| *** tpb has joined #timvideos | 00:00 | |
| mithro | then it has nothing to do with hdmi2usb capture then? | 00:00 |
|---|---|---|
| CarlFK | heck if I know. is uses videotestsrc | 00:01 |
| CarlFK | maybe I can just swap that, but generally that never just works. | 00:01 |
| mithro | CarlFK: what is the jpeg caps thing you need for dvsource-v4l2-other and why haven't you sent a patch to fix the documentation? | 00:04 |
| CarlFK | mithro: -c="image/jpeg,width=1280,height=720" | 00:06 |
| CarlFK | short on time | 00:07 |
| *** tvCommitBot has joined #timvideos | 00:10 | |
| tvCommitBot | [dvsource-v4l2-other] mithro pushed 1 new commit to master: http://git.io/vWbRJ | 00:10 |
| tvCommitBot | dvsource-v4l2-other/master 678131d Tim Ansell: Adding information about mjpeg based v4l2 devices | 00:10 |
| *** tvCommitBot has left #timvideos | 00:10 | |
| mithro | CarlFK: took me 2 minutes | 00:10 |
| mithro | xfxf: https://github.com/timvideos/dvsource-v4l2-other/blob/master/README.md#quick-start-with-mjpeg-devices-such-as-hdmi2usbtv | 00:10 |
| tpb | Title: dvsource-v4l2-other/README.md at master · timvideos/dvsource-v4l2-other · GitHub (at github.com) | 00:10 |
| mithro | v4l2src device=/dev/video0 ! image/jpeg,width=1280,height=720 ! decodebin ! | 00:12 |
| *** travis-ci has joined #timvideos | 00:12 | |
| travis-ci | [timvideos/HDMI2USB-misoc-firmware/master#255] (bb302c0): The build was broken. (https://travis-ci.org/timvideos/HDMI2USB-misoc-firmware/builds/88017669) | 00:12 |
| *** travis-ci has left #timvideos | 00:12 | |
| CarlFK | thats why you should do it and not me :) | 00:13 |
| CarlFK | mithro: xfxf isn't using dvswitch | 00:13 |
| CarlFK | he wants to save the full res to disk | 00:14 |
| CarlFK | so probably some of that and some of https://wiki.xiph.org/GST_cookbook#hdmi2usb_capture_.28timvideos.us.29 | 00:14 |
| tpb | Title: GST cookbook - XiphWiki (at wiki.xiph.org) | 00:14 |
| mithro | CarlFK: it would have taken you the same amount of time and you wouldn't had to ask me like 3 times | 00:15 |
| mithro | _florent_: Did you just break the firmware? :P | 00:15 |
| mithro | main.c:(.text.startup+0x2c): undefined reference to `print_board_dna' | 00:16 |
| mithro | main.c:(.text.startup+0x2c): relocation truncated to fit: R_LM32_CALL against undefined symbol `print_board_dna' | 00:16 |
| xfxf | mithro: the above doesn't really help me | 00:26 |
| mithro | xfxf: the above gives you the part you need to get the data out of the HDMI2USB, what do you want to do with it? | 00:29 |
| CarlFK | mithro: he wants to save the full res to disk | 00:29 |
| xfxf | it doesn't give me anything | 00:29 |
| xfxf | it already told me what i know | 00:30 |
| xfxf | gst-launch-1.0 v4l2src device=/dev/video1 ! jpegparse ! queue max-size-bytes=100000000 max-size-time=0 ! matroskamux name=mux alsasrc device='hw:1,0' ! audio/x-raw,channels=2,rate=48000 ! audioconvert ! vorbisenc ! queue ! mux. mux. ! queue max-size-bytes=100000000 max-size-time=0 ! filesink location=hdmi2usb_recording_`date +%s`.mkv | 00:30 |
| xfxf | ^ that | 00:30 |
| xfxf | is 'sort' of doing what i want | 00:30 |
| xfxf | except the framerate is like 5fps and out of sync with the audio (which sounds fine) | 00:30 |
| xfxf | shoving in a image/jpeg,width=1280,height=720 pipeline doesn't fix this | 00:31 |
| xfxf | the line in that wiki article is for dvswitch. not doing that, have negative interest in using dvswitch with hdmi2usb | 00:31 |
| mithro | Why are you using jpeg parse? | 00:31 |
| mithro | Your missing a decodebin too | 00:31 |
| xfxf | because without that i'd get errors | 00:31 |
| xfxf | sure, can you let me know the correct pipeline? | 00:31 |
| xfxf | my gstreamer pipeline knowledge is rudimentary at best | 00:32 |
| xfxf | i understand the concept, i'm not familiar with all of the pieces required | 00:32 |
| mithro | Replace everything before the matroskamux with the line I gave you | 00:32 |
| xfxf | which line? i only see a dvswitch line in that wiki article | 00:33 |
| CarlFK | xfxf: I'll try this and let you know if something seems to work | 00:33 |
| mithro | 11:09 a.m. <@mithro> v4l2src device=/dev/video0 ! image/jpeg,width=1280,height=720 ! decodebin ! | 00:33 |
| mithro | Remove the audio bit until you have the video working first | 00:34 |
| xfxf | sure, ta | 00:34 |
| mithro | xfxf: compare to what the encoder on the HDMI2USB is saying | 00:34 |
| xfxf | okay, that seems to save a file, but neither vlc or mplayer will play it | 00:36 |
| xfxf | https://www.irccloud.com/pastebin/Y04rRCTY/ | 00:37 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 00:37 |
| mithro | xfxf: well, you just piped raw frames into a file | 00:38 |
| mithro | gst-launch-1.0 v4l2src device=/dev/video1 ! image/jpeg,width=1280,height=720 ! decodebin ! matroskamux ! filesink location=hdmi2usb_recording_`date +%s`.mkv | 00:41 |
| xfxf | doesn't appear to work either | 00:41 |
| xfxf | when playing back | 00:41 |
| xfxf | https://www.irccloud.com/pastebin/LBh4KlLg/ | 00:42 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 00:42 |
| mithro | avprobe the file? | 00:42 |
| xfxf | https://www.irccloud.com/pastebin/h5lVrtzc/ | 00:42 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 00:42 |
| mithro | vlc != avprobe? | 00:42 |
| xfxf | https://www.irccloud.com/pastebin/xkkVHyT2/ | 00:42 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 00:42 |
| CarlFK | mithro: want to ssh into my box so you can try this stuff yourself ? | 00:42 |
| mithro | CarlFK: not particularly | 00:43 |
| mithro | xfxf: okay - so it is just vlc that doesn't like that file then | 00:43 |
| mithro | gst-launch-1.0 v4l2src device=/dev/video1 ! image/jpeg,width=1280,height=720 ! matroskamux ! filesink location=hdmi2usb_recording_`date +%s`.mkv | 00:43 |
| xfxf | or mplayer | 00:43 |
| mithro | xfxf: if you do that you'll get a mkv file with jpeg frames in it | 00:44 |
| xfxf | ok | 00:44 |
| xfxf | that one seems to work | 00:44 |
| mithro | xfxf: run avprobe on the file afterwards and check | 00:44 |
| *** Bertl_oO is now known as Bertl_zZ | 00:44 | |
| xfxf | https://www.irccloud.com/pastebin/oKJCZB6F/ | 00:45 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 00:45 |
| xfxf | okay,ta, i'll try recombining with my alsa muxer | 00:45 |
| mithro | xfxf: yeah | 00:45 |
| xfxf | wait, wat, that doesn't seem that much different to the original thing i pasted | 00:46 |
| mithro | xfxf: not much difference in off by 1 errors either :P | 00:47 |
| xfxf | ah, decodebin is removed | 00:47 |
| xfxf | heh inded | 00:47 |
| xfxf | indeed* | 00:47 |
| xfxf | i just didn't see the difference at all initially | 00:47 |
| mithro | decodebin converts the jpegs back into raw video | 00:48 |
| xfxf | ah, right | 00:48 |
| mithro | so if you want to reencode the video, you'll need it again | 00:48 |
| xfxf | right | 00:51 |
| xfxf | yay, this is working | 00:51 |
| xfxf | friggin' huge though | 00:51 |
| xfxf | 1 min was like 350mb | 00:51 |
| mithro | It is going at 480Mbit/s | 00:51 |
| xfxf | ya | 00:52 |
| mithro | so, now you just need to dump a encoder in between the matroska mux and the decodebin | 00:52 |
| mithro | xfxf: any luck? | 00:57 |
| xfxf | haven't done that yet but i can probably figure that part out fine | 00:58 |
| xfxf | just sorting other issues | 00:58 |
| xfxf | there's a total of 2 volunteers for this event | 00:58 |
| xfxf | which is making things interesting | 00:58 |
| GChriss | xfxf, CarlFK, mithro: I'm back. so, um, things are working? | 01:00 |
| CarlFK | no. | 01:00 |
| CarlFK | GChriss: want to ssh into the box so you can try things? | 01:00 |
| GChriss | I do suggest expanding decodebin into indiviual elements | 01:00 |
| mithro | GChriss: CarlFK and xfxf have Atlys boards, not Opsis boards | 01:00 |
| mithro | GChriss: and xfxf has things working as far as I can tell, don't know what carl is trying to do | 01:01 |
| CarlFK | mithro: same as xfxf | 01:01 |
| CarlFK | and no he does not have things working | 01:01 |
| mithro | 11:48 AM <xfxf> yay, this is working | 01:02 |
| mithro | 11:48 AM <xfxf> friggin' huge though | 01:02 |
| CarlFK | well, not working good enough. | 01:02 |
| CarlFK | GChriss: what's your gitub id? | 01:03 |
| mithro | CarlFK: the next part is trivial and has nothing to do with the HDMI2USB | 01:03 |
| mithro | CarlFK: it's just standard gstreamer | 01:04 |
| GChriss | I'm playing w/ self-hosting. might be up: code.openvideo.pro | 01:04 |
| CarlFK | then type it and I'll cut/paste it | 01:04 |
| mithro | GChriss: were did you get to testing the other night? | 01:04 |
| GChriss | (sorry, in a conversation with my brother, one sec) | 01:05 |
| CarlFK | bb in 10 min | 01:06 |
| mithro | Oh wait | 01:09 |
| mithro | I might be confusing GChriss and G33KatWork ? | 01:09 |
| mithro | bblr going to lunch | 01:11 |
| xfxf | mithro: whoa, can hdmi2usb work without external power (hdmi powered?) | 01:12 |
| xfxf | i unplugged its power trying to reset it and it's still shoving input0 to output0 | 01:12 |
| GChriss | ok, back for real this time | 01:12 |
| mithro | xfxf: I had that happen the other day too | 01:13 |
| mithro | xfxf: BTW you mean the "Atlys board" | 01:14 |
| xfxf | yes | 01:14 |
| mithro | xfxf: I think something is back driving the 5v through the HDMI port | 01:14 |
| mithro | Which is kind of bad | 01:15 |
| GChriss | mithro: re: "GChriss: were did you get to testing the other night?" creating a gstreamer pipeline for CarlFK/xfxf? I got as far as something that worked on my local machine w/o hardware, pasted to Xiph wiki | 01:15 |
| GChriss | mithro: ah, yes, you may have confused me w/ G33KatWork. no worries | 01:15 |
| mithro | GChriss: it appears so, you don't have any hardware right? | 01:16 |
| GChriss | different, semi-unrelated hardware | 01:16 |
| xfxf | mithro: what is the right order to get encoding working in a script? | 01:17 |
| xfxf | i have | 01:17 |
| xfxf | PROG=fpgalink make load; make load-fx2; make load-lm32 | 01:17 |
| xfxf | that gets me a prompt but no v4l device | 01:17 |
| xfxf | i quit, make load-fx2; make connect-lm32 | 01:17 |
| xfxf | then its fine | 01:17 |
| xfxf | by encoding i mean the encoder on hdmi2usb | 01:18 |
| xfxf | + the v4l device showing | 01:18 |
| xfxf | er, the latter only rather | 01:18 |
| mithro | PROG=fpgalink make load-gateware; make load-fx2; make connect-lm32; should work | 01:33 |
| CarlFK | prolly need to enable the encoder cuz it defaults to off | 01:34 |
| xfxf | it doesn't, have to do steps above | 01:54 |
| xfxf | and iirc the v4l device appearing is seperate to encoder being on/off | 01:55 |
| xfxf | oh, some interesting data for you as well mithro | 01:55 |
| xfxf | the projectors used at this venue are 16:10 | 01:55 |
| xfxf | they do wierd things to 16:9 resolutions | 01:55 |
| xfxf | 720p gets chopped off at the edges | 01:55 |
| xfxf | this is same venue as lca2017 | 01:56 |
| *** travis-ci has joined #timvideos | 02:02 | |
| travis-ci | [mithro/HDMI2USB-misoc-firmware/master#29] (bb302c0): The build failed. (https://travis-ci.org/mithro/HDMI2USB-misoc-firmware/builds/88029424) | 02:02 |
| *** travis-ci has left #timvideos | 02:02 | |
| mithro | xfxf: we can add scaling / padding stuff if we can get people to contribute | 02:02 |
| xfxf | yeah, figured - it's not critical now, i just shoved a scaler in front | 02:03 |
| xfxf | but just making you aware if you're not already that 720p isn't always 100% guaranteed to work | 02:03 |
| xfxf | but it's probably the closest res to being that | 02:03 |
| xfxf | hey | 02:03 |
| xfxf | i've written a super basic pyserial/tkinter frontend to change inputs | 02:03 |
| xfxf | is there anything special i need to do to talk to the device? | 02:04 |
| xfxf | opening a serial connection to /dev/ttyV-whatever it is at 115200 isn't sufficient | 02:04 |
| xfxf | + same as minicom | 02:04 |
| xfxf | i note flterm uses a --kernel paramater? | 02:04 |
| xfxf | parameter* | 02:04 |
| mithro | xfxf: that is only to load a kernel | 02:04 |
| xfxf | is that just for initial load, or every time we wish to connect? | 02:04 |
| xfxf | right | 02:04 |
| mithro | xfxf: put your code up somewhere? | 02:05 |
| xfxf | outside of 115200 what other serial options should i be setting? | 02:05 |
| xfxf | oh god its awful code | 02:05 |
| xfxf | but sure | 02:05 |
| xfxf | i can reproduce by trying to use minicom though | 02:05 |
| xfxf | https://www.irccloud.com/pastebin/HOL6v68f/ | 02:09 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 02:09 |
| mithro | xfxf: and what output do you get? | 02:09 |
| xfxf | the only output are my 'print's, which work as expected | 02:10 |
| xfxf | isn't doing anything to hdmi2usb though | 02:10 |
| xfxf | and tryingto connect to it with minicom doesn't work eithr | 02:10 |
| xfxf | whoa lag | 02:10 |
| mithro | I don't see you ever reading from the serial port? | 02:11 |
| xfxf | i'm not, just writing | 02:12 |
| xfxf | that said it's been a while since i've used pyserial | 02:12 |
| xfxf | i can figure that out | 02:12 |
| mithro | xfxf: so, how do you know it isn't giving you errors or something? | 02:12 |
| xfxf | i just need to know if anything special is required to talk to it | 02:12 |
| xfxf | if not, i'll keep debugging | 02:12 |
| xfxf | again, my first step after hitting this was to try with minicom | 02:12 |
| xfxf | i cannot talk to it with minicom | 02:13 |
| xfxf | hence pinging you | 02:13 |
| xfxf | and ya, next step is to implement a display that reads from the serial port, but i'm 10 minutes into writing this :P | 02:13 |
| xfxf | note EXPECT= is defined but i'm doing nothing with it | 02:14 |
| xfxf | which i should probs rename | 02:14 |
| mithro | hrm.... | 02:14 |
| mithro | SerialException: device reports readiness to read but returned no data (device disconnected?) | 02:14 |
| mithro | xfxf: what about "screen /dev/ttyVIZ0 115200" ? | 02:22 |
| mithro | xfxf: you might need to turn off hardware flow control? | 02:23 |
| xfxf | ah, possibly | 02:24 |
| mithro | xfxf: okay - I have minicom working here | 02:26 |
| xfxf | what settings? | 02:26 |
| mithro | hrm... minicom is acting weird.... | 02:28 |
| mithro | it works for a bit then stops.... | 02:29 |
| xfxf | fix it :P | 02:29 |
| xfxf | in all seriousness | 02:29 |
| xfxf | what is flterm doing differently to minicom? | 02:29 |
| mithro | xfxf: why are we using minicom? everything else seems to work fine? | 02:29 |
| xfxf | because i am trying to debug why pyserial isn't working | 02:33 |
| xfxf | going back to a known quantity (i.e. minicom) is how i'm debugging there's an issue at hdmi2usb somewhere | 02:33 |
| mithro | okay - the minicom was failing when screen was also attached to the serial port | 02:34 |
| mithro | where does minicom store its config files? | 02:36 |
| xfxf | ~/.minicom/ i'd assume | 02:39 |
| xfxf | what settings are you using to talk to it? | 02:39 |
| mithro | xfxf: nope | 02:40 |
| mithro | xfxf: no hardware flow control, 115200 | 02:40 |
| xfxf | aah | 02:41 |
| xfxf | yes | 02:41 |
| xfxf | turning off hardware flow control fixes it | 02:42 |
| xfxf | woo | 02:42 |
| xfxf | ghetto mixing | 02:42 |
| xfxf | haha | 02:42 |
| GChriss | ok, so working with CarlFK using real hardware we have the following updated pipeline: | 02:43 |
| GChriss | https://wiki.xiph.org/GST_cookbook#HDMI2USB.tv_capture_.28timvideos.us.29 | 02:43 |
| tpb | Title: GST cookbook - XiphWiki (at wiki.xiph.org) | 02:43 |
| mithro | xfxf: so your script works now? | 02:45 |
| GChriss | ^-- at least capturing at 15FPS, not sure if the board can go higher @ JPEG quality 85 | 02:46 |
| xfxf | mithro: yes | 02:46 |
| xfxf | i'm recording a talk right now with it | 02:47 |
| mithro | xfxf: are you reading anything back from the comm port? | 02:47 |
| xfxf | not implemented yet | 02:48 |
| xfxf | but i will | 02:48 |
| xfxf | also i know you told me not to, but i'm using a hdmi exender | 02:52 |
| xfxf | working fine | 02:52 |
| CarlFK | xfxf: hdmi over cat5? | 02:53 |
| xfxf | this just passes thru data tho, doesn't appear to have laptops negotiate directly with it | 02:53 |
| xfxf | yes | 02:53 |
| xfxf | although | 02:53 |
| xfxf | this is slightly diff | 02:53 |
| xfxf | because behind the hdmi extender is a scaler/splitter | 02:54 |
| mithro | xfxf: add the following to your script | 02:55 |
| mithro | https://www.irccloud.com/pastebin/NDexUgGI/ | 02:55 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 02:55 |
| mithro | In the keyHandler bit | 02:55 |
| xfxf | right, nod | 02:55 |
| xfxf | yeah this is mega early, handling everything better i'll do | 02:55 |
| xfxf | sorry was saying before, had to run to solve an issue before | 02:58 |
| xfxf | i'm usinga hdmi scaler/spitter | 02:58 |
| xfxf | scaler is forced to 720p and goes to hdmi2usb | 02:58 |
| xfxf | other output goes to projector | 02:59 |
| xfxf | so i guess i'm cheating around any issue a extender might introduce | 02:59 |
| xfxf | scaler will just be shoving out 720p50 no matter what hdmi2usb tells it | 02:59 |
| xfxf | or any devices between it | 02:59 |
| mithro | xfxf: yes you are | 03:03 |
| mithro | which means you aren't really testing anything all that useful | 03:03 |
| xfxf | i'm testing _a_ usecase for this | 03:05 |
| xfxf | i can't jam the hdmi2usb directly into the presenter chain due to venue equipment here | 03:05 |
| mithro | xfxf: you might as well be using something like the magwell device then? | 03:06 |
| xfxf | magwell device? | 03:06 |
| xfxf | i care about being able to use this *now* | 03:06 |
| xfxf | which is what i'm doing | 03:06 |
| CarlFK | what sort of fps should we be able to get from the atlys? | 03:10 |
| mithro | CarlFK: what do you mean? | 03:11 |
| CarlFK | mithro: at quality 50, should we see 30fps? | 03:12 |
| mithro | CarlFK: Depends on if the JPEG can compress the input at quality 50 to fit into 480Mbit or not | 03:12 |
| mithro | you could get 30fps at quality 100 if jpeg can compress it enough | 03:13 |
| CarlFK | k | 03:13 |
| CarlFK | does motion affect this? | 03:13 |
| GChriss | "motion" is a inter-frame thing, jpeg "image complexity" is another | 03:14 |
| GChriss | waterfalls and fire are the some of the hardest thing to encode on both counts | 03:15 |
| xfxf | can we test this? this needs to perform in all situations at some defined rate | 03:15 |
| xfxf | having the device freak out mid-conference won't be fun | 03:15 |
| CarlFK | xfxf: doing that now | 03:16 |
| xfxf | also if this is something that could change as part of development on the board, should testing this be part of the automated test suite? | 03:17 |
| mithro | xfxf: if you can figure out a way to automate the testing, then please go ahead and implement it | 03:24 |
| xfxf | what happens if the jpeg encoder can't keep up, the framerate changes? | 03:24 |
| xfxf | or something else? | 03:24 |
| GChriss | xfxf: at least on the gstreamer side there's an element to create a perfect stream even if the input framerate falters/increases: videorate | 03:32 |
| xfxf | ah, that's nice to know, thanks | 03:32 |
| GChriss | it's required working with elphel cameras, which produce anything from 0.5 - 60+ FPS on the fly depending on lighting and camera controls | 03:33 |
| GChriss | that said I'm not sure how to get the HDMI2USB board above 15FPS. maybe a v4l2-related issue? I'm new to the board | 03:35 |
| xfxf | framerate=25/1 or 30/1 didn't work for me | 03:41 |
| xfxf | question GChriss given you seem to be pretty switched on with gstreamer | 03:41 |
| mithro | xfxf: What does "didn't work for me" actually mean? | 03:42 |
| mithro | I can't seem to make this reliable -> https://gist.github.com/mithro/0717f9a0b39a538e1e6d | 03:42 |
| tpb | Title: Modified Ryans script · GitHub (at gist.github.com) | 03:42 |
| mithro | Occasionally pyserial just won't return data | 03:43 |
| xfxf | is there any way to implement functionality into a filesink so i can trigger an event (perhaps a key push, or something else) that'll close the current file, open up a new file, write whatever header data out is required for the encapsulation format, and start streaming new data to it? | 03:43 |
| xfxf | mithro: your read didn't work, i left that uncommented | 03:43 |
| xfxf | by reliable i mean i just recorded an entire talk with it | 03:43 |
| xfxf | re functionality above, i want to recreate a 'cut' without having to stop the whole gstreamer pipeline + run it again (which takes a few seconds to start) | 03:44 |
| GChriss | xfxf: yes, but you have to have python chops (or similar programing) skills | 03:45 |
| xfxf | i.e. push or do something, a new file starts being written out with minimal frames lost | 03:45 |
| xfxf | right, i can do python | 03:45 |
| CarlFK | xfxf: GChriss - what about sending a SIGHUP or some such signal? | 03:45 |
| CarlFK | xfxf: I have been thinking about this cut thing... | 03:46 |
| xfxf | er, s/uncommented/commented/ above, mithro - i'm still testing recording so i'll muck about with implementing more functionality later | 03:46 |
| xfxf | i'll turn it into a ghetto dvswitch | 03:46 |
| xfxf | so it'll invoke the gstreamer pipeline too | 03:46 |
| CarlFK | breaking the file is really just a way to send time stamps to veyepar | 03:46 |
| xfxf | ah, i guess so | 03:47 |
| xfxf | as you're using mlt | 03:47 |
| CarlFK | there are some handy side effects | 03:47 |
| xfxf | i'd like to still break the file though | 03:47 |
| xfxf | makes it easier to copy around files | 03:47 |
| CarlFK | I would like it too.. but it isn't happing | 03:47 |
| xfxf | and means your veyepar implementation works with third party devices that save down to files, not just this | 03:47 |
| CarlFK | meh | 03:47 |
| CarlFK | break the files during breaks when we don't care if we lose 5 seconds of content | 03:48 |
| xfxf | it's more volunteers breaking talks too early | 03:48 |
| CarlFK | meh | 03:48 |
| xfxf | i'd prefer to do this as properly as possible | 03:48 |
| xfxf | i'm okay with compromises but i'm sure there's a solution here | 03:49 |
| CarlFK | well do it! :D | 03:49 |
| CarlFK | till then.. I am trying to figure out how to work with what we have | 03:49 |
| CarlFK | and a text file of cut times might be good enough | 03:49 |
| *** GChriss has quit IRC | 03:49 | |
| *** GChriss has joined #timvideos | 03:50 | |
| GChriss | ok back, power outage | 03:50 |
| CarlFK | doh | 03:50 |
| xfxf | i kinda like cutting the file anyway | 03:50 |
| CarlFK | you didn't miss anything | 03:50 |
| CarlFK | xfxf: agree, but it isn't happening today. | 03:50 |
| xfxf | it is | 03:50 |
| xfxf | i'm restarting the script :P | 03:50 |
| GChriss | gst-launch is a rapid prototyping tool, gstreamer proper is really a programming thing | 03:50 |
| CarlFK | lol | 03:50 |
| xfxf | GChriss: yeah, understood, probably worth me learning more about it then | 03:50 |
| CarlFK | xfxf: you are losing frames. | 03:51 |
| xfxf | GChriss: if you say you can do this in python or whatever has gstreamer bindings tho then that's good, means this is achievable | 03:51 |
| CarlFK | if MaZderMind will add the cutbuton to vocto, and all it does is write time stamps to a file ... | 03:51 |
| CarlFK | I think we have a mixer | 03:52 |
| xfxf | CarlFK: i hear the argument you're making but i've been burnt enough in the past to know when i am shaving yaks | 03:52 |
| GChriss | _cleanly_. there's shared interest in fixing gstreamer up specifically for conferences but it's not implemented | 03:52 |
| xfxf | i think this is at least spending some time trying to solve | 03:52 |
| xfxf | GChriss: good to hear | 03:52 |
| CarlFK | xfxf: what? | 03:52 |
| xfxf | CarlFK: you're saying not to bother trying to implement the cut functionality, no? | 03:52 |
| xfxf | i've been burnt in the past trying to save entire days as a single file so i'd prefer to try implementing it still | 03:53 |
| CarlFK | xfxf: for now | 03:53 |
| GChriss | personally I'm excited about the prospect of openFrameworks integration with live pipelines. tap a face, type in who's speaking, add a few keywords, monitor the livestream, etc, from a tablet | 03:53 |
| CarlFK | xfxf: vocto chunks it into 5 min blocks | 03:53 |
| xfxf | erk | 03:54 |
| CarlFK | yeah, I am not thrilled about that. but again... it exists | 03:54 |
| xfxf | my goal is basic switching as rapidly as possible | 03:54 |
| xfxf | hence this, the switcher/scaler/etc | 03:54 |
| xfxf | once that works then i'll iterate out into more complexity | 03:55 |
| CarlFK | I don't think using hdmi2usb for mixing is a good idea | 03:55 |
| xfxf | why? | 03:55 |
| xfxf | it has two HDMI inputs | 03:55 |
| xfxf | we usually have two HDMI sources | 03:55 |
| xfxf | we can't get two streams from it at the same time | 03:56 |
| CarlFK | for the moment, I want to save the sources | 03:56 |
| xfxf | and post-produce? | 03:56 |
| CarlFK | no | 03:56 |
| xfxf | because f*k that, i don't :) | 03:56 |
| CarlFK | well.. for the moement, yes | 03:56 |
| xfxf | have fun :) | 03:56 |
| CarlFK | but I think we are weeks away from a mixer | 03:57 |
| xfxf | i would hope so too but i'm working off what i have now | 03:57 |
| xfxf | if my work is thrown away in 4 weeks then i don't care, better outcome for everyone | 03:57 |
| xfxf | if it's not at least i have an option i can confidently recommend to conferences moving forward | 03:57 |
| CarlFK | not if you soak up GChriss's time :0 | 03:57 |
| xfxf | being in "this will work anyday soon now" territory makes me extraordinarily uncomfortable | 03:58 |
| CarlFK | I see your project right up there with gst-switch | 03:58 |
| xfxf | none of this seems difficult to implement | 03:58 |
| CarlFK | well.. reality suggests othewise | 03:58 |
| CarlFK | vocto would have been used in production a week or so ago but client call or something yanked MaZderMind away | 03:59 |
| CarlFK | debconf is going to hack on it in a few weeks | 03:59 |
| xfxf | if it is i'm absolutely fine abandoning the effort | 03:59 |
| xfxf | that's fine, if that's ready, rad, will switch to it | 03:59 |
| CarlFK | well don't suck up reasources in your abandon ware :) | 03:59 |
| CarlFK | vocto builds and runs pretty cleanly | 04:00 |
| GChriss | I'll volunteer that the folks over at pitivi are really well versed with gstreamer pipelines (and, um, #gstreamer too but they're usually hacking away at the next gstreamer release) | 04:00 |
| CarlFK | GChriss: we need someone like you to actually do stuff | 04:01 |
| GChriss | i just glue things together! but happy to help | 04:01 |
| CarlFK | https://github.com/voc/voctomix | 04:01 |
| tpb | Title: voc/voctomix · GitHub (at github.com) | 04:01 |
| CarlFK | like add a button to it's UI | 04:01 |
| CarlFK | that does something | 04:02 |
| CarlFK | and MaZderMind (lead) will accept | 04:02 |
| CarlFK | which is currently up for debate | 04:02 |
| CarlFK | personally, I would like to see it log a time stamp to a text file | 04:03 |
| CarlFK | ideally it would send a signal to the process writing the file | 04:04 |
| CarlFK | and end/start a new file | 04:04 |
| CarlFK | the lack of Cut button is a blocker for quite a few people/events | 04:04 |
| GChriss | agreed | 04:07 |
| GChriss | I will volunteer that if you jot down the wall time when you know you will create a break in the file later that is most of the work | 04:07 |
| CarlFK | ? | 04:07 |
| GChriss | the `! filesink location=recording_`date +%s`.mkv part records the unix timestamp of when the recording was started as part of the file | 04:08 |
| mithro | xfxf: rotating files is hard | 04:08 |
| CarlFK | GChriss: take a look at: http://veyepar.nextdayvideo.com/main/recording_sheets/311/.pdfosdc2015_recording_sheets.pdf | 04:09 |
| GChriss | so convert "wall time of the break" to unix timestamp, do a subtraction, and you get durations to chop from the main file | 04:09 |
| CarlFK | GChriss: that pdf gets printed and put next to the mixer (in the room where the talk is ) | 04:10 |
| CarlFK | someone writes down the hh:mm:ss on the paper when the video should start | 04:10 |
| GChriss | ^== I promise I was never this organized | 04:10 |
| CarlFK | at the end of the day, that data gets put into a django app that has the hh:mm:ss in the db, along with the false starts (are we ready yes. start... start now.. hey!!! start talking!!! 20 seconds later... startnow! | 04:11 |
| CarlFK | so there may be a few start times to pick from in the db. the one written on the paper is what gets [x]apply | 04:12 |
| CarlFK | same for the end. it isn't always obvious, so there may be a few choices. the one on the paper [x]apply, hit save, onto the next one. | 04:12 |
| CarlFK | we could write a little bash script that basically does this | 04:13 |
| CarlFK | but having it in it's own ... gui? will be problematic | 04:13 |
| CarlFK | hmm... I may want to hack up that shell script for Node conf in 20 days | 04:16 |
| xfxf | https://usercontent.irccloud-cdn.com/file/nU0KW03w/20151029_151839.jpg | 04:34 |
| CarlFK | I see dvswitch :p | 04:34 |
| mithro | Is that Chris and Katie? | 04:35 |
| xfxf | yes | 04:39 |
| *** Sewar has quit IRC | 04:58 | |
| xfxf | https://youtu.be/iQhhdhz232Q | 04:59 |
| xfxf | be good to raise it over 15fps | 05:00 |
| mithro | xfxf: what are you using to do that? | 05:03 |
| xfxf | the gstreamer pipeline + my script | 05:03 |
| mithro | xfxf: your script to do the switching? | 05:04 |
| mithro | and just a gstreamer pipeline direct to youtube? | 05:05 |
| mithro | xfxf: so, I get the behavior here were if you don't read from the serial port, writes eventually stop working | 05:05 |
| mithro | xfxf: I'm guessing you are not seeing that? | 05:07 |
| *** Sewar has joined #timvideos | 05:16 | |
| xfxf | nope | 05:17 |
| xfxf | and nah I saved it to disk, encoded to mp4 using ffmpeg, then manually uploaded | 05:18 |
| xfxf | this is a veeery early test | 05:18 |
| mithro | xfxf: oh, so it wasn't a life stream | 05:18 |
| mithro | I just assumed it was live | 05:19 |
| GChriss | xfxf: you're using today (yesterday's) piplines? have you tried uploading the mkv directly to youtube? | 05:24 |
| GChriss | not sure if it works w/o transcoding but worth a try | 05:24 |
| CarlFK | GChriss: xfxf and mithro live in the future. everything we do we did yesterday ;) | 05:25 |
| GChriss | it's going to take me a while to wrap my head around this | 05:25 |
| xfxf | mithro: doing one step at a time. will def try streaming it next w/ flumotion | 05:28 |
| GChriss | "worth a try" ==> probably better with short videos | 05:29 |
| mithro | xfxf: what frame rate is the encoder on the atlys reporting? | 05:32 |
| mithro | seaLne: it would be good to understand why the rename script didn't work for you | 05:34 |
| mithro | anyone handy with ruby or python and know the github API? | 05:43 |
| CarlFK | mithro: github API rings a bell but I can't remember where. what are you looking to do? | 05:48 |
| mithro | CarlFK: I need a tool which will help manage the hooks for all the repos under the timvideos organisation | 05:48 |
| mithro | CarlFK: basically, want to make sure that all the HDMI2USB-*** repos are set up correctly with IRC and email hooks | 05:49 |
| CarlFK | mithro: hmm.. I am still drawing a blank. but it is 1am, and I am starting to blank on everything | 05:52 |
| mithro | CarlFK: Hooks are things which run when you push stuff into a github repo | 05:53 |
| mithro | CarlFK: the things which make the IRC announcements occur | 05:54 |
| CarlFK | mithro: yeah, I get what you want. I am pretty sure I have talked with someone in the last few weeks that has said "git api" | 05:54 |
| CarlFK | and they probably know python. | 05:54 |
| mithro | CarlFK: well, see if you can rope them in for me :) | 05:55 |
| mithro | I'll also poke in #github and see if something already exists | 05:55 |
| CarlFK | mithro: of course. but first I ahve to remember what country I was in ;) | 05:55 |
| CarlFK | #github is a good idea. I get good results normally | 05:55 |
| *** springermac has quit IRC | 07:17 | |
| *** springermac has joined #timvideos | 07:19 | |
| *** se6astian|away is now known as se6astian | 08:19 | |
| *** Bertl_zZ is now known as Bertl | 08:26 | |
| *** rohitksingh has joined #timvideos | 08:30 | |
| *** rohitksingh has quit IRC | 08:56 | |
| *** travis-ci has joined #timvideos | 09:11 | |
| travis-ci | [timvideos/HDMI2USB-misoc-firmware/master#256] (8a2bf53): The build was fixed. (https://travis-ci.org/timvideos/HDMI2USB-misoc-firmware/builds/88069255) | 09:11 |
| *** travis-ci has left #timvideos | 09:11 | |
| *** rohitksingh has joined #timvideos | 09:12 | |
| xfxf | mithro: sorry, equipment all unplugged now. now i can actually record something i'm gonna start attacking local user groups now | 09:16 |
| xfxf | so i can provide more debug info | 09:16 |
| *** rohitksingh has quit IRC | 09:26 | |
| *** rohitksingh has joined #timvideos | 09:39 | |
| mithro | Bertl: ping? It appears I can't count.... | 11:46 |
| *** Bertl is now known as Bertl_oO | 12:25 | |
| *** se6astian is now known as se6astian|away | 16:09 | |
| *** se6astian|away is now known as se6astian | 17:36 | |
| *** rohitksingh has quit IRC | 17:50 | |
| *** rohitksingh has joined #timvideos | 18:05 | |
| *** rohitksingh has quit IRC | 19:00 | |
| *** Bertl_oO is now known as Bertl_zZ | 20:18 | |
| *** se6astian is now known as se6astian|away | 22:17 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!
