[04:15:40 CET] <fpihl> o, my favourite debug feature is deprecated, `-debug vis_qp`. Then I found the codecview filter, `-vf codecview=qp=1` but that one is using the function av_frame_get_qp_table(), that function is however behid the FF_API_FRAME_QP define which in libavutil/version.h is removed. Enabling it causes an compile error. Is this feature deprectated or shall I spend some time getting it to work again? [04:16:45 CET] <pink_mist> fpihl: your question sounds more appropriate for #ffmpeg-devel [04:17:12 CET] <pink_mist> I at least haven't the faintest clue :P [04:19:28 CET] <fpihl> @pink_mist, thanks. I [04:19:40 CET] <fpihl> I'll try thre instead [04:44:53 CET] <kadiro> hello, a little question about delogo filter, I get this error when trying with band= : [Parsed_delogo_0 @ 0x7f8af01aa4c0] Option 'band' not found [04:45:20 CET] <kadiro> is it replaced or removed from delogo filter? [04:46:09 CET] <kadiro> I use something like: -vf "delogo=x=370:y=38:w=84:h=60:band=10" [04:46:48 CET] <kadiro> If I remove band=10 the command work but not as it was before [04:50:40 CET] <kadiro> !delogo [06:05:59 CET] <montana> anybody have ffmpeg that support HE-AAC ? [07:34:58 CET] <microchip_> montana: you need an ffmpeg compiled with libfdk-aac, which is not distributive due to license [07:35:15 CET] <microchip_> i don't think anyone offers such [07:35:30 CET] <montana> i've seen it [07:36:03 CET] <microchip_> then it violates the license [07:38:22 CET] <montana> but libfdk-aac worth it [07:41:16 CET] <edenist> https://www.freshports.org/multimedia/ffmpeg [07:41:35 CET] <edenist> freebsd version has the option available [off by default] if you want to built it yourself [07:48:41 CET] <montana> what do you people think about this rating [07:48:44 CET] <montana> Based on quality produced from high to low: [07:48:44 CET] <montana> libopus > libvorbis >= libfdk_aac > aac > libmp3lame >= eac3/ac3 > libtwolame > vorbis > mp2 > wmav2/wmav1 [07:55:33 CET] <microchip_> it's accurate? :) [07:59:50 CET] <montana> i just realized neroaac is better than opus [09:53:05 CET] <ryzenda> I am confused about something regarding using ffmpeg to create a video that is exactly 10 hours long, not a second less or a second more. [09:53:20 CET] <ryzenda> I created the video using command: ffmpeg -f concat -i list.txt -c copy out.webm [09:54:03 CET] <ryzenda> where list.txt contains 102,858 lines of "file 'filenamehere'" [09:54:40 CET] <ryzenda> and combined, all the video files together equals exactly 1,080,000 frames of video. [09:55:45 CET] <ryzenda> After I created the final output video, I confirmed with `ffmpeg -i ~/media/out.webm -map 0:v:0 -c copy -f null -` that the video contains exactly 1,080,000 frames [09:55:53 CET] <ryzenda> And the video is 30fps [09:56:42 CET] <ryzenda> $ mpv ~/media/out.webm ` (+) Video --vid=1 (*) (vp8 1280x720 30.000fps) [09:57:41 CET] <ryzenda> But for some reason the video appears as 10hrs 4mins 17secs, both with mpv and also uploaded to YouTube [10:03:35 CET] <ryzenda> https://www.zapstudio.net/framecalc/ confirms that even if somehow the video has become 29.97 fps, that 1,080,000 frames is still 10hrs and 36secs [10:05:45 CET] <ryzenda> It would have to be 29.78735 fps to be 10hrs 4mins 17secs [12:09:17 CET] <montana> i noticed ffmpeg's website is extreamly bias to free software: libopus > libvorbis >= libfdk_aac > aac > libmp3lame >= eac3/ac3 > libtwolame > vorbis > mp2 > wmav2/wmav1 [12:10:25 CET] <BtbN> That's just how those compare in quality per bitrate? [12:10:32 CET] <furq> all of those are free software [12:10:49 CET] <montana> btbn libfdk/apple-aac/nero-aac all outperform libopus [12:10:55 CET] <durandal_1707> lies [12:10:58 CET] <BtbN> I have my doubts about that. [12:10:59 CET] <furq> no they don't [12:11:17 CET] <montana> i have samples to prove it [12:11:33 CET] <BtbN> One sample proves nothing. There will always be special cases where one outperforms the other. [12:11:55 CET] <montana> that ranking should be changed [12:12:02 CET] <furq> no it shouldn't [12:12:34 CET] <montana> furq because you are bias to free software [12:12:39 CET] <BtbN> The only thing that surprises me a bit is that it claims libvorbis>=fdk, but I guess there's a bunch of blind testing behind that scale. [12:12:53 CET] <furq> all of those are free software [12:13:25 CET] <montana> it's more like nero-aac > apple-aac > libfdk_aac > libopus > libvorbis [12:13:43 CET] <BtbN> And your source for that claim is...? [12:13:55 CET] <BtbN> Also, how do I use nero or apple aac with ffmpeg? [12:13:58 CET] <montana> they sound better after 1 pass and 10 pass [12:14:02 CET] <BtbN> lol [12:14:25 CET] <furq> well this was a good conversation [12:14:52 CET] <montana> this is what opus sounds like after 10 pass: https://x0.at/w01.mp4 [12:15:11 CET] <furq> weren't we having this exact conversation months ago [12:15:17 CET] <furq> you were wrong then and you're still wrong for the exact same reason now [12:15:22 CET] <furq> i'm sensing a pattern [12:15:39 CET] <montana> furq month ago, i didn't do the test myself, i just believe what website told me [12:15:51 CET] <montana> now, i did the test myself [12:15:57 CET] <BtbN> re-encoding the same file again and again is not in any way a valid test [12:16:22 CET] <durandal_1707> transcoding already transcoded file is not good reason to claim anything [12:16:48 CET] <montana> nero-aac still outperforms libopus even after 1 pass [12:17:14 CET] <durandal_1707> just your opinion [12:17:15 CET] <montana> durandal_1707 i will admit somebody did 100 pass test and that's unrealistic, but 10 pass is not [12:17:41 CET] <montana> http://bernholdtech.blogspot.com/2013/03/Nine-different-audio-encoders-100-p... [12:18:22 CET] <durandal_1707> transcoding already transcoded file is stupid, and that is end of story. [12:18:31 CET] <BtbN> Even if nero aac is magically amazing as you claim, it's not supported by ffmpeg and thus does not belong into any comparison of ffmpeg supported encoders. [12:18:40 CET] <montana> yes 100 times is unrealistic, but 5 pass is not [12:18:46 CET] <durandal_1707> wrong [12:20:39 CET] <montana> it's not wrong [12:21:47 CET] <montana> durandal_1707 when you are transcoding a random audio file: that file has been transcoded before [12:22:02 CET] <furq> don't transcode random audio files then [12:22:10 CET] <montana> furq but what if you want to [12:22:20 CET] <furq> then that's your problem [12:23:30 CET] <montana> furq even audio from dvd/bluray are transcoded before [12:23:35 CET] <montana> commercial dvd/bluray [12:23:53 CET] <furq> most bluray audio is lossless now [12:23:59 CET] <montana> furq some yes [12:24:06 CET] <montana> not all [12:25:08 CET] <furq> so did you test the performance when transcoding from ac3 and dts [12:25:51 CET] <montana> my point is " <@durandal_1707> transcoding already transcoded file is stupid" this is unvoidable [12:26:21 CET] <furq> what does that have to do with transcoding to opus 10 times in a row [12:26:21 CET] <montana> Have you ever wondered how different codecs are affected by re-encoding / re-compressing? Of course, recompressing audio is a bad idea, but sometimes can't be avoided. Quality loss will inevitably occur, but are some codecs more resilient than others? To clear things up, I did a test with the following encoders: [12:32:41 CET] <BtbN> There are dedicated codecs with no generation loss if you need that [12:33:03 CET] <BtbN> though for audio that's rather uncommon, since you can just use a lossless codec instead, while still having a manageable bitrate [12:33:29 CET] <montana> btbn and none of them can do that with less than 64 kbps [12:33:48 CET] <BtbN> Nothing that sounds good can do with less than 64kb/s [12:36:51 CET] <montana> btbn i use 24kbps mono, and it sounds fine using HE-AAC [12:45:51 CET] <ryzenda> Oh, btw, if anyone noticed my messages from earlier, I am still trying to figure out why 1,080,000 frames at 30fps does not result in a 10:00:00 hour video, and instead results in a 10:04:17 video [14:28:36 CET] <GenTooMan> 1000000 frames is 9:15:33.33 in time I would guess 1080000 should be 10:00:00 (about0 [14:32:45 CET] <DHE> the math says yes, but depending on the file format or maybe there's actually a slightly off or variable framerate this could be incorrect or just mis-estimated [19:15:45 CET] <duude__> Hello everyone [19:17:00 CET] <duude__> Do you people know if it's possible to make ffmpeg work with shortcuts? [19:17:07 CET] <duude__> *.lnk files in windows [20:27:13 CET] <GenTooMan> duude__ well ffmpeg has a very extensive cli interface, I'm not sure a link file will do you much good to use it. It's not a GUI application. You will want to use it at the CLI. [20:28:13 CET] <GenTooMan> A better option is to find a GUI application that uses ffmpeg and sets everything up for you under windows if that's what you want to do. [20:29:28 CET] <GenTooMan> DHE good point with variable frame rate encoding frame rate becomes somewhat pointless. [22:12:45 CET] <kuresu> Hello. I'm writing an application that downloads a video split in several .ts chunks, joins them together, then finally encodes them to .mp4 using ffmpeg. [22:13:47 CET] <kuresu> I wanted to know if there's a way to pipe the chunks to ffmpeg "gradually" and let it do the encoding while the chunks are being downloaded. [22:14:47 CET] <kuresu> without having to download everything, then join the .ts files together, then start the ffmpeg encoding process. [22:34:10 CET] <KodiakIT[m]> So aside from brushing up on my trig/calc/matrix math, is there anything else I should be looking at (i.e. in the source of ffmpeg) to better grok how QP/RF affects bitrate and lossiness or lossless-ness? [22:35:09 CET] <furq> well you'd probably want to look in the source of x264 [22:35:15 CET] <KodiakIT[m]> s/affects/translates into/ [22:35:24 CET] <furq> if you're going to look into any source [22:36:37 CET] <KodiakIT[m]> furq: so https://github.com/FFmpeg/FFmpeg/blob/master/libavcodec/libx264.c ? [22:36:48 CET] <furq> no i mean the actual x264 code [22:37:28 CET] <furq> i don't think any of the ffmpeg internal encoders have a concept of qp or rf [22:37:42 CET] <KodiakIT[m]> Ah, whoops, https://code.videolan.org/videolan/x264.git [22:37:42 CET] <KodiakIT[m]> then [22:37:46 CET] <furq> sure [22:42:46 CET] <kuresu> also, just to be sure does the command 'ffmpeg -i in.ts -c copy -y out.mp4' automatically adds the '-bsf:a aac_adtstoasc' filter? [22:44:17 CET] <jemius> kuresu, I'd say that ffmpeg never automatically adds filters [22:44:28 CET] <furq> kuresu: yes [22:44:35 CET] <jemius> furq, ? [22:44:38 CET] <furq> it says in the docs and also it would throw an error otherwise [22:44:44 CET] <furq> https://www.ffmpeg.org/ffmpeg-bitstream-filters.html#aac_005fadtstoasc [22:45:11 CET] <jemius> hmm [22:45:33 CET] <KodiakIT[m]> Also, while I'm thinking of it, is there a way to easily concatenate a couple of DVD backups in an .mkv with chapter metadata, and preserve/transform that with just a few command line switches? or would I have to concatenate them, then add the chapters back in manually later? [22:45:46 CET] <furq> also it will autoinsert filters if what you asked for is impossible otherwise [22:46:09 CET] <furq> e.g. converting to a pixel format that's supported by the encoder [22:46:35 CET] <furq> KodiakIT[m]: there's no way of doing that that isn't annoying [22:46:49 CET] <furq> makemkv will do it but it has some crazy person shareware license [22:46:54 CET] <furq> and also it's gui only [22:47:02 CET] <KodiakIT[m]> furq: lovely. [22:47:10 CET] <kuresu> ok thanks, do you also know what 'Non-monotonous DTS' errors are caused by? Floating precision errors from the .ts file? [22:47:29 CET] <KodiakIT[m]> Guess that's an excuse to write one up or bash it together with mkvtoolnix + bash [22:48:04 CET] <furq> well you'd need something to actually dump the title and chapters from the ifo [22:48:12 CET] <furq> which mkvtoolnix can't do, or else i'd have suggested that [22:48:28 CET] <furq> i have a workflow for it on windows but not really on *nix [22:48:33 CET] <furq> maybe dvdbackup or something [22:48:39 CET] <KodiakIT[m]> Ah, I'd been mucking about with the .json for handbrake's GUI. [22:48:47 CET] <Sirisian|Work> Trying to wrap my head around astats and rms level. In the big picture I'm trying to find out if a video has any audible audio. I understand that RMS is relative to the max loudness at 0. Is it reasonable to say that any RMS greater than say -10 for RMS peak dB on audio channels is "no audio"? My thinking is a loud file will have an RMS peak db of like -100 and a quiet file will have around -40 which is consistent with my tests. [22:49:05 CET] <furq> tccat from transcode will dump the title to stdout but you'd need to get the chapters separately [22:49:09 CET] <Media_Thor> Greetings! can x265 pass 1 log files generated for 1080p output file be reused for other resolution file please, similar to X264? [22:50:11 CET] <Media_Thor> Sirisian|Work: "-af silencedetect" ? [22:52:45 CET] <Sirisian|Work> Fascinating. I did not even see that mentioned anywhere. Will test to see if it's viable. [23:22:13 CET] <Sirisian|Work> Is there a simple fix for "Application provided invalid, non monotonically increasing dts to muxer in stream". https://pastebin.com/w2Bqy9wd is the frames from ffprobe. I wrote a quick program to verify and all the dts values are increasing. [23:23:28 CET] <Sirisian|Work> Or is that not an important error? [00:00:00 CET] --- Tue Nov 12 2019