[Ffmpeg-devel-irc] ffmpeg.log.20200203

[burek]

[00:14:23 CET] <DHE> Heart_And_Chops: ffmpeg tries to maintain the aspect ratio when rescaling by changing the SAR/DAR. at the end of your filter pipeline add setsar=sar=1 to explicitly reset to square pixels
[00:14:39 CET] <Heart_And_Chops> thanks
[00:16:42 CET] <barattolino1993> Hi, the flag +delete was removed from segment_list_flags, how can i do?
[00:18:53 CET] <Heart_And_Chops> DHE thanks that fixed it.
[00:20:41 CET] <barattolino1993> What can i replace it with?
[00:22:15 CET] <BtbN> The hls muxer.
[00:26:50 CET] <barattolino1993> BtbN thanks
[00:33:36 CET] <barattolino1993> BtbN for re-stream of live m3u8, hls muxer or segment muxer?
[00:35:40 CET] <barattolino1993> What is the difference
[00:42:35 CET] <DHE> segment muxer is generic, able to split files pretty freely with or without an index file. HLS is a specific case requiring a specific index file output and stipulations on the output video file format
[00:50:02 CET] <barattolino1993> so nobody can help me?
[00:53:00 CET] <DHE> I just provided an answer. was it inadequate?
[00:54:39 CET] <barattolino1993> maybe because i'm crashed, can you repeat please?
[00:57:29 CET] <DHE> ah...
[00:57:29 CET] <klaxa> >segment muxer is generic, able to split files pretty freely with or without an index file. HLS is a specific case requiring a specific index file output and stipulations on the output video file format
[00:58:14 CET] <DHE> sorry I didn't notice
[01:03:53 CET] <barattolino1993> the segment muxer not delete segment files that are no longer part of the playlist, so i use hls muxer?
[01:05:45 CET] <BtbN> The segment muxer does not even generate a playlist by default
[01:05:49 CET] <BtbN> that's just a weird bonus feature
[01:05:56 CET] <BtbN> its primary focus is to segment a stream
[01:10:43 CET] <barattolino1993> because before with segment muxer you could
[01:11:00 CET] <barattolino1993> so i use hls muxer for re-stream?
[01:12:02 CET] <barattolino1993> DHE BtbN thanks really
[01:54:04 CET] <nickname123> hi
[02:03:44 CET] <nickname123> the static build doesn't work on kubuntu 19.10. there are 1 million fontconfig errors. the first one is: Fontconfig warning: line 5: unknown element "its:rules"
[02:05:41 CET] <nickname123> i only have used linux for 6 months and am no expert, but could it be that the static build contains an old fontconfig that can't read config files of kubuntu 19.10?
[02:06:36 CET] <nickname123> the older ffmpeg version from the ubuntu repository works fine
[03:57:18 CET] <Abbott> is there a suite of meta tools that people commonly use to encode video? Like applying filters or to have a bit more immediate control over how ffmpeg functions?
[12:53:35 CET] <funnybunny2> Can I get the buffer size needed for converting to PCM ahead of time? Right now I just realloc when full
[12:55:32 CET] <durandal_1707> missing context
[13:00:47 CET] <funnybunny2> durandal_1707: I get the first audio stream in a file and set up a codec. I send packets and then repeatedly call avcodec_receive_frame. With frame->linesize[0], I know how many samples each frame has. If it will overflow my buffer, I realloc my buffer and then  put frame->extended_data[0] in the buffer
[13:00:55 CET] <funnybunny2> Should I post a working example?
[13:01:37 CET] <funnybunny2> Instead of having to realloc, I want to know if there is a way before sending packets/calling avcodec_receive_frame if I can get the total buffer size needed for all the samples
[13:02:50 CET] <funnybunny2> Right now I'm only handling planar data and converting to S16 samples
[13:18:00 CET] <durandal_1707> i do not think so
[17:52:21 CET] <repo> hello there!
[17:53:07 CET] <repo> i'm using ffmpeg to record from a capture card and stitch the resulting video together with a static one.
[17:53:48 CET] <repo> so 2 seconds of capture card vid, 7 seconds of static vid, and then 1 second of capture card vid
[17:54:54 CET] <repo> here's the relevant code i'm using: https://p.jokke.space/YOc/
[17:55:35 CET] <repo> i'm seing the following output when i capture: [video4linux2,v4l2 @ 0x7f9e76e79400] Dequeued v4l2 buffer contains 16372 bytes, but 3110400 were expected. Flags: 0x00012001
[17:55:51 CET] <repo> and then when i concatenate the videos: Too many packets buffered for output stream 0:0
[17:56:41 CET] <repo> i already set the -max_muxing_queue_size to a pretty high value as you can see, but i still get the same error
[17:57:46 CET] <repo> oh and global_options = ["-y", "-loglevel", "error"]
[17:58:54 CET] <repo> in dmesg i see: uvcvideo: Non-zero status (-71) in video completion handler
[18:03:09 CET] <DHE> errno 71 = "Protocol Error"
[18:03:31 CET] <DHE> there might be a hardware or driver error, but if you're not constantly pulling video from the camera it might get upset
[18:04:27 CET] <repo> yeah i'm just pulling 10s of video
[18:04:58 CET] <repo> this is on ubuntu lts 18.04 btw
[18:07:14 CET] <repo> so that's 4.15.0 kernel
[18:08:29 CET] <JEEB> 18.04 now has also 5.3
[18:09:26 CET] <repo> ah
[18:09:31 CET] <repo> i'll try upgrading
[18:28:35 CET] <repo> JEEB: nice, that seems to have done the trick
[18:29:40 CET] <repo> i do have another issue now though... maybe it's the same one but in a different kind of way
[18:30:04 CET] <repo> now sometimes the resulting video is much longer than it should be
[18:33:18 CET] <repo> hm now i also see uvcvideo: Non-zero status (-32) in video completion handler
[19:57:13 CET] <nifker> Does ffmpeg rendering using CUDA?
[19:58:27 CET] <waleee-cl> nifker:  you need an ffmpeg compiled with support for it (and a nvidia card, the drivers, the cuda libs &c)
[19:58:47 CET] <nifker> waleee-cl: does it only work on nvidia hw?
[19:59:33 CET] <waleee-cl> well ... I haven't heard or read about someone using it on an amd/ati or integrated gpu
[19:59:43 CET] <waleee-cl> it's a search away
[20:00:17 CET] <waleee-cl> nifker:  https://trac.ffmpeg.org/wiki/HWAccelIntro
[20:00:41 CET] <waleee-cl> so yeah, you need an nvidia gpu
[20:01:53 CET] <BtbN> ffmpeg does not render anything though, so not sure what you're asking.
[20:02:26 CET] <nifker>  /s/render/encode
[20:02:26 CET] <BtbN> There's a handful of CUDA based filters
[20:02:41 CET] <BtbN> encoding via nvidia is nvenc, not CUDA
[20:05:19 CET] <nifker> how do I find out if my ffmpeg installation supports it?
[20:07:47 CET] <Hoffman> the cuda encoder/decoder is `cuvid`
[20:07:50 CET] <Hoffman> just fyi
[20:08:45 CET] <Hoffman> oh I guess they renamed it
[20:26:11 CET] <BtbN> cuvid never was an encoder
[22:05:38 CET] <CoreX> is there a way to output an encode to 2 folders?
[22:09:02 CET] <chriswells0> I'm trying to get an RTSP stream from a Garmin VIRB 360 camera and save it to a file, but I don't know what parameters to provide.  This is sample of what I'm doing (unsuccessfully) now: https://p.bsd-unix.net/view/2c9ae1bf
[22:09:47 CET] <chriswells0> Any suggestions for settings to try?
[22:16:07 CET] <CoreX> try the input with quotes
[22:16:51 CET] <CoreX> chriswells0 this thing when i google your problem https://stackoverflow.com/questions/53987141/eorror-when-try-recorder-rtsp-stream-ffmpeg
[22:19:04 CET] <chriswells0> I got the same error using quotes.  Let me change some params to match theirs.
[22:32:02 CET] <chriswells0> Ok, I can at least get a stream if I change the query string parameter from 0 to something normal like 1440.
[22:32:25 CET] <chriswells0> The "invalid parameter" was referring to that.
[22:46:28 CET] <DanielTheFox> hello again guys, how can I remove (as good as possible; I'm not expecting miracles) non-voice sound?
[22:47:39 CET] <DanielTheFox> like, which highpass/lowpass frequencies
[22:59:13 CET] <Hello71> nowadays you probably use "artificial intelligence"
[22:59:44 CET] <DanielTheFox> 'could be fine, sadly, I want to do that on my PC
[23:00:02 CET] <DanielTheFox> AI stuff doesn't appear to be available for offline systems, but it'd be great
[23:00:44 CET] <DanielTheFox> whatever, so one can't filter non-voice sound just by cutting non-voiceband frequencies, right?
[23:00:58 CET] <c_14> DanielTheFox: try -af stereotools=slev=0.015625
[23:01:29 CET] <DanielTheFox> the audio signal is mono
[23:06:05 CET] <c_14> do you have a version with only the instrumentals?
[23:06:16 CET] <DanielTheFox> nah, it's not a song
[23:06:29 CET] <DanielTheFox> it's just some random audio recorded with a cellphone
[23:07:22 CET] <c_14> if it's just background noise try something like afftdn or an audio denoiser
[23:07:50 CET] <c_14> https://ffmpeg.org/ffmpeg-filters.html#afftdn
[23:08:19 CET] <c_14> or just search in that page for denoise
[23:42:06 CET] <arhat> I'm trying to convert a DCP video file to apple prores with ffmpeg, it seems like it's working but as it's transcoding i'm getting a bunch of `jpeg 2000 end of mismatch 1` messages. Is this okay?
[23:48:50 CET] <cehoyos> It's likely that the warnings are buggy, if the j2k files are lossless, you can compare with other decoders to validate
[00:00:00 CET] --- Tue Feb  4 2020