[Ffmpeg-devel-irc] ffmpeg.log.20180108

burek burek021 at gmail.com
Tue Jan 9 03:05:02 EET 2018


[00:36:38 CET] <Hexian> even though this is probably not the best place to ask, I'm completely out of ideas at this point, so before giving up on this completely, it's worth a shot
[00:38:08 CET] <Hexian> when I output an ogg opus stream from ffmpeg to a janus rtp endpoint that expects opus/48000/2 and has a payload type 111, the quality of the very simple tone test that is played in a browser is extremely bad and constant drops out
[00:38:27 CET] <Hexian> if I output to a file instead of the RTP server, the quality is perfect
[00:39:31 CET] <Hexian> I'm testing with a VM on my local machine and the WebRTC connection is via the host-only network interface, so even if network conditions could explain such a huge quality issue on a simple tone test, that can't be the case here
[01:14:22 CET] <tex> Sorry I dropped off, but can anyone advise what is the procedure to create a MPEGTS from a text data file and a mp4 file.
[01:15:19 CET] <tex> I have tried -f bin and -f data but Im not sure if I can just "read" those files as such
[03:34:21 CET] <kazuma_> https://i.imgur.com/dv1FjBD.png
[03:34:35 CET] <kazuma_> anyway to supress the spam while maintaining the encoding line statistics?
[03:35:02 CET] <c_14> -q quiet -stats
[03:35:27 CET] <c_14> *-v quiet
[03:35:45 CET] <kazuma_> thanks :)
[04:22:20 CET] <hendry> hi, I have to process several large files with ffmpeg. Is there some el cheapo way of establishing some queue with ffmpeg? I guess once it's done, i need it scan an incoming directory for new files it needs to process
[04:35:36 CET] <DHE> seems highly shell scriptable...
[04:35:59 CET] <DHE> ffmpeg itself doesn't have that kind of intelligence. it just does the video work
[04:43:04 CET] <furq> if you need to watch a folder and you're on linux then use inotifywait
[04:43:23 CET] <furq> there are doubtless similar tools for other systems
[05:26:48 CET] <hendry> was wondering if there was some goto for this sort of automation / pipeline work
[06:21:39 CET] <dystopia_> hello
[06:21:53 CET] <dystopia_> i use "scale=720:-4" when resizing stuff to maintain aspect ratio
[06:22:24 CET] <dystopia_> im trying to encode an anamorphic dvd, and when using that to scale, it's becoming 4:3 instead of 16:9
[06:22:27 CET] <dystopia_> any ideas?
[06:22:47 CET] <dystopia_> -sws_flags spline -vf yadif=0:0,crop=716:552:2:12,scale=720:-4,setsar=1/1
[06:23:20 CET] <dystopia_> gived me an output of 716x552 instead of 716x384
[06:25:05 CET] <dystopia_> hmm manually set it to 716x384 and it's still not 16:9
[06:34:10 CET] <dystopia_> setting dar fixed it, nvm
[13:19:38 CET] <lozarythmic> Just had a convo with the guys in ffmpeg-devel that ended up getting punted here: https://pastebin.com/59yXi4sd
[13:21:08 CET] <lozarythmic> Long story short: I'm trying to work out if mine and a coworkers methods of extracting greyscale frames are equal. His way is long winded and ends up with a greyscale pgm file, mine is quicker and ends up with a greyscale .png file
[13:21:30 CET] <lozarythmic> Both files look pretty identical
[13:22:13 CET] <sfan5> you can also run checksums on the raw pixel data to determine whether they really *are* identical
[13:23:34 CET] <lozarythmic> Would I need to strip the headers from the files first?
[13:24:03 CET] <sfan5> no that wouldn't work as png and pgm files are totally different
[13:24:38 CET] <lozarythmic> Ah
[13:25:44 CET] <sfan5> you could use something like  ffmpeg -loglevel quiet -i file_one.png -f md5 -  to get a checksum over the raw pixel data
[13:26:44 CET] <sfan5> if you do that make sure ffmpeg decodes both files into identical pixel formats, e.g. add -pix_fmt rgb24
[13:31:16 CET] <lozarythmic> I'll have a play. Thanks sfan5
[13:42:58 CET] <atomnuker> lozarythmic: just use the extractplanes filter to get the y plane
[14:00:39 CET] <klaxa> >buffer	Memory block for input/output operations via AVIOContext. The buffer must be allocated with av_malloc() and friends. It may be freed and replaced with a new buffer by libavformat. AVIOContext.buffer holds the buffer currently in use, which must be later freed with av_free().
[14:00:41 CET] <klaxa> >It may be freed and replaced with a new buffer by libavformat.
[14:00:46 CET] <klaxa> has this always been the case? :/
[14:31:26 CET] <tagteam> I am having difficulty with installing the ffmpeg-php extension. I have done a lot of research and have determined that the issue was from a change in ffmpeg. My inability to get ffmpeg-php to compile started somehwere between three and six months back. Look at my pastebin info between lines 32 and 37. Here is my pastebin link. https://pastebin.com/jw7e2Qes
[14:33:55 CET] <klaxa> tagteam: ffmpeg-php is not officially supported in here, not sure if it is even actively maintained
[14:34:44 CET] <klaxa> you can also tell it was not really maintained by all the deprecation warnings
[14:35:09 CET] <tagteam> I know. I am trying to determine what change the ffmpeg team made to cause the failure I can get it to compile as it should on older ffmpeg releases.
[14:35:21 CET] <klaxa> you can try to replace instances of decoder_ctx->codec_name; with decoder_ctx->codec->name;
[14:36:04 CET] <tagteam> OK I will give that a go.
[14:36:37 CET] <BtbN> Basically, go back to the last ffmpeg version that still worked, and fix all the deprecation warnings.
[14:37:09 CET] <BtbN> There's an APIchanges doc that documents most/all of them, and their documentation usually mentions the replacement
[14:41:53 CET] <lozarythmic> atomnuker, i'll try that thanks!
[15:46:14 CET] <tagteam> Thanks BtbN I was able to get ffmpeg compile and work properly. I know ffmpeg-php is off topic but could you give me any guidance on fixing the depricated items showing up in my make?
[15:47:21 CET] <SortaCore> why is pts/dts completely inaccurate for the first 5 or so packets?
[15:47:26 CET] <SortaCore> way too low
[15:52:15 CET] <SortaCore> looks like it starts packet_pts_time at -1 for some reason
[15:52:52 CET] <SortaCore> off:-953333 off_time:-0.953333
[15:59:33 CET] <BtbN> It's pretty normal that the first few timestamps are negative
[15:59:59 CET] <BtbN> Caused by re-ordering
[16:04:12 CET] <SortaCore> I get warnings about non-monotonic DTS though
[16:04:57 CET] <BtbN> your source material is weird/damage then
[16:05:35 CET] <SortaCore> H264 over RTSP over TCP
[16:19:17 CET] <SortaCore> that sound normal for that protocol BtbN?
[18:05:52 CET] <therage3> so i was trying to encode an mkv stream copy from a dvd to x264 transcode, and during the conversion i got this https://pastebin.com/raw/67qbM88U
[18:06:10 CET] <therage3> the lines near the bottom that say "ac-tex damaged at 10 105.58"
[18:06:12 CET] <therage3> what exactly is that
[18:10:01 CET] <sfan5> there were errors in your input mkv
[18:11:09 CET] <therage3> m
[18:11:11 CET] <therage3> hm*
[18:11:12 CET] <therage3> I see
[18:11:19 CET] <sfan5> there will be minor graphical glitches at the affected location in the output file
[18:11:55 CET] <therage3> and will those also be there in the input file? so I can go look at a specific frame/timestamp to see what's going on there. the "10 105.58" figure, what is that?
[18:12:32 CET] <sfan5> yes they will be there in the input
[18:13:07 CET] <therage3> i see. what does 10 105.58 mean here? like, frames, or hours/minutes/seconds...?
[18:13:16 CET] <sfan5> that's not a timestamp or anything, it's a "corruption" of the status like and error message
[18:13:22 CET] <therage3> ah
[18:13:46 CET] <sfan5> also next time you encode you might want to use -tune film
[18:14:12 CET] <therage3> is that for a specific purpose?
[18:16:55 CET] <sfan5> it's named "film" but is reccommend for use with most kinds of video content
[18:17:35 CET] <therage3> ah i see
[18:17:44 CET] <therage3> i'll try to remember
[18:17:55 CET] <therage3> hm now how to find the problematic timestamps here
[19:31:55 CET] <stephen> Hello, I am getting Non-monotonous dts in output stream and past duration too large errors when trying using use_wallclock_as_timestamps 1 audio is also out of sync(behind) with the video
[19:32:09 CET] <stephen> ffmpeg.exe -f dshow -framerate 29.97 -rtbufsize 10M -use_wallclock_as_timestamps 1 -i video="Osprey-700 HD Video Device 1":audio="SDI Input 1 (Osprey-700 HD 1)" `
[19:32:09 CET] <stephen> -f flv -pix_fmt yuv420p -vcodec libx264 -x264opts no-scenecut -vprofile baseline -vf scale=640:360 -sws_flags lanczos -b:v 650k -preset veryfast -g 30 `
[19:32:09 CET] <stephen>  -level 3.1 -c:a aac -ab 64000 -ar 48000 -ac 2  'rtmp://boombingo.vorigin.scaleengine.net/boombingo-origin/live/boomtest_1'
[19:34:33 CET] <kepstin> stephen: the desync and past duration too large warning messages can probably both be attributed to the "-use_wallclock_as_timestamps" option - why did you add that?
[19:35:14 CET] <stephen> We want to use the system time instead of the device
[20:22:05 CET] <harpsicord1> is it possible to make an ISO file of a DVD that has some form of error in it? or would the program fail? reason i ask is, i have an ISO file that i mounted of a DVD i have back home (on vacation right now) and when i use makemkv to extract one of the titles and then use ffmpeg to transcode it, ffmpeg starts complaining about a damaged mkv input
[20:24:49 CET] <kepstin> harpsicord1: it would be helpful to see the error messages from ffmpeg
[20:25:11 CET] <harpsicord1> kepstin gotcha hold on
[20:27:52 CET] <harpsicord1> kepstin https://hastebin.com/poxocijuje.pas
[20:28:25 CET] <kepstin> hmm, so that's not an error, it's a warning
[20:28:59 CET] <kepstin> says that there's one corrupt frame that it's working around, should result in no more than a second of video with issues, since dvd has keyframes at really short intervals
[20:29:02 CET] <pmjdebruijn> harpsicord1: you have tried ddrescue ?
[20:29:35 CET] <harpsicord1> thats what i dont get, if the iso was copied perfectly what corrupt frame would there be even
[20:29:46 CET] <kepstin> harpsicord1: if you watch the video and don't see any issues then don't worry about it. Otherwise, you might try to re-rip the dvd to see if it was a bad read
[20:30:01 CET] <harpsicord1> because before i left home the md5 verify string matched the dvd
[20:30:07 CET] <harpsicord1> so i cant explain this
[20:30:14 CET] <kepstin> alright, then it's probably a video issue in the original dvd
[20:30:17 CET] <harpsicord1> what
[20:30:23 CET] <harpsicord1> on the original dvd??
[20:30:30 CET] <harpsicord1> this is a commercially pressed dvd
[20:30:42 CET] <therage3> yeah, it isn't unusual for mastering or pressing errors to occur
[20:30:44 CET] <therage3> it happens
[20:30:49 CET] <harpsicord1> what
[20:31:15 CET] <kepstin> as I said, the ffmpeg message is just a warning, it'll only affect a very small section of the video - if you don't notice anything when watching, then don't worry about it.
[20:31:37 CET] <harpsicord1> and whats even more annoying is that where this warning supposedly occurs theres not even any artifacting or anything in the audio going wrong its just so weird
[20:31:47 CET] <sfan5> what if your dvd drive is introducing an error?
[20:32:04 CET] <kepstin> nah, dvds are data encoding with ecc and stuff, that's very unlikely
[20:32:13 CET] <therage3> what the fuck? ok i just looked at your warning message.
[20:32:23 CET] <therage3> That exact message is what i just got for a thing I asked about here a while ago
[20:32:25 CET] <kepstin> harpsicord1: if there's no visual problem, that's probably why they just left it as is and didn't bother re-pressing
[20:32:25 CET] <therage3> o.O
[20:32:40 CET] <harpsicord1> therage3 haha for real?
[20:33:15 CET] <kepstin> it could even have been a bug in the mpeg2 encoder they used to make the dvd, who knows :/
[20:33:19 CET] <therage3> I mean the numbers are obviously different, but it's that same pattern, first the ac-tex damaged, then "Warning MVs not available", then the concealing stuff
[20:33:38 CET] <harpsicord1> ok heres what i need to do
[20:33:57 CET] <harpsicord1> i need to borrow this dvd from a friend who i know also has it and bought it with me, rip that too with maybe another drive to make extra sure
[20:34:08 CET] <harpsicord1> if the same error happens in the exact same place, the publisher messed up
[20:34:47 CET] <sfan5> if it doesn't it can still be a pressing error
[20:34:53 CET] <therage3> yeah, that would be a pressing/mastering issue, but again, since you can't notice it, someone at their studio may have even noticed the minor error but since it isn't visually detectable, they left it alone
[20:35:29 CET] <therage3> I don't know how the DVD pressing business works, but for vinyls first the pressing plant sends a sample vinyl to the record label
[20:35:36 CET] <therage3> who then play it and OK if it it's fine
[20:36:16 CET] <kepstin> I'd assume that doing the setup for a new different pressing is (relatively) expensive, so they want to avoid doing that if possible, yeah.
[20:36:25 CET] <therage3> I'm guessing something similar happens with DVDs unless they want some massive callbacks and lose a ton of money
[20:38:59 CET] <harpsicord1> yeah
[20:39:37 CET] <harpsicord1> ok so i messaged my friend and told him to make a verified copy of the DVD with his drive and do tests with it
[20:40:42 CET] <therage3> before your friend does that, ask him to wipe it down, handle it from the edges only, get rid of fingerprints and dust, etc.
[20:42:17 CET] <kepstin> dvds do have pretty strong ecc, so i would expect a reported IO error rather than a bad read, generally.
[20:42:33 CET] <therage3> yeah
[20:42:46 CET] <therage3> harpsicord1: this is a DVD meant to be played on a DVD player right?
[20:43:08 CET] <harpsicord1> yeah, its a DVD9 (dual layer) that you put into DVD players
[20:43:30 CET] <therage3> that's interesting because afaik, standalone DVD players are more finicky than computer optical drives
[20:43:50 CET] <therage3> and artifacting, if any, would be more likely to be seen on one of those than on a computer
[20:44:14 CET] <kepstin> well, they're supposed *not* finicky about media errors in general - they will generally try to play through errors rather than stopping
[20:44:31 CET] <kepstin> and artifacting is really limited by the keyframe interval being so short
[20:44:50 CET] <therage3> I see, so this is a question of looking at it frame by frame? o.o
[20:44:57 CET] <kepstin> (iirc <1s interval on ntsc, a bit longer on pal)
[20:45:40 CET] <harpsicord1> yeah lol i was gonna try but i dont have a DVD movie player here
[20:45:43 CET] <kepstin> most of the artifacting you'll see on a dvd is just due to the ridiculously low keyframe interval combined with the strict bitrate/buffer limits :(
[20:46:22 CET] <kepstin> dvds kinda suck for video quality, the main thing is that they *were* better than vhs in general.
[20:47:30 CET] <therage3> I see
[20:47:59 CET] <harpsicord1> lemme extract the frames in the vicinity of the warning
[20:48:03 CET] <harpsicord1> you gave me a good idea
[20:49:30 CET] <kepstin> lets see. this error is on a keyframe, so you'd be looking for a keyframe that has some smallish number of 16x16 macroblocks replaced with whatever ffmpeg did to conceal the error
[20:49:59 CET] <kepstin> if the frame is dark or doesn't have much detail, you might not be able to tell at all.
[20:50:02 CET] <harpsicord1> how did you know it was on a keyframe though? what part of the warning message showed that?
[20:50:30 CET] <kepstin> the message ends with "... in I frame"
[20:52:05 CET] <kerio> DVDs are horrible :|
[20:53:20 CET] <harpsicord1> well i didnt know "I frame" means keyframe you see
[21:00:49 CET] <alexp> kerio: i agree but probably for different reasons :) analog noise from VHS can't be captured without artifacting at the max DVD bitrate. that's a bigger problem for me
[21:03:47 CET] <harpsicord1> ok so i used "ffmpeg -i title.mkv -vf "select=eq(pict_type\,I)" -vsync vfr thumb%05d.jpg"   to generate only keyframes
[21:03:51 CET] <harpsicord1> is that the correct syntax?
[21:06:04 CET] <alexpigment> presumably this is interlaced, right? i wonder if it's going to be hard to see artifacts without deinterlacing or discarding fields
[21:06:53 CET] <harpsicord1> it has hard telecine yeah
[21:06:57 CET] <therage3> LOL
[21:07:03 CET] <therage3> my number one enemy.
[21:07:07 CET] <therage3> How I fucking hate telecine
[21:07:21 CET] <alexpigment> gotcha. well, i don't really understand all that's going on here, but i figured i'd throw that out there
[21:08:12 CET] <harpsicord1> well the tl;dr is that this seemingly perfectly copied iso has a specific vob file that has a minor error in one of the keyframes, and we are suspecting its something on the original dvd which indicates a pressing or a mastering issue
[21:09:32 CET] <alexpigment> harpsicord1: yeah, i got that vibe. i just don't know enough about the error to know if it should be visually obvious. but getting interlaced thumbnails will probably look like a mess anyway (e.g. half of the lines are from one shot half are from another)
[21:09:33 CET] <kepstin> harpsicord1: this sort of thing falls into the category of "if you don't see it when playing back the video at normal speed, it's not a problem"
[21:10:00 CET] <harpsicord1> kepstin definitely, still disconcerting when you see ffmpeg throwing warnings out which makes you worry
[21:10:09 CET] <alexpigment> welcome to ffmpeg ;)
[21:12:55 CET] <sfan5> harpsicord1: well you can add yadif before the select filter
[21:13:02 CET] <sfan5> but I'm not sure if that preserved the pict_type
[21:13:05 CET] <sfan5> preserves*
[21:13:05 CET] <harpsicord1> what does that do
[21:13:08 CET] <sfan5> deinterlace
[21:13:23 CET] <harpsicord1> ah
[21:14:16 CET] <therage3> Reminds me of RHCP's By the Way ... a song on there called The Zephyr Song has an autotune error near the end where the singer's voice unnaturally wobbles for a split second. it was a mastering error but it still made its way to consumer CDs
[21:14:29 CET] <alexpigment> yeah, i remember that
[21:14:31 CET] <therage3> and this is a) a major band, b) something you can clearly hear
[21:14:38 CET] <therage3> so if you can't even _see_ this, relax
[21:14:41 CET] <alexpigment> it's like "in the water where i *something* my emotions" etc
[21:14:52 CET] <alexpigment> or the next line
[21:14:53 CET] <therage3> yeah, "all the WORLD it passed me by"
[21:14:57 CET] <alexpigment> yep
[21:14:59 CET] <therage3> That next line, the "world" is fucked up
[21:15:03 CET] <alexpigment> yeah
[21:15:31 CET] <therage3> I bet Kiedis was all "fuck now everybody knows I use autotune ;)"
[21:15:32 CET] <alexpigment> there's a smashing pumpkins song that is similarly fucked up
[21:15:46 CET] <therage3> kind of surprising, because live he does seem to manage to sing in key
[21:15:50 CET] <therage3> not sure what that was about
[21:15:59 CET] <alexpigment> therage3: if you listen to by the way for 2-3 songs, you know autotune is being used ;)
[21:16:22 CET] <alexpigment> also, i don't know about singing in key live. the live at slane castle dvd was pretty bad
[21:16:29 CET] Action: kepstin notes that autotune can be used live
[21:16:35 CET] <therage3> was that the one where Flea is dressed up as a skeleton
[21:16:41 CET] <alexpigment> yeah
[21:16:43 CET] <therage3> in Ireland
[21:16:45 CET] <therage3> LMAO
[21:16:55 CET] <therage3> yeah, that was... I dunno, he sings well but you know, if you miss rehearsal
[21:16:57 CET] <therage3> Well, that happens
[21:17:07 CET] <alexpigment> anyway, john frusciante was the only person in that band who could carry a tune
[21:17:30 CET] <therage3> you mean vocally? or what. He was the guitarist up to Stadium Arcadium
[21:17:36 CET] Action: therage3 confused
[21:17:37 CET] <alexpigment> i just think anthony kiedis just doesn't "have it". some people can't sing well
[21:17:42 CET] <therage3> LOL
[21:17:45 CET] <alexpigment> yeah i meant vocally
[21:17:56 CET] <therage3> oh yeah, I mean Frusciante knows how to sing for sure
[21:18:08 CET] <therage3> it's just that you won't hear it that often in RHCP
[21:18:09 CET] <alexpigment> john frusciante sang backup on all the latter albums and of course put out like 10+ solo albums
[21:18:13 CET] <therage3> except as backing vocalist
[21:18:19 CET] <therage3> right
[21:18:29 CET] <alexpigment> he sings a few choruses as lead
[21:18:41 CET] <therage3> he is pretty good. I think all four of them are talented, I mean Flea... jeez, that guy can play too
[21:19:01 CET] <therage3> there's a show they did on SNL in the early 90s when they became really popular after Blood Sugar Sex Magik came out
[21:19:16 CET] <alexpigment> oh with john frusciante intentionally singing out of tune?
[21:19:21 CET] <therage3> LMFAO
[21:19:23 CET] <therage3> yeah
[21:19:30 CET] <therage3> And Kiedis kicked him at the end of Stone Cold Bush
[21:19:31 CET] <alexpigment> haha
[21:19:32 CET] <therage3> on live TV
[21:19:36 CET] <therage3> that was great
[21:19:48 CET] <alexpigment> yeah, i don't think john could sing well back then either though
[21:19:52 CET] <therage3> he tried to pass it off as accidental, but it's clear he aimed and slammed him
[21:19:52 CET] <alexpigment> even if he was exaggerating
[21:20:13 CET] <therage3> no, he did it deliberately I think because Frusciante was getting bored of all the attention the band was getting
[21:20:15 CET] <alexpigment> i'm not sure i wouldn't have done the same
[21:20:17 CET] <therage3> so he got into drugs
[21:20:20 CET] <therage3> and lost it for a while
[21:20:25 CET] <therage3> which of course pissed off everyone else in the band
[21:20:28 CET] <alexpigment> yeah, "lost it" is an understatement
[21:20:37 CET] <therage3> yeah, I know, the guy was fucked up
[21:20:53 CET] <harpsicord1> oh so thats why dave navarro showed up?
[21:21:00 CET] <alexpigment> yeah
[21:21:05 CET] <therage3> well, he was only on One Hot Minute
[21:21:08 CET] <alexpigment> navarro... [shakes head]
[21:21:09 CET] <therage3> love Aeroplane, great tune
[21:21:39 CET] <alexpigment> he just wasn't right for the band. they took a dude who was into heavy rock guitar and said "hey, play some funk guitar along to flea's bassline"
[21:22:15 CET] <harpsicord1> i cant get over how bad californication sounds
[21:22:26 CET] <therage3> blame Rick Rubin, the lord of the Loudness Wars
[21:24:06 CET] <harpsicord1> dude i heard it on vinyl and it was like night and day
[21:24:23 CET] <harpsicord1> everything on the vinyl is dynamic, its organic, it can breathe
[21:25:03 CET] <kepstin> if they'd used the same mastering levels on vinyl as on the cd, your needle would have jumped out of the track :/
[21:25:14 CET] <therage3> That's because vinyls require a different mastering, the lathe that cuts the master record can't go nuts so you can't record at those levels of loudness
[21:25:30 CET] <therage3> Otherwise the cutting head would jump out of the groove
[21:25:40 CET] <therage3> And that vinyl would be ruined
[21:26:17 CET] <therage3> Also, you can't have stereo bass on vinyls
[21:26:33 CET] <therage3> so mastering engineers for vinyls have to downmix that to mono
[21:27:46 CET] <harpsicord1> i see
[21:27:56 CET] <alexpigment> for the sake of preserving language, let's call "vinyls" records :)
[21:28:12 CET] <alexpigment> vinyl is a singular term
[21:28:24 CET] <harpsicord1> well records are made of PVC right? the "v" stands for vinyl
[21:28:28 CET] <alexpigment> and used as the format of records
[21:29:28 CET] <alexpigment> sorry, just a bit a pet peeve. this weird usage of "vinyl" just started popping up a few years ago
[21:29:38 CET] <alexpigment> rather, weird usage of "vinyls"
[21:30:34 CET] <therage3> I think it goes hand in hand with the resurgence of vinyl records
[21:30:40 CET] <therage3> probably due in part to the loudness wars
[21:30:53 CET] <alexpigment> https://s3.amazonaws.com/vf-images/wp-content/uploads/2016/03/plural-of-vinyl.png
[21:31:06 CET] <alexpigment> https://www.thepluralofvinyl.com/wp-content/uploads/2017/04/cropped-ThePluralOfVinyl_Podcast.jpg
[21:31:07 CET] <alexpigment> etc etc
[21:31:38 CET] <therage3> it's one of those things that could change down the line. Usage often shapes languages, who knows
[21:31:41 CET] <harpsicord1> yeah
[21:31:50 CET] <harpsicord1> i always used "vinyls" for the plural as well o.O
[21:31:50 CET] <kepstin> language prescriptivism has never worked with english anyways.
[21:31:55 CET] <harpsicord1> right
[21:32:32 CET] <alexpigment> yeah i agree. i just think young people who weren't around for records just kinda formed this misunderstanding of the word "vinyl" and its usage. a record is not "a vinyl"
[21:34:15 CET] <harpsicord1> goddamnit
[21:34:34 CET] <harpsicord1> the keyframe that has the problem is exactly the one darkest frame of a scene transition where on the broadcast they switch to commercials
[21:34:36 CET] <harpsicord1> are you kidding me
[21:34:52 CET] <kepstin> well, 'record' kind of refers to multiple different types of media, like the same term applies to 78rpm shellac and whatnot too
[21:35:21 CET] <alexpigment> kepstin: i agree. that's fine. but "vinyls" is never necessary or appropriate
[21:35:28 CET] <alexpigment> it's just "vinyl"
[21:35:32 CET] <harpsicord1> i think alexpigment is talking about the plural
[21:35:51 CET] <harpsicord1> anyway this just made me angry, the one keyframe with any artifacting is the one darkest frame
[21:35:55 CET] <harpsicord1> its pitch black in essence
[21:35:55 CET] <kepstin> hmm, it's a pretty natural extension of the usage given modern english.
[21:36:42 CET] <alexpigment> kepstin: maybe i'm an old man. but other old people stand with me in this crusade, as my google search just now confirmed
[21:36:56 CET] <alexpigment> https://pbs.twimg.com/media/CYJss4kWMAAPHfN.jpg
[21:37:09 CET] <alexpigment> see, young robin is mistaken. old batman is there to set him straight
[21:37:15 CET] <kepstin> yep, but like I said - language prescriptivism has never worked with english anyways.
[21:37:20 CET] <harpsicord1> ^
[21:37:30 CET] <harpsicord1> english is not one of those languages like spanish or french
[21:37:36 CET] <harpsicord1> that have these ridiculous global organizations
[21:37:39 CET] <alexpigment> kepstin: sure, but let's try to make this right before it gets out of hand
[21:37:42 CET] <harpsicord1> that nanny people
[21:37:48 CET] <kepstin> alexpigment: too late.
[21:37:51 CET] <harpsicord1> alexpigment: lol
[21:37:56 CET] <harpsicord1> alexpigment: it's already happening
[21:38:02 CET] <alexpigment> kepstin: i'm like batman. i'll just go around slapping people
[21:38:03 CET] <therage3> alexpigment: You're kind of late then :P
[21:38:31 CET] <alexpigment> alright, i give up
[21:40:43 CET] Action: kepstin hands alexpigment a few "glasses" of cold beverage as consolation for losing the vinyls war.
[21:40:53 CET] <alexpigment> thanks
[21:41:14 CET] <alexpigment> at the very least, when you go to the record store, you have a better chance of making friends with the workers there if you don't say "vinyls"
[21:41:33 CET] <harpsicord1> lol assuming the people that work there arent young people new to it themselves
[21:41:37 CET] <alexpigment> as a test, you can say "vinyls" to them and watch their body language as they wince and think less of you
[21:41:40 CET] <harpsicord1> if its some older dude maybe
[21:41:53 CET] <alexpigment> maybe you guys go to different record stores
[21:42:10 CET] <harpsicord1> well the ones i go to just hire people to man the cash register regardless of age
[21:42:13 CET] <alexpigment> unless you're talking about like best buy or hot topic or something
[21:42:37 CET] <harpsicord1> its not the 60s or 70s where you could ask the guys there for recommendations
[21:42:42 CET] <harpsicord1> you just go in buy stuff get out
[21:42:43 CET] <alexpigment> ??
[21:43:04 CET] <alexpigment> man, as a dude who worked at a record store for many years, and still goes to several record stores, i have a very different experience
[21:43:29 CET] <harpsicord1> i guess it depends on where you are, some places have a more musical history so to speak
[21:43:32 CET] <kepstin> there's still a few proper record stores around here, mostly dealing with used vinyl records.
[21:43:37 CET] <alexpigment> harpsicord1: record stores are a great place for getting recommendations from employees, if not just having great music conversations
[21:43:39 CET] <harpsicord1> so their stores have more of a personalized experience when you go to them
[21:45:01 CET] <therage3> i've been to both kinds of places, places with young staff that know FUCK ALL about the music they sell, and others with interested staff that give you recommendations and have you try out records you may enjoy
[21:45:14 CET] <therage3> it just depends on the store
[21:45:43 CET] <alexpigment> yeah, i recognize it's not the same everywhere. the record store in my neighborhood has been around since 1982 and is probably top 5 stores in the US in terms of vinyl. they're pretty selective about who they hire from what I gather
[21:46:00 CET] <therage3> one place I went to... i asked for some of the latest metal recommendations, the guy brings out a Linkin Park vinyl record, I'm like...
[21:46:10 CET] <harpsicord1> ... linkin park? metal?
[21:46:21 CET] <harpsicord1> i mean no offense to their lead singer who died a while ago, but dude
[21:46:28 CET] <harpsicord1> who the hell calls that band "metal"?
[21:47:01 CET] <therage3> well, they became famous during the heyday of the nu-metal movement, even its tail end, so to speak
[21:47:05 CET] <therage3> that's why
[21:47:12 CET] <alexpigment> i don't think anyone does, but yeah, it started with being labeled as nu metal
[21:47:33 CET] <alexpigment> and if you want to get real technical about it, they're heavier than a lot of the classic "metal" bands. that's not saying a lot... :)
[21:48:15 CET] <alexpigment> genres are fairly useless anyway
[21:48:25 CET] <therage3> well, their guitarist uses more gain for his guitar
[21:48:31 CET] <therage3> that's what makes it "seem" heavier
[21:48:48 CET] <therage3> but in the metal community if you go around saying a band like that is metal...............................
[21:48:56 CET] <alexpigment> well, don't forget the "screaming at the top of his lungs" part
[21:49:15 CET] <therage3> hardcore punk vocalists also do that lol
[21:49:19 CET] <alexpigment> that goes a long way for getting labeled as "metal", even if that band has 0% metal cred
[21:49:19 CET] <therage3> that doesn't make them metal
[21:49:37 CET] <alexpigment> i won't disagree with that
[21:50:58 CET] <therage3> sorry if I come across as a metal elitist, I was just blasting some Mercyful Fate ;D
[21:51:12 CET] <harpsicord1> youre shitting me right?
[21:51:14 CET] <harpsicord1> king diamond owns
[21:51:19 CET] <therage3> o.O
[21:51:32 CET] <therage3> i was certainly not expecting that. yes, indeed he does :D
[21:52:29 CET] <alexpigment> this sounds weird, but i'm pretty sure king diamond lives in dallas now. i always have this weird hope that i'll see him around
[21:52:37 CET] <alexpigment> (i don't live in dallas but visit often)
[21:52:50 CET] <therage3> you won't recognize him even if he does, i doubt he goes out with the black and white facepaint
[21:53:02 CET] <harpsicord1> LOL
[21:53:03 CET] <alexpigment> i've been the several record conventions there. it just seemed like he might show up
[21:53:17 CET] <alexpigment> i've seen him on That Metal Show. i have a decent idea what he looks like without makeup
[21:53:53 CET] <therage3> i have the Melissa album with him without makeup and other band members on the back
[21:54:10 CET] <therage3> of course this was a 1983 album, he doesn't look like that in 2018 I bet
[21:54:28 CET] <harpsicord1> i was born in a cemetery...
[21:54:32 CET] <harpsicord1> under the sign of the MOOOOOOOOOOOOOOOOOON
[21:54:33 CET] <alexpigment> last i saw him, he looked like a younger joe pesci
[21:54:40 CET] <harpsicord1> raised from my grave.... by the deeeeeeeeeead
[21:54:47 CET] <alexpigment> :)
[21:54:52 CET] <therage3> that note he hits on "moon" is insane
[21:55:07 CET] <therage3> other than Painkiller-era Rob Halford, arguably the highest notes i've heard in a metal song
[21:55:18 CET] <alexpigment> what about jim gillette?
[21:55:19 CET] <alexpigment> :)
[21:55:27 CET] <alexpigment> the band Nitro had the dumbest gimmicks of anyone
[21:55:28 CET] <harpsicord1> what band was he in again?
[21:55:41 CET] <therage3> isn't Nitro the band that had that super fast guitarist
[21:55:44 CET] <therage3> Michael Angelo Batio
[21:55:46 CET] <therage3> or am i confused
[21:55:57 CET] <alexpigment> it was basically this band where the singer sang very high notes and the guitarist played these double and quadruple neck guitars
[21:56:03 CET] <alexpigment> yeah
[21:56:09 CET] <alexpigment> michael angelo batio makes me laugh every time i see him
[21:56:22 CET] <therage3> well, arguably doing what Batio does is no laughing matter, that takes a looooot of practice
[21:56:33 CET] <furq> https://www.youtube.com/watch?v=CtGY6Ucseso&t=2m15s
[21:56:34 CET] <alexpigment> his instructional dvd was great. not sure if it's readily available on the internet
[21:56:36 CET] <furq> looking good alexpigment
[21:57:19 CET] <alexpigment> yeah, sometimes you can be amazing at something, but you completely counteract it with complete lack of style and taste
[21:58:24 CET] <alexpigment> https://www.youtube.com/watch?v=hb5QaCfm7bg <-- this is the one i'm thinking of
[21:58:42 CET] <alexpigment> "i'm going to give *you* the keys to the lamborghini"
[21:58:44 CET] <alexpigment> hahahah
[21:58:52 CET] <harpsicord1> lmao
[21:58:55 CET] <therage3> LOL
[22:00:06 CET] <alexpigment> anyway, to my earlier point, there's a reason why david gilmour is a household name and yngwie malmsteen isn't
[22:00:24 CET] <harpsicord1> oh that fat slob yeesh
[22:00:37 CET] <harpsicord1> i almost got into a fist fight with him once
[22:00:39 CET] <harpsicord1> malmseteen i mean
[22:00:44 CET] <therage3> LOL what
[22:00:55 CET] <harpsicord1> the guy thinks hes hot shit
[22:01:29 CET] <alexpigment> that's the vibe i've always gotten too
[22:01:37 CET] <harpsicord1> i was at a concert and he stops the show and points at my friend who was taking pictures with his cellphone of his girlfriend, not even the stage
[22:01:48 CET] <harpsicord1> he was all "stop fucking recording"
[22:02:05 CET] <harpsicord1> when I said something he said "Not talking to you fuckboy"
[22:02:12 CET] <harpsicord1> I was like, "Oh, is that right motherfucker?"
[22:02:17 CET] <harpsicord1> security blocked me
[22:02:25 CET] <harpsicord1> otherwise id have rammed his guitar up his ass
[22:02:32 CET] <therage3> LOL
[22:03:13 CET] <alexpigment> a lot of shows are moving to extremes with cell phone usage. it's a hassle, but people don't heed the "no cell phones" rule at all
[22:03:28 CET] <harpsicord1> yeah though this was a long ass time ago though
[22:03:32 CET] <harpsicord1> before that was a thing
[22:03:49 CET] <alexpigment> not defending him or anything
[22:03:58 CET] <therage3> harpsicord1: at least it wasn't Axl Rose
[22:04:06 CET] <therage3> if Rose called you out, trust me, you were a goner
[22:04:07 CET] <alexpigment> just saying that artists are very serious about it
[22:04:14 CET] <therage3> that guy is known to leap down into the crowd and beat the shit out of people
[22:04:17 CET] <therage3> i am not kidding
[22:04:23 CET] <alexpigment> yeah, that's on film too ;)
[22:05:00 CET] <harpsicord1> yeah that was a riot. in that show hes singing and then abruptly stops, his drops a whole octave when he starts addressing the guy in the crowd
[22:05:18 CET] <harpsicord1> his voice i mean
[22:05:30 CET] <alexpigment> when i saw ryan adams they had a woman come out and speak before the show explaining that he has a disease that makes camera flashes fuck him up in some way
[22:06:06 CET] <alexpigment> it was pretty weird, but hey, *not taking pictures* isn't a hard thing for me to do
[22:06:20 CET] <alexpigment> it requires less effort than taking pictures
[22:06:21 CET] <therage3> not sure if that's the case here, but that legitimately may be something. epilepsy and the like
[22:06:34 CET] <alexpigment> no it's like menieres disease or something
[22:06:37 CET] <therage3> ah
[22:06:53 CET] <alexpigment> i don't remember the details of it, but i guess i'll trust that flashes actually mess him up
[22:07:16 CET] <alexpigment> anyway, that dude put out a pretty great classic metal album a while back
[22:07:28 CET] <therage3> yeah, it's possible for sure. i know for some people exposure to bright light causes them to sneeze so things involving intense flashes of light could do weird stuff
[22:07:35 CET] <alexpigment> it's like a sci-fi themed classic metal album - fairly tongue in cheek but great
[22:07:41 CET] <therage3> i see
[22:07:51 CET] <alexpigment> it's called orion
[22:08:01 CET] <alexpigment> anyway, i'll stop getting off track
[22:08:03 CET] <therage3> fitting name for a sci-fi album
[22:10:22 CET] <harpsicord1> so anyway
[22:11:07 CET] <harpsicord1> im not 100% sure this is the one keyframe where the weird stuff happens, is there a way to ask ffmpeg to halt extraction of frames right after a warning?
[22:11:41 CET] <harpsicord1> problem is it continues to plow through till the last keyframe and i can only guess which frames it could be by keeping an eye on the frame counter as its extracting
[22:11:52 CET] <furq> maybe -xerror
[22:11:56 CET] <furq> i forget if that bails on warnings
[22:12:00 CET] <harpsicord1> hm
[22:12:14 CET] <kepstin> hmm, even if it does, that would stop it *before* the problematic frame is output, i'd think?
[22:12:34 CET] <harpsicord1> well then i just know it's the very next frame
[22:12:36 CET] <harpsicord1> it still gives me info
[22:12:39 CET] <harpsicord1> so thats fine
[22:23:25 CET] <MrNaz> ok so i have h264/mov files coming off my camera in 4gb files. they are all part of the same stream, and i can join them in premiere no problem. however, it's inconvenient to manage them as separate files, and i was wondering if it's possible to join them into a single container without transcoding.
[22:24:48 CET] <kepstin> MrNaz: https://www.ffmpeg.org/ffmpeg-formats.html#concat-1
[22:25:40 CET] <kepstin> MrNaz: write the concat script file, use that as an input to ffmpeg, use "-c copy", and it'll write the output to one continuous file
[22:27:02 CET] <adrusi> I'm trying to debug some distortion happening in my rtp audio stream
[22:28:21 CET] <adrusi> I'm using ffplay to receive a 48000 Hz mulaw stream, the same distortion occurs when I use opus.
[22:29:08 CET] <alexpigment> adrusi: a sample of the distortion is probably more helpful than a description of it
[22:29:30 CET] <adrusi> alexpigment: static.adrusi.com/rec.flac
[22:29:44 CET] <adrusi> the audio starts about 30 seconds into the recording
[22:30:00 CET] <alexpigment> ah, yeah, that sounds like an encoding issue
[22:30:09 CET] <alexpigment> like it's not keeping up in real time and dropping samples
[22:30:21 CET] <adrusi> so the stream works fine for about 3040 minutes
[22:30:26 CET] <adrusi> and then it starts doing that
[22:30:29 CET] <alexpigment> gotcha
[22:30:41 CET] <alexpigment> what does your CPU look like when you're doing this?
[22:30:48 CET] <alexpigment> (even before it starts to have the problem)
[22:31:42 CET] <alexpigment> there's a chance increasing a buffer could help (that's out of my area of knowledge though, so i'll defer to others here)
[22:32:24 CET] <alexpigment> but if the problem is that your CPU is taxed too much, increasing a buffer is only going to be a bandaid, i think
[22:33:02 CET] <adrusi> There might be other buffers I'm not aware of, but jack isn't reporting any xruns, so I don't think it's buffer size that's the problem, and doubling jack's buffer size didn't do anything to help either
[22:34:40 CET] <alexpigment> so the system running the stream isn't under full CPU load?
[22:34:44 CET] <alexpigment> or near-full?
[22:35:56 CET] <adrusi> no, all four cores are at less than 50% usage
[22:36:01 CET] <alexpigment> k
[22:36:07 CET] <alexpigment> i'll defer to the others here
[22:36:20 CET] <alexpigment> furq or kepstin - you guys got any ideas what's causing that sound?
[22:36:28 CET] <kepstin> (keep in mind that audio encoders & ffmpeg core is signle-threaded, so you should look at process cpu usage totals)
[22:36:56 CET] <kepstin> if the ffmpeg process is using near 100% cpu, you're in trouble.
[22:37:31 CET] <adrusi> no, ffmpeg is using at most 20% at any time
[22:37:59 CET] <alexpigment> 20% = 80% singled threaded for a 4-core system
[22:38:23 CET] <alexpigment> i wonder if it hits more when the problem starts
[22:38:28 CET] <adrusi> although distortion isn't happening at the moment, I will have to check when it is happening
[22:38:48 CET] <alexpigment> anyway, it sounds like the CPU is struggling to my ears in your flac sample
[22:39:05 CET] <kepstin> hmm. just sounds like packet loss concealment to me
[22:39:08 CET] <alexpigment> maybe someone else will know that symptom more intimately and have a better idea
[22:39:53 CET] <adrusi> it's not packet loss, because ffplay reports occasional packets lost, but their frequency doesn't increase when the distortion begins
[22:41:00 CET] <adrusi> none of my 4 cores is going above 35% at any time according to htop, I think the 20% it reports for ffplay is single-threaded
[22:49:18 CET] <alexpigment> admittedly, i don't know much about how packet loss works with rtp streams, but i wonder if there's a buffer that it has, and the audio problem starts when the packet loss eventually adds up to exhausting your buffer
[22:50:05 CET] <harpsicord1> nope, -xerror plows through just like always
[22:50:25 CET] <harpsicord1> i'm a bit surprised this doesn't have a command, theres no way to stop a command when it outputs a warning?
[22:50:33 CET] <harpsicord1> that sounds like something that should be there
[22:50:46 CET] <alexpigment> harpsicord1: i had to edit the source code for an encoder for that
[22:52:03 CET] <alexpigment> the qsv_h264 encoder has a deal where in some scenarios it will warn that it can only use a software fallback, which way slower than x264. so i edited the source code to fail on that error so i could ensure that it was only available when hardware was going to be in use
[22:52:22 CET] <adrusi> alexpigment: that sounds plausible, but I don't *think* that's what's happening. I do notice the slightest stutter when it logs a lost packet.
[22:52:30 CET] <harpsicord1> i see
[22:53:01 CET] <alexpigment> harpsicord1: so in other words, i never found a way to stop on a warning for my particular usage
[22:53:26 CET] <alexpigment> adrusi: hmmm, not sure what it could be then :(
[22:53:41 CET] <harpsicord1> yeah this is annoying, youd think ffmpeg would have something like that, since some people would want to not have it finish encoding on specific warnings
[22:54:21 CET] <alexpigment> right. in some scenarios, a warning is very important
[22:54:22 CET] <adrusi> I've tried testing the stream with mplayer, and I don't think I ever saw any distortion (although admittedly I didn't keep it running for very long because the latency was terrible). But judging by mplayer's latency, it probably has some kind of buffer that might be stopping it from encountering this distortion
[22:54:59 CET] <alexpigment> adrusi: that logic makes sense
[22:55:30 CET] <kepstin> oh, huh, you're using ffplay?
[22:55:40 CET] <kepstin> yeah, I could see a real player doing a better job
[22:55:47 CET] <alexpigment> haha
[22:55:49 CET] <alexpigment> :)
[22:56:12 CET] <alexpigment> not a day goes by where you don't see someone talk smack about ffplay around here
[22:56:20 CET] <harpsicord1> wait what
[22:56:23 CET] <harpsicord1> "real player"
[22:56:25 CET] <harpsicord1> uh
[22:56:33 CET] <harpsicord1> what makes ffplay not a real player :S
[22:56:34 CET] <alexpigment> *a real player*, not realplayer
[22:56:45 CET] <kepstin> ffplay is a perfectly useful tool for some sorts of troubleshooting, and as a demo of api usage
[22:56:51 CET] <kepstin> but other than that :/
[22:57:07 CET] <adrusi> kepstin: do you have a suggestion for one, the latency I'm getting with mplayer is unacceptable, even with -benchmark -nocache
[22:57:42 CET] <adrusi> I haven't tried vlc yet
[22:57:57 CET] <furq> mpv?
[22:58:12 CET] <kepstin> hmm, the general recommendation around here is usually mpv, but i dunno if it'll do less buffering than mplayer
[22:58:13 CET] <harpsicord1> "vlc" lol
[22:58:24 CET] <harpsicord1> the player that in 2018 still cant do gapless playback
[22:58:49 CET] <alexpigment> yeah, i'll echo "vlc lol" as well, although i'd guess that it should be fine for this purpose
[22:58:55 CET] <harpsicord1> stick with mpv yeah, thats a good light one
[22:59:23 CET] <harpsicord1> alexpigment when i used it earlier this year i was shocked because it gets recommended quite a lot
[22:59:45 CET] <adrusi> I use mpv on desktop, didn't think it would have much advantage over mplayer for this headless audio player, but I can try it
[23:00:12 CET] <kepstin> adrusi: for your usage, you might be able to get away with running ffmpeg rather than ffplay (I assume it's audio-only?), but I don't know if that'll behave all that much better
[23:00:18 CET] <harpsicord1> ... on desktop? huh wait
[23:00:23 CET] <harpsicord1> isnt that like a command line player
[23:00:26 CET] <harpsicord1> i swear it was CLI
[23:00:49 CET] <adrusi> it is, but it has a "pseudo-gui"
[23:01:03 CET] <therage3> the "desktop" is just this simple window you drag stuff to, harpsicord1
[23:01:11 CET] <therage3> It isn't like *shudders* VLC
[23:01:44 CET] <kepstin> hmm. I wonder if the issue you're hitting is simply that your sound card clock doesn't match the stream source
[23:02:07 CET] <alexpigment> kepstin: you'd think that would happen before 30 minutes though, right?
[23:02:10 CET] <therage3> you mean the sampling frequency? or an actual jitter?
[23:02:11 CET] <kepstin> a simple player like ffplay has no compensation for that, and it'll eventually just desync enough that you'll get lost packets or other distortion
[23:02:16 CET] <kepstin> sample frequency
[23:02:55 CET] <therage3> that'd be something ALSA/PulseAudio manages yeah, some cards arent smart enough to switch their modes and resample, and often the resampling is done using a one-point algorithm which makes it sound bad
[23:03:01 CET] <alexpigment> admittedly 48000khz is overkill for most audio, since the masters are usually in 44100
[23:03:16 CET] <therage3> I've had to manually edit PulseAudio's config to change it a few times
[23:03:20 CET] <kepstin> 30 minutes is off by & 0.0005 hz on the clock rate, which isn't all that much
[23:03:43 CET] <kepstin> the player basically has to watch its buffer and start adding/dropping samples if the clocks drift :/
[23:04:43 CET] <harpsicord1> wait
[23:04:46 CET] <kepstin> basically, the source is 48kHz and the playback soundcard is 48kHz, but they're both off by a tiny amount that would normally be considered within tolerances& and so you get drift.
[23:04:53 CET] <harpsicord1> masters are done at CDDA redbook frequency?
[23:05:23 CET] <kepstin> CD masters are, part of the process of mastering for CD is resampling to 44.1kHz
[23:05:44 CET] Action: kepstin would expect internal studio stuff to usually be higher rates tho
[23:06:00 CET] <therage3> Well the gripe is with "most audio", I don't know if that's the case anymore since streaming is the usual scenario nowadays, and CDDA is slowing being phased out
[23:06:07 CET] <therage3> I was confused by the statement as well
[23:06:14 CET] <harpsicord1> exactly
[23:06:48 CET] <harpsicord1> I mean 44.1kHz was made up because the Nyquist frequency is just above the human hearing range and because 60Hz even divides, good for American frequencies
[23:06:57 CET] <alexpigment> sorry, i meant "masters" meaning that whatever your source audio is
[23:07:04 CET] <harpsicord1> OK thats different
[23:07:07 CET] <therage3> Ahhh I see
[23:07:09 CET] <alexpigment> as opposed to your streaming frequency
[23:07:32 CET] <adrusi> well mpv adds about 30ms latency relative to ffplay, but it's still... ok. I can compensate for latency since I play all my video through mpv which has a scriptable A/V Delay setting. No way I can play DRM video with this though
[23:07:34 CET] <alexpigment> but since most audio is sourced from CD still, it's usually 44.1
[23:08:21 CET] <harpsicord1> it is? the stuff that's streamed I would assume is sourced from whatever masters the record labels have, I don't think their highest quality tends to be CDDA Redbook quality
[23:08:32 CET] <kepstin> I've downloaded a few bandcamp releases that had weird mixes of frequencies
[23:08:41 CET] <harpsicord1> i mean not that you can tell the difference anyway from double blind tests
[23:08:42 CET] <harpsicord1> but still
[23:09:12 CET] <kepstin> assuming the audio signal is bandlimited to <22kHz or so, converting between the two should be indistinguishable with any decent resampler.
[23:09:37 CET] <therage3> the "decent resampler" part is important; by default, PulseAudio has a crappy one
[23:09:41 CET] <therage3> you need to change it
[23:10:01 CET] <therage3> naturally that means your CPU will be hammered more
[23:12:48 CET] <harpsicord1> also speaking of that, be very very careful with so called "HD tracks" you buy online, some are shamelessly upsampled, which a spectrogram instantly reveals as theres a brickwall at 22.05kHz
[23:13:37 CET] <kepstin> well, throwing a lowpass on during mastering to avoid any issues from ultrasonic harmonics is good sense anyways
[23:13:40 CET] <therage3> that's the sampling frequency; the bit depth/word length can also be easily upscaled by first padding the least significant bits with zeroes, and then doing something very subtle
[23:14:00 CET] <kepstin> (which means of course that high-sample-rate digital releases are of course basically useless)
[23:14:09 CET] <therage3> this fills those least significant bits and gives the illusion of a 24-bit true recording
[23:14:13 CET] <therage3> which it isn't
[23:15:05 CET] <adrusi> I'm technically getting something out of 48000 Hz since I run a midi synth on the device, but I doubt it makes much of a difference
[23:15:35 CET] <harpsicord1> kepstin: no i mean, what they do is they take the cdda PCM or FLAC or whatever and just upsample it, its a brickwall at exactly that frequency
[23:16:24 CET] <harpsicord1> if you want a lowpass filter, you do that to the actual high frequency master so that it tapers off at something like 30kHz or whatever
[23:17:33 CET] <kepstin> no reason to even go that high, 20kHz with rolloff filling another ~4kHz is great
[23:17:48 CET] <harpsicord1> sure, what im trying to say is, its clear theyre fooling the crowd
[23:17:57 CET] <therage3> indeed
[23:18:22 CET] <kepstin> high sample rate is useless anyways, so anyone who buys it is already being fooled? :/
[23:18:34 CET] <therage3> not that a double blind test could distinguish a CDDA quality master from a "high-resolution" true master, but the fact is, if you pay for a 24-bit FLAC at 96kHz, that is what you should _get_
[23:19:23 CET] Action: kepstin is more annoyed that a 24bit digital releases usually still have the same maxed volume and clipping as the cd versions
[23:19:43 CET] <kepstin> they have extra headroom... and then don't use it
[23:19:50 CET] <therage3> exactly, that is a problem indeed
[23:20:47 CET] <sfan5> might be because 24-bit 96kHz is often just a marketing trick to suck money out of uninformed "audiophiles"
[23:21:27 CET] <harpsicord1> indeed, and the problem is amplified when what they get isnt even a genuine 24-bit 96kHz file, but something shamelessly upscaled from a CD master quality file
[23:21:36 CET] <harpsicord1> and they cant even tell the difference
[23:21:38 CET] <harpsicord1> and theyre happy!!!
[23:21:55 CET] <therage3> lol yeah
[23:23:01 CET] <kepstin> if the only way you can tell the difference is by looking at a spectrogram, it probably wasn't worth it :)
[23:23:55 CET] <therage3> i remember watching this video on youtube once about a guy that reviewed that bullshit Jitterbug thing that supposedly cleans up signal for before it enters your DAC... he compared the DAC's output with and without the Jitterbug, and found that the signals were identical in both cases
[23:24:22 CET] <therage3> yet since he "clearly heard a difference", his explanation was that the Jitterbug was doing something to the signal at frequencies higher than 22kHz!!!
[23:24:31 CET] <therage3> this is the kind of snake oil we're dealing with here
[23:24:37 CET] <therage3> who the fuck does that guy think he's fooling
[23:24:45 CET] <sfan5> himself
[23:24:52 CET] <harpsicord1> LOL
[23:25:43 CET] <durandal_1707> you can easily add non noise to >22khz
[23:26:10 CET] <durandal_1707> fake spectrals are real
[23:26:22 CET] <therage3> yeah, with harmonic exciters
[23:26:24 CET] <therage3> that's also true
[23:26:32 CET] <therage3> and I do suspect some fakes were done that way
[23:27:56 CET] <Durandal> how can I tell h265 to encode in black & white?
[23:28:19 CET] <Durandal> is there an x265-param for it or do I just like, set pix_fmt=yuv400 or something
[23:28:24 CET] <sfan5> -vf format=gray
[23:28:55 CET] <sfan5> it will probably be converted back to yuv420p before entering the encoder, but that should work
[23:30:09 CET] <Durandal> well that's what I'm wondering. I thought I saw an option somewhere to tell the encoder to just ignore chroma entirely
[23:30:23 CET] <Durandal> I don't know. Does it matter?
[23:30:35 CET] <Durandal> if i'm encoding, like, Citizen Kane and want to minimize size?
[23:30:49 CET] <alexpigment> i don't think you'll see much (any?) benefit in size tbh
[23:30:54 CET] <sfan5> oh x265 actually seems to support gray input
[23:30:59 CET] <sfan5> at least x264 --fullhelp says so
[23:30:59 CET] <kepstin> ffmpeg -h encoder=libx265 says it supports gray (and the 10+12 bit if your x265 is compiled appropriately) as input format
[23:31:02 CET] <harpsicord1> therage3 i find it fascinating how this business has people spending thousands of dollars and its all built on snake oil and chicanery
[23:31:13 CET] <kepstin> so I'd assume just setting the pixel format to gray would be enough
[23:31:54 CET] <SortaCore> when is mov atom written for copy codec + mp4?
[23:32:14 CET] <kepstin> SortaCore: at the end of the file, like always
[23:32:17 CET] <alexpigment> i do seem to recall someone talking recently about ffmpeg having an "enable-grayscale" compiler flag recently. not sure if that's specific to x265 or is used in conjunction
[23:32:30 CET] <SortaCore> faststart flags move it to header?
[23:32:40 CET] <SortaCore> or does it move it to front but only writes it when file ends
[23:32:44 CET] <kepstin> SortaCore: faststart runs a second pass after the encode that moves it to the start.
[23:33:06 CET] <SortaCore> alrighty
[23:33:15 CET] <kepstin> SortaCore: the problem is that the moov atom contains the seek index info, which can't be known in advance
[23:33:30 CET] <SortaCore> yea, I'm trying to hack together a C++ copy codec
[23:33:35 CET] <therage3> harpsicord1: yeah
[23:33:49 CET] <SortaCore> literally stepping through ffmpeg.c and trying to work out what needs to be done
[23:34:04 CET] <SortaCore> I thought you needed the AVCodec for copying, apparently NULL is fine
[23:34:25 CET] <kepstin> no avcodec needed, you just take the packets from the demuxer and pass them to the muxer
[23:34:39 CET] <kepstin> ffmpeg does some timestamp cleanup and whatnot even in copy mode, tho
[23:34:47 CET] <kepstin> (ffmpeg.c that is)
[23:36:01 CET] <SortaCore> yea, it seems odd to me, because you can have multiple video codec types in a mp4, so I thought it would have to specify somehow
[23:36:15 CET] <SortaCore> yea, the timestamp stuff has me puzzled
[23:36:33 CET] <SortaCore> it seems to start at -1.0 offset, then complain about dts/pts until it hits >0
[23:36:44 CET] <therage3> all right, let me ask this here, since people in the Xiph channel were puzzled. is there a way to copy all metadata from one FLAC file to another? not just the Vorbis tags, but all metadata
[23:37:39 CET] <SortaCore> I dunno, I'm currently finding all my metadata is being stripped and tossed aside silently
[23:37:48 CET] <SortaCore> even my ones that match ffmpeg's ones, like "encoder"
[23:38:47 CET] <SortaCore> that's on the fix list, eventually
[23:39:00 CET] <Durandal> therage3: what metadata is there that is outside of vorbis tags?
[23:39:50 CET] <therage3> Durandal: these https://xiph.org/flac/format.html#metadata_block
[23:40:07 CET] <therage3> Durandal: so, stuff like seektable, padding, pictures for cover art, etc
[23:42:08 CET] <therage3> to be specific, since this is an X Y problem, let me make it clear: i have some FLAC files and i wish to determine what level of compression the encoder used to generate them. however when i decode them to WAV and reencode them at differing compression levels, the size is never the same as the original FLAC because the full metadata is never copied over
[23:42:52 CET] <sfan5> surely it's easier to strip the metadata first
[23:43:21 CET] <Durandal> If I were to try this would be my command line: ffmpeg -i File.flac -i Metadata.flac -c:a copy -map_metadata 1 "Output.flac"
[23:43:23 CET] <therage3> i see, so take the original FLAC and strip it all, and then compare?
[23:43:26 CET] <sfan5> yea
[23:43:38 CET] <therage3> hmmm i wonder if there's a way to do that
[23:43:41 CET] <Durandal> and also -map 0:a:0 in there too
[23:43:54 CET] <therage3> Durandal: oh interesting
[23:46:01 CET] <therage3> sfan5: i'm thinking for your solution I may have to rig up a bash script with metaflac
[23:47:57 CET] <Durandal> -map_metadata -1 should also strip metadata
[23:48:25 CET] <Durandal> so you can do that with -c:a copy on your original files and it should clean them without touching the audio
[23:48:55 CET] <therage3> Durandal: is that a metaflac parameter or an ffmpeg parameter
[23:49:00 CET] <Durandal> ffmpeg
[23:49:01 CET] <GuiToris> Hey! Did anything happen to concat? It used to work, ffmpeg -i 'concat:input1|input2' output but it's not working now
[23:49:04 CET] <therage3> I can't see it in metaflac's man pages
[23:49:06 CET] <therage3> that's why
[23:49:32 CET] <Durandal> ffmpeg -i "original.flac" -c:a copy -map_metadata -1 "original_without_metadata.flac"
[23:49:40 CET] Action: therage3 nods
[23:49:42 CET] <therage3> thanks let me try
[23:49:52 CET] <therage3> i'll confirm with metaflac if indeed all metadata's been stripped
[23:49:55 CET] <alexpigment> GuiToris: start a pastebin with your full command. it'll probably be obvious what's happening
[23:50:29 CET] <alexpigment> GuiToris: for what it's worth, you can use concat as a demuxer or a filter, and there are limitations for each
[23:50:42 CET] <Durandal> GuiToris: I've had better luck using the alternate method that reads an external file list
[23:51:05 CET] <Durandal> seems like last time I tried to just do -i concat it had was really fickle as well
[23:52:27 CET] <alexpigment> Durandal: yeah i think unless you are using ts files the concat -i [whatever|whatever] won't work
[23:53:00 CET] <alexpigment> i know i kinda botched that syntax, but i think the point was still valid
[23:53:02 CET] <Durandal> right, like I think I once got it to work joining DVD VOBs together and that was it
[23:53:23 CET] <GuiToris> alexpigment, https://ptpb.pw/hPw7
[23:53:23 CET] <alexpigment> makes sense. PS and TS are similar enough
[23:53:39 CET] <alexpigment> ok you're trying to do it with mp4
[23:54:11 CET] <alexpigment> https://trac.ffmpeg.org/wiki/Concatenate
[23:55:03 CET] <GuiToris> I've already looked this page over
[23:55:04 CET] <alexpigment> i think the method under "Concat demuxer" should still work though; you'll just need a text file
[23:55:25 CET] <alexpigment> GuiToris: well, you did see the part about the method you used not working with MP4 files, right?
[23:55:26 CET] <GuiToris> my problem is that output1 is basically input1
[23:55:44 CET] <GuiToris> alexpigment, no that actually escaped my attention
[23:56:55 CET] <GuiToris> as a matter of fact, I still can't find this part
[23:57:11 CET] <alexpigment> look under the "concat protocol" heading
[23:57:29 CET] <alexpigment> read from that section though where it says "using intermediate files"
[00:00:00 CET] --- Tue Jan  9 2018


More information about the Ffmpeg-devel-irc mailing list