the first 2 lines of vainfo output
[13:17:08] jheizer: hmm, that was -v playback,libav
[13:17:31] stuarta: it comes out at normal logging so you should have seen it
[13:17:31] jheizer: IIRC, did it last night then went to bed.
[13:17:38] jheizer: ah
[13:17:45] stuarta: oh well, doesn't matter too much, certainly looks like it worked
[13:18:16] jheizer: Alright, chalk it up to cruddy source blown up to 4k.
[13:19:07] stuarta: but why is it needing to drop frames
[13:19:28] jheizer: I skipped a bunch to get to where it was ugly testing things
[13:19:31] jheizer: I assume it was from that
[13:20:43] stuarta: prob, look like only a very short period of time
[13:23:03] jheizer: if paused in kodi is displaying a single deinterlaced frame, can see here how bad it was . . . 08-22-46.jpg
[13:23:31] stuarta: ug
[13:28:55] ooshlablu (ooshlablu!~ooshlablu@ has joined #mythtv
[14:10:44] ** hampton waves **
[14:23:52] ** stuarta waves back **
[14:49:27] SteveGoodey: stuarta: Has there been a change in the amount of time a user can stay logged in to the forum?
[15:21:23] stuarta: SteveGoodey: not that i have done
[17:45:13] natanojl (natanojl!~jonatan@mythtv/developer/natanojl) has joined #mythtv
[17:45:13] Mode for #mythtv by ChanServ!ChanServ@services. : +v natanojl
[18:53:49] willcooke (willcooke!~willcooke@ubuntu/member/willcooke) has quit (Quit: Do your hobbies)
[18:55:21] jpabq: jheizer: it is sad. Once upon a time, to get the highest quality TV signal, you used OTA. Now, I am not sure they are not worse than what you get over satellite or cable.
[19:00:17] jpabq: peterbennett: I have noticed that a I sometimes have to INVERT the interlacing to get it to play smooth on the Shield. In other words: (M)enu -> Video -> Advanced -> Video Scan -> Interlaced (Reverse). I only have that issue on the Sheild, and it does not matter if I am using OpenGL or Mediacodec. I have not had time to dig into why that might be the case.
[19:03:27] peterbennett: jheizer: yes – codec-deinterlaced frames means that they are deinterlaced in vaapi hardware.
[19:04:11] jheizer: jpabq, yeah it was nice until they all merged here. Did mean I could drop 2 old PCI cards I have been using since they didn't really matter anymore.
[19:05:20] jheizer: peterbennett, great work by the way. Swap sources and done. Took me longer to navigate the new settings menus more than anything. After all these years my hands are just use to the old setup still.
[19:05:38] jheizer: err, swap profiles.
[19:06:57] peterbennett: jpabq: I don't understand how inverting the interlacing helps the shield. You should set deinterlacer to none because the shield deinterlaces automatically.
[19:07:41] jpabq: peterbennett: until I tell it to reverse it, it has a bad "combing" effect.
[19:08:00] peterbennett: jpabq: What do you have set as deinterlacer?
[19:08:13] jpabq: Not sure. Let me go loock.
[19:09:39] peterbennett: jpabq: If you use mediacodec you should not set a deinterlacer because it is already deinterlaced.
[19:11:27] jheizer: peterbennett, anything specific/outlier you'd like checked still? I have a bay trail and an older 3xxx series ultra mobile chip I could try it on.
[19:11:59] jheizer: Meant to load this up a week or two ago but been busy.
[19:13:26] peterbennett: jheizer: I am sot sure which CPUs support vaapi. Documentation seems to imply it is only i3, i5, i7 and perhaps core.
[19:14:06] jpabq: peterbennett: okay, this may have only been with OpenGL. I had it set with Kernel 2X. If I changed that to Interlaced 2x, the problem goes away. Back before VDPAU I remember Kernel being the best. I don't know how "Interlaced 2x" compares.
[19:14:21] jheizer: One I tried last night is a "Pentium" what ever that means anymore. J4205
[19:14:42] jpabq: peterbennett: Because of the artifacts, I have it configured to use OpenGL for my local mpeg2 sources.
[19:15:55] peterbennett: jheizer: did the pentium work with vaapi? Make sure by checking playback-> playback data to see what decoder it is really using because it may silently fail over to software (ffmpeg) decoder
[19:16:20] jheizer:
[19:17:02] jheizer: Step up from the old Atoms kind of with the embedded graphics included.
[19:17:17] peterbennett: jpabq: Yes I have moticed some very strange things with the OpenGL deinterlacers in general
[19:19:41] peterbennett: I have a pentium fro 2012 that I have not tried it on, currently it has an Nvidia card and the embedded graphics disabled, I winder whether it will support vaapi.
[19:29:54] jpabq: peterbennett: do you know what the difference is between "Kernel 2X" and "Interlaced 2x" ?
[19:30:03] jpabq: As far as "performance" goes?
[19:34:56] peterbennett: jpabq: I don't know, I have not looked into it.
[19:35:37] peterbennett: jpabq: I have not had much success with kernel.
[19:37:00] gigem: jpabq: As I understand it, Interlaced 2x isn't really a deinterlacer. It's waht you should use if your video card is capable of outputing interlaced video and you have it set that way.
[19:37:38] jpabq: gigem: okay, that confirms what I was seeing with it — it did not look near as good as kernel — once I "inverted" kernel, that is.
[19:37:39] gigem: Kernel 2x is working pretty well for me on the shield.
[19:37:40] peterbennett: jpabq: If it works, go for it.
[19:39:29] peterbennett: I do not watch much that is interlaced. I have a transcode job that transcodes everything that is mpeg2 into h264.
[19:40:30] peterbennett: Only for news or live TV do I see interlaced, and that mainly on Raspberry Pi which has good deinterlace and does not suffer from Local TV station pixelation like the NVidia
[19:45:02] gigem: Does anyone know if "Kernel" deinterlacing is a specific algorithm? It's much better on the shiled and with vaapi than I remember on other systems when opengl rendering was first added. Better enough that it makes me wonder if "Kernel" is just the generic name to refer to the implementations "best" deinterlacer.
[19:46:34] gigem: I still want to get surface rendering with mediacodec, but for now, I'm actually pretty happy with software decoding and kernel deinterlacing on the shield. Like with the rpi, there's no pixelation either.
[20:00:11] peterbennett: gigem: I saw in the explanation that kernel is linear blend but only applied when necessary. Experience does not seem to confirm this.
[20:05:23] natanojl: gigem: I really don't know much of the details of our kernel deinterlacers but I think it refers to doing a convolution between a kernel and the fields
[20:07:11] jpabq: I just tried to compile latest master for the shield, and got "vaapi2context.h: No such file or directory". Do I have to rebuild the library dependancies?
[20:10:33] peterbennett: jpabq: It builds ok of you do a clean build
[20:11:10] jpabq: peterbennett: I did a fresh checkout, but just of mythtv. I did not rebuild the dependancy libraries.
[20:11:13] gigem: peterbennett: I just remember never being satisfied with anything but nvidia deinterlacing for football and kernel seems fine now. Hockey starts in a month and that will be a bigger test.
[20:11:50] peterbennett: jpabq: Just di fresh
[20:12:02] peterbennett: then
[20:12:24] jpabq: Okay, I will give that a try.
[20:12:26] jpabq: Thanks.
[20:25:04] jpabq: peterbennett: doing the ` fresh` fixed it.
[20:34:52] jpabq: Is there a way to "side load" a MythTV theme onto the shield?
[20:40:21] peterbennett: jpabq: There is probably some way to push teh files with adb
[20:40:57] hampton (hampton!~hampton@mythtv/developer/hampton) has left #mythtv ()
[20:41:54] peterbennett: jpabq: adb push
[20:42:01] natanojl: peterbennett: I think the OpenGL kernel deinterlacers are more of a Gaussian weighted average kind
[20:44:12] peterbennett: natanojl: Thanks for the info, I don't understand them or how they work. I guess someday I will look at the OpenGL code.
[20:47:13] peterbennett: jpabq: Look at for an example of push – unfortunately you cannot push directy to the mythfrontend files directory you have to push to a temp diretory and then ru a cp to copy teh files across
[20:48:45] jpabq: I am trying to figure out why fonts looks SOOO different on the Shield compared to my linux machines.
[20:48:53] stuarta: jpabq: the usual way of just selecting a new theme doesn't work?
[20:49:17] jpabq: stuarta: I want to tinker with the theme without committing it.
[20:49:32] jpabq: I also need to figure out how to see frontend logs.
[20:50:05] peterbennett: Frontend logs – run adb logcat mfe:D *:S |& tee tmp/android.log
[20:50:16] peterbennett: oops
[20:50:38] peterbennett: frontend logs : adb logcat mfe:D *:S
[20:50:52] peterbennett: adb logcat mfe:D * : S
[20:51:06] peterbennett: impossible to paset teh command
[20:51:08] jheizer: Latest ansible doesn't install libbluray-dev on Mint 19
[20:51:11] peterbennett: paste
[20:51:40] peterbennett: adb logcat mfe : D *:S |& tee tmp/android.log
[20:52:07] peterbennett: that filters only the mythfrontend mesages
[20:52:52] jpabq: actually, the first one worked for me, not the last.
[20:53:17] jpabq: Intresting.... on Android it is NOT loading the embedded fonts in the theme.
[20:55:54] jpabq: gigem: Did you say there was a way to figure out the android "hostname" ?
[20:55:57] ooshlablu (ooshlablu!~ooshlablu@ has quit (Remote host closed the connection)
[20:56:38] jpabq: Or, I should say, set the hostname?
[20:59:11] peterbennett: I set a host name in the config.xml. Otherise you get andoid with some random number.
[20:59:54] jpabq: peterbennett: you do that before packing it for android?
[21:00:18] jpabq: ./packaging/android/build64/mythtv/contrib/config_files/config.xml ?
[21:00:19] peterbennett: No, you can do it in the MythTV frontend setup
[21:00:55] peterbennett: Or by editing config.xml on the android
[21:04:12] peterbennett: The random android host name remains the same as long as the app is installed, but changes if you uninstall and install again, so its best to set your own.
[21:05:10] jpabq: Yes, I just found that out. Trying to (re)install, it complained about the environment changing in some way, so I had to uninstall first.
[21:10:00] jpabq: markspieth: should android be loading the fonts from the embedded theme, or was that not implemented?
[21:20:34] jpabq: Doing some googling, it looks like the font names may show up differently under Android, and that could be why they are not loading.
[21:27:47] jpabq: peterbennett: how do I set the verbose flags when running mythfrontend on android?
[21:30:11] peterbennett: jpabq: Update the manifest then build
[21:31:00] jpabq: ?
[21:31:34] peterbennett: Yes that sounds right
[21:32:34] jpabq: I see it, thanks.
[21:49:03] SteveGoodey (SteveGoodey! has quit (Quit: Konversation terminated!)
[21:52:17] jheizer: vaapi2 seems all good on an old i5 3317U
[21:57:27] natanojl (natanojl!~jonatan@mythtv/developer/natanojl) has quit (Ping timeout: 240 seconds)
[22:02:18] jpabq: Indeed — the font names show up different under Android than they do Linux, so the load fails.
[22:09:21] peterbennett (peterbennett!~Peter_Ben@mythtv/developer/peterbennett) has quit (Quit: WeeChat 1.6)
[22:21:32] gigem: jpabq: Why do you need the hostname? If you want it for settings, you can set the custom database identifier (or similar) in the frontend settings. That's what I use to get a sane hostname in mythconverg.
[22:21:52] jpabq: gigem: that is what I ended up doing. I just never needed to before.
[22:24:24] gigem: jpabq: Also be aware that the copying work around in might not work. I had to pass gdbserver via pipe to adb shell and redirect stdout to get it installed. I don't see that in anymore so that change might have gotten lost.
[22:25:07] gigem: jpabq: For my own, pet theme, I just include it in my package.
[22:25:40] markspieth2 (markspieth2!~yaaic@mythtv/developer/markspieth) has joined #mythtv
[22:25:40] Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth2
[22:28:29] jpabq: gigem: not a bad idea.
[22:28:31] markspieth: Yes kernel deinterlace is a specific algo. See the opengl implementation. I have some tweeks for the opengl deinterlacers when I was investigating the green lines but could not be sure they didnt affect the PC version detrimentally.
[22:32:00] peterbennett (peterbennett!~Peter_Ben@mythtv/developer/peterbennett) has joined #mythtv
[22:32:00] Mode for #mythtv by ChanServ!ChanServ@services. : +v peterbennett
[22:38:20] markspieth2 (markspieth2!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 244 seconds)
[22:41:55] markspieth2 (markspieth2!~yaaic@mythtv/developer/markspieth) has joined #mythtv
[22:41:55] Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth2
[22:59:19] markspieth2: jpabq: yes the fonts dhould be loaded but I have seen that it uses the android fonts instead of the myth provided ones. may be a font path / find issue. never worried about it because bigger fish to fry.
[23:09:28] jheizer (jheizer!~jheizer@2601:246:8201:7bfc:eddc:a2db:4685:59c4) has quit (Read error: Connection reset by peer)
[23:16:19] mad_enz (mad_enz!~mad_enz@2607:fea8:bdf:e784:6d4c:82e6:e0ad:3c05) has quit (Quit: Leaving)
[23:45:28] jpabq: markspieth: It is a Qt issue — it provides the font NAMES in a slightly different way between Linux and Android — However, I believe I see a "common ground" way of getting it to work. It will require me to change my font definitions in my theme.
[23:56:17] mad_enz (mad_enz!~mad_enz@2607:fea8:bdf:e784:45c3:2d10:6110:df04) has joined #mythtv

IRC Logs collected by BeirdoBot.
Please use the above link to report any bugs.