[Ffmpeg-devel-irc] ffmpeg.log.20161111

[burek]

[00:10:43 CET] <FishPencil> Is it possible to get the duration of a media file in milliseconds?
[00:12:57 CET] <bens218> thanks again furq
[00:13:08 CET] <bens218> i'll try it out and see how it works
[00:16:12 CET] <llogan> FishPencil: not that I know of. you can convert it yourself though.
[00:16:30 CET] <llogan> ffprobe -loglevel error -show_entries format=duration -of default=nw=1:nk=1 input.foo
[00:16:41 CET] <llogan> or stream=duration
[00:48:14 CET] <FishPencil> llogan: Thanks I'll work with that
[00:49:06 CET] <FishPencil> When including subtitles, the first one is automatically enabled when playing back. How could I include them, but default to the "forced" subtitles?
[00:50:24 CET] <FishPencil> The subtitles are in format "hdmv_pgs_subtitle" and I'm -c:s copy ing them
[00:51:30 CET] <c_14> FishPencil: -disposition:1 default -disposition:0 0
[00:51:32 CET] <c_14> something like that
[00:52:29 CET] <c_14> Where 1 and 0 are stream ids for the output stream you want default/not default
[00:52:48 CET] <c_14> so -disposition:s:0 for example
[00:52:55 CET] <c_14> (why isn't this documented...)
[00:53:44 CET] <FishPencil> c_14: I've had a lot of fun figuring out chapter information and subtitle information.
[00:53:53 CET] <llogan> c_14: was just about to say, "this option needs to be documented"
[00:54:14 CET] Action: c_14 is working out if that's how it actually works and will submit a patch
[00:54:23 CET] <llogan> thanks
[00:54:32 CET] <FishPencil> I can test
[00:56:34 CET] <c_14> It does appear to work like that
[00:56:51 CET] <bens218> time to go home. thanks again for the help furq. i'll try your method tonight when i get free time.
[00:59:08 CET] <FishPencil> c_14: It does appear to select to overall track (the language), but not forced. http://i.imgur.com/SIQsCLw.png
[00:59:43 CET] <FishPencil> And -disposition:3 0 -disposition:4 0 does not set it to "No subtitles"
[01:00:13 CET] <c_14> FishPencil: can you upload the contents of ffprobe on that file to a pastebin service?
[01:02:26 CET] <FishPencil> c_14: https://paste.ubuntu.com/23458680/
[01:03:13 CET] <c_14> None of those streams are marked as default (according to that output)
[01:04:58 CET] <FishPencil> c_14: Playing from the source using MPC-HC (which uses libbluray as well I believe) results in http://i.imgur.com/OVw7y9G.png
[01:05:33 CET] <c_14> I think the disposition is printed (in more detail) with -show_frames as an option to ffprobe
[01:06:19 CET] <c_14> eh
[01:06:25 CET] <c_14> no, -show_streams
[01:06:30 CET] <c_14> Getting confused
[01:07:28 CET] <FishPencil> c_14: https://paste.ubuntu.com/23458696/
[01:08:06 CET] <c_14> Both of those have DISPOSITION:default=0
[01:08:19 CET] <c_14> so it should be off
[01:09:20 CET] <FishPencil> I'm pretty certain there is a lot of info getting lost from libbluray->libav*
[01:09:40 CET] <FishPencil> For example, chapter information is missing.
[01:11:34 CET] <FishPencil> https://trac.ffmpeg.org/ticket/3953
[01:11:40 CET] <TheXzoron> https://wiki.freebsd.org/WacomTablet this article apears to be outdated as input-wacom doesn't exist
[01:12:00 CET] <furq> i think you're looking for #freebsd
[01:12:10 CET] <TheXzoron> yes
[01:12:12 CET] <TheXzoron> sorry
[01:19:55 CET] <c_14> FishPencil: what's your output format?
[01:20:06 CET] <FishPencil> c_14: mkv
[01:20:32 CET] <c_14> And setting the disposition for that isn't working?
[01:23:09 CET] <FishPencil> c_14: what do you mean by working? I can use it to set a "default" subtitle stream, which does work, but I cannot: select the "forced" stream, or select no stream by using -disposition:3 0 -disposition:4 0
[01:24:32 CET] <c_14> Yeah, I just noticed you can't set no default stream either.
[01:24:41 CET] <c_14> But you can't set the forced stream as default?
[01:27:25 CET] <FishPencil> c_14: No. Something tells me the subtitle format has some sort of flag for forced subs that MPC-HC is able to pick up on, but VLC and FFmpeg cannot. FFmpeg only sees two subtitle steams in the input, whereas MPC-HC is giving me three (forced being one).
[01:29:34 CET] <c_14> hmm, ok. Yeah it's probably missing something then
[01:30:43 CET] <FishPencil> c_14: It might also be that if I was able to set no default subtitle stream that MPC-HC would select forced, also achieving the same goal.
[01:31:07 CET] <c_14> Either way, it's probably a deficiency in ffmpeg and worthy of a ticket on trac (if there isn't already one).
[01:32:57 CET] <FishPencil> I'll take a look at the code to see if I can figure it out. I'm guessing this isn't mkv specific?
[01:33:40 CET] <c_14> Probably not (getting similar things with mp4 at least in regards to setting no default stream)
[01:36:24 CET] <c_14> FishPencil: https://trac.ffmpeg.org/ticket/2402 probably related
[01:43:51 CET] <FishPencil> c_14: Nice find, I'll use that as a starting point.
[02:33:02 CET] <kurufu> Maybe someone can point me in the right direction, when using libav I want to demux rtsp wth the "-rtsp_flags listen", so I should be trying to set "listen" in the AVDictionary in avformat_open_input or where should I be looking for setting these demuxer specific flags/settings
[02:33:20 CET] <kurufu> Im pretty much entirely new to working with libav, so sorry if im looking in completely the wrong place.
[02:38:59 CET] <kurufu> Oooh Mailing list archives sure are a blessing. Looks like I was right. Setting the rtsp options can be done through the AVDictionary parameter.
[02:47:19 CET] <DHE> kurufu: loosely speaking any leftover parameters on ffmpeg's command-line go into the AVDictionary of the encoder, container, and so on.
[06:00:20 CET] <Ertain> I have built ffmpeg and installed it to a specific directory on my computer.  I then built the MLT libraries and installed them on my computer.  Now I have one binary, melt, which, for some odd reason, is not linked with the right file.  And I haven't a clue why.
[11:09:41 CET] <SouLShocK> occasionally ffmpeg 3.2 will generate corrupt mp4 files. when I try to decode them with "ffmpeg -xerror -i file.mp4 -f null -" they will say [aac @ 0000000000980280] channel element 0.0 is not allocated
[11:13:15 CET] <SouLShocK> http://pastebin.com/SDfBD0Wz
[11:14:39 CET] <BtbN> sounds like it wasn't allowed to finish writing that mp4, so it's broken.
[11:14:48 CET] <BtbN> [mov,mp4,m4a,3gp,3g2,mj2 @ 00000000008724a0] Found duplicated MOOV Atom. Skipped it
[11:15:22 CET] <BtbN> writing to a network drive sounds like a good reason for that to randomly fail
[11:43:22 CET] <SouLShocK> but the first pass says it exited correctly?
[11:48:36 CET] <SouLShocK> does that mean that I can't really count on the exit code from ffmpeg?
[15:22:32 CET] <pgorley> is this page out of date: https://trac.ffmpeg.org/wiki/HWAccelIntro
[15:22:52 CET] <pgorley> it says vaapi decoding is not supported, but hasn't this been implemented a while ago?
[15:24:02 CET] <jkqxz> There is no standalone vaapi decoder, it attaches to the normal decoder as a hwaccel.
[15:24:18 CET] <pgorley> oh, thanks
[15:27:07 CET] <jkqxz> The real hwaccels (dxva, d3d, vaapi, vdpau) are more general and have better support (including transparent software fallback) than the things which require a specific full decoder and then have a dummy hwaccel to attach to it (cuvid, qsv, mediacodec).
[15:29:32 CET] <jkqxz> I.e. hwaccel = Y, decoder = N is the "best" case in that table.  The page could be clearer about the differences.
[15:33:21 CET] <pgorley> that explains it, and yes, it should be clearer
[15:44:59 CET] <nyuszika7h> is there a way to take exactly 3 snapshots at specific positions from a file with one command?
[15:45:42 CET] <nyuszika7h> ffmpeg -ss 10:00 -i in.mkv -vframes 1 thumb01.png - I want e.g. two more at 20:00 at 30:00, but it won't always be the same time because the snapshot may turn out to be blurry or something in which case I have to change it
[16:06:37 CET] <pgorley> is using vdpau on a system with only vaapi support that inefficient? decoding 100 frames takes 2 seconds longer on average than using software decoding...
[16:55:55 CET] <c_14> nyuszika7h: ffmpeg -ss 10:00 -i in.mkv -frames:v 1 out.png -ss 10:00 -frames:v 1 out1.png -ss 20:00 out2.png
[17:24:12 CET] <ozette> when i google for 'ffmpeg resample' it's like i only find results related to  audio, not about resampling video streams. is there a reason for that?
[17:25:09 CET] <ozette> only 'ffmpeg resize' will find image (video) stuff
[17:26:38 CET] <SchrodingersScat> what is resample?
[17:28:35 CET] <ozette> to resample a data signal
[17:33:18 CET] <kurufu> well are you trying to work on a video or audio?
[17:33:31 CET] <kurufu> Resampling a video is in essence, resizing it.
[17:37:07 CET] <ozette> yea, but it's hard to find sources which atually all it resampling
[17:37:15 CET] <ozette> call
[17:37:46 CET] <ozette> and i wonder why that is
[17:38:49 CET] <ozette> even the ffmpeg docs only mention audio
[17:39:58 CET] <ozette> so, why is changing ar called resampling, but changing video size called rescaling or resizing
[17:41:54 CET] <furq> probably because libswresample is for audio and libswscale is for video
[17:42:01 CET] <furq> at least in the context of ffmpeg
[17:42:17 CET] <ozette> i see
[17:43:13 CET] <ozette> yea, because generally searching for 'video resampling' does provide expected info
[17:43:33 CET] <ozette> but 'ffmpeg resampling' (only) provides audio
[17:45:35 CET] <ozette> mystery solved, i think
[18:08:26 CET] <flappy> hey, anyone got an idea on how to pixelate a video even more than dropping overall bitrate to 80k and scaling to 240p
[18:12:28 CET] <DHE> could just use -qp 50 (for h264)
[18:12:58 CET] <furq> you could drop the bitrate to 80k and scale it to 1080p
[18:25:52 CET] <flappy> rescaling it to 1080 with -b 80k now
[18:35:06 CET] <flappy> oh that last mangling broke the file heh
[18:53:38 CET] <mrechte> Hello. Trying to publish an HD video on youtube / vimeo or dailymotion gives an horrendous result (pixelization), whereas it looks pretty good on my PC. I used: ffmpeg -i ../$1.avi -codec:v libx264 -crf 21 -bf 2 -flags +cgop -pix_fmt yuv420p -codec:a aac -strict -2 -b:a 384k -r:a 48000 -movflags faststart $1.mp4 to encode it. An extract is here : https://www.youtube.com/watch?v=pkwVe4cBKvk Thanks
[18:57:30 CET] <furq> mrechte: what's the source
[18:57:40 CET] <furq> if it's small enough to upload then just upload that, youtube will reencode it anyway
[18:58:40 CET] <nyuszika7h> c_14: oh thanks
[18:58:59 CET] <nyuszika7h> I assume there should be a "-frames:v 1" before out2.png as well?
[18:59:56 CET] <furq> nyuszika7h: yes
[19:00:19 CET] <c_14> yep, just forgot it
[19:01:19 CET] <mrechte> furq: The source is another blender avi/h264 encoded file with "lossless output selected": Stream #0:0: Video: h264 (High 4:4:4 Predictive) (H264 / 0x34363248), yuv420p(progressive), 1920x1080 [SAR 1:1 DAR 16:9], 295235 kb/s, 30 fps, 30 tbr, 30 tbn, 60 tbc
[19:01:49 CET] <nyuszika7h> (why would you put H.264 video in .avi? you can, but usually .mp4 or .mkv is used for that)
[19:02:51 CET] <furq> i guess that is probably too big to upload then
[19:03:49 CET] <mrechte>  nyuszika7h: Not very sure. I followed a tutorial on how to encode from blender to youtube (actually it is using handbrake to do the final conversion)
[19:05:07 CET] <furq> i wonder if disabling psy and aq would help minimise generation loss
[19:05:08 CET] <mrechte> furq: this what I uploaded: Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p, 1920x1080 [SAR 1:1 DAR 16:9], 36962 kb/s, 30 fps, 30 tbr, 15360 tbn, 60 tbc (default)
[19:05:33 CET] <furq> well yeah youtube's 1080p is about 4mbit
[19:05:43 CET] <furq> going from 36mbit to 4mbit is probably going to look bad
[19:06:15 CET] <furq> assuming it is actually 36mbit and that's not just ffprobe going crazy
[19:07:21 CET] <furq> if you have enough upload then youtube will process the source just fine
[19:07:42 CET] <furq> and it should look at least a bit nicer
[19:08:33 CET] <mrechte> furq: do you mean I should increase -crf ? and what do you mean by enough upload ?
[19:08:52 CET] <furq> enough upload bandwidth to make it viable to upload your source file
[19:09:50 CET] <mrechte> furq: I uploaded the video, but the result is not acceptable
[19:10:14 CET] <DHE> if the source is lossless, than 36 megabits seems pretty good
[19:10:30 CET] <furq> is that the source or the one you reencoded with ffmpeg
[19:10:31 CET] <DHE> lower CRF is better quality, with 0 being (close to) lossless
[19:13:12 CET] <mrechte> No the source was encoded with blender (probably throug ffmpeg). Then I encoded with ffmpeg cli with above mentioned parameters. That final encoded file looks quite OK on my computer but the uploaded version (I mean after being published by YT and watched online) looks ugly.
[19:14:05 CET] <DHE> youtube can take a while to encode all variants. you're sure you were watching the 1080p version?
[19:14:20 CET] <DHE> once uploading is finished there might only be a 480p version at that exact instance
[19:14:31 CET] <furq> the 720p version looks really bad
[19:14:31 CET] <mrechte> Yes: https://www.youtube.com/watch?v=pkwVe4cBKvk
[19:14:44 CET] <mrechte> same on vimeo: https://vimeo.com/191156496
[19:15:10 CET] <furq> upload the avi if you can
[19:15:21 CET] <mrechte> and dailymotion: http://www.dailymotion.com/video/x51mynu
[19:15:33 CET] <furq> how much of that is generation loss and how much is bit starvation i don't know
[19:15:38 CET] <DHE> weird, youtube's only giving me 720p but I see a 1080p variant with youtube-dl...
[19:15:42 CET] <furq> but if you can eliminate the former then you should
[19:16:19 CET] <furq> generally the only reason to reencode before uploading to youtube is if it would take forever to upload the original
[19:16:30 CET] <mrechte> AVI is 5.7 GB for 3 minutes !
[19:16:42 CET] <furq> (or your video is hevc, interlaced, etc)
[19:17:20 CET] <DHE> well, vimeo makes a ~30 megabyte video for a 6 second segment.. which is also crazy high
[19:17:32 CET] <DHE> oh wait, that's the original
[19:18:40 CET] <mrechte> Yes 30MB is the size of the file I upload for this test
[19:19:00 CET] <DHE> I have to say vimeo does look better by a little bit..
[19:20:50 CET] <mrechte> I am uploading the file to an ftp site
[19:21:21 CET] <DHE> well we have the 6 second 30MB file. vimeo offers it straight up
[19:27:13 CET] <mrechte> But it looks completely different when watched from their site !
[19:28:03 CET] <DHE> yes, likely because the crazy high bitrate isn't served to users by default. with tools like youtube-dl you can force the variant of your choosing and save it to disk
[19:32:45 CET] <FishPencil> What is the fastest way to encode 4 videos? One after another? Two at a time? All 4 at a time? Assuming all conversions use all cores/threads
[19:32:54 CET] <mrechte> Published as is here: http://mrechte.free.fr/tmp/0000-0200%28ffmpeg%29.mp4
[19:33:06 CET] <furq> probably all four at a time because then you only have to decode it once
[19:33:20 CET] <furq> also using multiple cores doesn't scale linearly
[19:33:36 CET] <DHE> if it's the same video 4 different ways, yeah do them all in one shot
[19:33:38 CET] <FishPencil> furq: Huh? These are four different input
[19:33:39 CET] <FishPencil> s
[19:33:45 CET] <furq> oh
[19:33:57 CET] <furq> i guess you'll have to benchmark it then
[19:34:27 CET] <DHE> too many at a time might be bad for CPU cache, but more than 1 might ensure better CPU utilization
[19:35:01 CET] <FishPencil> That was my thought. Without benchmarking it myself I was curious if this was known
[19:35:03 CET] <furq> it'll depend on a lot of factors
[19:35:22 CET] <furq> e.g. if you're using an encoder with weak multithreading like libvpx then parallel jobs will definitely be faster
[19:35:36 CET] <DHE> I have major multi-core CPUs (20 cores + hyperthreading). utilizing all cores doesn't really happen. here there's advantages.
[19:35:52 CET] <FishPencil> All four use prettymuch the same encoding settings. x264 and Opus
[19:36:28 CET] <furq> i doubt it'll be a big difference either way
[19:39:53 CET] <damarusama> I have a mp4 that doesn't play in quicktime, I want to change it with -pix_fmt yuv420p
[19:40:25 CET] <damarusama> not sure how to go about it, ffmpeg this.mp4 -pix_gmt yuv420 that.mp4 doesn't work
[19:40:43 CET] <damarusama> Output file #0 does not contain any stream
[19:40:49 CET] <sstangl> I have a 6GB video file that sometimes contains images of an Excel spreadsheet. I'm looking at using ffmpeg-filters to detect those spreadsheet images, and output jpg. But I can't figure out how to isolate them. They tend to be green in the top-left corner, is there expr syntax where I could select on pixel color?
[19:43:38 CET] <FishPencil> sstangl: That's an interesting one
[19:43:42 CET] <DHE> damarusama: read the manual. you put "-i" in front of input videos
[19:44:09 CET] <damarusama> ._. ho... thanks
[19:44:41 CET] <damarusama> it's just that I can't read...
[19:44:52 CET] <DHE> also it's going to get a lot more complicated than that. you will need to specify some encoding parameters because there aren't really defaults (or rather, the defaults aren't that good)
[19:45:25 CET] <damarusama> hum, even if I'm happy with the mp4 that was created?
[19:45:57 CET] <damarusama> I actually don't really understand that the -pix_fmt yuv420p does to the video  (apart from making it playable in quicktime)
[19:46:32 CET] <FishPencil> sstangl: I don't know if FFmpeg can do this all in one go. You may need to extract the images to process them later
[19:47:15 CET] <sstangl> FishPencil: I don't have enough disk space to convert all the keyframes to jpg at once, but if I process them as they come out that would be doable.
[19:47:43 CET] <damarusama> sstangl: exporting everyframe to image then using something like imagemagick with a bash script
[19:47:53 CET] <damarusama> you could export all keyframe to really low quality
[19:48:05 CET] <sstangl> alright, I'll try that. Thanks!
[19:49:27 CET] <mrechte> I tried to reduce the bitrate (increased -crf to 32). Now the uploaded file is 10047 kb/s and 8.4MB. Quality has become worse... https://youtu.be/oLmaBxqOzvc
[19:50:00 CET] <DHE> damarusama: rather than RGB, pixels are defined by brightness and two colour components. but when encoded several pixels share a single colour parameter. the numbers in yuv420 describe how it's distributed.
[19:50:22 CET] <DHE> yuv420 is most common, so some decoders only support that
[19:51:44 CET] <damarusama> thanks - that's quite a transfer... that makes quicktime player quite limited isn't it?
[19:54:58 CET] <damarusama> DHE: so yeah the video is really mashed up after that transition -
[19:56:18 CET] <damarusama> it's plays ok in vlc - but is really grainy in quicktime
[20:12:49 CET] <intracube> does anyone know why ffmpeg/libmp3lame reduces the maxiumum volume of an input file when encoding?
[20:13:00 CET] <intracube> ffmpeg reports "max_volume: -2.0 dB" for the source .wav
[20:13:23 CET] <intracube> but "max_volume: -7.7 dB" for the final encoded video file with mp3 audio
[20:22:22 CET] <mrechte> Good bye. Thanks.
[20:23:26 CET] <Cheater_> Hi! Players cant start live stream untill i restart ffmpeg process. Here are settings
[20:23:27 CET] <Cheater_>   sudo ffmpeg \     -f video4linux2 \     -i /dev/video0 \     -filter:v "crop=250:480:195:0" \     -intra-refresh 1 \     -pix_fmt yuv420p \     -framerate 30  \     -c:v libx264 \     -preset ultrafast \     -threads 8 \     -tune zerolatency  \     -analyzeduration 0 \     -an \     -s 640x480 \     -probesize 32 \     -x264opts crf=20:vbv-maxrate=12000:vbv-bufsize=100:intra-refresh=1:slice-max-size=1500:keyint=30:ref=1 \     -f 
[20:24:55 CET] <JEEB> 2/32
[20:42:30 CET] <mwg> hello
[20:42:37 CET] <mwg> i am looking for some advice
[20:43:01 CET] <mwg> i have a barex device which streams (voice) audio over a network
[20:43:08 CET] <mwg> it can stream UDP or RTP
[20:43:31 CET] <mwg> i can use ffmpeg to record the audio and record it to wav or mp3
[20:43:46 CET] <mwg> (it's PCM 24-bit from the device)
[20:44:08 CET] <mwg> however, it only broadcasts packets when somebody presses the button on the device
[20:44:35 CET] <mwg> meaning, the recording does not have the correct timing (e.g. without the pauses where there was no broadcast)
[20:45:08 CET] <mwg> is there anyway to record the stream and have it capture the period when there was no broadcast? e.g. fill with silence? so that the time base for the audio is correct?
[20:52:35 CET] <mwg> anyone here?
[20:53:30 CET] <intracube> mwg: are you just dumping the audio or converting?
[20:54:19 CET] <mwg> intracube: i'm converting to mp3 (but i can do either)
[20:54:33 CET] <mwg> i'm using the following command: ffmpeg.exe -loglevel debug -y -protocol_whitelist "rtp,file,udp" -i .\foo.sdp foo.mp3
[20:54:54 CET] <intracube> mwg: this is not something I'm familiar with
[20:55:02 CET] <intracube> was just looking at the documentation to see if there was something obvious
[20:55:10 CET] <mwg> no worries
[20:55:13 CET] <mwg> thanks
[20:55:45 CET] <xeuari> hello people.. I'm looking for a way to rotate the hue linearly on a short clip, such that hue is proportional to time. Can ffmpeg do this?
[20:56:03 CET] <intracube> I was wondering if it could be done indirectly by having an ffmpeg audio filter generate silence and mix this network stream in on top
[20:56:16 CET] <intracube> which might be enough to coax ffmpeg to keep encoding throughout
[20:56:27 CET] <intracube> I can't find such a silent filter though
[20:56:40 CET] <mwg> that's a good idea
[20:56:43 CET] <furq> xeuari: https://ffmpeg.org/ffmpeg-filters.html#hue
[20:57:46 CET] <furq> intracube: https://ffmpeg.org/ffmpeg-filters.html#anullsrc
[20:57:54 CET] <intracube> mwg: ^ :)
[20:58:06 CET] <mwg> okay!
[20:58:07 CET] <mwg> so
[20:58:13 CET] <mwg> ...sorry for being an idiot
[20:58:23 CET] <mwg> how do I record both the sdp and the filter?
[20:59:20 CET] <furq> presumably something like -i foo.sdp -f lavfi -i anullsrc -filter_complex amix=inputs=2
[20:59:25 CET] <furq> whether that'll do what you want i have no idea
[20:59:30 CET] <mwg> ha
[20:59:33 CET] <mwg> i'll try it
[20:59:37 CET] <intracube> what furq said :P
[21:01:36 CET] <intracube> mwg: you might need the -re option (realtime encoding) or the anullsrc might race away as fast as your computer can process
[21:01:39 CET] <xeuari> sweet! I'll have a play, thanks. Should have checked there first really :/
[21:02:22 CET] <intracube> -re goes as the very first parameter
[21:02:35 CET] <mwg> okay, standby - just brew'ing ffmpeg
[21:03:37 CET] <furq> -re is an input option so it'd go before -i anullsrc
[21:03:42 CET] <furq> otherwise it'll apply to the wrong input
[21:03:50 CET] <xeuari> Am I right to assume that the hue is an offset ie it doesn't sample the video clip?
[21:07:55 CET] <xeuari> ..I'm off to find out.. good day to you *doffs cap*
[21:08:19 CET] <intracube> furq: ah, yep. it depends which input you specify first
[21:22:59 CET] <mwg> it might be working...
[21:37:21 CET] <xeuari> hmm .. hope you will be patient with me. My hue rotation command looks like this: hue="h=360*(t/7.018)" ie go from 0 to 360 degrees across the length of the clip. But it throws invalid argument..
[21:37:39 CET] <xeuari> can anyone spot the mistake in there?
[21:42:35 CET] <xeuari> I also tried radians, because that can defo deal with floats.. but same error
[21:45:27 CET] <dragmore88> Is ffmpeg able to convert SRT file to DVB-TTX and mux it into a TS mux?
[21:52:24 CET] <xeuari> does ffmpeg need the hue filter to be explicitly enabled or disable at build?
[21:54:11 CET] <intracube> xeuari: not sure, but maybe the doublequote is in the wrong place?
[21:54:29 CET] <intracube> what about either: hue=h="360*(t/7.018)"
[21:54:54 CET] <intracube> or encompassing the whole video filter set: -vf "hue=h=360*(t/7.018)"
[21:56:33 CET] <xeuari> I have followed the syntax in the example on the documentation to see if it works: hue="H=2*PI*t"
[21:56:46 CET] <xeuari> and that didn't work .. I'll give those a go
[21:57:23 CET] <xeuari> oo is -vf reuired for filtering?
[21:57:34 CET] <intracube> if it's a video filter, yep
[21:57:40 CET] <intracube> and hue is
[21:57:57 CET] <intracube> or -filter_complex
[21:58:02 CET] <intracube> but that uses a different syntax
[21:58:06 CET] <xeuari> ahh!! well there's my fuckup. Nice one
[21:58:28 CET] <intracube> :)
[21:58:50 CET] <intracube> and at least one of the examples in the manual page encompasses the whole filter string:
[21:58:59 CET] <intracube> -vf "hue=H=2*PI*t:s=sin(2*PI*t)+1"
[21:59:07 CET] <intracube> which is what I'd expect
[22:00:03 CET] <xeuari> yeah so would I .. certainly makes more sense
[22:01:20 CET] <xeuari> Huzzah! it worked like a dream
[22:01:32 CET] <intracube> nice :)
[22:02:13 CET] <xeuari> ps. that was without globally quoting the string
[22:03:14 CET] <intracube> yep, not always needed if there aren't unusual characters or whitespace
[22:08:50 CET] <xeuari> cool cool .. reckon it'll take some time for me to figure the workings of this beast
[23:11:57 CET] <levo> i have a bunch of webm files, and for all of them there are seeking problems, meaning that most of them time i can't fast forward or rewind and for the occasions I can it goes like 8 minutes forward and that's it, like there is two or three time points  i can reach by fast forwarding. is there any way to fix this?
[23:13:01 CET] <furq> probably not without reencoding
[23:13:34 CET] <furq> you can try remuxing with -i foo.webm -c copy -map 0 bar.webm
[23:14:00 CET] <DHE> hope it's not a case of huge GOPs. then you're in trouble and need to reencode it as indicated
[23:14:08 CET] <furq> yeah i'm not sure what else it would be
[23:14:10 CET] <DHE> well, not with that command
[23:14:32 CET] <furq> it's probably worth trying to remux first though because it's basically free
[23:22:22 CET] <levo> furq, DHE: it was unsuccessful. what is the fastest encoding around? i tried with -i foo.webm bar.mp4   but the speed was 1x, it's very slow for me
[23:23:18 CET] <kerio> what's your hardware?
[23:23:54 CET] <kerio> i think that defaults to x264 at default settings, it shouldn't be that slow
[23:25:08 CET] <levo> kerio: intel Core 2 Duo, 2GHz. 2GB RAM DDR2
[23:25:22 CET] <kerio> LOL
[23:25:30 CET] <levo> don't lol me :))
[23:26:06 CET] <levo> i love it btw although he's a bit old
[23:26:40 CET] <kerio> old architecture tho
[23:26:43 CET] <kerio> like very old
[23:26:57 CET] <kerio> no wonder this super optimized stuff doesn't run that well
[23:27:25 CET] <IntelRNG> I am running like that but with twice as much RAM. Show some respect to antiques.
[23:28:33 CET] <kerio> levo: you could gpu it up a bit
[23:28:40 CET] <kerio> either for the decoding or for the encoding
[23:30:21 CET] <levo> kerio: how can i do so?
[23:30:40 CET] <kerio> do you have a gpu, first of all?
[23:31:29 CET] <kerio> it's usually OS-dependent
[23:35:17 CET] <levo> dedicated gpu i don't think so. my graphics is one of those chipsets you can barely do anything other than watching movies with
[23:35:27 CET] <kerio> then nope
[23:35:42 CET] <FishPencil> Is there a "proper" way to normalize 5.1 audio? Or should I just leave it
[23:35:44 CET] <furq> you can use -preset superfast
[23:35:58 CET] <furq> you probably also want to keep the audio as it is
[23:36:12 CET] <furq> so -i foo.webm -c:v libx264 -preset superfast -c:a copy bar.mkv
[23:37:22 CET] <kerio> have you tried remuxing it tho
[23:37:44 CET] <furq> i think he just said he did
[23:37:44 CET] <levo> kerio: yeah, didn't work
[23:37:55 CET] <kerio> try superfast yea
[23:38:14 CET] <furq> there's also ultrafast but that turns a bunch of important stuff off
[23:38:19 CET] <levo> furq: superfast gave me 1.9x well compared to the previous one seems better
[23:38:21 CET] <furq> i wouldn't advise that unless you're really desperate
[23:38:47 CET] <furq> i'm not aware of a faster encoder than x264 which gives comparable filesizes
[23:39:37 CET] <kerio> ffv1? :^)
[23:39:52 CET] <furq> those last four words are there for a reason
[23:39:55 CET] <kerio> they're comparable in the sense that they only differ by about two orders of magnitude
[23:40:03 CET] <furq> i'd also be surprised if ffv1 was faster tbh
[23:40:19 CET] <furq> maybe ffvhuff would beat it
[23:43:13 CET] <kerio> levo: also, why does it matter?
[23:43:25 CET] <kerio> this is a thing that you'll only do once anyway
[23:43:59 CET] <levo> kerio: it's a course like hours ... and files each 10minutes and so
[23:44:16 CET] <kerio> how fast is your internet?
[23:44:29 CET] <levo> pretty slow
[23:44:34 CET] <kerio> rip
[23:44:35 CET] <furq> you could upload them to something like google photos
[23:44:36 CET] <levo> 512kbit
[23:44:39 CET] <furq> or maybe not
[23:45:05 CET] <kerio> yeah i was going to suggest some high perf ec2 instance
[23:45:41 CET] <furq> google photos will transcode them all for free, but it sounds like it'd be slower than just doing it yourself
[23:46:32 CET] <levo> thanks guys  for your help
[23:46:54 CET] <kerio> levo: is disk space an issue?
[23:46:58 CET] <kerio> rawvideo is a thing
[23:47:04 CET] <levo> not that much
[23:47:18 CET] <levo> although the files are only 20mb max eacch
[23:47:19 CET] <furq> you can try with -c:v rawvideo or -c:v ffvhuff
[23:47:26 CET] <furq> the files will end up at least 10x bigger
[23:47:30 CET] <furq> maybe closer to 100
[23:47:41 CET] <furq> it'll be fast though
[23:48:27 CET] <kerio> incredibly snappy to seek through, too
[23:49:06 CET] <levo> meaning fast?
[23:49:12 CET] <kerio> yeah
[23:49:30 CET] <levo> what should i set the output to?
[23:49:55 CET] <kerio> are you going to watch them with mpv?
[23:50:15 CET] <kerio> .mkv is a good choice
[23:50:25 CET] <kerio> .nut is ffmpeg-specific
[23:51:41 CET] <levo> gnome-mplayer but i got mpv too
[00:00:00 CET] --- Sat Nov 12 2016