[Ffmpeg-devel-irc] ffmpeg.log.20170106

burek burek021 at gmail.com
Sat Jan 7 03:05:01 EET 2017

[00:37:14 CET] <Lac3rat3d> when encoding mp3 stream, is there any way to tell it to use whatever bitrate the source stream has?
[00:37:18 CET] <Lac3rat3d> or do you ahve to manually set a value
[00:37:32 CET] <c_14> manually
[00:37:53 CET] <Lac3rat3d> damn
[00:38:16 CET] <Lac3rat3d> i suppose it might be possible if using an API, like a python library or the like?
[00:38:30 CET] <BtbN> why would you even want to do that?
[00:39:11 CET] <BtbN> Unless you are highly size limited, just go for best or second best quality VBR
[00:39:43 CET] <Lac3rat3d> ya, i could use -q:a 0-3
[00:39:49 CET] <Lac3rat3d> but if the source stream is only 128k
[00:39:52 CET] <Lac3rat3d> that becomes overkill
[00:39:57 CET] <Lac3rat3d> not space limited
[00:40:02 CET] <Lac3rat3d> just efficiency oriented :D
[00:40:56 CET] <klaxa> if the source is mp3, you could copy the stream
[00:41:05 CET] <klaxa> and keep the same quality + encoding
[00:41:06 CET] <Lac3rat3d> but, because space is not so limited, i don't know if writing a python script to programmaticly set the VBR rate is worth the effort :)
[00:41:15 CET] <Lac3rat3d> can't copy
[00:41:23 CET] <Lac3rat3d> need to re-encode
[00:41:31 CET] <BtbN> encoding as 128kbps will make it even worse
[00:42:13 CET] <Lac3rat3d> you think so BtbN? source is mp3 128 (different encoder), and you ask it to encode using constant bitrate of 128 for output file, you think it turns out worse than source?
[00:42:24 CET] <BtbN> of course it does
[00:42:33 CET] <Lac3rat3d> sorry maybe i'm niave, but why is that?
[00:42:51 CET] <BtbN> because you lossy encode something that's already been lossy encoded
[00:43:03 CET] <BtbN> every generation only gets worse
[00:43:17 CET] <Lac3rat3d> ahhh
[00:43:19 CET] <Lac3rat3d> yes i see your point
[00:43:34 CET] <Lac3rat3d> so despite bitrate of source file, just use -q:a 0 ?
[00:43:48 CET] <BtbN> so might as well minimize that loss if size is not an issue, and use a high vbr
[00:44:05 CET] <Lac3rat3d> ya, so -q:a 0 then?
[00:44:13 CET] <Lac3rat3d> that would be highest VBR setting i think
[00:44:14 CET] <BtbN> Or even a diffrent codec
[00:44:21 CET] <Lac3rat3d> can't do different codec
[00:44:22 CET] <BtbN> if you don't absolutely need mp3
[00:44:27 CET] <Lac3rat3d> need mp3
[00:44:35 CET] <Lac3rat3d> for device compatibility via DLNA
[00:44:45 CET] <Lac3rat3d> clients are picky about what encoding of mp3 they handle
[00:45:03 CET] <BtbN> yeah, just use vbr 0 then
[00:45:06 CET] <Lac3rat3d> ok cool
[00:45:15 CET] <Lac3rat3d> thank you for the info and suggestions
[00:45:48 CET] <llogan> i don't get why you're re-encoding MP3 to MP3 in the first place.
[00:46:16 CET] <Lac3rat3d> different encodings
[00:46:18 CET] <Lac3rat3d> as i explained above
[00:46:31 CET] <Lac3rat3d> some devices are very picky about what encoding of mp3 they can handle via DLNA
[00:46:49 CET] <llogan> sounds miserable
[00:46:55 CET] <Lac3rat3d> same mp3 file with two different encodings but still mp3 stream, one can't be played, one can
[00:47:07 CET] <Lac3rat3d> lame seems to be handled properly, others dont
[00:47:33 CET] <Lac3rat3d> it's only annoying prior to writing the script to convert it :D
[00:47:39 CET] <Lac3rat3d> after that it's a transparent process
[00:52:50 CET] <Lac3rat3d> ok thanks again guys
[00:52:51 CET] <Lac3rat3d> cya
[01:45:59 CET] <phillipk> I have a source  .flv (audio only) containing speex at a sample rate of 16000--but based on the timing I used when recording it, the .flv is about 3 minutes shorter than it should (but it's an hour long--so it's off by 3 minutes/hour)
[01:46:30 CET] <phillipk> I think I just need to extend the duration a tiny tiny bit.
[01:46:55 CET] <phillipk> the sound is not noticeable fast/slow (nothing like the difference betweeen 44100 and 48000)
[01:49:01 CET] <phillipk> I'm layering the sound ontop of a .ts file containing a silent audio track with rate: 44100 Hz and bitrate 256kb/s
[01:50:00 CET] <phillipk> ffmpeg -y -i silent_video.ts -i problem.flv -strict experimental -filter_complex amix - profile:v main -level 3.1 -movflags +faststart -c:a aac -b:a 64K -f mp4 final.mp4
[01:50:32 CET] <phillipk> the silent_video.ts is the correct duration, so the audio ends a bit early.
[04:02:36 CET] <BotoX> has anyone here used ffmpeg to feed raw samples to an audio buffer before, building my own "audio player" using ffmpeg
[04:03:17 CET] <BotoX> Built an extension with audio buffer and mixer for the source engine, so I can transmit audio over the steam voice protocol
[04:04:03 CET] <BotoX> now that extension is buffering the audio (~1s), encoding it and correctly sending it to the clients, ffmpeg is feeding it raw samples over tcp
[04:04:33 CET] <BotoX> does that sound like a sane idea to use ffmpeg for this
[04:05:30 CET] <BotoX> I'd basically want an audio player but audio players only push out audio at the right speed, so I'd introduce delay if I wanted to buffer it D:
[04:26:30 CET] <BotoX> I need to build my own audio player, time to look into integrating libffmpeg or whatever there is it offerw
[04:58:14 CET] <DHE> BotoX: maybe check ffplay, which is a rather rudimentary player using ffmpeg libs
[04:58:45 CET] <BotoX> I think I will pipe the raw samples from ffmpeg into python
[04:58:53 CET] <BotoX> (which is my audio controller)
[04:59:30 CET] <BotoX> and then do the sending from there, so I know for sure how many samples I sent, how much time elapsed and I can kill the socket with linger 0 to stop the transmission right away and drop all buffers
[08:15:20 CET] <Jesperhead> Im converting a mp3 to a wav using http://ffmpy.readthedocs.io/en/latest/ffmpy.html (my conversion basically follows the base example, with input.mp3 to output.wav) the mp3 is 24 kHz but I want the output wav to be @ 16 kHz. I looked on https://ffmpeg.org/ffmpeg.html#Audio-Options and tried to implement '-ar 16000' but get "At least one output file must be specified" as an output error. A...
[08:15:21 CET] <Jesperhead> ...regular conversion with no arguments for the output.wav works just fine. Posted in the python channel as well, but this sounds like im messing up the command argument.
[08:16:33 CET] <Jesperhead> ignore... i just figured it out. typo.
[08:46:59 CET] <DHE> Jesperhead: yes, order of parameters matters a lot
[08:51:41 CET] <Jesperhead> DHE: Im converting an mp3 to a wav for a project. The resulting wav has a lot of whitenoise that I can eliminate with -acodec dts. only problem is it messes with my python module that only likes pcm wav data
[08:52:18 CET] <DHE> what's the complete commandline you use?
[08:52:53 CET] <Jesperhead> im making the call with a python wrapper. the output is variable is: outputs={'temp.wav': ['-ar', '44100', '-ac', '1', '-acodec', 'dts', '-strict', '-2', '-f', 'wav']}
[08:55:02 CET] <c_14> If you use pcm you get white noise?
[08:55:14 CET] <Jesperhead> yep. a good bit
[08:55:33 CET] <Jesperhead> well im assuming pcm... if i dont add any arguments i get a wav with a lot of white noise
[08:55:44 CET] <Jesperhead> that i can also use with my project...
[08:57:15 CET] <DHE> I'm wondering if you're using the wrong sample format for your appilcation. maybe check if -sample_fmt is applicable to you?
[08:57:30 CET] <DHE> if you got the signedness wrong or something it'd be a problem
[08:57:57 CET] <c_14> What are you playing the wav with?
[08:59:36 CET] <Jesperhead> https://paste.pound-python.org/show/9EiLWF9wx4Rn9m9h8qdl/
[08:59:41 CET] <Jesperhead> sounddevice lib
[09:00:13 CET] <Jesperhead> which as I understand, only takes numpy array audio data, which as I understand, for whatever reason has to be wav format
[09:00:46 CET] <c_14> have you tried playing the wav with ffplay/mpv or something?
[09:01:31 CET] <Jesperhead> vlc?
[09:01:35 CET] <c_14> but like DHE said, probably signedness or float size
[09:01:45 CET] <c_14> should work
[09:02:21 CET] <Jesperhead> prior to implementing dts i used vlc and heard the white noise. after implementing dts the conversion still works, and the actual wav file is much cleaner... i suppose this is more a python module problem
[09:04:14 CET] <Jesperhead> thanks for the help regardless
[10:59:51 CET] <sebbl> Hi, I'm trying to Stream a video I get from a RaspberryPi as a RTSP Stream to YouTube. It works fine if I use not audio, as soon as I add audio (with ALSA under Linux) the frame rate drops dramatically, these are the two commands I use: http://pastebin.com/fTbNLuK3 There is enough processing power for the audio encoding, the CPU isn't running at full load. Does anyone have an idea what I can change?
[11:01:20 CET] <sebbl> so with "& -f lavfi -ac 1 -i anullsrc &" it works, with "& -f alsa -ac 1 -i hw:2,0 &" the frame rate drops
[11:26:34 CET] <nobody44> Hello,
[11:27:36 CET] <nobody44> is the quality with h264_nvenc encoder noticable? All I see are old reviews which say that nvenc basically does not produce good results. Did that change?
[11:28:32 CET] <nobody44> For comparison: Usually I transcode with x264 -preset slow -crf 20
[12:59:05 CET] <Kirito> Does x264 encoding with multiple threads really have an impact on quality, or is this just a myth? And if not, how significant is the impact?
[13:04:05 CET] <JEEB> if you are not using VBV/HRD or slice threading, the number of threads does not matter
[13:04:26 CET] <Kirito> https://birds-are-nice.me/publications/extremex264_5.shtml a lot of the references on this I've seen seem old also, so I'm not sure if this is still as much of a concern today. I've been capping at 2 threads when encoding animation, but I'm not sure if there's any benefit to me doing that
[13:04:28 CET] <JEEB> if you are using VBV/HRD not every run will give you the exact same result with threading
[13:04:38 CET] <Kirito> ahhhh
[13:04:44 CET] <Kirito> that makes sense
[13:05:05 CET] <Kirito> so that's completely irrelevant with CRF encoding, yes?
[13:05:07 CET] <JEEB> and with slices the image is cut into pieces :P
[13:05:31 CET] <JEEB> Kirito: CRF is a primary rate control thing, while VBV/HRD are an additional rate control limiter
[13:05:34 CET] <Kirito> (I'm guessing VBV is variable bitrate, not sure what HRD is)
[13:05:42 CET] <Kirito> oh VBV
[13:05:44 CET] <Kirito> I need sleep
[13:06:00 CET] <JEEB> for example if you're encoding for a limited bandwidth network, VBV/HRD is pretty much a must
[13:06:24 CET] <JEEB> so that you can say that the maxrate is your lowest supported transfer speed, and bufsize is the amount of data you want the players to buffer
[13:06:38 CET] <JEEB> thus VBV/HRD limits the bandwidth to maxrate over bufsize
[13:06:57 CET] <Kirito> ahh, right, got it. Thank you!
[13:07:30 CET] <JEEB> anyways, never saw *frame* threads being mentioned as lowering quality
[13:07:43 CET] <JEEB> *slice* threading on the other hand uses slices, so d'uh
[13:07:54 CET] <JEEB> guess which has been the default for ages
[13:08:04 CET] <JEEB> (like ten years or so)
[13:08:50 CET] <Kirito> I can't even remember where I originally heard that, ahah. I guess I've just been tormenting myself with horribly slow encoding times for no reason the past few days then >_>
[14:05:23 CET] <Kirito> a bit of an odd question, but are there any options  that can be toggled to help boost performance when encoding enormous raw-video input streams? It's a 500GB 6 minute video stream >_>. It's stored on a RAID 0 array, but I think the disk performance is still probably choking the encoder.
[14:05:59 CET] <Kirito> Wasn't sure if there was a way to pass in a custom read-ahead value or something of that nature, so it caches more of the stream in-memory to avoid constant disk seeks
[14:49:19 CET] <ghost1> anyone who wanna help me build a line for ffmpeg using x264 with some diff settings and crop stuff into it? im basicly a newbie and i need alot of diffrent settings implemented in a bashscript?
[14:51:56 CET] <Polochon_street> Hi! I'm trying to use the swr_convert() function to downsample songs from some sample rate to 22.05 kHz using this function I made http://sprunge.us/dZbH , it works for interleaved format but crashes (ret = 0 for an unkown reason) using a planar song format
[14:52:21 CET] <Polochon_street> is there a widely known thing I am not doing when using swr_context() for planar files ?
[17:52:50 CET] <raineys> is it possible to use the ffmpeg api to encode audio with twolame?
[17:54:27 CET] <kepstin> raineys: assuming it's enabled in the ffmpeg build, yes... codec name to use is 'libtwolame'
[17:56:07 CET] <raineys> ah I see, I didn't see the find encoder by name function
[17:56:09 CET] <babadoc> Hi, is there a fast way to convert mkv to mp4 using ffmpeg?
[17:56:12 CET] <raineys> was using the enums before
[17:56:36 CET] <babadoc> Or are all options slow?
[17:57:38 CET] <JEEB> you can -c copy quite a few types of track
[17:57:43 CET] <JEEB> as long as it fits mp4, you can just copy streams
[17:57:48 CET] <JEEB> which is pretty damn fast :P
[17:57:55 CET] <babadoc> Okay, I will try that
[17:57:57 CET] <JEEB> ffmpeg -i input -c copy output.mp4
[17:58:35 CET] <babadoc> Yeah, that was extremely fast...
[17:58:36 CET] <ghost1> ok so im trying to encode with this line
[17:58:40 CET] <babadoc> Thank you!
[17:58:40 CET] <ghost1> ffmpeg -i $path/to-encode/* -vcodec libx264 -crf 20 -preset slow -level 3.1 -x264opts force-cfr:stitchable -r 25 -vf yadif=0,crop=out_85:out_85:0:0 -s 718:404 -aspect 16:9 $path/encoded/$lowrls.mkv
[17:58:59 CET] <ghost1> tried numurous of things with crop
[17:59:20 CET] <ghost1> but nothing gets the crop right and the resolution on the output still gets like the source file
[17:59:32 CET] <ghost1> tried to add the crops from cropdetect aswell
[17:59:59 CET] <ghost1> anyone who can point me in some direction please
[18:02:28 CET] <kepstin> ghost1: what is "crop=out_85:out_85:0:0" supposed to do? "out_85" isn't a number...
[18:02:55 CET] <ghost1> tried to follow some old crap in a google guide
[18:03:56 CET] <kepstin> so, what are you actually trying to do?
[18:05:04 CET] <ghost1> encode and crop the borders
[18:05:18 CET] <ghost1> get this in cropdetect
[18:05:21 CET] <ghost1> [Parsed_cropdetect_0 @ 0x128b300] x1:0 x2:719 y1:1 y2:575 w:720 h:560 x:0 y:10 pts:12504 t:12.504000 crop=720:560:0:10
[18:05:42 CET] <ghost1> so i guess i should add crop=720:560:0:10
[18:05:49 CET] <kepstin> given the other options, I guess you have some PAL 575-line interlaced video that you're trying to clean up?
[18:06:01 CET] <ghost1> no not really
[18:06:15 CET] <ghost1> i wanna encode from 1 format to another with some other settings in it
[18:06:35 CET] <kepstin> well, what you want to use for the crop filter depends on what the input video format is...
[18:07:30 CET] <ghost1> mpeg
[18:07:34 CET] <ghost1> if thats enough?
[18:07:47 CET] <kepstin> no, stuff like video resolution and source
[18:08:18 CET] <ghost1> res on source is 720x576
[18:08:24 CET] <kepstin> given the fact that you have a deinterlacing filter and the source is 720x575 according to that crop filter, I'm guessing your source is a european dvd or tv capture
[18:08:38 CET] <ghost1> yes it is
[18:08:38 CET] <kepstin> is the video widescreen (letterboxed)?
[18:08:43 CET] <ghost1> letterboxed
[18:08:52 CET] <ghost1> its not ws
[18:09:11 CET] <ghost1> Active Format Description                : Letterbox 16:9 image (top)
[18:09:33 CET] <ghost1> and its a tv capture
[18:11:27 CET] <kepstin> alright, so you probably want crop=704:432:8:72 assuming it is actually 1 16:9 video, and is centered
[18:11:44 CET] <kepstin> and remove the '-s' option, that's adding an extra rescale on the end
[18:12:03 CET] <ghost1> ok so let me test
[18:12:42 CET] <kepstin> you could optionally rescale to 768x432 if you want the output to have square pixels
[18:14:24 CET] <ghost1> nah
[18:14:59 CET] <ghost1> how did you get to that crop?
[18:15:20 CET] <ghost1> since i wanna implenment cropdetect into this bashscript to know what crops that need to be done
[18:15:45 CET] <ghost1> basicly want an autocrop feature into this
[18:16:10 CET] <kepstin> those crop values are based on the specs of the pal video system and a little bit of background knowledge
[18:17:04 CET] <kepstin> in this case, it's pal video letterboxed (so 4:3 containing 16:9), and I used the modern digial sar of 12:11 and active video width of ~704px to calculate the size of a 16x9 image area, centered.
[18:17:46 CET] <kepstin> best you could do automatically is probably have a few preset cropping values and use heuristics to pick one that should be appropriate, i guess.
[18:18:40 CET] <ghost1> cd nothing
[18:18:48 CET] <ghost1> still same w and h as source
[18:19:00 CET] <ghost1> 720x576
[18:19:16 CET] <kepstin> can yo paste your complete, updated ffmpeg command *and* the output please?
[18:19:28 CET] <kepstin> to a pastebin, not in the channel
[18:24:57 CET] <kepstin> Looks like it's working, FFmpeg days it's encoding a file with the correct size
[18:25:07 CET] <kepstin> Probably an issue elsewhere in your script
[18:25:51 CET] <ghost1> :/
[18:27:31 CET] <ghost1> well i made it like 2 lines
[18:27:41 CET] <ghost1> one for audio 1 for video then muxing them
[18:28:57 CET] <ghost1> you wanna check the script and see if you can see anything?
[19:47:45 CET] <darsain> I thought I was reading the docs correctly. why is this filter upscaling? :( ->
[19:47:47 CET] <darsain> -vf scale="w=1024:h=600:force_original_aspect_ratio=decrease"
[20:02:14 CET] <kepstin> darsain: that just means that it'll reduce either the provided w or h value in order to match the same aspect ratio as the original video, it doesn't stop it from upscaling
[20:02:46 CET] <kepstin> e.g. a 200x200 video would be turned into 600 tall, but 600 wide (the 1024 width will be decreased to 600)
[20:04:03 CET] <kepstin> if you want to not upscale, you should make the w= and  h= values into expressions, e.g. you might do w=min(iw,1024) and similarly for height.
[20:05:47 CET] <darsain> oh, I thought that the force_... argument was applied to video dimensions, not arguments before
[20:05:54 CET] <darsain> the expression makes sense now. thx! :)
[20:39:15 CET] <cynn> heya
[20:39:45 CET] <cynn> anyone got a sec to help me with something quicly?
[20:40:06 CET] <cynn> Im trying to get that tutorial code working
[20:41:17 CET] <cynn> I've fixed all the deprecated stuff in the first tutorial, just now when I try run my code I get "deprecated pixel format used, make sure you did set range correctly" followed by a whole lot of "bad src image pointers" from swscaler
[20:41:25 CET] <cynn> any ideas?
[20:42:26 CET] <cynn> pastebin of the code can be found here http://pastebin.com/Gj2GYP3K
[21:13:50 CET] <cynn> hey hey
[21:13:51 CET] <cynn> qq
[21:13:55 CET] <cynn> regarding av_image_get_buffer_size
[21:14:08 CET] <cynn> what is align meant to be>
[21:14:11 CET] <cynn> what is align meant to be?
[21:15:47 CET] <DHE> cynn: it specifies what level of alignment you want for each row. CPUs perform best when pointers are on exact multiples of certain numbers.
[21:17:27 CET] <faLUCE> Hello. With 2.8.7 version, how can I grab v4l /dev/video0,  encode it with x264 and stream it?  I tried  ffmpeg -f dshow -i video="Virtual-Camera" -vcodec libx264 -tune zerolatency -b 900k -f mpegts udp://  but 1) it says "Unknown input format: 'dshow'" 2) how can I find the string to place instead of "virtual-camera" for /dev/video0 ?
[21:17:52 CET] <cynn> hmm
[21:17:53 CET] <cynn> ok
[21:18:39 CET] <cynn> dhe: any idea why I would be getting "bad src image pointers"?
[21:18:50 CET] <DHE> cynn: can you be more specific?
[21:18:55 CET] <DHE> I have limited API support
[21:18:55 CET] <cynn> sorry, yeah
[21:18:59 CET] <DHE> s/support/experience/
[21:19:04 CET] <cynn> http://pastebin.com/Gj2GYP3K
[21:19:10 CET] <cynn> this is my pastebin
[21:19:24 CET] <cynn> Its pretty much just the example found at dranger.com
[21:19:28 CET] <cynn> just fixed up
[21:19:41 CET] <cynn> fix all the deprecated issues etc..
[21:19:59 CET] <faLUCE> nor I find "dshow" in ./configure -h
[21:20:13 CET] <cynn> Problem is, when I run it the sws_scaler says that all the time
[21:20:34 CET] <cynn> the 5 frames that I output are just black
[21:22:38 CET] <furq> faLUCE: it should appear in ffmpeg -formats
[21:24:05 CET] <DHE> cynn: avcodec_decode_video2 was split from 1 function into 2. You have to send packets and receive frames. you are not sending packets
[21:25:05 CET] <faLUCE> furq: sorry, just solved with   ffmpeg -f video4linux2 -i /dev/video0  -vcodec libx264 -tune zerolatency -b 900k -f mpegts udp://localhost:1234
[21:26:04 CET] <furq> oh
[21:26:22 CET] <furq> well yeah directshow is windows
[21:29:22 CET] <cynn> DHE: sorry
[21:29:27 CET] <cynn> I got DC'ed
[21:29:43 CET] <cynn> not sure if you responde
[21:34:04 CET] <faLUCE> furq: now, how can I change from upd to http (or tcp or rtp) ?   I tried  ffmpeg -f video4linux2 -i /dev/video0  -vcodec libx264 -tune zerolatency -r 30 -b 100k -f mpegts http://localhost:1234 but it's not the right syntax
[21:35:43 CET] <DHE> cynn: avcodec_decode_video2 was split from 1 function into 2. You have to send packets and receive frames. you are not sending packets
[21:36:32 CET] <DHE> on line 141 you receive a packet (encoded bytes) from the video, and literally do nothing with the content until it's freed on line
[21:36:34 CET] <DHE> 163
[21:36:45 CET] Action: DHE beats on the numlock of his keyboard until it apologizes
[21:39:50 CET] <serb> hy guys
[21:39:52 CET] <serb> is tis channel active
[21:39:55 CET] <serb> i need some Urgent Information
[21:40:06 CET] <c_14> Just ask, if someone can help you they will.
[21:40:27 CET] <serb> do u guys know whats the best way to extract audio from youtube videos without any loss of quality
[21:40:58 CET] <serb> a youtube downloader i use thinks the best option is to save to original "m4a" format but youtube dl thinks opus is best
[21:41:07 CET] <serb> idk if they care more for quality or size/quality ratio
[21:41:17 CET] <c_14> ffmpeg -i video -c copy -map 0:a out.mka
[21:41:29 CET] <c_14> If you know what audio codec the input is in you can use a more fitting output format than matroska
[21:41:29 CET] <serb> im asking this here because im guessing you guys know how audio inside video Werks better than Me,
[21:41:33 CET] <c_14> But matroska will always work
[21:41:47 CET] <serb> ya knowin youtube audio codec would be useful but idk where to ask
[21:42:05 CET] <c_14> most youtube videos have multiple audio codecs
[21:42:22 CET] <serb> Most of the ones i download are 720/1080p vids altho Not All.
[21:42:49 CET] <c_14> If you have youtube-dl just use youtube-dl -f bestaudio url
[21:42:51 CET] <faLUCE> I can't find a right syntax for the http protocol
[21:43:01 CET] <c_14> faLUCE: to do what?
[21:43:06 CET] <serb> c_14: thanx
[21:43:13 CET] <faLUCE> c_14: to stream http:
[21:43:23 CET] <faLUCE> ffmpeg -f video4linux2 -i /dev/video0  -vcodec libx264 -tune zerolatency -preset ultrafast -r 30 -b 200k localhost:5552/stream.ffm
[21:43:24 CET] <c_14> stream how
[21:43:33 CET] <c_14> ffserver?
[21:43:42 CET] <faLUCE> c_14: no, only one go command
[21:44:23 CET] <c_14> do you want ffmpeg to act as a http server or send to a http server?
[21:44:36 CET] <faLUCE> c_14:  I can stream udp with this:   ffmpeg -f video4linux2 -i /dev/video0  -vcodec libx264 -tune zerolatency -b 900k -f mpegts udp://localhost:1234
[21:44:46 CET] <faLUCE> c_14: act as a http server
[21:44:49 CET] <cynn> DHE: Ah...looks like I missed something here
[21:44:52 CET] <cynn> Will check it out
[21:45:59 CET] <c_14> faLUCE: https://ffmpeg.org/ffmpeg-protocols.html#http scroll down for listen
[21:46:02 CET] <c_14> there's a few examples
[21:46:09 CET] <DHE> cynn: on a related note, deprecation notices are warnings, not errors. you can still keep using the old APIs. at least for now.
[21:46:12 CET] <c_14> Note, ffmpeg doesn't make a great http server
[21:47:06 CET] <cynn> Hmmm, Im using Visual Studio 2015
[21:47:13 CET] <cynn> Its treating them as errors
[21:47:21 CET] <cynn> I tried turning it off but nothing :/
[21:48:16 CET] <JEEB> uhh, if it was then FATE on MSVC wouldn't pass?
[21:48:21 CET] <c_14> do you have -Werror somewhere?
[21:48:32 CET] <JEEB> or the equivalent of that
[21:48:37 CET] <JEEB> which makes errors from warnings
[21:48:39 CET] <faLUCE> c_14: can't find an example for http live streaming
[21:48:58 CET] <c_14> faLUCE: http live streaming = HLS?
[21:49:00 CET] <faLUCE> c_14:
[21:49:01 CET] <faLUCE> # Server side (sending):
[21:49:03 CET] <faLUCE> ffmpeg -i somefile.ogg -c copy -listen 1 -f ogg http://server:port
[21:49:28 CET] <DHE> JEEB: looks like the old avcodec_decode_video2 API was causing him grief. he tried to port to the new API but didn't fully understand what was expected of him.
[21:49:34 CET] <faLUCE> c_14: ?
[21:49:37 CET] <DHE> does FATE still test old/deprecated APIs?
[21:49:47 CET] <c_14> faLUCE: when you say "http live streaming" do you mean HLS?
[21:49:59 CET] <JEEB> I'm pretty sure during compilation you have deprecation warnings pop up
[21:50:06 CET] <JEEB> so those shouldn't cause his errors
[21:50:12 CET] <JEEB> that was the point with mentioning FATE
[21:50:14 CET] <faLUCE> c_14: I mean that I have to grab from a video4linux dev and stream it through http
[21:50:26 CET] <faLUCE> I can stream udp with this:   ffmpeg -f video4linux2 -i /dev/video0  -vcodec libx264 -tune zerolatency -b 900k -f mpegts udp://localhost:1234
[21:50:29 CET] <JEEB> that if deprecation warnings were fatal, they'd stop those MSVC FATE hosts :P
[21:50:44 CET] <faLUCE> I jonly need to adapt that command to http, but I can't find the right syntax
[21:51:47 CET] <c_14> faLUCE: why doesn't `ffmpeg -f v4l2 -i /dev/video0 -c:v libx264 -tune zerolatency -b:v 900k -listen 1 -f mpegts http://localhost:5556' work?
[21:52:06 CET] <faLUCE> c_14: I tried with listen without args :-)
[21:52:38 CET] <cynn> DHE: So what should I use instead of avcodec_decode_video2?
[21:52:51 CET] <faLUCE> c_14: great, thanks
[21:53:27 CET] <cynn> I tried avcodec_send_packet but that results in the same
[21:53:34 CET] <cynn> just blank images
[21:53:49 CET] <DHE> cynn: you submit the AVPacket with avcodec_send_packet first, then use avcodec_receive_frame. Also check the return value because the rules have changed. those two functions can fail with error AV_ERROR(EAGAIN) which is it asking you to start using the other function
[21:56:00 CET] <faLUCE> c_14: there's a problem. It doesn't start the stream until a client is connected. but meanwhile it grab frames and stores them into a buffer. How to avoid that?
[21:56:08 CET] <crelloc> hey everyone, how can i add subtitles (.srt) to a FLV file?
[21:56:33 CET] <cynn> DHE: I found that VS was set to treat C4996 as an exception
[21:56:42 CET] <cynn> Fixed it and used the deprecated way
[21:56:51 CET] <c_14> faLUCE: not aware of any method that wouldn't make it a lot more complicated
[21:57:18 CET] <faLUCE> c_14: what do you mean?
[21:57:38 CET] <DHE> sounds like maybe apache and either DASH or HLS might be the better solution, as long as you don't mind what gamers refer to as "twitch lag"
[21:58:15 CET] <c_14> I'm not aware of any ffmpeg-internal method to not open the input device and begin buffering until the output is opened
[22:00:58 CET] <faLUCE> c_14: I see
[22:02:23 CET] <zumba_ad_> hi folks. Is the word transcode synonymous to encode?
[22:02:32 CET] <serb> no.
[22:02:46 CET] <JEEB> transcode usually means that you decode and encode. but usually when people talk of encoding they mean transcoding :P
[22:04:11 CET] <zumba_ad_> ah
[22:04:13 CET] <zumba_ad_> so it's both
[22:04:36 CET] <zumba_ad_> yeah, when I say transcoding, my brain think it's encoding
[22:05:21 CET] <zumba_ad_> not sure what the transcode feature in mythtv does. I couldn't remember since it was from 2005
[22:07:50 CET] <furq> transcode typically means that the source is a lossy format
[22:08:08 CET] <furq> technically you still have to decode lossless formats, but nobody uses it like that in practice
[22:14:18 CET] <DHE> transcoding means something like going from MPEG-2 to H.264. Both have deteriorated the image quality some.
[22:14:48 CET] <DHE> but us end users don't usually get access to raw, pristine sources for our videos without high end cameras and the like
[22:28:57 CET] <furq> yeah the distinction is more useful with audio
[22:48:42 CET] <Phi_> heyo folks
[22:48:57 CET] <Phi_> I'm using H264 QSV, and finally got it to initialise an encoder
[22:49:10 CET] <Phi_> should I be passing YUV420P or NV12?
[22:51:07 CET] <kerio> aren't they the same
[22:51:18 CET] <BtbN> whatever works and is faster.
[22:51:23 CET] <Phi_> nah, they're in different order
[22:51:41 CET] <Phi_> https://ffmpeg.org/doxygen/2.8/libavcodec_2qsvdec_8c_source.html implies both, so I'm puzzled
[22:51:53 CET] <Phi_> line 171 and 174
[22:52:12 CET] <zumba_ad_> thanks furq!
[22:52:19 CET] <BtbN> so use whatever works best and avoid a needless conversion
[22:52:30 CET] <BtbN> or leave it unspecified, and don't bother.
[22:52:57 CET] <Phi_> how will I tell if it breaks?
[22:53:01 CET] <zumba_ad_> DHE, what if we rip our bluray? Is it still RAW?
[22:53:08 CET] <Phi_> the output video looks extremely odd but it's outputting
[22:53:12 CET] <kerio> lmao
[22:53:23 CET] <kerio> zumba_ad_: bluray is lossy
[22:53:30 CET] <zumba_ad_> ok so it's raw
[22:53:33 CET] <kerio> no
[22:53:40 CET] <zumba_ad_> oh lossless
[22:53:45 CET] <zumba_ad_> i got confused. Thanks
[22:53:51 CET] <zumba_ad_> I thought bluray was lossless
[22:54:09 CET] <kerio> bluray is h264 i believe
[22:54:19 CET] <kerio> at a pretty damn high bitrate, mind you
[22:54:30 CET] <zumba_ad_> k
[22:54:39 CET] <zumba_ad_> i remember it's avchd
[22:55:12 CET] <zumba_ad_> i'd like to rip my bluray disks so it's easier to play
[22:55:37 CET] <zumba_ad_> but I don't have bluray drive yet
[22:55:40 CET] <kerio> zumba_ad_: apparently it's either mpeg2, h264 or vc-1
[22:55:53 CET] <zumba_ad_> k
[22:58:36 CET] <furq> lol mpeg2
[22:58:48 CET] <furq> are there actually discs in the wild using mpeg2
[22:59:40 CET] <JEEB> furq: yes
[22:59:57 CET] <JEEB> 2005-7 or maybe even later depending on the region
[23:00:19 CET] <JEEB> I have the underwater ray romano BD release which is 40mbps MPEG-2 :P
[23:00:25 CET] <JEEB> *MPEG-2 Video
[23:00:31 CET] <kerio> also, if what wikipedia says is true, there's no 1080p60 spec in there
[23:00:34 CET] <kerio> only 1080p59.94
[23:01:05 CET] <TD-Linux> JEEB, but, does it look good?
[23:01:07 CET] <JEEB> in blu-ray? as far as I know blu-ray only supports 60/1.001 in 720p
[23:01:09 CET] <JEEB> TD-Linux: no
[23:01:15 CET] <JEEB> the MPEG-2 Video encoder did a crappy job
[23:01:23 CET] <JEEB> didn't help that it was hard telecined
[23:01:23 CET] <TD-Linux> incredible
[23:01:27 CET] <kerio> :D
[23:01:32 CET] <JEEB> so it had to cope with the extra fields
[23:01:33 CET] <furq> christ
[23:01:48 CET] <JEEB> I was hoping it would later get a re-release with AVC
[23:01:51 CET] <JEEB> but noope
[23:02:03 CET] <kerio> that's what you deserve for buying media
[23:02:04 CET] <TD-Linux> would you rate it worse than youtube 1080p?
[23:02:09 CET] <JEEB> no
[23:02:11 CET] <furq> that's what you deserve for buying a ray romano product
[23:02:20 CET] <TD-Linux> well I guess being hard telecined already makes it worse
[23:02:22 CET] <kerio> bluray romano
[23:02:29 CET] <furq> nice
[23:02:30 CET] <JEEB> TD-Linux: at leats you can handle that
[23:02:34 CET] <JEEB> *least
[23:02:37 CET] <Phi_> "LOG: h264_qsv : Specified pixel format yuv420p is invalid or not supported"
[23:02:40 CET] <JEEB> even if it's bad
[23:02:40 CET] <Phi_> whale den
[23:03:04 CET] <Phi_> also, is there a way to receive RTSP H264 frames and feed that into H264_QSV?
[23:03:07 CET] <kerio> does youtube have "source" as a thing
[23:03:14 CET] <kerio> or does it always reencode everything
[23:03:15 CET] <JEEB> kerio: it's still the least awful version of something so if I like the series I'll support it
[23:03:19 CET] <TD-Linux> JEEB, could be worse, they could have telecined then bobbed :^)
[23:03:22 CET] <JEEB> kerio: it will always re-encode everything
[23:03:28 CET] <JEEB> they do keep originals internally
[23:03:30 CET] <furq> kerio: it keeps the source but you can only download it from your own account afaik
[23:03:37 CET] <JEEB> TD-Linux: lal
[23:03:54 CET] <JEEB> also usually hard telecine in Japanese releases is due to shit like "we edited the credits interlaced"
[23:03:58 CET] <furq> that's how there are 4k videos from before youtube officially supported 4k
[23:04:08 CET] <furq> and 60fps etc
[23:04:57 CET] <furq> i think it depends on the codecs though
[23:05:12 CET] <furq> google photos uses the same infra and it only keeps the originals if you upload h264/stereo aac in mp4
[23:06:01 CET] <kerio> furq: also apparently only if you don't get a copyright strike :|
[23:06:41 CET] <furq> yeah thankfully that's not an issue on google photos
[23:07:08 CET] <furq> not that i'd know anything about that. i'm storing 750GB of videos of my precious family
[23:07:32 CET] <kerio> hmm
[23:07:57 CET] <kerio> is it "download mp4" in My Videos
[23:08:11 CET] <kerio> because i'm pretty sure i uploaded this as a ffvhuff nut, not as a h264 mp4
[23:10:09 CET] <furq> i don't think any of my uploads had aac audio so i can't check
[23:10:30 CET] <furq> also i think they were all mkv
[23:10:40 CET] <furq> this definitely isn't the original video track though
[23:11:15 CET] <serb> Hy guys
[23:11:32 CET] <serb> are there any popular audio codecs i should be aware of besides opus/ogg/mp3 ?
[23:11:37 CET] <furq> aac
[23:11:38 CET] <serb> [Lossi]
[23:11:51 CET] <kerio> ogg isn't popular
[23:12:00 CET] <kerio> also by ogg i assume you mean vorbis
[23:12:02 CET] <furq> ogg is also a container, the codec is vorbis
[23:12:04 CET] <serb> o
[23:12:04 CET] <furq> yeah what he said
[23:12:08 CET] <kerio> furq: jinx
[23:12:34 CET] <serb> Is it ok to ask about general audio storage stuff here?
[23:13:12 CET] <furq> well nobody else needs help right now so sure
[23:13:33 CET] <serb> i dont Get the codec/container Relationship
[23:13:49 CET] <serb> U can have same codec in different container ?
[23:14:09 CET] <furq> depends
[23:14:25 CET] <serb> Why do we need Containers ?
[23:14:27 CET] <furq> mp3 is normally just the audio stream with some hacked-on bits for tagging
[23:14:52 CET] <furq> m4a and ogg are containers, they can have multiple codecs, multiple streams, video tracks etc
[23:14:56 CET] <serb> or is "codec" a process rather than a Thing
[23:16:13 CET] <kerio> that too
[23:16:27 CET] <serb> so the Results of a "Codec" are stored then inside a Container
[23:16:43 CET] <kerio> h264 is a codec
[23:17:19 CET] <kerio> aac is also a codec
[23:17:24 CET] <kerio> mkv is a container
[23:17:58 CET] <serb> Is there a place where i can thoroughly compare audio codecs ?
[23:18:07 CET] <furq> you can have a raw aac stream, but you won't be able to store metadata etc
[23:18:27 CET] <furq> as far as audio goes that's normally the job of the container
[23:19:06 CET] <furq> a video is typically something like an h264 stream and an aac stream in a container
[23:19:52 CET] <serb> nice
[23:20:46 CET] <serb> h264 stream meaning data made w/ h264 codec ?
[23:20:53 CET] <furq> yes
[23:21:21 CET] <furq> also http://listening-test.coresv.net/results.htm
[23:21:38 CET] <furq> that's a few years old now but it's the most comprehensive test i know of
[23:22:48 CET] <serb> Thanx
[23:23:03 CET] <kerio> furq: why 96
[23:23:09 CET] <furq> if you have free choice then opus is the obvious winner, but it's not as widely compatible as aac
[23:23:20 CET] <furq> kerio: probably because opus and aac would score all 5s at 128k
[23:23:25 CET] <furq> and probably vorbis tbh
[23:23:44 CET] <kerio> but itunes store ships 256kbps aac
[23:23:50 CET] <furq> it sure does
[23:24:09 CET] <furq> people say v2 mp3 is transparent for a reason
[23:24:29 CET] <serb> Yeha im lookin to use opus but idk how widely Compatible it is
[23:24:32 CET] <kerio> but it's not :|
[23:24:43 CET] <furq> it is for most people
[23:24:55 CET] <kerio> i can accept that V0 LAME is perceptually lossless
[23:24:59 CET] <furq> i never tried with a particularly big corpus but i can't abx v0 and flac
[23:25:11 CET] <TD-Linux> mp3 is a bit "special" in that in corner cases it falls over badly even at really high bitrates
[23:25:19 CET] <kerio> hm
[23:25:22 CET] <furq> i imagine most people who say they can were testing with pathological samples
[23:25:32 CET] <furq> or some ancient lame version
[23:25:41 CET] <serb> Def a lot of factors to be taken into consideration wen doin tests
[23:25:56 CET] <serb> also i think abx tests dont reflect real listening situations
[23:26:33 CET] <TD-Linux> serb, they are designed to as much as possible.
[23:26:35 CET] <serb> if u focus enough durin a test and notice a difference it doesnt mean you'll probably notice it also wen listening passibly
[23:26:45 CET] <serb> vely
[23:26:47 CET] <serb> i guess
[23:26:58 CET] <TD-Linux> right well it does mean that you can probably use a lower bitrate than implied by the abx
[23:27:20 CET] <TD-Linux> note in the linked test mp3 is given 128kbps instead of 96kbps like all the other codecs, so that it can be competitive at all
[23:27:37 CET] <furq> yeah
[23:28:20 CET] <TD-Linux> serb, see also this chart: https://en.wikipedia.org/wiki/Opus_(audio_format)#/media/File:Opus_quality_comparison_colorblind_compatible.svg
[23:28:23 CET] <furq> but it's also good to know that v5 scores 4.2
[23:28:31 CET] <TD-Linux> the lower the bitrates the more obvious the differences
[23:28:39 CET] <furq> you can imagine v2 is going to be pretty close to 5
[23:28:43 CET] <serb> i have no idea what that chart conveys
[23:28:52 CET] <serb> oh.
[23:29:35 CET] <kerio> can opus do 24bit audio
[23:29:36 CET] <furq> that chart is opus marketing, idk if it's based on any real data
[23:29:41 CET] <furq> although it is pretty accurate
[23:29:44 CET] <kerio> :^)
[23:30:08 CET] <TD-Linux> furq, it was originally a whiteboard drawing. it was made by plotting some real points and then guessing on the rest
[23:30:50 CET] <TD-Linux> kerio, libopus by default is float. so "yes" :^)
[23:30:51 CET] <serb> ya prob marketing
[23:30:53 CET] <kerio> does it really make sense to give it the "opus" name across all bandwidths
[23:31:01 CET] <furq> that's the whole point really
[23:31:02 CET] <kerio> since it actually changes quite a bit in nature
[23:31:16 CET] <TD-Linux> kerio, it is one codec. and aac also changes.
[23:31:24 CET] <paolo_> ffmpeg -f video4linux2 -input_format mjpeg -i /dev/video0  <--- how can I specify the other params of video4linux ? I need to set size and fps
[23:32:00 CET] <furq> yeah he-aac is a separate profile, not a separate codec
[23:32:12 CET] <TD-Linux> and he-aacv2
[23:32:14 CET] <TD-Linux> and xhe-aac
[23:32:15 CET] <furq> and that's a bit of a pointless distinction
[23:32:28 CET] <furq> i guess it's only needed because of compatibility
[23:32:40 CET] <furq> whereas opus was designed this way from the start so that's not an issue
[23:33:43 CET] <paolo_> found it, nm
[23:34:08 CET] <kerio> TD-Linux: single-precision, or double-precision? :D
[23:34:46 CET] <paolo_> this ffserver is wonderful, and allows me to avoid the unnecessary and messy chain of vlc
[23:35:18 CET] <kerio> should i configure my mumble server to allow 128kbps opus
[23:35:18 CET] <adayzdone> Hi All. Is there a way of determining if the compression completed without errors, without reading stdout
[23:35:19 CET] <adayzdone> ?
[23:35:25 CET] <furq> kerio: why wouldn't you
[23:35:34 CET] <furq> other than that 128kbps is way too high for voice
[23:35:55 CET] <furq> i think mine was running 64kbps when i still ran one
[23:36:05 CET] <furq> not that it helped my friends' mics sound any less shit
[23:36:25 CET] <furq> adayzdone: the exit code
[23:36:29 CET] <kerio> apparently i have 288kbps as a limit
[23:36:31 CET] <kerio> why 288
[23:36:59 CET] <furq> it's not an opus limit
[23:37:02 CET] <kerio> i know
[23:37:06 CET] <kerio> but did i set that
[23:37:14 CET] <furq> probably just mumble being weird
[23:37:28 CET] <furq> or sensible, because if you're using 510kbps opus for your mumble server then you're a bad person
[23:37:37 CET] <furq> i could say the same about 288kbps though
[23:37:51 CET] <kerio> furq: positional audio, the hard way
[23:38:00 CET] <adayzdone> furq I need to send the command and background the output. I then need to check something to determine if it completed properly, like a text file it would create once compression completes succesfully
[23:38:01 CET] <kerio> just send a 7.1 stream
[23:38:08 CET] <kerio> :^)
[23:38:14 CET] <furq> adayzdone: like i said, the exit code
[23:38:43 CET] <furq> if you're on windows then i guess something like && echo OK > ok.txt
[23:38:55 CET] <adayzdone> furq Can you give me an example? On mac.
[23:39:24 CET] <furq> that should work on osx and linux as well
[23:39:48 CET] <furq> i forget how you get the exit status from a detached process now, it's not something i do very often
[23:40:53 CET] <adayzdone> furq usr/local/bin/ffmpeg -y -i '/Users/me/input.MP4' -s hd480 -c:v libx264 -crf 23 -c:a aac -strict -2 '/Users/me/output.MP4'
[23:40:56 CET] <furq> `ffmpeg ... && foo` will only run foo if ffmpeg exits 0
[23:41:16 CET] <adayzdone> ok.. let me try
[23:41:21 CET] <furq> or `ffmpeg ... && foo || bar` will run foo on success and bar on failure
[23:41:53 CET] <TD-Linux> kerio, single, but no reason you couldn't implement it with double :)
[23:42:08 CET] <kerio> other than the sheer waste of it?
[23:42:13 CET] <furq> ^
[23:42:49 CET] <kerio> how are ints mapped to floats, for the purpose of PCM audio? just shove them in the mantissa and have it be a number between 0 and 1?
[23:42:52 CET] <TD-Linux> hey you're the one that asked about 24 bit audio
[23:43:09 CET] <TD-Linux> kerio, between -1 and 1
[23:44:40 CET] <furq> the highest consecutive integer for single precision floats just so happens to be 2^24
[23:46:08 CET] <furq> but soon enough the 24-bit gold will dry up and they'll have to invent 32-bit downloads
[23:46:31 CET] <adayzdone> furq is there a way to throw an error within ffmpeg to test?
[23:46:55 CET] <kerio> furq: we can all agree that 48khz is superior to 44.1khz right
[23:46:57 CET] <furq> -c:v notarealcodec
[23:47:03 CET] <furq> or something
[23:47:11 CET] <adayzdone> thx
[23:47:12 CET] <furq> kerio: shrug
[23:47:14 CET] <furq> probably!
[23:47:28 CET] <furq> as long as my music comes off a cd i don't care
[23:47:39 CET] <kerio> ...but then your music is 44100hz
[23:47:57 CET] <furq> well yeah so i don't care if it's resampled ahead of time or by my soundcard
[23:48:51 CET] <furq> that reminds me, i still need to figure out how i have 48khz mp3s of a cd-only release
[23:49:02 CET] <furq> i wonder if it's a transcode from opus or something
[23:49:13 CET] <kerio> :|
[23:49:19 CET] <furq> alternatively i could just grab it in flac and never think about it again
[23:49:57 CET] <kerio> grab it in 48/24 flac
[23:49:59 CET] <kerio> ;o
[23:50:11 CET] <furq> transcode from a homemade vinyl pressing
[23:50:20 CET] <furq> warmth~
[23:50:30 CET] <kerio> low frequency distortion~
[23:50:36 CET] <furq> no it's warmth
[23:50:52 CET] <kerio> i'm triggered
[23:50:54 CET] <furq> if it was bad then i wouldn't have spent $10,000 on it
[23:50:57 CET] <furq> quid pro quo
[23:51:09 CET] <kerio> ¿qué?
[23:51:25 CET] <furq> it's latin for "please stop saying things i disagree with"
[23:51:58 CET] <kerio> for 10k$ you could've gotten the original DAT :3
[23:52:10 CET] <furq> well yeah i could have got a lot of things for this imaginary $10,000
[23:52:46 CET] <kerio> what even is a DAT
[23:53:03 CET] <kerio> can i store my backups on it
[23:53:27 CET] <furq> it's a tape which stores 44.1k pcm
[23:53:28 CET] <TD-Linux> helical recording tape with uncompressed audio, great if a) you want to be massively hipster
[23:53:33 CET] <kerio> *48k
[23:53:36 CET] <TD-Linux> b) you watched evangelion too many times
[23:53:51 CET] <furq> oh it does do 48k
[23:54:00 CET] <furq> you probably can store backups on it if you encode them as pcm
[23:54:11 CET] <TD-Linux> furq, it has a special mode for computers
[23:54:12 CET] <furq> inasmuch as that can be called encoding
[23:54:16 CET] <furq> oh does it really
[23:54:30 CET] <furq> 80GB eh
[23:55:36 CET] <furq> christ, 2009
[23:55:44 CET] <TD-Linux> I want a DAT walkman but they are $100+ on ebay
[23:56:05 CET] <paolo_> I'm feeding ffserver with a native mjpeg stream from a webcamera:   ffmpeg -f video4linux2 -input_format mjpeg -video_size 640x480 -i /dev/video0 -c:v copy http://localhost:8090/feedSistema.ffm   <---- unfortunately, it forces a re-encoding from mjpeg to mjpeg. I don't understand why: s it a bug?  If I replace the http url of ffserver with "myfile.avi" it works correctly and doesn't re-encode the input stream.... where can
[23:56:06 CET] <paolo_> be the issue ?
[23:56:42 CET] <furq> i don't think there's a way to not transcode with ffserver
[23:56:48 CET] <furq> also ffserver is bad and unmaintained
[23:58:31 CET] <paolo_> furq: then, instead of using ffserver, I should use another http server ?
[23:58:52 CET] <paolo_> ffmpeg -f video4linux2 -input_format mjpeg -video_size 640x480 -i /dev/video0 -c:v copy http://otherhttpserver ?
[00:00:00 CET] --- Sat Jan  7 2017

More information about the Ffmpeg-devel-irc mailing list