[Ffmpeg-devel-irc] ffmpeg.log.20151107

[burek]

[00:37:30 CET] <DHE> ChocolateArmpits: put a constraint on how far off the target bitrate it's allowed to deviate by
[00:38:21 CET] <DHE> to clarify, the target bitrate is almost like CBR, but with more short term variation allowed. short being less than 1 or 2 seconds
[00:42:59 CET] <ChocolateArmpits> DHE: so you want it to hit 5500k only for 2 seconds tops ?
[00:43:13 CET] <ChocolateArmpits> What rate limit test software are you using to check the bitrate ?
[00:44:03 CET] <ChocolateArmpits> Bitrate Viewer is pretty good, though not updated anymore http://www.winhoros.de/docs/bitrate-viewer/download.html
[00:44:20 CET] <ChocolateArmpits> But it supports your output
[00:44:39 CET] <ChocolateArmpits> has second/GOP/frame bitrate views
[00:44:45 CET] <ChocolateArmpits> Check what the second one is
[00:44:50 CET] <ChocolateArmpits> seconds
[00:45:04 CET] <DHE> ChocolateArmpits: just a perl script parsing "ffprobe -show_frames". I add the size of the video frames to a running counter and compare it against a running counter at my set bitrate
[00:45:36 CET] <ChocolateArmpits> Are you sure it averages the seconds correctly ?
[00:45:47 CET] <DHE> in my mind the difference should never exceed the bufsize
[00:46:11 CET] <DHE> oh, I also used 'tsplay' to actually play the video to multicast and watching a per-second bandwidth report on the NIC
[00:47:50 CET] <DHE> less empirical but easier to read
[00:48:26 CET] <DHE> someone suggested just going cbr (minrate, maxrate and bufsize all identical) and got better looking picture and better bitrate control. which surprised me
[00:50:30 CET] <ChocolateArmpits> bufsize should be set higher than the rest, you might get buffer overflows else
[00:52:21 CET] <ChocolateArmpits> Anyways, bufsize doesn't act as a hard number you can add to other rate control settings and expect something
[00:53:45 CET] <DHE> I imagined the bufsize as a bucket with bits. it fills at the b:v rate, drains up to the maxrate:v, and all bits spend on the output must come from the bucket.
[00:54:16 CET] <ChocolateArmpits> nah it fills up at the maxrate
[00:54:23 CET] <DHE> oh?
[00:54:43 CET] <DHE> and -b:v is just the target file size?
[00:54:47 CET] <ChocolateArmpits> the b:v simply determines the bitrate the content is at
[00:55:08 CET] <ChocolateArmpits> maxrate at what maximum rate the content can be served to the buffer
[00:55:24 CET] <ChocolateArmpits> as in the demuxed (but not decoded) video stream data
[00:55:43 CET] <DHE> that actually explains a lot...
[00:56:03 CET] <DHE> well, I have some parameters to tweak...
[00:57:09 CET] <ChocolateArmpits> the buffer then serves data to the decoder, if decoding isn't fast enough, buffer overflow happens. If maxrate isn't big enough to serve the data fast enough the buffer gets empty and buffer underflow happens as the decoder can't decode frames
[00:57:43 CET] <DHE> the decoding hardware is more than up to the task
[00:57:54 CET] <DHE> but the connection speed is very limited
[01:03:12 CET] <ChocolateArmpits> it might be practical to recognize particular stream errors
[02:12:36 CET] <jorb> so i asked this question last night, and actually ChocolateArmpits helped me a bit to get some things going a few days ago.
[02:13:02 CET] <jorb> https://gist.github.com/anonymous/4cac887664a6d44813c5
[02:13:23 CET] <jorb> the final result of this concat operation is screwed up a bit
[02:13:46 CET] <jorb> the resulting video ends up with gaps between the concat'd streams
[02:13:48 CET] <jorb> any advice?
[02:14:23 CET] <jorb> *caugh* use some other program to do this?
[04:11:57 CET] <Prelude_Zzzzz> good day everyone ( evening )
[04:13:30 CET] <Prelude_Zzzzz> can anyone help me.. i seem to ahve a problem with a 2Mbit/s video that i am encoding. On motion ( mostly people and their faces ) .. when they move i get sort of a frame rate issue which looks like motion blur or stutter or something. Almost like the frame rate is below 24 frames per second and also blury as they move.  but after they stop moving its clean again. bit rate is 2m but jumps up to liek 4mbit/s with vbr .. not sure what the
[04:13:31 CET] <Prelude_Zzzzz> problem is :( can anyone help ?
[04:13:37 CET] <carlitux> hey, there is a way to put many images on a video? each image on different positions of the video in one command or should I have do many commands?
[04:16:21 CET] <Prelude_Zzzzz> carlitux that woudl be cool .. i wish i coudl put lots of videos on like 720p... is there some way ?
[04:16:38 CET] <Prelude_Zzzzz> like i fyou get a screen and you use up 1080p of space but make videos of 10x10 or somehing like that
[04:16:56 CET] <Prelude_Zzzzz> and devide into say 20 videos per 1080p
[04:18:15 CET] <Prelude_Zzzzz> i guess they call it mosaic
[04:19:19 CET] <Prelude_Zzzzz> check out contact command for ffmpeg
[04:19:23 CET] <Prelude_Zzzzz> there may be something we can use
[05:45:28 CET] <Prelude_Zzzzz> hey guys.. vstats log file... can anyone tell me the best way to speficy just what i want to output ? eg.. it doesn't show me dropped frames.. just sends at p=?
[14:05:10 CET] <leleobhz> A non-ffmpeg related question, but i dont know where to ask: Someone know what Sony thinked when they choose export AAC audio to a 3GP container?
[14:05:23 CET] <leleobhz> And more info about the way PS3 encodes AAC and 3GP
[14:11:04 CET] <JEEB> I have no idea PS3 encodes anything during the course of usage :/ also 3gpp file format is pretty much a subset with some extensions of ISOBMFF
[14:11:13 CET] <JEEB> (which is colloquially called "mp4")
[14:18:51 CET] <furq> "Audio imported from CD onto a PlayStation 3 when it is set to encode to the MPEG-4 AAC format copies onto USB devices in the 3GP format."
[14:18:56 CET] <furq> everything about that sentence is terrible
[14:18:59 CET] <nyuszika7h> I have subtitles in TX3G format but the text is messed up in some places, is there any way to fix it? https://i.imgur.com/QcS4ywv.png
[14:26:49 CET] <nyuszika7h> actually since it makes sense in some places I guess Hub Network just messed up with their closed captions
[14:27:06 CET] <leleobhz> JEEB: CD audio encoding..
[14:27:20 CET] <leleobhz> furq: right. I have these files here and i found it very weird
[14:27:47 CET] <leleobhz> furq: and just for curiosity, PS3 does not encode the ID3 tag together, so no audio metadata
[14:32:57 CET] <furq> speaking as someone who used to own a hi-MD player and thus had to use sonicstage and atrac3+
[14:33:03 CET] <furq> the answer is almost certainly that sony are terrible
[14:33:40 CET] <furq> just be glad they didn't invent another proprietary audio format which can only be played back on sony phones
[14:34:44 CET] <JEEB> well, 3gp circa ~2005 was what phones took in, which is most probably why that brand of "mp4" was picked
[14:34:55 CET] <JEEB> at least it encoded in AAC and not AMR
[14:36:43 CET] <JEEB> and yeah, sony gave up on only having their own proprietary thing supported years ago thankfully. even MD players started supporting plain PCM audio if nothing else (probably later other compressed formats, too - the MD format as a disc still lives in japan due to multiple companies licensing the physical format)
[14:36:54 CET] <JEEB> although they still use their own proprietary shit for games consoles
[14:37:10 CET] <JEEB> like PSVita and PS4 have standard functionality to decode ATRAC9
[14:37:39 CET] <furq> oh man there's an atrac lossless too
[14:39:00 CET] <leleobhz> PS3 also supports atrac
[14:39:37 CET] <leleobhz> JEEB: there is a way with ffmpeg to dissect this file?
[14:41:09 CET] <JEEB> leleobhz: if it's crypted no
[14:41:17 CET] <JEEB> if it's not crypted then it should be supported out of the box
[14:41:22 CET] <leleobhz> JEEB: ti plays normal
[14:41:24 CET] <leleobhz> *it
[14:41:43 CET] <leleobhz> JEEB: I just want to find out more about the way it encodes, psychoacoustic, etc.
[14:42:06 CET] <JEEB> not much you can gather from the decoded output about that
[14:42:14 CET] <JEEB> you'd be better off reverse engineering the encoder library
[14:42:25 CET] <JEEB> that way you would know exactly what it does
[14:44:17 CET] <leleobhz> JEEB: last but not least. I used PS3 - like said - to extract CDDA. I did not know any good (really good) audio extractor/ripper/encoder for Linux. Can you say something about it?
[14:46:50 CET] <TD-Linux> cdparanoia is good. but literally anything on linux, like rhythmbox or any of the other GUIs, would be better than what you get with the ps3
[14:46:56 CET] <furq> i hear abcde is good if you're happy with a cli tool
[14:47:02 CET] <furq> or you can run eac in wine
[14:59:29 CET] <leleobhz> TD-Linux: why? I mean, encoder is ok, AAC @320kbps
[15:00:01 CET] <leleobhz> But last time i extracted a CD, what makes more diference in CDDA extraction is the way software handles CDRom drive. I dont know if it still appliable today
[15:00:19 CET] <leleobhz> Also, the lattest software in linux i know is rubyripper
[15:01:12 CET] <furq> what's wrong with that
[15:01:35 CET] <furq> anything which uses cdparanoia should be fine
[15:17:19 CET] <darktoad> Anyone has those errors sometime will rebroadcasting an rtmp feed? WriteN, RTMP send error 10054 (42 bytes) RTMP_ReadPacket, failed to read RTMP packet body. len: 27520
[15:17:25 CET] <darktoad> *while
[15:18:13 CET] <darktoad> most of the time it's just fine, then it's noobs out of nothing.
[15:22:14 CET] <darktoad> both side use nginx-rtmp, I don't think it's bandwidth related
[16:05:32 CET] <AlexYeCu> Hi everybody.
[16:05:32 CET] <AlexYeCu> [alsa @ 0x209cb00] cannot set sample format 0x10000 2 (Invalid argument)
[16:05:32 CET] <AlexYeCu> hw:2,1: Input/output error
[16:05:32 CET] <AlexYeCu> How to resolve?
[17:56:44 CET] <Prelude_Zzzzz> hey , anyone around ?
[17:57:03 CET] <Prelude_Zzzzz> can anyone help me.. i seem to ahve a problem with a 2Mbit/s video that i am encoding. On motion ( mostly people and their faces ) .. when they move i get sort of a frame rate issue which looks like motion blur or stutter or something. Almost like the frame rate is below 24 frames per second and also blury as they move.  but after they stop moving its clean again. bit rate is 2m but jumps up to liek 4mbit/s with vbr .. not sure what the
[17:58:28 CET] <waressearcher2> Prelude_Zzzzz: hallo
[22:15:57 CET] <waressearcher2> niemand is hier
[22:42:11 CET] <ChocolateArmpits> Using wallclock as an input for timestamps on Windows results in broken playback. Is there any particular way to use it ?
[22:53:47 CET] <zumba_ad_> hey folks, any ideas how to fix this issue? I'm using mac osx - http://pastebin.com/1miFY8Pt
[22:53:59 CET] <zumba_ad_> i don't like to install it using brew
[22:54:51 CET] <waressearcher2> zumba_ad_: hallo
[22:55:15 CET] <zumba_ad_> hi
[22:55:38 CET] <waressearcher2> zumba_ad_: wie gehts ?
[22:55:44 CET] <zumba_ad_> what?
[22:55:57 CET] <zumba_ad_> i don't understand what you are saying
[22:56:56 CET] <c_14> zumba_ad_: Can you pastebin the last 50 or so lines from your config.log ?
[22:58:22 CET] <zumba_ad_> sure
[22:59:33 CET] <zumba_ad_> http://pastebin.com/WaVN0Q7v c_14
[22:59:57 CET] <c_14> >Agreeing to the Xcode/iOS license requires admin privileges, please re-run as root via sudo.
[22:59:59 CET] <c_14> lel
[23:00:24 CET] <c_14> eeeeh
[23:00:31 CET] <zumba_ad_> so i just need to run it as sudo?
[23:00:34 CET] <c_14> Try running `sudo cc' or something
[23:00:40 CET] <c_14> Don't actually compile as root
[23:00:48 CET] <c_14> Just accept that random license agreement
[23:00:52 CET] <zumba_ad_> k
[23:00:59 CET] <c_14> then try again
[23:01:33 CET] <zumba_ad_> still failed w/ same error
[23:01:54 CET] <zumba_ad_> WARNING: Unknown C compiler gcc, unable to select optimal CFLAGS
[23:02:08 CET] <zumba_ad_> from config.log
[23:02:15 CET] <c_14> That's not an error
[23:02:18 CET] <zumba_ad_> k
[23:02:43 CET] <c_14> Does it still have the license stuff?
[23:03:00 CET] <zumba_ad_> yes
[23:03:09 CET] <zumba_ad_> You have not agreed to the Xcode license agreements, please run 'xcodebuild -license' (for user-level acceptance) or 'sudo xcodebuild -license' (for system-wide acceptance) from within a Terminal window to review and agree to the Xcode license agreements.
[23:03:15 CET] <zumba_ad_> maybe it's different
[23:03:21 CET] <c_14> run that
[23:03:23 CET] <zumba_ad_> k
[23:04:23 CET] <zumba_ad_> i think it's working now :)
[23:04:34 CET] <zumba_ad_> thanks
[23:04:52 CET] <zumba_ad_> yes, configure was successfull. Running make && make install now
[23:05:20 CET] <c_14> ChocolateArmpits: what's your command?
[23:06:29 CET] <ChocolateArmpits> c_14: well I tried to use -use_wallclock_as_timestamps 1 input option in one case and on another setpts=time(0);asetpts=time(0) . The last one results in the video track not playing, the wallclock produces absolutely choppy output
[23:07:42 CET] <ChocolateArmpits> I'm figuring ffmpeg is requesting time from Windows for the option?
[23:09:51 CET] <c_14> It should be using gettimeofday
[23:12:26 CET] <ChocolateArmpits> as I said I'm on windows.
[23:12:33 CET] <c_14> So?
[23:12:35 CET] <c_14> gettimeofday is posix
[23:12:53 CET] <c_14> So it should be provided by msys or whatever
[23:21:25 CET] <c_14> If you can reproduce it, (especially with something like -i testsrc) open a bug I guess.
[23:23:51 CET] <ChocolateArmpits> You presume the ouptut should be "ok" ?
[23:23:55 CET] <ChocolateArmpits> output
[23:23:57 CET] <ChocolateArmpits> *
[23:24:06 CET] <c_14> hmm?
[23:24:21 CET] <ChocolateArmpits> Well I mean not screwed up by trying to use current time for timestamps
[23:24:50 CET] <ChocolateArmpits> You think that should not be happening ?
[23:31:25 CET] <c_14> No idea. The worst thing that can happen is that the bugreport is closed as invalid.
[23:33:40 CET] <zumba_ad_> I'm trying to adjust the contrast, brightness, etc. I remember using parameters somewhat similar to this but I'm getting the error "no such filter: mp". Here is the command I executed - ffmpeg -i WP_20151106_16_58_47_Pro.mp4 -c:v libx264 -c:a aac -vf "mp=eq2=1:1.68:0.3:1.25:1:0.96:1" new.mp4
[23:34:05 CET] <c_14> drop the mp=
[23:34:10 CET] <zumba_ad_> ok
[23:34:11 CET] <c_14> and eq2 is now just eq
[23:34:13 CET] <c_14> iirc
[23:34:17 CET] <zumba_ad_> got it
[23:34:37 CET] <zumba_ad_> i have it in my notes but i couldn't remember where I saved it
[23:44:33 CET] <basisbit> hey, using ffmpeg 2.8.1, I am having issues with video playback of certain avi files when using CODEC_FLAG_LOW_DELAY. It mixes up the frames. So you see frame2, then frame 1, then frame 4 then frame 3 and so on. When not using CODEC_FLAG_LOW_DELAY, everything plays nicely. This is a karaoke software, so a log audio to video delay is essential for good experience. Any idea how to fix this?
[23:48:46 CET] <zumba_ad_> thanks c_14, it worked using eq
[23:49:06 CET] <zumba_ad_> is it possible to remove some noise on our video?
[23:50:07 CET] <zumba_ad_> i mean image noise on the video
[23:50:10 CET] <zumba_ad_> like grain
[23:50:24 CET] <c_14> atadenoise, dctdnoiz, hqdn3d, owdenoise, looks like pp also has a denoizer,  removegrain,
[23:50:27 CET] <c_14> Pick one
[23:50:35 CET] <c_14> Or 2 or 3
[23:50:49 CET] <c_14> Never used any of them, so you'll probably just have to test.
[23:51:02 CET] <zumba_ad_> k
[00:00:00 CET] --- Sun Nov  8 2015