[Ffmpeg-devel-irc] ffmpeg.log.20200106

[burek]

[03:48:21 CET] <wondiws> hi furq
[04:02:09 CET] <CounterPillow> hi furq
[05:38:18 CET] <Roden> the -to parameter decides the length of the video, is there a way to decide the ending point instead?
[05:54:02 CET] <furq> that is what -to does
[05:54:10 CET] <furq> it needs to be on the same side as -i
[06:04:29 CET] <Roden> Can you check my command for your off-side error? `> ffmpeg -ss 00:00:05 -i i.mp4 -to 00:02:02 -c copy o.mp4`
[06:05:22 CET] <Roden> You mean it must come before -i if -ss is before -i    and after -i if -ss is after -i?
[06:06:39 CET] <furq> on the same side as -ss i mean
[06:06:42 CET] <furq> of -i
[06:07:31 CET] <Roden> `> ffmpeg.exe -ss 00:00:24 -to 00:01:10 -i .\1xmas.mp4 -c copy 0xmas.mp4` works.
[06:10:24 CET] <Roden> I don't know if you feel like talking about ffmpeg stuff, but if you do, I'm also wondering if it's possible to use ffmpeg to quickly make an animated gif from 2 or more image files (png/jpg).
[06:10:53 CET] <Roden> maybe it's an imagemagik question
[08:32:11 CET] <jaggz> I have a 10mb mp3, and a 600k jpg, and I'm making a video out of them with the image looped..
[08:33:10 CET] <jaggz> it ends up being a bit over 20mb in size though (it's actually more like 15mb mp3, 700k jpg, and ends up as a 33mb mp4, but probably no matter)
[08:33:27 CET] <jaggz> ffmpeg -loop 1 -r 18 -i "$img" -i "$snd" -c:v libx264 -tune stillimage -c:a aac -b:a 160k -pix_fmt yuv420p -shortest -q:a .1 "$out"
[08:34:01 CET] <jaggz> I was testing with that possibly horrible -q:a (I haven't even listened yet), but am wondering what options are available to make the final mp4 much smaller
[08:50:15 CET] <jaggz> ok.. setting video quality helps.. now I'm down at 900k and it still sounds and looks nice :)
[08:50:23 CET] <jaggz> out-ba160k-qa1.8-qv31-r18.mp4
[08:50:33 CET] <jaggz> I wonder if -loop 1 is a problem
[09:25:15 CET] <Roden> jaggz that sounds really strange to me, man, but I don't have a deeper explanation of why I expect things to go differently other than an instinct: I feel like the same jpg file should be referenced at multiple times in an optimized video rather than having the file redundantly loaded throughout the video.
[09:27:24 CET] <Roden> Like it should only load frame 1 and frame 2 and activate like this: (t=1 fr=1), (t=2 fr=2), (t=3 fr=1), (t=4 f=2) ... (t=last_odd_time f=1), (t=last_even_time f=2).
[09:28:16 CET] <Roden> I've also been very curious about how to composite images into video format without having them load redundantly.  If you see me around and happen to come upon an answer, or if you have a URL linking to your question online, please let me know.
[10:01:45 CET] <helloimdumb> hello
[10:01:57 CET] <helloimdumb> I'm wondering if someone can help me
[10:02:39 CET] <helloimdumb> I'm trying to get ffmpeg running on a new pc
[10:03:01 CET] <helloimdumb> but unlike prior versions, this one doesn't come with ff-prompt.bat
[10:04:16 CET] <helloimdumb> how am I supposed to open ffmpeg with the new version?
[11:09:45 CET] <wondiws> hi furq
[11:51:32 CET] <pagios> hi all, is it possible to create an rtmp redirect like an http redirect? Idea is to completely send a broadcaster to a new server instead ofrelaying netween servers and consuming server to server bandwidth
[11:52:20 CET] <BtbN> no
[11:52:44 CET] <BtbN> rtmp is an entirely different protocol
[11:56:35 CET] <nimaje> how does being a diffrent protocol imply it can't have a similar feature, BtbN?
[11:57:13 CET] <BtbN> rtmp is pretty damn primitive
[11:57:27 CET] <BtbN> and somehow over-complex at the same time, but not in a useful way
[11:58:55 CET] <nimaje> so the anwser is just that it doesn't have such a feature
[12:01:10 CET] <TheAMM> pagios: see nginx-rtmp
[12:01:20 CET] <furq> nginx-rtmp doesn't do it either
[12:01:25 CET] <furq> there's an ancient open issue for it
[12:01:47 CET] <furq> FMS has some kind of redirect functionality but it's not clear if anything actually supports it
[12:01:51 CET] <TheAMM> nevermind then
[12:01:57 CET] <furq> lavc/librtmp don't
[12:02:03 CET] <furq> or lavf rather
[12:30:53 CET] <pagios> so what options do i have
[13:11:33 CET] <BtbN> nginx-rtmp can re-stream it
[14:21:06 CET] <kepstin> pagios: the best option is probably to use dns to load balance, possibly in combination with running the traffic from one ip to multiple severs with a frontend like haproxy.
[14:27:14 CET] <wondiws> hi furq
[14:29:33 CET] <CounterPillow> ^- what is the "hi furq" all about?
[14:34:18 CET] <wondiws> CounterPillow, I wanna talk with him, a followup on an earlier conversation
[14:34:41 CET] <CounterPillow> maybe you should just say what you want to say.
[14:51:04 CET] <wondiws> Then I will ask another question that I can't find a more appropriate active channel for: I took an old VCD from the shelf. I wanted to copy it to the harddrive, in linux I needed to use cdrdao in raw mode. Windows gives me a filesystem an I can copy the (virtual) DAT file to C:\ but the resulting file is in RIFF format.
[14:51:15 CET] <wondiws> This is not the format on the disc itself I imagine
[14:51:29 CET] <wondiws> what kind of format is this? It isn't AVI either I suppose
[14:52:35 CET] <DHE> I've burned VCDs in the past. it's just .mpg format video on a track containing mpeg1video (!)
[14:53:07 CET] <DHE> though AVI is a variant of RIFF I think...
[14:53:17 CET] <wondiws> that is true
[14:54:46 CET] <wondiws> In linux you can't copy this *.dat file
[14:55:38 CET] <DHE> linux doesn't work that way. there's not really a filesystem. I think it might literally by a .mpg file burned directly as a track on the CD
[14:57:05 CET] <wondiws> DHE, yes on track 2. Track 1 is a filesystem though
[14:57:28 CET] <wondiws> but the problem is, it is in CDXA mode
[14:58:26 CET] <wondiws> I think the mpeg does use 2 byte CRC per sector though, I'm afraid that if I readraw track 2 I will get the CRC intertwined in the resulting file, right?
[14:59:33 CET] <DHE> this is going beyond my level of knowledge. but since unrepaired corruption would be highly visible on playback I would guess VCD would use the full ECC of data CD modes
[15:01:52 CET] <wondiws> DHE, no, it doesn't use ECC for the actual video
[15:05:46 CET] <kepstin> yeah, vcd did a tradeoff where they accepted video corruption in favour of getting more video on disk
[15:06:13 CET] <DHE> interesting..
[15:14:26 CET] <kepstin> wondiws: fwiw, if your cdrdao toc file says the track is MODE2/2352 then it should be just the data, no crcs
[15:14:44 CET] <kepstin> er, no cd-rom format junk specifically
[15:15:26 CET] <wondiws> kepstin, it says MODE2_FORM_MIX
[15:15:56 CET] <kepstin> oh, i'm getting the cue and toc format mixed up, whoops
[15:17:16 CET] <kepstin> hmm, that includes the subchannel data (cdrom stuff), it's not just the video :/
[15:23:39 CET] <wondiws> kepstin, yes, I don't understand why they would want the to save those few bytes on the tiny track 1
[15:32:00 CET] <devinheitmueller> Is there still a ffmpeg meeting at 11:00 EST?  Has anyone posted meeting info?
[16:08:33 CET] <durandal_1707> devinheitmueller: nope, j-b is skiing
[16:08:53 CET] <devinheitmueller> @durandal_1707 Ok, thanks.
[16:11:33 CET] <durandal_1707> devinheitmueller: it will probably be in next few weeks, follow -devel mailing list
[16:14:42 CET] <devinheitmueller> ok
[16:31:02 CET] <unixabg> Greetings, I launch a process to grab images from a rtsp stream and it works great. I would like to leave the process running
[16:31:29 CET] <unixabg> and send a sighup or something to send it new params and save the overhead of launch with new params. Is this possible?
[17:47:22 CET] <andymaier> I am using ffprobe to get the title from an AVI file, outputting it in JSON format. When the title in the file contains non-ASCII characters, ffprobe represents these with the Unicode replacement character (U+FFFD). Is there any way to tell ffprobe to either assume a particular codepage for the metadata, or to output it in bytes as read from the file (i..e. without converting it to Unicode)?
[19:15:27 CET] <relaxed> andymaier: I don't think so, look at string_validation_replacement in the man page
[19:26:03 CET] <dostoyevsky> > [mp4 @ 0x72211c0] AV1 muxing is currently not supported.
[19:26:49 CET] <andymaier> @relaxed Thanks much, this helped! Using -of json=sv=ignore made the difference for me, then i can process the result myself.
[19:27:01 CET] <dostoyevsky> Does ffmpeg support av1 muxing with mp4 these days?  (I think the error comes from an older version of ffmpeg)
[19:41:25 CET] <Hello71> did you search for that string in the source code
[20:04:49 CET] <cehoyos> av1 muxing in mp4 is supported
[22:31:27 CET] <^Neo> Hello, in ff_spdif_probe in spdifdec.c, why does it look to see if the data type (buf[1]) is less than 0x37/55?
[22:46:50 CET] <BtbN> The data type?
[22:46:54 CET] <BtbN> You mean... the value?
[22:47:01 CET] <BtbN> Probably because the format looks like that.
[22:50:28 CET] <^Neo> yeah, I'm reading through the 61937 standard and it's not obvious
[22:55:02 CET] <^Neo> the datatype field is supposed to be 7 bits, but is limited to 4 bits  (0x1F)
[22:55:12 CET] <^Neo> there was an ammendment in 2011 to add sub data types
[22:55:16 CET] <^Neo> and the code is from 2010
[23:02:35 CET] <cehoyos> Do you see a sample for which probing fails?
[23:18:27 CET] <^Neo> No, not really. I'm just trying to figure out why it's 37...
[23:18:30 CET] <^Neo> I think I figured it out though
[23:18:52 CET] <^Neo> 0x37 = 10111 for the datatype, and in the 2010 version of the standard, it only went up to datatype 22 (10110)
[00:00:00 CET] --- Tue Jan  7 2020