[00:16] <dashcloud> Hi, I'm looking for a commandline to encode video from my capture card to SD h264 in real time. If it matters, I'll be editing it afterwards to trim out things. [00:19] <c_14> https://trac.ffmpeg.org/wiki/Encode/H.264 [00:24] <dashcloud> thanks! do you think an AMD Athlon II X2 B22 is fast enough to do real-time encoding of SD resolution h264? [00:25] <c_14> Hmm, you'd have to try. [00:25] <c_14> Using -preset ultrafast will help. [00:26] <dashcloud> thanks [00:28] <iive> i'm with i3 2.9GHz (2 physical cores, 2 virtual). I can encode SD at 3 times the realtime speed. [00:28] <iive> that is, with the normal default setting. [00:29] <iive> btw, are you capturing though v4l2 (aka raw image) or dvb (aka digital tv) [00:33] <dashcloud> v4l2 [00:35] <iive> libx264 it is then :) [01:01] <Peter_Occ> here is a command line which seems to work fine [01:01] <Peter_Occ> ffmpeg -f mjpeg -r 8 -i http://user:pass@192.168.0.6/video/mjpg.cgi -r 10 /var/www/html/videos/videoMain2.mp4 [01:01] <Peter_Occ> Why would I have set it up this way, with the rate for input at 8 while the rate for output is 10? I remember doing that on purpose but I don't remember why. [02:52] <waressearcher2> I have one video 10 minutes long that is with 24 fps, and other 10 minutes long that is with 29.970 fps, when I concatenate them together using command: "ffmpeg -f concat -i list.txt -c copy out.avi" in place where they are joined there is a lag or pause for few seconds and if I add audio to that out.avi when it reaches 10 minutes there goes that [02:52] <waressearcher2> lag then video continue playing but the audio seek forward for few minutes, what I did to remedy it is added "-r 24" to ffmpeg command but will it worsen the quality ? or should I've added "-r 29.970" instead ? [02:53] <c_14> Wait, that helped? [02:53] <c_14> -r as an output option should not do anything if -codec copy was used. [02:53] <c_14> iirc [02:54] <waressearcher2> I don't remember if I removed "copy" option [02:55] <waressearcher2> but anyway, if I concatenate two videos, 24fps and 29.970fps, should I increase to upper value or decrease to lowest value fps ? [02:56] <c_14> I'd increase just to avoid frame dropping. [02:56] <c_14> But it doesn't really matter. [02:56] <c_14> Might depend on if you want to stay more PAL or more NTSC conform. [04:45] <sebastianlacuest> Hi everyone! [04:45] <sebastianlacuest> does anybody knows about qmg file format support in ffmpeg? [05:00] <c_14> Don't think so. [05:05] <active8> in linux, when ffmpeg runs and all those output messages fly, do the errors (in red) go to stderr ( fd 2) or stdout (fd 1) - i take it all normal output text goes to stdin; and what about those warnings that I see in yellowish orange? I'd like to catch them in some info dialogs I can pop up with zenity (pretty cool, that) [05:06] <JodaZ> active8, all messages go to stderr [05:07] <active8> thanks JodaZ. I'll try catching it. be nice to catch the errors separately. guess I'll have to filter on key words in the output. [05:10] <JodaZ> active8, use the return code [05:13] <active8> JodaZ, won't that just return success or failure. I was going to use two info popups. one for warnings and/or errors (the dialog icon is specific to the zenity dialog type) and one for normal output. be interesting to see if zenity preserves the message colors. [05:13] <JodaZ> active8, thing is, since encoding is supposed to be deterministic, you can always rerun it to get the error :) [05:15] <JodaZ> oh, you making some kinda gui [05:15] <JodaZ> hmm [05:15] <active8> the script I'm writing should figure out whether the dialog needs to have and exclamation icon, info, or error icon [05:16] <JodaZ> doesn't new ffmpeg output colored error info? (or was that libav) [05:16] <c_14> It does. [05:16] <JodaZ> colour would correspond nicely to your error classification [05:17] <JodaZ> but you propably shouldn't show anything poping up for anythign below red, and even for red it shouldn't abort ffmpeg or interrupt it or anything [05:17] <active8> bash script calls zenity for output info. zenity does list, checkboxes, text views, etc. radios - I could set codec, formats, pick files, all that good stuff and build the command froom the selections and not have to process commandline options to the script. [05:18] <c_14> JodaZ: I'm pretty sure red is fatal, isn't it? [05:18] <JodaZ> dunno man [05:18] <active8> i might find a way to ... easy q for quit echo q >&1 - i think [05:19] <active8> or >&0 stddin [05:19] <JodaZ> zenity sounds annoying >_> [05:19] <louis__> what should encode lossless faster, huffyuv or ffv1? [05:21] <active8> what's annoying is trying to remember that option t (or whatever) means a different thing for every stinkin linux cmd that has that option. -t for list in tar, -l for list in unzip. when I come home with a 3gp or amr on my phone, I want to just open the script, set whatever and do the comversion to mp3 or mp4 [05:21] <active8> or whatever i want without having to look up a command in my note. the script is my notebook [05:22] <JodaZ> active8, theres handbrake [05:22] <active8> you get handbrake to run on CentOS [05:22] <JodaZ> maybe i can interest you in a desktop os that is not trying to be rhel? [05:24] <JodaZ> but there are rpm's for it aparently [05:24] <active8> well, in a VM guest, debian sux and that travesty BKlinux died, IIRC. but it ran like a pig in mud when it wasn't locking up - cinelerra just locked. [05:25] <JodaZ> ?_? [05:25] <JodaZ> vm? [05:25] <JodaZ> handbrake is available for windows [05:26] <JodaZ> and centos [05:26] <JodaZ> dunno what more you could want [05:26] <active8> what was before that? AV Linux. buncha locking up. they wouldn't install right from a livecd, either. [05:27] <c_14> louis__: According to several short tests, they both take about the same length of time. But you should probably test with your source and your settings as I didn't take all variables into account (and didn't test that thoroughly). [05:28] <active8> handbrake is NOT available for CentOS in any of the repos I trust - the ones that won't break things. Now in windows, on the laptop, I can make exceptions. Doze is made to break anyway and I can keep XP in one piece with no effort [05:28] <JodaZ> ... [05:28] <c_14> The ffv1 output was about half the size of the huffyuv output though (again, might depend on the source and encoding settings). [05:28] <JodaZ> great os you got there if packages break the system [05:28] <JodaZ> but ya, suit yourself :) [05:29] <active8> thanks JodaZ . I'll keep handbrake in mind along with some of the other 'doze video and sound apps I've found usable. [05:33] <louis__> c_14: kdenlive says 45 minutes to encode ffv1, but 13 for huffyuv [05:33] <louis__> i am skeptical since the huffyuv video came out to be 29 gb for 27 minutes [05:33] <active8> JodaZ, it's for cert training - not meant for gui fooey. server and devel box. prob is that without carefully maintaining exclude lists in the repo.d files, it can accidentally replace a dependency from another app. at least that's what would happen with VLC til i found the nux-dextop repo. [05:33] <active8> but what distro / desktop OS did you have in mind, JodaZ ? [05:34] <c_14> louis__: I only tried with a short sample file, but both took about 35 seconds. [05:34] <c_14> I can try with something longer. [05:35] <active8> sumthing that will run in a VM guest without trouble? note that USB passthrough can be flakey with KVM/qemu [05:35] <JodaZ> replace a dependency? never had dep trouble with debian, while all the rpm plebs always go around installing random rpm's to save their lifes [05:35] <c_14> louis__: let me use the git version of ffmpeg while I'm at it. [05:36] <active8> JodaZ, debian ran like crap in a VM. Not sure why [05:36] <active8> but YAY! [05:36] <JodaZ> maybe because you brought up a gui on the shitty emu'd gpu [05:36] <JodaZ> or you didn't have kvm but original real deal qemu xD [05:36] <active8> I got Ardour to compile into a safe dire where I can run it directly without hosing anything [05:37] <active8> lots of ifs and and buts with VMs , lol 8) [05:37] <louis__> how are you c_14 ? [05:37] <active8> Maybe one day I can package my Ardour into an RPM and script that for easy updates [05:37] <louis__> thanks for all the help with the v4l stuff a few weeks back [05:39] <c_14> I'm fine, no problem. [05:39] <louis__> is it possible for huffyuv to do lossless video at only 169106 kilobits per second? [05:39] <louis__> that seems a little low [05:40] <nmz787> when I use libx264 trying to write video from opencv my computer freezes, but IYUV and DIV3 codecs work [05:40] <louis__> 1080p30 source [05:41] <c_14> I'm getting 130000k about for huffyuv 720p24 [05:41] <c_14> And the bitrate aside, the compression ratio is usually the more interesting part. Is the source raw? [05:42] <louis__> c_14: rawvideo capture [05:42] <louis__> i work with rawvideo because my wimpy laptop processor likes it better [05:42] <louis__> capturing lossless x264 works, but trying to edit it is a nightmare [05:43] <louis__> there's 5 seconds of lag just when you hit play in kdenlive [05:55] <c_14> Right, I've got some huffyuv encodes running now. Will do the ffv1 tests once they're done. [06:13] <c_14> Hmm, looking at some io stats, huffyuv at least might be IO limited on my machine. [06:46] <c_14> louis__: At least on my machine, both ffv1 and huffyuv take about the same amount of time and ffv1 creates files that are ~40% as large as the huffyuv encode, but as I said above I might be IO-limited. [07:35] <socksandsandals> is there an API in ffmpeg to take one AVPicture and draw it on top of another one at a given (x,y) offset? [08:26] <aditya3098> Hi, I have a live audio stream, and would like to use ffmpeg to relay this stream, with a minuite of latency [08:27] <aditya3098> In other words, if a a sound is sent over the main stream, then in a minuite, the same sound is sent over the new stream [08:27] <aditya3098> I tried other channels and was asked to use liquidsoap/gstreamer, but can I do it with ffmpeg [15:55] <svkoskin> I'm having an issue with -shortest option. I'm trying to convert a .ts file which has around 7œ minutes of H.264 video and 9œ minutes of audio to an .mp4 file, which should have 7œ min length (so that the part which only has audio track gets discarded, as I suppose -shortest should work that way). However, I get a .mp4 file with a longer audio stream than [15:55] <svkoskin> the video stream, so it seems -shortest is not working correctly. Here is the command line command and its output http://pastebin.com/k8r7dva6 [17:57] <quebre> hi [17:58] <quebre> i joined two movies into one, which is now: NVEExport.mpg, File size: 7.44gb, Audio: AC-3, Audio bitrate: 448kbit/s, Video codec: MPEG-2, Video bit rate: 3,0Mbit/s, Resolution 720x576 [17:58] <quebre> so how do i put it on my double layer DVD to play on dvd plaer ? [19:04] <sky99> I am trying to get a question answered about ffmpeg and I have tried the mailing list for for some reason anything I send there seems to fall into a black hole and never gets posted. [19:05] <sky99> My question is this: I have video files in .ts format that were obtained via a satellite receiver. [19:06] <sky99> They contain one video stream but mulyiple audio streams. [19:06] <sky99> multiple [19:07] <sky99> Each audio stream appears to have two channels (left and right) but one contains background sounds, another dialogue and so on. [19:08] <sky99> Is there any way using ffmpeg to merge all those audio streams down to a single stream, without changing the video at all? [19:09] <sky99> What i need to do is actually slightly more complicated than that nut that is the key part that I cannot figure out how to do. [19:09] <sky99> nut=but [19:29] <jberger1> hello [19:34] <jberger1> I've just upgraded to ffmpeg 2.4.1 and adjusted to the modified syntax of the overlay feature. Nearly all seems to be fine, except a input/output error messages about the (different) overlay .png files. The resulting .mov file works as expected, but I am unable to get more information about the reason, -loglevel debug does not help. [19:39] <jberger1> Debug output: http://pastebin.com/BYXdtUdh [19:54] <JodaZ> sky99, https://trac.ffmpeg.org/wiki/AudioChannelManipulation [20:12] <vlatkozelka> hi ... how to change ffprobe output format to xml instead of the plain text default ? i read its possible but can't figure out how [20:12] <JEEB> http://www.ffmpeg.org/ffprobe.html [20:13] <vlatkozelka> i read that [20:13] <JEEB> then you should know [20:13] <vlatkozelka> but doesnt show what option to use [20:13] <JEEB> with little scrolling I see the option [20:14] <JEEB> and for the value I can see it tells me to scroll further [20:16] <vlatkozelka> i get it the value is xml [20:16] <vlatkozelka> but what comes before that ? [20:16] <JEEB> hint: the example in the documentation in that point has the value "json" [20:16] <JEEB> I mean, seriously. it's written there and I found it with little scrolling. [20:17] <JEEB> and I don't even use ffprobe like that so it's not like I already knew what I was looking for :| [20:17] <vlatkozelka> maybe im just tired ... [20:17] <someone-noone> hello. I capture my mic & webcam with avfoundation framework, but resulting file is out of sync. Here's what I run: http://pastebin.com/EZRRCXST [20:18] <someone-noone> I think it's due to the fact that my webcam starts not immediately as mic and that's why video little bit late [20:18] <someone-noone> Can ffmpeg fix it without offsets? [20:18] <JEEB> vlatkozelka, ctrl+f for "json" upwards from the listing of the formats :P you should get the example there [20:19] <vlatkozelka> yeah found that [20:19] <vlatkozelka> but [20:19] <vlatkozelka> its not really usefull i though im doin something wrong [20:19] <vlatkozelka> all it does is <ffprobe> </ffprobe> [20:19] <vlatkozelka> everything else is still printed as it would always do [20:20] <JEEB> the xml format has a comment in there [20:20] <JEEB> Note that the output issued will be compliant to the ffprobe.xsd schema only when no special global output options (unit, prefix, byte_binary_prefix, sexagesimal etc.) are specified. [20:20] <JEEB> anyways, no idea otherwise, that's general ffprobe usage then and I have nfi about that \o/ [20:31] <Polarwinds> JEEB, I finished my transcoding with the script you gave me. it reduced the whole series from 27 gigs to 13 :) [20:46] <gopi> I am trying to add watermark to a video [20:46] <gopi> this is my code [20:46] <gopi> ffmpeg -i bigbunny.mp4 -i /media/movies/watermark/wm_720p.png -filter_complex 'overlay=10:main_h-overlay_h-10' -t 1 test.mp4 [20:47] <gopi> I am getting following error "/media/movies/watermark/wm_720p.png: Input/output error' [20:48] <gopi> what is this error? How do I debug it.. [20:51] <vlatkozelka> try [0:v][1:v] overlay [20:52] <gopi> what does those parameters do? [20:52] <vlatkozelka> specifies what ur overlaying over what [20:53] <vlatkozelka> 0:v is the video , 1:v is the png [20:53] <gopi> oh..mapping.. [20:53] <gopi> i am getting same error [20:53] <gopi> this is the code ffmpeg -i bigbunny.mp4 -i /media/movies/watermark/wm_720p.png -filter_complex '[0:v][1:v] overlay=10:main_h-overlay_h-10' -t 1 test.mp4 [20:55] <vlatkozelka> -map 0:v:0 -map 0:a:0 -c:v copy -c:a copy [20:55] <vlatkozelka> 1 sec ill show u a test i just did [20:57] <vlatkozelka> gopi , try this [20:57] <vlatkozelka> http://pastebin.com/QfnVtMMZ [20:58] <gopi> ok.. [20:58] <vlatkozelka> thats the result i got [20:58] <vlatkozelka> http://oi59.tinypic.com/1428tau.jpg [20:58] <gopi> give me a minute [21:01] <gopi> vlatkozelka, I am getting same [21:01] <gopi> ffmpeg -i bigbunny.mp4 -i /media/movies/watermark/wm_720p.png -filter_complex "[1:v] scale=20:20 [s] ; [0:v][s] overlay=10:10" -map 0:v:0 -map 0:a:0 -c:v libx264 -c:a copy test.mp4 [21:02] <gopi> what does this error means? "/media/movies/watermark/wm_720p.png: Input/output error' [21:02] <gopi> can't it find the file? [21:02] <vlatkozelka> maybe [21:03] <vlatkozelka> try putting -f mpeg right before the test.mp4 [21:04] <gopi> i ll trye [21:04] <gopi> http://pastebin.com/7WGvdvhP [21:05] <gopi> that's the error report [21:05] <vlatkozelka> hmm [21:05] <vlatkozelka> can u try [21:05] <vlatkozelka> ffplay /media/...../ .. jpg [21:05] <vlatkozelka> just ffplay the picture see if it works [21:08] <gopi> i can't do ffplay..since i am running it as server.. [21:08] <gopi> i did ffprobe [21:09] <svkoskin> do you have right to write to /media/movies/watermark/ [21:09] <gopi> http://pastebin.com/J5WcdfSh [21:09] <gopi> yeah [21:09] <gopi> i set 777 [21:10] <gopi> folder watermark have 777 permission [21:18] <someone-noone> hello. I capture my mic & webcam with avfoundation framework, but resulting file is out of sync. Here's what I run: http://pastebin.com/EZRRCXST [21:19] <someone-noone> I think it's due to the fact that my webcam starts not immediately as mic and that's why video little bit late [21:19] <someone-noone> Can ffmpeg fix it without offsets? [21:22] <someone-noone> or: can I tell ffmpeg to recalculate pts values not how they actually written with avfoundation or some muxer, but when the first packet comes [21:22] <gopi> i got the watermark working.. [21:23] <gopi> I am still getting Input/Output error.. [21:23] <gopi> but watermark and encoding working.. [21:23] <gopi> thanks a lot.. [21:24] <vlatkozelka> ? [21:24] <vlatkozelka> its working but giving an error ? how ? and what did you do ... im interested [21:24] <vlatkozelka> gopi ^ [21:28] <gopi> vlatkozelka, watermark and encoding is working.. [21:28] <gopi> but I am still getting this message /media/movies/watermark/wm_720p.png: Input/output error [21:29] <gopi> I don't know why [21:29] <vlatkozelka> wierd but [21:29] <vlatkozelka> what did u do to make it work ? [21:29] <gopi> I didn't make any changes.. [21:30] <vlatkozelka> lol [21:30] <vlatkozelka> well if its working dont break it i guess [21:30] <gopi> It must have been working before as well...i was trying to figure out the error message...i didn't see the output [21:30] <gopi> lol [21:30] <gopi> i should have looked at the output.. [21:31] <gopi> stupid me :P [21:31] <gopi> lol [21:31] <vlatkozelka> programmers always care about errors , its warning we dont give a fuck about ;) [21:31] <gopi> But, I am still curious..why I am getting input/Output error.. [21:31] <gopi> lol...debugging is the fun side of programming.. [21:32] <gopi> :) [21:32] <gopi> Do you have any idea, how we can trace this error? [21:32] <gopi> i am curious to figure it out.. [21:32] <vlatkozelka> no not really ... [21:32] <vlatkozelka> maybe post on the forums [21:32] <vlatkozelka> or in the mailing list [21:33] <gopi> yeah..i will do.. [21:37] <gopi> vlatkozelka, I have posted to mailing list.. [21:38] <gopi> lets wait.. [21:49] <spectralsun> I created some video files from images, but I am having issues when concatting them with existing video files that were created by phones [21:50] <c_14> https://trac.ffmpeg.org/wiki/How%20to%20concatenate%20(join,%20merge)%20medi... [21:50] <spectralsun> c_14: https://dpaste.de/si7M [21:51] <spectralsun> it's complaining about pixel format when I use this method [21:51] <c_14> That won't work. [21:51] <c_14> Have you tried reading the link I gave you? [21:51] <c_14> Try using the concat demuxer. [21:51] <c_14> That or the filter. [21:52] <c_14> Since you're reencoding anyway. [21:52] <c_14> The only video format (I know of) that you can concat with `cat' is mpegts. [21:52] <spectralsun> I'll try with the filter [21:52] <spectralsun> I also tried with the mmcat scrip [21:53] <spectralsun> https://dpaste.de/BB9T [21:53] <spectralsun> Output file #0 does not contain any stream [21:53] <c_14> >/tmp/mcs_v_all: Operation not permitted [21:53] <c_14> Something's going wrong. [21:54] <spectralsun> indeed. [21:54] <c_14> I also would not recommend that script. I've had mixed results with it. [21:54] <c_14> Sometimes it works fine for people, sometimes you need to patch certain parts, and at other times it just doesn't work. [21:55] <c_14> Also, it might not break anything, but are you sure you want this: [..]b/nsextreme/../nsextreme/[..] ? [21:57] <spectralsun> c_14: those were paths generated by the django app [21:57] <spectralsun> getting better results with the concat demuxer [21:58] <c_14> Better or good? [21:58] <spectralsun> it made a file, but the video clip that is an actual clip from a phone, is all messed up [21:58] <spectralsun> I'm going to see if converting it all to the same scale helps [21:59] <spectralsun> the file plays in ffplay [21:59] <c_14> Scale shouldn't matter that much, (except for maybe with codec copy). What has to match is things like pixel format etc. [22:01] <spectralsun> the video clip created from the image plays correctly, but the video from the actual video file looks like this: http://i.imgur.com/7Y1O76A.png [22:04] <c_14> Ye, looks like something doesn't match up between the 2 videos. [22:04] <c_14> Try with the filter or pastebin the ffprobe of both files. [22:06] <spectralsun> Okay, would it help if all the files I'm concatting are the same resolution? [22:06] <c_14> Are you reencoding during concat or not? [22:07] <spectralsun> Yeah, I want them all to come out as .mp4 [22:08] <spectralsun> basically I'm taking a 6 second clip from videos, and having splash image + 6 second clip + another splash image [22:08] <spectralsun> sometimes the 6 second clip is a .mov, but it will usually be .mp4 [22:10] <c_14> If you're reencoding, differing sizes shouldn't be an issue, if you're only remuxing (ie with -codec copy) it might be. [22:13] Action: hawken looks around [22:14] <hawken> Wondering if there are any plans with making ffmpeg read bluray mvc (.ssif-files) and if I can contribute :P [22:16] <c_14> Related to OPW or on your own? [22:23] <hawken> c_14: OPW? I just want to be able to read those files :P but I won't mind putting some work in ffmpeg if I know where to start :P [22:23] <hawken> I'm just your regular computer science student [22:27] <ubitux> hawken: ask in #ffmpeg-devel instead [22:27] <ubitux> this is where the development on FFmpeg happens [22:27] <hawken> Mkay :P [22:44] <voip1__> Hello guys, please help me [22:44] <voip1__> I have problem and i am not sure my problem related with my source or with ffmpeg parameters. [22:44] <voip1__> When i am trying to play my source with VLC, video stops and starts. Stream source provider says on his side everything is ok, and he is able normal play with VLC [22:44] <voip1__> Please chek: http://pastebin.com/dEmLZYEz [22:46] <spectralsun> c_14: https://dpaste.de/kMfc#L23 [22:47] <c_14> splash.mp4 needs an audio track, even if it's just silence. [22:48] <spectralsun> ok, how can i add one? [22:48] <c_14> you can use aevalsrc=0 [22:49] <spectralsun> c_14: thanks, trying that now [22:49] <spectralsun> c_14: Unable to find a suitable output format for 'aevalsrc=0' [22:50] <c_14> For the current command you can just use -filter_complex 'aevalsrc=0:d=2[as];[0:0][as][1:0][1:1] concat=n=2:v=1:a=1[v][a]' instead of what you have as a filter_complex. [22:51] <spectralsun> ah [22:52] <spectralsun> c_14: Input link in1:v0 parameters (size 568x320, SAR 0:1) do not match the corresponding output link in0:v0 parameters (568x320, SAR 640:639) [22:54] <spectralsun> that was when i tried to use a splash.mp4 that was converted to the same resolution as the video file [22:54] <spectralsun> here is when I use the original splash: [22:54] <spectralsun> Input link in1:v0 parameters (size 568x320, SAR 0:1) do not match the corresponding output link in0:v0 parameters (1280x720, SAR 1:1) [22:55] <c_14> Try piping in1 through setsar=1 [22:55] <c_14> '[1:0]setsar=1[vs]' [22:55] <c_14> and then use [vs] instead of [1:0] in the concat [22:56] <spectralsun> c_14: 'aevalsrc=0:d=2[as];[1:0]setsar=1[vs]; [0:0] [as] [vs] [1:1] concat=n=2:v=1:a=1 [v] [a]' ? [22:56] <spectralsun> still same error [22:58] <c_14> scale=iw*sar:ih,setsar=1 [22:58] <c_14> ie [1:0]scale=iw*sar:ih,setsar=1[v1];[0:0]scale=iw*sar:ih,setsar=1[v2] [22:58] <c_14> then use v1 and v2 in the concat [22:59] <spectralsun> c_14: could you show me a full example in my context please? [23:01] <c_14> 'aevalsrc=0:d=2[as];[1:0]scale=iw*sar:ih,setsar=1[vs];[0:0]scale=iw*sar:ih,setsar=1[v2s]; [v2s] [as] [vs] [1:1] concat=n=2:v=1:a=1 [v] [a]' [23:01] <spectralsun> ok, we are getting somewhere [23:01] <spectralsun> that worked [23:01] <spectralsun> :) [23:02] <spectralsun> I did this: [23:02] <spectralsun> 'aevalsrc=0:d=2[as];[1:0]scale=iw*sar:ih,setsar=1[v1];[0:0]scale=iw*sar:ih,setsar=1[v2]; [v1] [as] [v2] [1:1] concat=n=2:v=1:a=1 [v] [a]' [23:02] <spectralsun> it rendered the clips in the wrong order though [23:02] <c_14> switch v1 and v2 [23:03] <c_14> either after the setsar (for correctness) or in the concat input. [23:04] <spectralsun> 'aevalsrc=0:d=2[as];[0:0]scale=iw*sar:ih,setsar=1[v1];[1:0]scale=iw*sar:ih,setsar=1[v2]; [v1] [as] [v2] [1:1] concat=n=2:v=1:a=1 [v] [a]' -map '[v]' [23:04] <spectralsun> I changed the input streams before the scale parameter [23:05] <c_14> That works too. [23:05] <spectralsun> now its in the correct order, but the splash plays for too long and you only see a few seconds of the video clip [23:05] <c_14> Can you pastebin the command output? [23:07] <spectralsun> c_14: https://dpaste.de/ucnw [23:09] <c_14> That one's still in the wrong order, isn't it? [23:09] <c_14> Also, you should change d=2 to d=4 since splash.mp4 is 4 seconds long. [23:09] <spectralsun> yes, you're right [23:09] <spectralsun> ok, I'll try that [23:10] <c_14> You might just want to remux the splash mp4 directly with the silent audio [23:11] <c_14> ffmpeg -i splash.mp4 -af aevaslrc=0 -c:v copy -c:a [whatever] -shortest out.mp4 [23:12] <spectralsun> I accidently made the splash 4 seconds, it really is supposed to be 2. Can I add silent audio with this command: [23:12] <spectralsun> ffmpeg -f image2 -r 1/2 -i Downloads/nsextreme_logo.jpg -vcodec libx264 -pix_fmt yuv420p -b:v 128k ~/code/nsextreme/nsextremeweb/nsextreme/video/clips/splash.mp4 [23:13] <c_14> yeah, just add -af aevalsrc=0 -shortest [23:14] <spectralsun> I did that, but still only seeing stream 0:0 for video [23:15] <c_14> In splash.mp4? [23:15] <spectralsun> Yep [23:15] <spectralsun> ffmpeg -f image2 -r 1/2 -i Downloads/nsextreme_logo.jpg -vcodec libx264 -pix_fmt yuv420p -b:v 128k aevalsrc=0 ~/code/nsextreme/nsextremeweb/nsextreme/video/clips/splash.mp4 [23:16] <spectralsun> sorry that was an old try [23:16] <spectralsun> it has the -af flag [23:16] <c_14> Try using -f lavfi -i aevalsrc=0 -shortest [23:16] <c_14> You should place that before -vcodec [23:18] <spectralsun> okay, that worked. [23:22] <spectralsun> Still the same result on the ouput [23:22] <spectralsun> for the concatted file [23:29] <voip1__> guys, please help me [23:29] <voip1__> I have problem and i am not sure my problem related with my source or with ffmpeg parameters. [23:29] <voip1__> When i am trying to play my source with VLC, video stops and starts. Stream source provider says on his side everything is ok, and he is able normal play with VLC [23:29] <voip1__> Please chek: http://pastebin.com/dEmLZYEz [23:29] <spectralsun> c_14: output.mp4 should be around 4 seconds, but ffprobe says it's 8 [23:30] <spectralsun> https://dpaste.de/kKOH [23:30] <spectralsun> line 81 is when I do ffrobe [23:31] <c_14> ah, the fps don't match up [23:32] <c_14> append an ,fps=29.97 before the [v1] [23:33] <c_14> right after the setsar=1 [23:34] <spectralsun> Okay, thats better, but the splash screen is only up for a second [23:34] <spectralsun> the video clip plays fine after a flash of the splash screen [23:34] <spectralsun> ffrobe reporting correct video time [23:36] <c_14> So the splash screen shows for 1 second instead of 2? [23:37] <spectralsun> not even a second, its just a flash of it [23:37] <spectralsun> plays in vlc! thats a good sign [23:38] <spectralsun> but yeah, it shows the splash for a second, then it immediately shows the first frame of the 2nd clip [23:38] <spectralsun> during when it should be showing the splash [23:38] <c_14> Might now why, hmmm. Recreate the splash with -r 29.97 as an output option and then run the command again without the ,fps= [23:39] <c_14> *know [23:41] <spectralsun> c_14: ffmpeg -f image2 -r 1/2 -i Downloads/nsextreme_logo.jpg -f lavfi -i aevalsrc=0 -shortest -vcodec libx264 -r 29.97 -pix_fmt yuv420p -b:v 128k -vf scale=568:320 ~/code/nsextreme/nsextremeweb/nsextreme/video/clips/splash568x320.mp4 [23:41] <spectralsun> I am using -r twice ? [23:42] <c_14> Once as an input option, once as an output option. [23:42] <spectralsun> is that correct ^? [23:42] <c_14> should be [23:42] <spectralsun> I used that splash output and it doesn't hang on the frame of the next video, however it still only shows the splash screen for a split second [23:43] <c_14> the duration of the splash video is as it should be though? [23:43] <spectralsun> no [23:43] <spectralsun> it's just a flash of it [23:43] <c_14> I mean the file. [23:43] <spectralsun> then into the next clip [23:43] <spectralsun> no [23:43] <spectralsun> its 00:00:00.12 [23:45] <spectralsun> c_14: how do I set the length of that file? Are you sure I should be using -r twice? [23:45] <spectralsun> I want the splah to be 2 seconds long [23:46] <c_14> I'm testing right now. [23:48] <spectralsun> ok [23:51] <c_14> According to the wiki what I said _should_ work, but I found an alternative. [23:51] <c_14> ffmpeg -loop 1 -i nsextreme_logo.jpg [foobar] -t 2 splash.mp4 [23:51] <c_14> keep the -r 29.97 as an output option [23:52] <spectralsun> c_14: sweet, working now :) [23:54] <spectralsun> so, this requires me to render the splash to the same size as the video I'm concatting to [23:55] <spectralsun> if I attempt to use a larger resolution splash, I get this error again: [23:55] <spectralsun> Input link in1:v0 parameters (size 568x320, SAR 1:1) do not match the corresponding output link in0:v0 parameters (1280x720, SAR 1:1) [23:55] <spectralsun> is it possible to have ffmpeg resize the splash during the combining of the clips? [23:57] <c_14> you can try activating the unsafe flag in concat [23:57] <spectralsun> I would prefer to do it this way, since we may get new higher resolutions in the future. (as opposed to having premade splash clips at identicle expected resolutions) [23:59] <spectralsun> c_14: I don't think that will solve my question [23:59] <spectralsun> I want to have it convert all video streams to a common resolution, I'll be using the video clip as that resolution [00:00] --- Tue Sep 30 2014
participants (1)
-
burek