August 2019 - Ffmpeg-devel-irc

ffmpeg.log.20190817
by burek 18 Aug '19

18 Aug '19

[05:02:43 CEST] <argent0> hi, why is this command, intended to generate white noise audio, also generating video when I pipe it to ffplay? ffmpeg -ar 44100 -ac 2 -f s16le -i /dev/urandom -c:a copy -t 2 -f nut - | ffplay -f nut - [05:03:26 CEST] Action: argent0 wants to generate white noise audio and play it at the same time [05:04:56 CEST] <furq> argent0: ffplay -nodisp [05:05:27 CEST] <argent0> furq: but is the output audio only? [05:05:30 CEST] <furq> also you can just use ffmpeg -f lavfi -i anoisesrc [05:05:39 CEST] <furq> and yes [05:06:22 CEST] <nicolas17> you could output to a file and check with ffprobe :P [05:08:24 CEST] <argent0> furq: nicolas17: Thanks [05:12:36 CEST] <argent0> now ffplay keeps running [05:13:01 CEST] Action: argent0 ffmpeg -f lavfi -i anoisesrc -c:a copy -t 2 -f nut - | ffplay -f nut -nodisp - [05:16:36 CEST] Action: argent0 i used `ffmpeg ... | mpv -` [12:20:30 CEST] <Foaly> i'm using the blend filter with addition [12:20:35 CEST] <Foaly> but it comes out purple [12:20:40 CEST] <Foaly> why is that? [12:49:07 CEST] <Foaly> :( [13:26:28 CEST] <goel> Hi is it possible to pipe in compressed video data from an RTMP stream to ffmpeg and then possibly for it to output the decoded images? The compressed data from an RTMP stream would be FLV1. [13:27:49 CEST] <goel> I accidentally disconnected. [13:28:12 CEST] <goel> I just wanted to know if ffmpeg supported piping in data to decode? [13:28:45 CEST] <klaxa> i don't see why not [13:28:46 CEST] <DHE> yes, you can use stdin for a source [13:29:29 CEST] <goel> Ok, so I would just have to specify the type of video data coming for the decoder? E.g. H263/FLV1 or H264? [13:29:56 CEST] <goel> That's brilliant, thank you anyway! [13:30:55 CEST] <DHE> ffmpeg attempts autodetection if it's detectable and wrapped in something parsable [13:31:28 CEST] <goel> I mean it wouldn't really by in a container, it would be just raw bytes. [13:31:42 CEST] <goel> A bytestream which needs to be decoded as FLV1 or H264 [13:32:04 CEST] <DHE> h264 elementary stream is a thing... not sure about flv1 [13:32:59 CEST] <goel> Ok, that's fine, it's not a problem at all. Thank you everyone. [14:43:13 CEST] <Foaly> is there a way to make the "blend" filter run in rgb color format? [14:44:12 CEST] <Foaly> even when i convert both inputs to rgb24 or rgba, it still adds an "auto_scaler" that convert it to yuv [14:45:24 CEST] <c_14> looking at the query_formats list in vf_blend.c makes it look like bgra should work [14:45:37 CEST] <c_14> eeh, gbr [14:45:48 CEST] <Foaly> thx, i will try [14:47:06 CEST] <c_14> gbrp or gbrap [14:47:47 CEST] <fling> Is there an ultimate guide on muxing multiple live sources together? [14:48:00 CEST] <Foaly> great, it works! [14:48:02 CEST] <Foaly> thank you! [14:48:04 CEST] <fling> Like multiple v4l devices into different streams in a single file [14:48:09 CEST] <fling> Maybe with sound too [14:48:17 CEST] <fling> Can't I sync two pulse devices together btw? [15:15:37 CEST] <snatcher> is there such a thing as "volume level" of audio? [15:18:13 CEST] <DHE> relatively speaking yes. there is a "maximum volume" that can be produced, though the measurement is normalized since it doesn't directly correspond to real-world volume levels [15:23:28 CEST] <snatcher> what's the unit of measurement in such a case? for example: audio1 "volume level" is 1, audio2 is 2, so audio2 louder than audio1, is there a filter/tool? [15:25:11 CEST] <furq> snatcher: https://en.wikipedia.org/wiki/DBFS [15:26:48 CEST] <furq> not to be confused with dB SPL [15:28:58 CEST] <snatcher> is there a way to extract it with ffmpeg? [15:29:53 CEST] <furq> https://ffmpeg.org/ffmpeg-filters.html#volumedetect [15:47:44 CEST] <fling> audio snatcher! [17:01:07 CEST] <kepstin> snatcher: note that if you're concerned about perceptual loudness - i.e. if a person would think A or B is louder - you want https://en.wikipedia.org/wiki/LKFS (also called LUFS). The ebur128 filter in ffmpeg can measure this. [17:08:08 CEST] <DHE> sounds like a competitor to ZFS or btrfs... [17:08:12 CEST] <DHE> (joke) [17:16:20 CEST] <durandal_1707> really bad one [17:23:37 CEST] <pink_mist> it wasn't particularly good, but at least it wasn't offensive ... saying it was really bad is a bit harsh [17:24:41 CEST] <durandal_1707> i feel insulted [18:14:24 CEST] <Foaly> is there a trick to correctly overlay one video with alpha ontop of another [18:14:27 CEST] <Foaly> ? [18:14:30 CEST] <Foaly> or rather, png sequences [18:14:58 CEST] <durandal_1707> tried trick with overlay filter? [18:15:24 CEST] <Foaly> i use overlay=format=gbrp:alpha=straight [18:17:14 CEST] <Foaly> wait, i might be really dumb [18:21:28 CEST] <Foaly> yeah, i just had the order wrong :( [18:21:36 CEST] <Foaly> sorry, it works fine of course [18:22:20 CEST] <durandal_1707> i thought ffmpeg was wrong all this years [18:23:41 CEST] <durandal_1707> just for one second [19:17:55 CEST] <Oldest> why is ffprobe.exe file big as ffmpeg.exe ? [19:18:54 CEST] <durandal_1707> because it is static build [19:19:12 CEST] <Oldest> i understand that but ffprobe doesn't do anything like ffmpeg [19:19:14 CEST] <DHE> because it has all the core ffmpeg features. really the encoders and muxers are about the only things you could omit [19:19:36 CEST] <Oldest> ffmpeg is the file that does everything, ffprobe file is just for probing [19:20:02 CEST] <DHE> you say that, but you can write a filter chain and have ffprobe tell you about the resulting output [20:03:58 CEST] <Krock> Hello everyone. Is there an option to playback or process videos with GLSL using ffmpeg/ffplay? [20:04:21 CEST] <furq> have you tried mpv [20:05:23 CEST] <Krock> yes, but is it also capable processing videos? [20:05:36 CEST] <BtbN> It's pure render-shaders for all I'm aware. [20:05:39 CEST] <BtbN> ffmpeg has no such support [20:06:08 CEST] <Krock> so it's up to the front end software to add shader support on top of ffmpeg? [20:06:44 CEST] <BtbN> ffmpeg generally does not interface with OpenGL at all [20:06:56 CEST] <BtbN> There used to be opengl stuff, but I think it's all gone? [20:07:00 CEST] <Oldest> krock what is GLSL and why do you need that [20:07:03 CEST] <BtbN> And it was only some weird renderer [20:07:03 CEST] <Krock> originally came here from VLC but apparently yet only MPV suports that [20:07:15 CEST] <BtbN> FFmpeg focuses on OpenCL [20:09:51 CEST] <Krock> Oldest: GLSL are shaders which can be applied on videos for visual effects such as sharpen or upscale [20:11:39 CEST] <BtbN> They require OpenGL though, as the name suggests [20:12:29 CEST] <furq> there are thirdparty glsl filters for ffmpeg [20:12:37 CEST] <furq> i assume they won't get merged because of dependencies [20:13:00 CEST] <BtbN> There is no OpenGL interop at all, no compatible frames. So those are bound to be pretty inefficient [20:14:03 CEST] <Krock> I see. Thanks for the answers :) [20:25:14 CEST] <jph> howdy. anyone know a way to get ffmpeg (or really, ffprobe) to show if an audio stream is VBR? [20:26:04 CEST] <der_richter> don't ask in mpv, it can't use the rende routput as encoding input [20:26:10 CEST] <der_richter> *render [20:26:25 CEST] <der_richter> your best bet, if you can program, might be libplacebo [20:27:38 CEST] <der_richter> i was assuming encoding since you asked here, if it's just playback, yeah mpv.. [23:07:38 CEST] <poutine> Is there a guaranteed order with the optional PES header's data? if pts/dts indicator, escr flag, and es rate information had their proper bits set, would I expect them to be in that order? [23:08:17 CEST] <DHE> mpegts headers? [23:08:36 CEST] <poutine> DHE: yes, MPEG-2 PES header's Optional PES Header [23:13:14 CEST] <Thomas_J> Is there a way to get ffmpeg to import part of it's command line from files or external input? [23:13:58 CEST] <nicolas17> I don't think so, but if you're on non-Windows you can let the shell do that for you [23:14:42 CEST] <nicolas17> ffmpeg $(cat args.txt) [23:15:00 CEST] <durandal_1707> you can set filtergraphs from file [23:15:54 CEST] <Thomas_J> I was just thinking that. I should be able to build the ffmpeg command string with a bash script. Is there any documentation on using args? [23:18:24 CEST] <Aerroon> i don't understand what loudnorm does [23:18:31 CEST] <Aerroon> i read about it and it all makes sense to me [23:18:39 CEST] <Aerroon> but when i use it the results are nothing like what i read about [23:18:59 CEST] <Aerroon> the idea is great: it makes audio sound roughly at a similar loudness, right? [23:19:21 CEST] <Aerroon> but when i actually test it, i have two audio clips with almost the same integrated loudness, yet one sounds MUCH louder [23:19:48 CEST] <Aerroon> and even the peak loudness is lower on the louder clip according to loudnorm [00:00:00 CEST] --- Sun Aug 18 2019

1 0

ffmpeg.log.20190728
by burek 17 Aug '19

17 Aug '19

[01:40:28 CEST] <barg> if i do ffmpeg -i myvideo.mp4 then i see it says rotate : 90 and and displaymatrix: rotation of -90.00 degrees. How do I get rid of that? I notice also when I upload it to youtube then it shows rotated. [01:52:29 CEST] <nine_milli> blb [04:30:00 CEST] Last message repeated 1 time(s). [06:30:07 CEST] <barg> got it ffmpeg -i input.mp4 -c copy -metadata:s:v:0 rotate=0 output.mp4 [06:32:02 CEST] <nine_milli> blb [10:52:00 CEST] Last message repeated 1 time(s). [13:43:48 CEST] <nikio_> how can i switch stdin input streams at will without breaking the ffmpeg outputstreams. [13:44:12 CEST] <nikio_> Im streaming in audio from youtube and streaming it out as mpeg-dash [13:45:02 CEST] <nikio_> ? [16:45:29 CEST] <nikio_> anyone? [16:46:12 CEST] <durandal_1707> simply said you can not do that [16:46:16 CEST] <JEEB> nikio_: you need an API client that has that functionality [16:46:35 CEST] <JEEB> the libraries let you do something like that, but you need a way to sticking it together that the ffmpeg.c command line client doesn't handle [16:46:47 CEST] <nikio_> well i am using nodejs [16:46:49 CEST] <nikio_> and streams [16:46:53 CEST] <nikio_> i could do it myself? [16:47:18 CEST] <nikio_> i have to add the right bytes or somethng [16:47:37 CEST] <nikio_> concat the end of the stream and the beginning of the new one the right way? [16:48:02 CEST] <JEEB> in some ways you might or might not be able to use ffmpeg.c and stitching, but not sure how well that'd work [16:48:18 CEST] <nikio_> i just need some pointers into the rigth direction [16:48:20 CEST] <JEEB> I've yet to have time to check upipe for their dynamic input stuff [16:48:29 CEST] <nikio_> i dont like using extra libraries and let them hide all that is going on [16:48:42 CEST] <nikio_> id rather just learn the minimum amount of theory and understand why the stream breaks [16:48:47 CEST] <nikio_> and add some code to make it work [16:49:11 CEST] <nikio_> ive read that using named pipes is a solution somehow [16:51:06 CEST] <DHE> mkfifo pipe1 pipe2 pipe3 [...] [16:51:21 CEST] <nikio_> these days people rely too much on external libraries to do everythng for them, when often the problem that needs to be fixed istn even that hard once you dig into it a bit [16:51:30 CEST] <nikio_> and then they end up working with 100 libraries, stuff breaking everywhere [16:51:33 CEST] <nikio_> nobody understands why [16:51:34 CEST] <DHE> one program writes to a file like "cmd ... > pipe1" and ffmpeg can read from it with "ffmpeg -i pipe1 ..." [16:51:56 CEST] <DHE> I did something similar with mencoder with discrete audio and video pipes WAY back in the day [16:53:03 CEST] <nikio_> yeah there is something with named pipes that supposably makes it work, i tried simply using nodejs [16:53:10 CEST] <nikio_> unpiping a stream and then piping a new one [16:53:17 CEST] <nikio_> but that just breaks it [16:54:55 CEST] <DHE> you basically have two options. output each of your things to a discrete file descriptor, or have your application output in a streaming-friendly container format that ffmpeg can read [16:56:25 CEST] <nikio_> thx I will keep that in mind. im gonna hang around here a bit while i work on my project. [20:51:38 CEST] <kepstin> well, there's the third option, write your application in a language that can use C abi libraries, then directly use libav* instead of going through the ffmpeg cli wrapper :/ [20:55:02 CEST] <nikio_> hmm [20:58:01 CEST] <nikio_> bbl, ill keep that in mind too, thx [00:00:00 CEST] --- Mon Jul 29 2019

1 0

ffmpeg-devel.log.20190816
by burek 17 Aug '19

17 Aug '19

[00:04:03 CEST] <Compnn> Lynne, i ask because its strange to just have some new dev show up and insult other developers within the same project [00:17:26 CEST] <tmm1> is there any place ffmpeg does runtime checking to see if NEON is available [00:21:40 CEST] <mkver> tmm1: ff_h264dsp_init_arm [00:22:11 CEST] <mkver> git grep have_neon turns up lots of matches. [00:23:36 CEST] <tmm1> ah cool, i missed the lowercase define [00:23:36 CEST] <tmm1> thank you [00:33:52 CEST] <nevcairiel> jamrial: the special thing about truehd is that frames are sub 1ms in length, which can screw players up if they are careless with their timings [00:34:03 CEST] <nevcairiel> (they typically are 0.8333ms) [00:45:39 CEST] <Lynne> Compnn: what, you too want to question my amount of git commits? [00:47:48 CEST] <durandal_1707> ah, i always forget that mplayer dev... [00:49:44 CEST] <durandal_1707> bunch of mplayer devs, doing nothing but complains [01:38:22 CEST] <jamrial> nevcairiel: could be that. i'm getting a lot of stutter with ffplay, but not with other players like mpc-hc [06:34:09 CEST] <Kevin43> Hi...is this just like a chatroom? I'm interested in potentially contributing. Got here from the trac page for GSoC [06:39:57 CEST] <BradleyS> Kevin43: yes, but most people are probably sleeping now [06:40:27 CEST] <BradleyS> europe will be awake in 5-6 hours roughly [06:41:14 CEST] <philipl> BtbN: I don't think h.263 works with nvdec, even now. [06:41:25 CEST] <BradleyS> if you're interested in submitting patches, see the mailing list http://ffmpeg.org/mailman/listinfo/ffmpeg-devel [06:41:36 CEST] <philipl> I get some weird errors with the samples I have (generated by an old nokia 10+ years ago) [06:47:12 CEST] <Kevin43> thanks BradleyS, I will check that out. You got any other tips for getting familiar with ffmpeg besides working through the code? [06:47:54 CEST] <BradleyS> i'm probably the wrong person to ask but when people are around, i'm sure someone can help you [06:50:14 CEST] <Kevin43> (y) thanks! [07:06:14 CEST] <Compnn> Lynne, just curious who you are. dont care about your commits [07:07:05 CEST] <Compnn> Kevin43, what part of ffmpeg are you interested in? [07:07:53 CEST] <Compnn> Kevin43, most people just work on what interests them the most. some people like audio tech so they work on codecs. other people like to optimize code. others enjoy reverse engineering [07:09:07 CEST] <Compnn> Kevin43, what are you doing awake at 4am heh [07:11:02 CEST] <Kevin43> Compnn I've pretty much only used ffmpeg for cutting video. I was having issues cutting video at timestamps between keyframs without reencoding, and I saw a comment somewhere saying that it wasn't possible? got me thinking about working on it if it's really not implemented [07:11:24 CEST] <Kevin43> Compnn also, it's only 1 am where I am :P [07:11:47 CEST] <Compnn> oh :D [07:13:37 CEST] <TheAMM> Cuts have to be at keyframes though [07:15:14 CEST] <TheAMM> Unless you're thinking of implementing something that reencodes only the part before the next keyframe and copies the rest [07:15:35 CEST] <TheAMM> (which sounds a bit messy and afaik you could do without adjustments, although that's a bit messy too) [07:15:44 CEST] <Kevin43> TheAMM ^This is what I was thinking. Not sure if that's possible / would even save much time... [07:17:59 CEST] <TheAMM> It won't be impossible but imo it's fairly niche and it's going to touch the... user experience code in many places for a small thing [07:18:23 CEST] <TheAMM> Though I'm no expert ffmpeg dev, I only have some local changes [07:18:56 CEST] <TheAMM> It's worth to take a look at ffmpeg, then you'll know what it looks like and might get other ideas to implement or fix [07:19:52 CEST] <TheAMM> but if I had to make a partical copy like that, I'd rather experiment with running two commands and then concatenating the encoded new and old data together [07:21:08 CEST] <TheAMM> AFAIK there's no ffmpeg development primer, if you're working on some new thing look at existing things on how they do the thing (codecs, formats, devices) and read the ffmpeg headers for structs and stuff [07:22:07 CEST] <Compnn> Kevin43, what format codec are you cutting [07:23:47 CEST] <Kevin43> Ah gotcha, TheAMM. Yeah I suppose playing with two commands would probably save a lot of development time...I have been sorta looking for an excuse to get into some sort of open source project so thought I'd at least poke my head around haha [07:23:52 CEST] <Kevin43> Compnn mp4 [07:24:30 CEST] <Compnn> h264 then ? [07:25:42 CEST] <Compnn> probably not possible unless its all iframes [07:26:07 CEST] <Kevin43> sorry, yes h264 [07:32:50 CEST] <Kevin43> hmm...given my limited understanding (wiki just now...), wouldn't iframes = inter frames meaning they're "diff" frames from a recent keyframe? Wouldn't that mean having all iframes (and 1 keyframe at the beginning?) be the worst case scenario? [07:33:30 CEST] <JEEB> i is intra usually [07:36:05 CEST] <JEEB> p/b are inter in classic mpegnspeak [07:40:45 CEST] <Kevin43> ah, gotcha. Thanks JEEB. so doing 'diff' frames for intra uses spatial redundancy instead of temportal. Hmm. with spatial redundancy, I'm starting to think this isn't possible without reencoding [07:58:00 CEST] <Compnn> Kevin43, so a h264 frame usually goes ibpbpbpbpppibpbppbpbpi etc [07:58:03 CEST] <Compnn> the iframe is the keyframe [08:03:08 CEST] <Compnn> the pbpbpb are all fragments of the keyframe [08:03:12 CEST] <Compnn> only the motion changes [08:19:05 CEST] <Compnn> not a problem when keyframes are per 30 frames but maybe 250 between frames, almost 10 secs [08:28:16 CEST] <Compnn> so you want to cut between keyframes, you have to rebuild the frame from the keyframe ... without reencoding [08:28:19 CEST] <Compnn> very complicated [08:28:33 CEST] <Compnn> some would say ... impossible [08:28:51 CEST] <Compnn> and mp4 is a terrible format :D [09:39:25 CEST] <kurosu> Kevin43: in >=h264, the coding order of the pictures may differ a lot from the display order. It's doable to cut on a particular frame, but the trailing pictures can cause quite a mess (concatenation of bitstreams/VBV stuff/display) [09:41:47 CEST] <kurosu> tl;dr: hardly doable, often breaks [11:34:00 CEST] <_bluez> Hello! I am facing an issue where I have allocated a new frame, called av_image_alloc() on it passing width, height, format and such, and then called av_image_fill_black() on it... but I keep getting a seg fault inside av_image_fill_black :( [11:34:35 CEST] <_bluez> Does anyone have any idea what the problem could be? [11:35:22 CEST] <_bluez> I've been at it for 2 days now... [11:35:37 CEST] <durandal_1707> Lynne: michaelni put transcript on ml, he is good citizen [11:38:24 CEST] <_bluez> It faults somewhere while accessing the frame->data... [11:48:59 CEST] <durandal_1707> which pixel format you use? [11:49:22 CEST] <_bluez> Its YUVJ420p I guess [11:51:10 CEST] <durandal_1707> whatever you do, you allocate stuff wrong way iirc [11:51:34 CEST] <_bluez> How to allocate correctly then? [11:52:06 CEST] <J_Darnley> What does your debugger say? Is it a null dereference? Or just out of bounds? Have you tried asan? [11:52:25 CEST] <J_Darnley> or valgrind [11:52:28 CEST] <_bluez> It says cannot access memory at that location [11:53:54 CEST] <_bluez> i guess its out of bounds [11:55:03 CEST] <_bluez> no i have never used asan or valgrind [11:56:30 CEST] <_bluez> my debugger says the fault happens inside __memset_avx2_erms, inside memset_bytes, inside av_image_fill_black [11:58:33 CEST] <durandal_1707> it should work if you use normal pointers to image alloc [11:58:54 CEST] <durandal_1707> not sure if you pass avframe ones [11:59:19 CEST] <durandal_1707> avframe data is reference counted [11:59:39 CEST] <durandal_1707> at least inside libs [12:01:08 CEST] <_bluez> how do i allocate avframe data then? I tried also using av_frame_get_buffer, same issue [12:03:00 CEST] <durandal_1707> can you show actual codr? [12:08:02 CEST] <durandal_1707> make sure Avframe is pointer on stack and not full on stack [12:08:28 CEST] <_bluez> https://pastebin.com/NLdvRKaU [12:08:39 CEST] <_bluez> Here is a portion of the code [12:09:15 CEST] <_bluez> durandal_1707: what do you mean by not "full on stack".. can u explain a bit? [12:13:41 CEST] <durandal_1707> you can not cast int to ptrdiff [12:14:26 CEST] <_bluez> but it says incompatible pointers [12:14:30 CEST] <durandal_1707> at least not on 64bit [12:14:42 CEST] <_bluez> should i cast it to const long * ?? [12:14:58 CEST] <durandal_1707> cadt can not fix incompatible pointers [12:15:10 CEST] <durandal_1707> *cast [12:15:28 CEST] <durandal_1707> this is basic programming [12:16:49 CEST] <_bluez> oh sry.. but the avframe linesize is int* i guess...how do i get around it then? [12:16:52 CEST] <nevcairiel> you cant really cast arrays like that, you'll need to make a new one in the proper type and copy the values over [12:17:10 CEST] <durandal_1707> yes what nev said [12:18:37 CEST] <_bluez> isn't there a function I can use to simply allocate frame buffers and then fill black?? [12:19:15 CEST] <durandal_1707> yes what you already use [12:19:45 CEST] <durandal_1707> and i think stuff is initialized already upon allocation [12:20:20 CEST] <durandal_1707> so no need for fill black, assuming you will copy other data over [12:20:49 CEST] <_bluez> oh.. okay then [12:21:03 CEST] <J_Darnley> Isn't filling with black usually an extra step the user must take?waste [12:22:31 CEST] <durandal_1707> not any more, its zeroed iirc, but that may not be same as black [12:22:33 CEST] <_bluez> also if you can look at the code... even removing the fill black... i get segfault while copying data from tiles into the frame after certain amount of iterations... it led me to think if there was some fault in frame buffer allocation [12:23:21 CEST] <_bluez> Cuz the code seems to be correct... [12:25:56 CEST] <_bluez> Can you please look into it once? [12:28:07 CEST] <durandal_1707> make sure you use same format for copy [12:28:36 CEST] <_bluez> yes i am using the same format [12:29:17 CEST] <_bluez> I did, frame->format = tile->format [12:33:05 CEST] <durandal_1707> same issue for linesize like for copy? [12:33:21 CEST] <durandal_1707> you used cast again? [12:33:45 CEST] <_bluez> no no I did not have to cast it [12:34:11 CEST] <_bluez> You can look at the loop where I copy data from tiles into the frame [12:34:55 CEST] <J_Darnley> If it's 420, you need different widths and heights per plane [12:36:21 CEST] <J_Darnley> currently it looks like one fixed value for all planes [12:36:31 CEST] <_bluez> oh i see... will the widht be same as tile->linesize[plane]? [12:37:43 CEST] <durandal_1707> no [12:37:55 CEST] <J_Darnley> That is the stride, not the width. It will include padding. [12:38:34 CEST] <_bluez> ohh.. how do i get the variable width and height for each plane then? [12:39:05 CEST] <_bluez> I saw similar code in the vf_tile, and it wasn't done there i guess... [12:39:15 CEST] <J_Darnley> The pixel format will give you all the details of the planes. [12:41:27 CEST] <J_Darnley> Sorry that should be "AVPixFmtDescriptor":vs [12:41:59 CEST] <J_Darnley> Use av_pix_fmt_desc_get() to get it from the enum AVPixelFormat [12:43:15 CEST] <_bluez> sorry i am a bit confused... what info do i get from the fmtdescriptor? And where in the loop should i make the changes? [12:43:49 CEST] <J_Darnley> In the paste, line 81, the height needs to be half for 420 [12:44:12 CEST] <J_Darnley> line 82, the width [12:44:25 CEST] <J_Darnley> The chroma planes are half height and half width [12:44:32 CEST] <_bluez> oh i get it [12:45:01 CEST] <_bluez> so those values only for plane 1 and 2 right? Plane 0 has full height and width? [12:45:33 CEST] <J_Darnley> The format descriptor has log2_chroma_w and log2_chroma_h fields [12:45:49 CEST] <J_Darnley> yes [12:46:19 CEST] <_bluez> okay... thanx a lot!!! [12:46:41 CEST] <_bluez> how do i use log2_chroma_w/h ? [12:48:27 CEST] <J_Darnley> Read libavutil/pixdesc.h "Amount to shift the luma width right to find the chroma width." [12:52:29 CEST] <_bluez> okay i will.. thanks again! [14:30:47 CEST] <durandal_1707> michaelni: subscribe me to security ml [14:36:07 CEST] <thardin> is porting the mxf imx patch to 4.1 and 4.2 ok? [14:36:15 CEST] <thardin> ping michaelni I guess [14:36:17 CEST] <durandal_1707> no [14:36:42 CEST] <durandal_1707> features are never backported [14:37:06 CEST] <durandal_1707> only fixes for already available features [14:40:29 CEST] <durandal_1707> write this somewhere so you do not ask again same questions, this is also developer policy [14:53:34 CEST] <thardin> I know, but this is more a case of a feature having stopped working. or wait.. it's only broken i master [14:53:41 CEST] <thardin> in [14:54:07 CEST] <BtbN> regression fixes are backported, but yeah, only if the regression is actually in the release [15:00:01 CEST] <thardin> after that I'm going to investigate if the legacy cinepak decoders are fine with some simple but effective optimizations I have in mind [15:00:15 CEST] <thardin> for that ~cutting edge~ performance [15:24:36 CEST] <Lynne> michaelni: you're an asshole [15:25:41 CEST] <Lynne> I did provide criticism, and yet your default response is that I am aggressive and doing an ad hominem attack [15:27:21 CEST] <Lynne> you haven't paid a single piece of attention to anything I've ever said and are always looking around to dodge facts and twist the discussion [15:27:34 CEST] <Lynne> like bringing up undefined behaviour in timeouts [15:27:59 CEST] <Lynne> and you always go with your bullshit rhetoric of "why can't all of us cooperate" [15:28:25 CEST] <Lynne> how do you expect anyone to cooperate when you and nicolas refuse to listen? [15:29:17 CEST] <Lynne> your behaviour in the first place is what I find most insulting and aggressive [15:34:01 CEST] <jamrial> Lynne: your first line invalidates anything after it [15:38:09 CEST] <kierank> something something code of conduct [15:39:19 CEST] <jamrial> yes, just with the stuff michaelni quoted in this morning's email plus the above he should be kicked [15:42:49 CEST] <jamrial> none of this stuff is justified. a bunch of one-line patches shouldn't devolve into this [15:43:41 CEST] <thardin> durandal_1707: just confirmed that they are regressions in 4.1 and 4.2 compared to 4.0.1 [15:46:05 CEST] <durandal_1707> no, if you kick him i go to [15:46:49 CEST] <durandal_1707> than you must comply with agpl code of all around codebase [15:48:01 CEST] <jamrial> durandal_1707: you're ok with people throwing insults around? [15:48:32 CEST] <kierank> like carl does? [15:48:37 CEST] <kierank> is it ok when carl does it? [15:49:03 CEST] <jamrial> kierank: i'm sure you remember i was one of the people that supported the vote to ban carl [15:49:30 CEST] <kierank> I don't blame durandal_1707 and Lynne, they are clearly at the end of their tether [15:50:16 CEST] <jamrial> and as i sadi, there's zero justification for it [15:50:33 CEST] <kierank> not true, their objections to patches are just ignored [15:50:37 CEST] <kierank> and changes are bullied through [15:51:24 CEST] <jamrial> you don't insult people when you think they are not looking because you think you're being ignored [15:51:46 CEST] <kierank> I don't agree with it but it has been months and months of this [15:51:49 CEST] <kierank> they are human [15:52:35 CEST] <durandal_1707> they have obviously unlimited time and resources to do their work and ignore others [15:54:56 CEST] <jamrial> i am also human, this project has given me a fair share of stress and still does, yet you don't see me throw spite and insults around [15:55:03 CEST] <durandal_1707> there is no need for insults at all, they appears to just easy up their situation [15:55:24 CEST] <jamrial> and for someone that joined a few months ago, he sure ran out of patience fast [15:55:41 CEST] <jamrial> durandal_1707: at least we agree that insults only make things worse [15:56:40 CEST] <kierank> durandal_1707: I think we should ban fuzz patches unless agreed via technical committe [15:56:48 CEST] <BradleyS> is there any current effort toward a formal governance model with a voting council [15:57:08 CEST] <durandal_1707> it is one big chaos [15:58:24 CEST] <durandal_1707> some obvious patches are ok [15:59:01 CEST] <durandal_1707> but you need to pick them between less useful ones [15:59:19 CEST] <BtbN> fuzzing simply should not go to the security ML [15:59:25 CEST] <BtbN> most of that stuff is not critical [15:59:34 CEST] <BtbN> Make a dedicated, public ML for it [16:01:00 CEST] <durandal_1707> and what about decoders and skip frames- making something vfr when it is cfr [16:01:39 CEST] <BradleyS> thefuzz(a)ffmpeg.org [16:01:53 CEST] <durandal_1707> justification that it allows easy DOS is unrealistic to me [16:02:21 CEST] <jamrial> durandal_1707: suggest an user settable option/flag to choose between the two [16:02:43 CEST] <jamrial> default to vfr since it's faster and the output will be the same, but can be changed to cfr behavior [16:03:38 CEST] <durandal_1707> no, that contradicts with decoder specification and compatibility [16:03:43 CEST] <BtbN> I'm also still confused about the -safe 0 of the concat demuxer [16:03:46 CEST] <thardin> DOS is already an issue with any kind of thing that takes random files from the internets [16:03:53 CEST] <kierank> 15:03:47 <thardin> DOS is already an issue with any kind of thing that takes random files from the internets [16:03:54 CEST] <kierank> this [16:04:01 CEST] <durandal_1707> i think this was discussed on one of meetings [16:04:08 CEST] <thardin> just set the resolution really high, possibly using resolution updates so ffprobe doesn't catch it [16:04:34 CEST] <thardin> nothing stops me from uploading a 100x100 .ts file to youtube that has a change to 30000x30000 in it (I think) [16:04:54 CEST] <thardin> and if it's all black it compresses really well too [16:05:04 CEST] <BtbN> You can probably also build compression-bombs with hand-crafted h264. Tiny bitstream, expands to enormous images [16:05:32 CEST] <thardin> it's possible with a lot of game codecs without even trying [16:05:58 CEST] <thardin> you can also do a lot of nasty crap with all formats that (a) use length fields or (b) use pointers [16:06:25 CEST] <thardin> that and other things make me very nervous about basically anything having to do with parsing [16:08:05 CEST] <thardin> RE: fuzz stuff, separating crashes from slowdowns makes sense. I don't think we should rely on just michael to fix every performance issue, that's terrible bus factor [16:09:04 CEST] <thardin> sweet, I got win3.1 rolling [16:09:25 CEST] <BtbN> I'm also confused why a 2s hang is a major issue now [16:11:16 CEST] <thardin> I have a few fixes I intend to submit that I've been pondering this exact thing about [16:11:17 CEST] <jamrial> not major, but why a fuzzer cares to report something taking 2 seconds to finish is a good question [16:11:31 CEST] <kierank> i think the fuzzers have contention as well [16:11:47 CEST] <thardin> jamrial: something taking unexpectedly long time to decode is possibly an issue, depending [16:12:12 CEST] <thardin> I fuzzed gzip a while back with afl-fuzz and thought I'd found a bug [16:12:31 CEST] <thardin> turns out afl just figured out it can generate output that expands a lot, meaning suspiciously longer runtime [16:13:47 CEST] <thardin> since no one uses any tooling for complexity analysis, this tends always to be an issue [16:35:56 CEST] <thardin> heck yeah, got vfw, audio and video set up [16:40:23 CEST] <J_Darnley> Any particular reason you're doingthat? Or just for shits and giggles? [16:46:33 CEST] <thardin> fun! [16:53:00 CEST] <thardin> gotta have something to take the edge off dealing with mxf [18:01:53 CEST] <BBB> kierank: so I've been watching BBC while in London, and boy, the banding on their heads is pretty awful, is that a known issue? [18:02:27 CEST] <kierank> BBB: hd? [18:02:30 CEST] <BBB> yes [18:02:36 CEST] <kierank> BBB: which channel [18:02:41 CEST] <BBB> bbc 1 one hd [18:02:53 CEST] <BBB> morning show (around 9am) [18:02:53 CEST] <kierank> do you know what it came from (OTA or satellite?) [18:02:57 CEST] <BBB> no idea [18:03:01 CEST] <kierank> because the OTA looks like complete crap they do some strong lowpass on it [18:03:03 CEST] <BBB> i turned on the tv [18:03:14 CEST] <kierank> satellite looks a bit better but it's h.264 at 4-5meg interlaced [18:04:00 CEST] <kierank> the biggest problem is the bbc are obsessed with psnr [18:04:08 CEST] <kierank> so probably psnr is bloody brilliant [18:04:12 CEST] <kierank> but it looks like crap [18:06:30 CEST] <kierank> BBB: I was going to say are you doing the "American tourist in London" thing but then I remembered you are not American :) [18:07:29 CEST] <BBB> ah that would explain the banding yes [18:09:00 CEST] <kierank> BBB: http://a.files.bbci.co.uk/media/live/manifesto/audio_video/webcast/dash/uk/… [18:09:04 CEST] <kierank> that should work if you are in the uk [18:09:07 CEST] <kierank> 35mbit 4k [18:09:34 CEST] <BBB> why would I want that? :-p [18:09:45 CEST] <kierank> to see how bad hevc encoders are [18:10:07 CEST] <BBB> someone told me to not badmouth competitors if they suck [18:10:08 CEST] <BBB> oops [18:10:52 CEST] <nevcairiel> is it badmouthing if its factual [18:11:29 CEST] <BBB> ikr [18:11:37 CEST] <BBB> anyway [18:11:41 CEST] <BBB> how do I play that file? [18:11:44 CEST] <BBB> ffplay doesn't like it [18:11:46 CEST] <BBB> vlc doesn't play it [18:11:49 CEST] <BBB> mplayer? [18:12:29 CEST] <nevcairiel> its dash, most browsers would probably be able to [18:12:45 CEST] <jamrial> BBB: you need libavformat compiled with libxml2 support for ffplay to read it [18:12:52 CEST] <nevcairiel> well, with a proper website around it [18:14:00 CEST] <kierank> BBB: vlc plays it here [18:14:42 CEST] <BBB> I suck :( [18:15:01 CEST] <BBB> let's try to update [18:15:09 CEST] <durandal_1707> compile with libxml2 [18:16:03 CEST] <durandal_1707> write native xml parser while you at it [18:17:17 CEST] <BBB> the grass is kind of blur [18:17:20 CEST] <BBB> otherwise it looks ok [18:17:28 CEST] <BBB> but I am on a laptop so no 4k scree [18:17:46 CEST] <BtbN> Tell that to Lenovo, who put 4K screen in 14" Laptops [18:17:52 CEST] <BtbN> With a GPU barely able to drive it [18:17:54 CEST] <kierank> it uses very few hevc features [18:18:01 CEST] <durandal_1707> grass is always blur in RL [18:18:54 CEST] <BBB> i would've looked at feature usage if my analyzer opened it [18:18:59 CEST] <BBB> but it doesn't support xml as a container [18:19:00 CEST] <BBB> :-p [18:19:38 CEST] <durandal_1707> isnt it xml with links to partial files? [18:20:26 CEST] <BBB> dunno [18:22:07 CEST] <jamrial> sure is. just download the init.mp4 file (moov atom) plus any segment, concatenate them, and you have a readable mp4 file [18:23:06 CEST] <durandal_1707> can youtubedl do that? [18:26:48 CEST] <jamrial> for supported websites at least. not sure if it will if you just tell it to get a random mpd [18:26:50 CEST] <jamrial> never tried [18:29:12 CEST] <BtbN> It might invoke ffmpeg to do that [18:54:12 CEST] <philipl> BtbN: So I don't see how h.263 hwaccel works. The initialisation for h.263 follows a path where it ends up doing hwaccel in (from ff_get_format) before avctx->codec_[width|height] is set, so this always fails with 0x0 dimensions. [18:54:38 CEST] <BtbN> So it's just broken in itself, entirely, no matter what hardware? [18:54:55 CEST] <philipl> That's my impression - I don't understand how the nvidia guy tested his change. [18:55:27 CEST] <BtbN> neat [18:55:37 CEST] <philipl> It looks fixable, and I will try. [18:59:40 CEST] <durandal_1707> should we do new ff meeting? [19:11:49 CEST] <jamrial> durandal_1707: imo, not with the current animosity [19:12:49 CEST] <durandal_1707> it will be even worse if nothing changes [19:28:19 CEST] <durandal_1707> current state is unbearable [19:47:22 CEST] <lotharkript> durandal_1707: btw, regarding oss-fuzz, it seems we can set the view_restrictions to "security" to have only the bugs mark as security closed, the rest will be open, which should include timeout bugs. [20:04:36 CEST] <durandal_1707> that still doest explain why some entities like carl and reimar have security access [20:06:42 CEST] <lotharkript> I was trying to find a solution where all the timeout bug will be open, so any one could look at them and maybe reproduce them, but keep the security to a strict amount of people. Now, the people who will have access to it, i will let the community to decide about it. [21:13:15 CEST] <lotharkript> I confirm with the oss fuzz that it should be possible to have all the NON security bugs to be open to every one. Is this what the community wants? [21:17:09 CEST] <lotharkript> in this case, i twill think that any one will be able to reproduce the problem.. [23:18:19 CEST] <mkver> nevcairiel: Is there really consensus that sizeof(AVPacket) should eventually be removed from the public API? [23:19:12 CEST] <nevcairiel> any such struct with its size as part of the ABI is very limiting [23:19:32 CEST] <nevcairiel> we cant even add a new field [23:21:05 CEST] <nevcairiel> and yeah this was definitely discussed before [23:25:03 CEST] <jamrial> we should start doing some cleaning in that regard, then. i see AVPacket used in stack and heap in a lot of lavf files [23:29:16 CEST] <nevcairiel> well in lavf its "fine" [23:29:21 CEST] <nevcairiel> oh wait [23:29:27 CEST] <nevcairiel> no it would be fine in lavc :D [23:29:35 CEST] <nevcairiel> i still think its weird its defined in lavc [23:29:39 CEST] <nevcairiel> but anyway, yeah. [23:41:35 CEST] <mkver> Ok, and how would the need for temporary packets outside of lavc be fulfilled in this scenario? [23:41:59 CEST] <nevcairiel> how do temporary frames work? [23:42:02 CEST] <nevcairiel> you allocate them. [23:42:30 CEST] <mkver> But that's way more expensive than what is currently done. [23:42:37 CEST] <nevcairiel> irrelevant cost, though [23:43:34 CEST] <nevcairiel> if its a commonly called function it could also keep one temp packet that it reuses [23:45:05 CEST] <mkver> And these temp packets would be stored in AVFormatInternal/AVStreamInternal? [23:46:43 CEST] <nevcairiel> thats an impossible to answer question since there could be thousand different uses [23:46:53 CEST] <nevcairiel> with different applicable contexts [23:48:02 CEST] <nevcairiel> but its really quite an obvious solution, if its a hot code path and it has a context it has control over, then thats where it can store temporary resources, not sure whats unclear about that [23:49:30 CEST] <mkver> Your "it [a commonly called function] could also keep one temp packet that it reuses" sounded like you wanted to have a static pointer in a hot function to a packet that is allocated once. [23:50:38 CEST] <nevcairiel> in what world does static anything make sense? please. [23:52:04 CEST] <mkver> That's why I was confused. It didn't really made sense. [00:00:00 CEST] --- Sat Aug 17 2019

1 0

ffmpeg.log.20190816
by burek 17 Aug '19

17 Aug '19

[00:11:29 CEST] <LigH> Hi [00:12:06 CEST] <LigH> Are there any guidelines regarding including the ffmpeg logo into the logo of a project that is tightly related to ffmpeg? [00:22:54 CEST] <pink_mist> LigH: https://en.wikipedia.org/wiki/File:FFmpeg_Logo_new.svg claims that the logo is too simple to meet requirements for copyright, and thus it is in the public domain ... whether or not wikipedia is correct in this assessment is something I'd leave to a lawyer though [00:24:31 CEST] <LigH> Wouldn't mind asking an official for agreement if I knew who. [00:24:56 CEST] <LigH> I hope there are some official lines about it. [00:24:56 CEST] <pink_mist> I believe the original creator of the logo is Hervé Flores [00:25:28 CEST] <pink_mist> ah yes, wikipedia agrees [00:26:19 CEST] <LigH> Not today anymore ... GN8 [00:26:26 CEST] <LigH> Thanks so far [02:13:12 CEST] <nicolas17> how far back can a frame reference another in h264 and h265? [02:20:41 CEST] <furq> nicolas17: 16 frames (or 32 fields) [02:20:53 CEST] <furq> that's the maximum, it's normally less [02:22:18 CEST] <furq> i think hevc maxes out at 8 [02:22:29 CEST] <nicolas17> they reduced it? :/ [02:23:43 CEST] <furq> looks like it [02:24:42 CEST] <furq> you can set >8 in x265 but it'll produce a noncompliant bitstream with no profile/level set [02:25:20 CEST] <MoziM> i edited a video, exported it, then ran a hash check on it, i made another video the same exact way and it fails the hash check... any idea why they differ? [02:25:46 CEST] <furq> are you checking the hashes of the files or the streams [02:26:22 CEST] <MoziM> files [02:26:26 CEST] <MoziM> the output videos [02:26:30 CEST] <furq> there's probably some creation date metadata then [02:26:38 CEST] <nicolas17> I wonder if there's any codec with even farther reference frames [02:26:50 CEST] <MoziM> furq: but the video itself should be the same right? [02:26:57 CEST] <furq> you'd hope [02:27:00 CEST] <furq> you can check with the hash muxer [02:27:19 CEST] <MoziM> hash muxer? [02:27:32 CEST] <furq> -i video1 -i video2 -map 0:v -f hash - -map 1:v -f hash - [02:27:52 CEST] <nicolas17> I was thinking, if the encoder could selectively mark frames as "this will be a reference frame", the decoder could keep the last 8 *marked* frames in memory (instead of last 8), and that would allow them to be arbitrarily far back (well, limited to *encoder* memory) [02:28:15 CEST] <nicolas17> but I'm probably missing some obvious reason why all this wouldn't work ^^ [02:28:20 CEST] <furq> vp9 is capped at 8 as well [02:28:34 CEST] <furq> it's rare to actually be able to use >8 in h264 anyway [02:28:47 CEST] <furq> since you're normally stuck at level 4.1 for hwdec compliance [02:28:53 CEST] <furq> which is like 3 refs at 1080p [02:29:26 CEST] <furq> i assume hevc/vp9 capping at 8 is just to avoid the whole mess of hardware decoders that plagued h264 for a long time [02:29:31 CEST] <furq> or help avoid [02:31:10 CEST] <nicolas17> I think av1 limits at 8 too [02:31:27 CEST] <furq> seems likely considering how much lineage it takes from vp9 [02:31:59 CEST] <furq> you normally hit diminishing returns pretty hard at that point anyway [02:33:42 CEST] <nicolas17> I was thinking of cases where scene 1 and scene 3 are very similar, such that the first frame of scene 3 could predict from the last frame of scene 1 instead of being an I frame, but currently that only works if scene 2 is shorter tha 8 frames [03:48:04 CEST] <nicolas17> x265 docs say a keyframe interval of -1 will trigger "A special case of infinite-gop (single keyframe at the beginning of the stream)" [03:48:12 CEST] <nicolas17> but passing -g -1 to ffmpeg didn't seem to do that [03:49:08 CEST] <kepstin> MoziM: note that x264 isn't deterministic when using frame threading. so you can get different video streams from that. [03:57:30 CEST] <nicolas17> -x265-params "keyint=-1" this works [04:45:23 CEST] <AiNA_TE> is there someway to specify input resolution with a direct show device? [06:11:07 CEST] <bobbertson> Hello, how is it going. [06:11:25 CEST] <bobbertson> getting an error when trying to stream to vaugnlive with ffmpeg "av_interleaved_write_frame(): Operation not permitted" [06:12:20 CEST] <bobbertson> the command I am running "/usr/bin/ffmpeg -f x11grab -s 1280x768 -framerate 30 -i :0.0 -c:v libx264 -preset fast -s 1280x768 -b:v 1000k -f flv "rtmp://live.vaughnsoft.net/live/live_***" [06:47:39 CEST] <logicWEB> hello :-) I have been asked to make a video compilation of about 40 short videos. I just took a poke at them, and they're a random distribution of 23.976, 25 and 29.97 FPS. what is my best option for combining these into one consecutive video? [07:48:00 CEST] <pk08> hi [07:48:21 CEST] <pk08> sorry for the off topic question [07:48:53 CEST] <pk08> I am trying to compile and install mpv in headless Centos 7, but i am getting error in opengl code: /video/out/opengl/common.h error: unknown type name GLchar in "GLsizei, const GLchar *,const void *);" [07:48:53 CEST] <pk08> configure command and output: https://pastebin.com/dEM2S599 [07:48:53 CEST] <pk08> compilation output: https://pastebin.com/7wBz3Uvi [07:48:53 CEST] <pk08> i can compile of i disable gl but i want to use nvidias hardware acceleration (CUDA hwaccel) so i need to have opengl enabled. [07:48:55 CEST] <pk08> can anyone help me to solve this? [07:49:23 CEST] <pk08> and i have installed ffmpeg with hwaccell using this https://trac.ffmpeg.org/wiki/CompilationGuide/Centos [16:01:16 CEST] <logicWEB> so I asked my question last night and then I went to bed and found my computer had restarted automatically for updates overnight. hooray! the public logs don't seem to be entirely complete/up-to-date, so I'm not aware of any way to see if I got replies overnight :-( [16:05:01 CEST] <pink_mist> you didn't get any replies overnight [16:06:29 CEST] <logicWEB> thanks :-) [16:07:28 CEST] <danilo82> hello, i want to record the screen with h264_vaapi, but how can i record the audio too? [16:16:26 CEST] <durandal_1707> os? [16:26:16 CEST] <danilo82> shoot [16:26:26 CEST] <danilo82> i lost the chat [16:27:21 CEST] <danilo82> so i asked for help to record the screen with vaapi and record the internal audio [16:30:16 CEST] <durandal_1707> os? [16:34:51 CEST] <danilo82> linux [16:48:46 CEST] <c_14> danilo82: alsa or pulse? [16:51:35 CEST] <danilo82> i want to record pulse audio [16:51:48 CEST] <danilo82> to not mess with alsa [16:52:12 CEST] <c_14> create a monitor device on the pulse output using pavucontrol or something and then use -f pulse -i <device_name.monitor> [16:55:02 CEST] <danilo82> :T wot [16:56:14 CEST] <kepstin> in most cases the monitor should already exist, just look for the monitor device for your main audio device in `pacmd list-sources` and grab the stuff out of the <> in the name field, use that as the input device to ffmpeg. [17:16:09 CEST] <danilo82> ok got the idea [17:16:09 CEST] <danilo82> i'm using this command [17:16:09 CEST] <danilo82> ffmpeg -vaapi_device /dev/dri/renderD128 -f x11grab -video_size 1366x768 -i :0 -vf 'hwupload,scale_vaapi=format=nv12' -c:v h264_vaapi -qp 24 output.mp4 [17:16:09 CEST] <danilo82> where should i put the pulse thing? [17:24:05 CEST] <durandal_1707> after ffmpeg [17:25:06 CEST] <danilo82> :P this way will have two " -f " [17:25:09 CEST] <danilo82> is this right? [17:25:14 CEST] <furq> yes [17:25:17 CEST] <furq> one for each input [17:26:12 CEST] <danilo82> ok [17:28:36 CEST] <danilo82> good good [17:28:40 CEST] <danilo82> now it works [17:28:41 CEST] <danilo82> :) [17:29:12 CEST] <danilo82> aye [17:29:13 CEST] <danilo82> should i use x11grab or KMS? [17:29:13 CEST] <danilo82> what is better? [17:54:23 CEST] <danilo82> [kmsgrab @ 0x55a0b7bdcf80] No handle set on framebuffer: maybe you need some additional capabilities? [17:54:24 CEST] <danilo82> pipe:: Invalid argument [17:54:24 CEST] <danilo82> i get this every time [18:37:51 CEST] <durandal_1707> i do not think you can use vaapi with kmsgrab [18:39:41 CEST] <BtbN> vaapi should be able to take drm handles as input somehow [18:39:50 CEST] <BtbN> There might be a mapper-filter for that [18:40:41 CEST] <danilo82> i think so [18:40:41 CEST] <danilo82> but in the site sais that kmsgrab uses less CPU so i want to use [18:40:41 CEST] <danilo82> but i sdon't know how [18:49:00 CEST] <DHE> ksmgrab requires DRM, so I suspect it wouldn't work if you were using an nVidia card with the binary nvidia driver for example. [18:52:12 CEST] <BtbN> nvidia also uses drm though [18:52:19 CEST] <BtbN> It literally comes with a Kernel module called nvidia-drm [19:28:39 CEST] <kepstin> someone tried this a couple weeks ago, the nvidia drm module doesn't implement the required kms interfaces for the kmsgrab stuff to work [19:28:50 CEST] <kepstin> couple of weeks? maybe a month or so, anyways. [21:51:16 CEST] <degenerate> hey guys [21:52:19 CEST] <degenerate> lets say i've got a junky video with a lot of noise in it (from a small CCD and low light conditions) sample here: https://uccdn.bid13.com/original_videos/i/L/iLW4ic6u7jhcDpf8ZjTrrP8CRvpBO13… [21:52:35 CEST] <degenerate> is there any sort of filters i could run with FFMPEG to make it less noisy? [21:53:13 CEST] <nicolas17> there's several denoising filters, I guess you'll have to experiment [21:53:46 CEST] <degenerate> secondary question is there a way to detect "noise"... so i don't denoise every video, just the ones that "need" it [22:37:17 CEST] <Oldest> why does ffmpeg support wmv to mkv but mkvtool does not [22:37:47 CEST] <c_14> because mkvtool isn't meant to support wmv [22:38:15 CEST] <Oldest> c14 it supports mp4 to mkv [22:38:31 CEST] <degenerate> harumph: [AVFilterGraph @ 0x7f97225013e0] No such filter: 'fftdnoiz' [22:38:39 CEST] <degenerate> do i need to compile it for support of this filter? [22:39:13 CEST] <danilo82> me back here [22:39:14 CEST] <danilo82> anyone know how to use kmsgrab? [22:39:21 CEST] <furq> degenerate: it's new in 4.1 [22:39:24 CEST] <degenerate> ah [22:39:37 CEST] <degenerate> i see: ffmpeg version 3.4.2 [22:39:46 CEST] <degenerate> time to upgrade :P [22:39:46 CEST] <furq> try hqdn3d or nlmeans [22:39:49 CEST] <furq> those are both popular [22:44:22 CEST] <degenerate> Thanks furq, btw, do you know the answer to my question above as well, what would be the best way to denoise a video, and is there a way to detect noise before applying the filter? [22:46:07 CEST] <kepstin> i can't really think of anything other than running the denoise filter and checking the psnr between the funfiltered and filtered video. bigger numbers means the filter did more work [22:46:34 CEST] <kepstin> you could do that on sample segments of the video, and throw out the decoded video rather than re-encode it. [22:48:25 CEST] <degenerate> thanks [22:49:54 CEST] <Oldest> why does ffmpeg support wmv to mkv but mkvtool does not? [22:50:45 CEST] <danilo82> what is mkvtool? [22:51:37 CEST] <Oldest> https://en.wikipedia.org/wiki/MKVToolNix [22:52:03 CEST] <pink_mist> Oldest: because mkvtool is for mkvs [22:52:07 CEST] <pink_mist> hint: it's in the name [22:52:27 CEST] <kepstin> Oldest: ffmpeg is a general multimedia tool with a goal of being able to read and decode pretty much every video format ever. mkvtoolnix is a tool for making mkvs, which can read some popular other formats as a convenience. [22:52:37 CEST] <Oldest> pink_mist that's silly, you are implying that mkvtool should only able to do mkv to mkv then [22:52:52 CEST] <Oldest> mkvtool can do mp4 to mkv [22:53:13 CEST] <durandal_1707> ask mkvtool devs [22:53:38 CEST] <Oldest> durandal_1707 you are right, i should [23:44:23 CEST] <danilo82> ok, i used "sudo ffmpeg -f kmsgrab -i - -vf 'hwdownload,format=bgr0' output.mp4" because it needed something like that, i can make a video but the image is completely scranbled :| i'm almost there [23:45:56 CEST] <danilo82> oh "Capture from the first active plane, download the result to normal frames and encode. This will only work if the framebuffer is both linear and mappable - if not, the result may be scrambled or fail to download." [23:46:38 CEST] <danilo82> so how can i make the framebuffer linear and mappable [00:00:00 CEST] --- Sat Aug 17 2019

1 0

ffmpeg-devel.log.20190724
by burek 16 Aug '19

16 Aug '19

[01:25:23 CEST] <BradleyS> is a tarball for 4.2 coming soon, i hope? [01:43:37 CEST] <BtbN> It'd have to be released first. [02:48:40 CEST] <Compnn> BradleyS, first is the announce, then the backports, then the tarball and testing, then the release. i think [04:12:23 CEST] <BradleyS> Compnn: ah, thanks [14:20:17 CEST] <durandal_1707> kierank: please apply my recent dynaudnorm patch on ML if you can [14:21:25 CEST] <kierank> Yes when in office [14:26:10 CEST] <daniel__> Hi everyone ! Does someone of you know how to move "data" from an AVPacket to the corresponding AVFrame(s) (H.264) ? I have a specific bitstream filter, which does some processing and outputs a "decision" and I would like to make use of this "decision" in a vf filter. It looks like the other way around is trivial since bsf filters are applied after vf filter but in this situation I am a bit lost. Thank you in advance :) [14:27:16 CEST] <J_Darnley> Isn't it a decoder that "moves" data from packets to frames? [14:28:10 CEST] <J_Darnley> do you mean metadata or side data or similar? [14:28:29 CEST] <nevcairiel> you probably want side data for that [14:30:54 CEST] <daniel__> Yes, I would av_packet_add_side_data(...) in my bsf filter but I am not able to make use of this side data in a vf filter. [14:34:18 CEST] <daniel__> Is there a way to make this side data persistent, so that I can apply the bsf filter to put the side data in the AVPacket and then modify the decoder code to extract the side data from AVPacket(s) and inject it in the AVFrames before filtering ? [15:34:32 CEST] <cone-099> ffmpeg 03Andreas Rheinhardt 07master:b56031b992ab: vp3data: Make some arrays unsigned to prevent overflow [15:34:33 CEST] <cone-099> ffmpeg 03Andreas Rheinhardt 07master:a081a6d20117: avcodec/mpc8huff: Make some arrays unsigned to prevent overflow [15:34:46 CEST] <durandal_1707> so is there way to do FFT of very long length (2646000) by doing FFT of smaller length? [15:44:51 CEST] <Lynne> durandal_1707: 2646000 = 16 x 27 x 125 x 49 [15:45:14 CEST] <Lynne> so yeah, if you implement a 27-point, 125-point and a 49-point fft [15:47:39 CEST] <Lynne> and excluding the 3 and 5-point ones, you'll need a 9-point one, a 25-point one and a 7-point one (7*7) [15:48:25 CEST] <Lynne> actually no, all you need is a 7 point one [15:50:53 CEST] <Lynne> thinking again no, you will need a full 27-point, 25-point and a 49-point, since 3x3, 5x5 and 7x7 are not coprimes [15:52:41 CEST] <durandal_1707> Lynne: https://www.dsprelated.com/showarticle/63.php [15:53:10 CEST] <durandal_1707> cant i just add some padding and use this ^ [15:56:28 CEST] <Lynne> durandal_1707: this is what tx.c does but less efficiently, the input map is identical but the output map is ommited in favor of a full set of twiddles [15:57:37 CEST] <Lynne> what filter requires 44100 x 6 fft? [15:57:52 CEST] <durandal_1707> 44100 x 60 [15:58:08 CEST] <durandal_1707> or any sample_rate * 60 [15:58:37 CEST] <Lynne> yeah, why? [15:58:54 CEST] <durandal_1707> it is for determining IR frequency response [15:59:22 CEST] <durandal_1707> for automatic gain and correct display for long lenght IRs [16:01:30 CEST] <durandal_1707> Lynne: is tx.c limited to max len of 131072 ? [16:02:10 CEST] <Lynne> no, 131072 * 15 = 1966080 [16:03:33 CEST] <durandal_1707> how hard would be to add support to bigger stuff? [16:05:11 CEST] <Lynne> to add higher power of two transforms is trivial, add another COSTABLE(X); add DECL_FFT(X, X - 1, X - 2) and add fftX in fft_dispatch[] [16:06:31 CEST] <durandal_1707> another point two transform is costly, i'm lookiing for using 131072 * X approach [16:07:52 CEST] <Lynne> to make X larger than 15 you'll need to implement a non-comprime to 3 and 5 transform, like 7 to get 15 * 7 [16:08:19 CEST] <Lynne> 11, 13, 17, 19, etc would work too [16:09:00 CEST] <durandal_1707> you can do 24 FFT with three 8 FFTs [16:09:56 CEST] <Lynne> yes, inefficiently, you could implement a hybrid system here too but you're on your own [16:10:30 CEST] <Lynne> and this case just happens to be this simple since 3 is the lowest factor after 2 [16:11:26 CEST] <durandal_1707> so you not interested in doing 1024 * 131072 FFT in tx.c ? [16:13:47 CEST] <Lynne> you can do 1024 * 131072 already with a few minor line changes [16:15:02 CEST] <durandal_1707> i'm interested in that, if that approach is not really slow [16:15:41 CEST] <Lynne> its a power of two fft, you need effort to make it slow [16:16:45 CEST] <durandal_1707> just need this complex multiplication .. and summing [16:17:01 CEST] <Lynne> keep in mind you will end up with a 10, maybe 20, probably more megabyte tx.o, so its not really practical [16:17:12 CEST] <Lynne> and you can't get away from this really [16:17:51 CEST] <durandal_1707> i'm talking about approach when only 131072 is used [16:17:59 CEST] <durandal_1707> no new tables are added to tx.c [16:18:26 CEST] <Lynne> just write a script that writes COSTABLE(131072 * i), DECL_FFT(X * i, X * i / 2, X * 1 / 4) and fft ## X * i [16:19:24 CEST] <Lynne> ptwo FFTs require all FFTs down to 4-point in our case [16:19:44 CEST] <durandal_1707> what fftw uses for such big cases? [16:20:15 CEST] <Lynne> heap memory requirements will be the same, they still need memory for all twiddles [16:21:00 CEST] <Lynne> its the code size that will be different, since they probably stop fusing FFTs into functions above some size [16:22:10 CEST] <Lynne> you could write a recursive function to do that though [16:23:36 CEST] <Lynne> look at static void fft##n(FFTComplex *z) and just make it recurse until fft_dispatch[N] exists [16:24:08 CEST] <cone-099> ffmpeg 03Paul B Mahol 07master:73afea3cccbd: avfilter/af_dynaudnorm: add more descriptive aliases for options [16:24:16 CEST] <Lynne> that way 1024 * 131072 will only be a single 5-odd line function [16:25:29 CEST] <Lynne> but you will need to make the COSTABLE init for 131072 - 1024 & 131072 be allocated on heap [16:26:00 CEST] <durandal_1707> yes, those numbers ... how are they calculated? [16:26:14 CEST] <Lynne> by init_ff_cos_tabs [16:26:49 CEST] <durandal_1707> and those are this e^i power thingy article mentions? [16:26:57 CEST] <Lynne> add COSTABLE_MALLOC() which declares the arrays as allocated during runtime and make init_ff_cos_tabs allocate them [16:27:06 CEST] <Lynne> yes, that's the twiddles [16:27:31 CEST] <Lynne> but the code in that function only works for ptwo sizes [16:29:04 CEST] <Lynne> got a paper or something for the filter? [16:30:32 CEST] <durandal_1707> you mean article or filter feature? [16:32:24 CEST] <Lynne> yes [16:32:53 CEST] <durandal_1707> the idea for filter feature is to make impulse response with FFT (instead of current code in liavfilter/af_afir.c) [16:33:34 CEST] <Lynne> okay, I've just started to benchmark libfftw3f for a 44100x60 transform with SIMD and inplace turned on, the fastest settings [16:33:39 CEST] <durandal_1707> which is incorrect and deadly slow for big impulse response files [16:33:44 CEST] <Lynne> its been 30 seconds and its still computing [16:33:50 CEST] <Lynne> on an i7 [16:34:44 CEST] <durandal_1707> shit, i need time machine to go to far future [16:36:13 CEST] <Lynne> segfault in my code, trying again, just init, and execute, no filling of arrays or anything [16:36:35 CEST] <Lynne> 46 seconds for 44100*60 [16:37:45 CEST] <durandal_1707> that is for worst case scenario [16:37:57 CEST] <durandal_1707> now try for just: 44100*10 [16:38:34 CEST] <Lynne> wait, I forgot I had FFTW_MEASURE, which takes time, trying the simple ESTIMATE version [16:38:52 CEST] <Lynne> 30 seconds flat -_- [16:39:28 CEST] <Lynne> trying 1024 * 131072 currently [16:39:50 CEST] <durandal_1707> if 10*44100 is in <=5 seconds its acceptable [16:42:19 CEST] <Lynne> 1024 * 131072 is still computing, 4 minutes in or so [16:42:32 CEST] <durandal_1707> that will never finish [16:43:35 CEST] <Lynne> yeah, cancelled, trying 44100 * 10 [16:46:03 CEST] <Lynne> a single forward transform takes 9 seconds, just above realtime [16:46:12 CEST] <Lynne> *below [16:46:42 CEST] <Lynne> inplace, out of place takes 25 seconds [16:47:50 CEST] <Lynne> wait, machine is downclocking, will try again in a few minutes [16:48:02 CEST] <durandal_1707> too hot [16:48:28 CEST] <durandal_1707> well, i think i will wait for better times instead [16:50:16 CEST] <durandal_1707> even the results are good, because the gain for IR is calculated only once and used many times after that, but our users are not going to wait that long [16:51:53 CEST] <Lynne> tried again, no downlocking, 5.8 seconds [16:53:21 CEST] <Lynne> dunno, I used to not be big fan of filters but often they're just the only pratical solution [16:55:02 CEST] <Lynne> is there something you can simd there? [16:56:42 CEST] <durandal_1707> the audio filtering part is already using SIMD a lot, also there are options to control: latency vs speed ratio [16:57:18 CEST] <durandal_1707> where it for convolution uses by default 8192 window [16:57:37 CEST] <durandal_1707> for all partitions [16:58:00 CEST] <durandal_1707> which on my i3 cput gives best speed [16:58:09 CEST] <durandal_1707> *cpu [16:58:36 CEST] <durandal_1707> maybe on faster cpu, it can get bigger 16384/32769 [16:59:51 CEST] <durandal_1707> you could write SIMD for af_biquads.c or for af_aiir.c if you feel bored [17:04:11 CEST] <Lynne> I am occasionally using biquads, and the float version doesn't seem that ba [17:04:14 CEST] <Lynne> *bad [17:04:54 CEST] <durandal_1707> use double version always [17:05:08 CEST] <durandal_1707> we need long long double too [17:05:43 CEST] <J_Darnley> You sure you don't want to use Google's new Brain format? [17:07:37 CEST] <durandal_1707> Lynne: i forgot, need double support in tx.c [17:08:20 CEST] <durandal_1707> with simd for float, and i will then convert af_afir.c (and other FFT filters) to tx.c [17:09:46 CEST] <Lynne> yeah, ok, I'll try to template that tonight, should be easier than that 8-point fft simd [17:11:10 CEST] <durandal_1707> Lynne: the simd does not need to be perfect, enough is to be almost as fast as current rdft code [22:10:51 CEST] <J_Darnley> Wow. Looks like I was missing a real shit-show today featuring VLC, clickbait scammers, and "security researchers". [22:24:52 CEST] <J_Darnley> Oh... Also, a starring roll is played by my favorite "stable = old" character. [22:54:18 CEST] <jamrial> what was it? [22:59:39 CEST] <nevcairiel> oh i read some headline to d ay about some claimed security issues in VLC, i didnt actually bother to read the article, i wonder if that was related [23:35:20 CEST] <ubitux> https://mobile.twitter.com/videolan/status/1153963312981389312 [23:35:33 CEST] <ubitux> jamrial, nevcairiel ^ [23:43:58 CEST] <jamrial> deja vu [23:45:35 CEST] <nevcairiel> irresponsible security people that just want their name on something that sounds important, you should be able to sue people like that for defamation [23:55:38 CEST] <Compn> hah welcome to "security" wild wild west [23:55:50 CEST] <Compn> "we cant change the system its all we have" [23:55:57 CEST] <Compn> catch 22 [23:58:18 CEST] <nevcairiel> the sysmte is fine [23:58:27 CEST] <nevcairiel> its people not following it [23:59:23 CEST] <J_Darnley> yeah... from what I read: follow the SOP of your organization and you won't have angry twitter messages flying about [00:00:00 CEST] --- Thu Jul 25 2019

1 0

ffmpeg-devel.log.20190730
by burek 16 Aug '19

16 Aug '19

[00:21:06 CEST] <cone-420> ffmpeg 03Andreas Rheinhardt 07master:4e7e30bbe0fb: cbs: Don't set AVBuffer's opaque [00:21:07 CEST] <cone-420> ffmpeg 03Andreas Rheinhardt 07master:ae49993ce6e5: cbs_h264: Improve adding SEI messages [00:21:08 CEST] <cone-420> ffmpeg 03Andreas Rheinhardt 07master:0e66e1b61ea2: cbs_mpeg2: Decompose Sequence End [00:21:09 CEST] <cone-420> ffmpeg 03Andreas Rheinhardt 07master:276b21a58690: cbs_mpeg2: Rearrange start code search [00:21:10 CEST] <cone-420> ffmpeg 03Andreas Rheinhardt 07master:fd93d5efe642: cbs_mpeg2: Fix parsing the last unit [05:10:14 CEST] <Compnn> hehe [05:10:29 CEST] <Compnn> bayer talk has me interested in this r3d bug again https://trac.ffmpeg.org/ticket/2690 [05:11:04 CEST] <Compnn> Lynne, who are you , welcome to ffmpeg development! [05:35:25 CEST] <cone-318> ffmpeg 03James Almer 07master:502aff91a769: avformat/av1: fix AV1CodecConfigurationBox name in doxy [06:38:56 CEST] <rcombs> clang diagnosed an issue in vf_thumbnail_cuda that was presumably missed by NVCC (operator precedence) [09:25:36 CEST] <cone-673> ffmpeg 03Linjie Fu 07master:b3b7523feb5a: lavu/hwcontext_qsv: fix the memory leak [09:41:55 CEST] <durandal_1707> nobody managed to test dsddec patch? [10:00:41 CEST] <rcombs> alright my stuff works, patch sent [10:05:38 CEST] <philipl> rcombs: amazing. What inspired you to do it? [10:06:12 CEST] <rcombs> nvidia didn't respond within like an hour, I was bored, and it turned out to be easy [10:06:27 CEST] <rcombs> ffmpeg really doesn't use very much of the API and the parts that it does use are trivial [10:07:20 CEST] <rcombs> the hardest parts were texture sampling (which is a single instruction I just had to write a wrapper for) and a few special register accesses (blockIdx, blockDim, threadIdx) [10:07:27 CEST] <rcombs> and as you can see from the code, none of those was complex either [10:08:03 CEST] <rcombs> I was a little worried that the compiler might actually emit all the redundant ASM in the macros, but no it's smart enough to avoid that [10:08:27 CEST] <philipl> Nice. [10:08:34 CEST] <philipl> I totally support switching to clang by default. [10:08:49 CEST] <rcombs> it's also apparently got better diagnostics [10:08:58 CEST] <rcombs> judging by that thumbnail bug it spotted [10:09:06 CEST] <nevcairiel> do you need like a special version of clang that supports cuda? [10:09:22 CEST] <philipl> Separately, the issue with nvcc requiring libs even when producing a ptx is that, as far as I can tell, they do a host-side compilation during the flow even if you don't want it and then discard it. [10:09:32 CEST] <rcombs> ubuntu's clang has it by default [10:09:35 CEST] <philipl> nevcairiel: no. It's all upstream now [10:09:41 CEST] <rcombs> the one that ships with Xcode doesn't, but the one in brew does [10:09:41 CEST] <nevcairiel> does the msvc clang have it? [10:09:42 CEST] <nevcairiel> :D [10:09:55 CEST] <rcombs> dunno about msvc, didn't test that [10:10:04 CEST] <nevcairiel> incidentally, they also gave msvc-clang full IDE support recently [10:10:22 CEST] <philipl> Also, who cares. If you have neither a working clang nor a working nvcc, you go and get one. That's the same as what the story is today but with less licencing hassl.e [10:10:30 CEST] <rcombs> yup [10:10:46 CEST] <rcombs> I was also a little worried about abs(), since the SDK calls some extern function for that [10:11:04 CEST] <philipl> rcombs: I really appreciate this. I got as far as seeing the need to write a header but never tried to work out what it needed to contain. [10:11:13 CEST] <rcombs> but nvcc magically turns that extern call into an abs instruction, and clang does the same thing with my code [10:11:35 CEST] <rcombs> philipl: my procedure was just "build, look at the errors, and add the stuff it's complaining about" [10:12:17 CEST] <rcombs> hmmmmmmmm [10:12:22 CEST] <philipl> a problem? [10:12:29 CEST] <rcombs> I think I could implement color format conversion pretty easily [10:12:45 CEST] <philipl> As a cuda filter? [10:12:50 CEST] <rcombs> as part of the scale filter [10:13:15 CEST] <philipl> probably. It's not the smartest filter right now. [10:13:35 CEST] <rcombs> hell, I can probably macro-ize a lot of this [10:13:49 CEST] <rcombs> or template-ize [10:15:07 CEST] <rcombs> why do cu_func_ushort4 and cu_func_uchar4 even exist, they're not used [10:15:29 CEST] <philipl> Need to ask the nvidia engineers who wrote the original filter [10:15:55 CEST] <rcombs> maybe for RGBA support, I guess [10:16:41 CEST] <philipl> Yeah, but they never hooked it up [10:17:29 CEST] <rcombs> I don't especially care to either tbh [10:17:41 CEST] <rcombs> but I guess if I template-ize I'll keep it generic [10:18:38 CEST] <rcombs> y'know what this would actually be easier to do if I didn't have to support the real SDK [10:18:47 CEST] <rcombs> though I can do it anyway [10:18:51 CEST] <philipl> I'm fine with dropping support for it. [10:19:11 CEST] <philipl> At least once we're comfortable with the functionality and know what our macos and windows stories are. [10:19:21 CEST] <philipl> s/macos// [10:22:11 CEST] <philipl> rcombs: Someone will probably argue that the compat header should go in ffnvcodec too. Do you mind? [10:22:25 CEST] <rcombs> fine by me [10:22:25 CEST] <rcombs> and I don't particularly care what it's licensed under [10:22:30 CEST] <rcombs> though, it might be preferable to keep it in ffmpeg [10:22:38 CEST] <rcombs> because it might need additions as new stuff's added [10:22:50 CEST] <durandal_1707> cehoyos: do you have fast multicore CPU near you? [10:22:58 CEST] <rcombs> it's really less API declaration and more utility library tbh [10:23:17 CEST] <cehoyos> I have access to a multicore hardware, how can I help you? [10:23:36 CEST] <philipl> rcombs: we've had to add a bunch of stuff to the other headers over time. It's a tax on doing business, to be sure, but it's managable. [10:23:53 CEST] <philipl> And it allows other projects to benefit. I use the headers in mpv, although I seriously doubt we'd ever have a kernel in there. [10:24:19 CEST] <durandal_1707> cehoyos: just test dsddec patch speed improvement if you want [10:24:19 CEST] <philipl> Anyway, I'm personally fine with it in compat but the same arguments that lead to the original headers getting moved out apply here too. [10:24:19 CEST] <cehoyos> Could you point me to an input file? [10:25:34 CEST] <rcombs> philipl: also, we -include it, as you can see in configure [10:25:43 CEST] <philipl> yeah. [10:25:50 CEST] <rcombs> which could be done with a file in ffnvcodec but it's a bit more awkward [10:26:05 CEST] <philipl> As I said, I'm fine pushing as-is, but I'd bet on grumbling in due course. [10:26:07 CEST] <rcombs> I see this as more of an x86inc-style thing really [10:26:16 CEST] <durandal_1707> cehoyos: http://www.lindberg.no/hires/test/2L-145/2L-145_mch_DSF_2822k_1b_01.dsf [10:26:29 CEST] <rcombs> I'm hoping to get a bit more testing from people who know this code better than I do before pushing [10:26:42 CEST] <durandal_1707> in up directory you could find smaller mch dsf files i think [10:26:55 CEST] <philipl> OK. As I said in the email, I'm still two weeks out from being able to test. [10:27:51 CEST] <philipl> Hopefully BtbN can take a look sooner. [10:31:37 CEST] <cehoyos> durandal_1707: Is this the patch? avcodec/dstdec: add slice threading support [10:31:50 CEST] <durandal_1707> cehoyos: not that one, dsddec one [10:32:31 CEST] <cehoyos> avcodec/dsddec: add slice threading support ? [10:32:40 CEST] <durandal_1707> yes, that one [10:36:09 CEST] <cehoyos> It gets slower and it heats like crazy... [10:36:28 CEST] <durandal_1707> lol, how much slower? [10:37:45 CEST] <cehoyos> It only gets two times slower, but it heats up to ten times [10:38:11 CEST] <durandal_1707> what is your CPU? [10:38:20 CEST] <cehoyos> With -threads 1: real 0m6.090s, user 0m6.005s [10:39:06 CEST] <cehoyos> With automatic threads: real 0m13.079s, user 0m59.798s [10:39:15 CEST] <durandal_1707> omg [10:39:36 CEST] <durandal_1707> this is with 8 cores or? [10:40:18 CEST] <cehoyos> This is on an old 4-core, hyperthreading Intel cpu, on Power (with many cores), its: user 0m17.020s vs real 0m31.435s, user 2m2.515s [10:40:50 CEST] <cehoyos> Please ask Stefano for new hardware... [10:40:52 CEST] <cehoyos> (Or Intel) [10:41:17 CEST] <cehoyos> Yes, 8 hyperthreading cores [10:41:38 CEST] <durandal_1707> i have new hardware, but not here, and another one here too but that computer is dead.... [10:42:15 CEST] <durandal_1707> here i have 2 core Celeron CPU, and it is also 2 times slower with multiple threads [10:42:40 CEST] <durandal_1707> this is really strange [10:54:02 CEST] <cehoyos> Does anybody know why the irc logging does not work anymore? [11:52:29 CEST] <durandal_1707> michaelni: do you know why execute2 would make decoding slower for audio with >1 threads? [12:07:14 CEST] <michaelni> know? no, but maybe the overhead for thread activation and sync is too expensive in relation to the smaller computational units in audio [12:07:47 CEST] <durandal_1707> michaelni: i doubt so, the decoder is pretty slow even with 1 threads [12:08:24 CEST] <durandal_1707> michaelni: or you mean because packet sizes feed to decoder are too small? [12:09:29 CEST] <michaelni> more guessing than , meaning, i dont know what the packet sizes are or what testcase you use ... but yes thats what i was guessing [12:34:59 CEST] <nevcairiel> how big are dsd packets? if they are too small, the overhead would indeed kill it [12:36:39 CEST] <durandal_1707> 8192 bytes size with 1044 duration, for 352800 sample rate [12:37:36 CEST] <durandal_1707> around 87 packets per second [12:38:41 CEST] <durandal_1707> for flac ratio is around 2x times lower [12:39:25 CEST] <nevcairiel> thats not terribly small, but since slice threading was made for video, perhaps still too much overhead [13:28:45 CEST] <durandal_1707> i doupled packet sizes and slowdown is still 2x [13:33:39 CEST] <cone-673> ffmpeg 03Stephan Hilb 07master:b761ae072a16: lavd/v4l2: produce a 0 byte packet when a dequeued buffer's size is unexpected [15:03:09 CEST] <BtbN> I have an AMD card in my system now, so I can test building it, but my Nvidia Linux box had its PSU fail or something, so I can't test it in action. [15:07:52 CEST] <philipl> sad panda. You'll fix that eventually right? :-) [15:10:19 CEST] <BtbN> It's a special form-factor, so I can't just buy a new one, and will have to buy a whole new PC. I planned that for a while now anyway, but it's not something I'll do within the next couple days. [15:13:41 CEST] <BtbN> Does any regular clang build work for this? Or does it need special flags? [15:14:09 CEST] <BtbN> I'd even argue that we can drop the non-free for the nvcc builds as well, cause there is a way to product a functionally equivalent build now. [15:17:41 CEST] <cone-673> ffmpeg 03Rodger Combs 07master:a0c19707811c: lavfi/vf_thumbnail_cuda: fix operator precedence bug [15:18:48 CEST] <cone-673> ffmpeg 03Rodger Combs 07release/4.2:6a5ed71d36f7: lavfi/vf_thumbnail_cuda: fix operator precedence bug [15:21:17 CEST] <JEEB> BtbN: I think a regular build with the target enabled [15:22:11 CEST] <BtbN> Gentoo has a llvm_targets_NVPTX flag, I guess I need that [15:33:17 CEST] <philipl> BtbN: it worked ootb for rcombs and in my incomplete clang experiments, it worked with my stock ubuntu clang too. I imagine Fedoras and SUSEs etc turn it on by default too. [15:34:37 CEST] <philipl> I agree on removing nvcc from non-free, although I obviously supported that before too. :-) [15:40:45 CEST] <j-b> Waw. [15:40:56 CEST] <j-b> Someone deserves a bg hug [15:40:59 CEST] <j-b> big [15:51:01 CEST] <BtbN> j-b, it should be perfectly fine to drop the non-free on everything but libnpp now, should it? GPL is happy with clang providing functionally equivalent builds to nvcc? [15:51:58 CEST] <BtbN> Specially as one can download nvcc without needing an account now. They gave up that part, and it's a straight forward download now. Still not under a free license though. [17:36:07 CEST] <j-b> BtbN: I would remove the old one, or at least deprecate it [17:36:43 CEST] <BtbN> Removing nvcc support will greatly increase the annoyance-level of making a Windows-Build [17:37:12 CEST] <BtbN> Specially when building with msvc and the like [17:40:25 CEST] <j-b> Why? MSVC supports clang-cl now [17:40:36 CEST] <BtbN> Because you need to set it up first [17:41:01 CEST] <BtbN> Download binaries from somewhere, put it in PATH, keep it up to date, ... [17:41:24 CEST] <BtbN> They do not appear to provide official Windows binaries [17:41:42 CEST] <j-b> Sorry, but licensing clarity > annoyance of building. [17:42:04 CEST] <j-b> Especially, since now you have WSL and WSL2 on Windows. [17:42:16 CEST] <BtbN> Those produce linux binaries and cannot access GPUs [17:42:28 CEST] <j-b> Untrue. [17:42:39 CEST] <j-b> You can cross-compile from WSL [17:42:41 CEST] <BtbN> ... [17:42:47 CEST] <BtbN> Very useful for MSVC builds [17:43:15 CEST] <BtbN> I don't see any issue with the license anymore for nvcc, so no reason to drop it and break existing setups [17:43:36 CEST] <j-b> I disagree. [17:43:42 CEST] <j-b> but we will disagree here. [17:44:28 CEST] <BtbN> Dropping that would be just another case of hurting users on the back of some weirdly perceived freedom from the evil commercial overlords. [17:45:02 CEST] <j-b> commercial overlords that contribute a lot to this community... [17:45:04 CEST] <j-b> ...not [17:45:28 CEST] <j-b> This is just helping nVidia, and validating their strategy of doing nothing open source [17:45:36 CEST] <j-b> and moving away from open source [17:45:40 CEST] <BtbN> "Company XY doesn't seem very nice, let's do something that hurts thousands or more users, but doesn't fuzz the company one bit." [17:45:46 CEST] <j-b> That's not true. [17:46:04 CEST] <ubitux> -/g 14 [17:46:05 CEST] <BtbN> Nvidia clearly does not care if the CUDA/nvenc stuff is non-free or not [17:46:18 CEST] <j-b> "Company XY does something fishy wrt to open source. We have an alternative, that works. Let's push that one" [17:46:33 CEST] <j-b> This has nothing to do with being nice or not [17:46:54 CEST] <BtbN> A functional 100% equivalent alternative, so by GPL, it's fine to drop the non-free flag on those filters, and offer both ways to build them [17:48:32 CEST] <j-b> I disagree. But as I said, we will just disagree here. [17:48:51 CEST] <BtbN> so, llvm does have Windows binaries. But only for older releases...? [17:49:11 CEST] <BtbN> And for 9.0 snapshots. But not current 8.x. What? [18:00:11 CEST] <nevcairiel> how do i find out what clang s upports? [18:06:20 CEST] <nevcairiel> the help output includes cuda stuff, is that enough? :d [18:08:33 CEST] <BtbN> it needs nvptx support [19:58:49 CEST] <durandal11707> what i need to install to get cl.exe for VS 2019? [20:02:21 CEST] <BtbN> Visual Studio 2019 [20:02:44 CEST] <BtbN> The Community Edition is enough. Just make sure to actually select C/C++ dev tools in the installer [20:08:48 CEST] <durandal11707> how can i use VS profiler? [20:27:58 CEST] <vel0city> Debug -> Profiler or Alt+F2 on vs2017 [20:34:25 CEST] <durandal_1707> but i need to profile ffmpeg, is that possible at all? [20:36:51 CEST] <JEEB> I would guess this just lets you run an exe [20:36:51 CEST] <JEEB> https://docs.microsoft.com/en-us/visualstudio/profiling/how-to-install-the-… [20:37:35 CEST] <JEEB> https://docs.microsoft.com/en-us/visualstudio/profiling/command-line-profil… [20:43:06 CEST] <durandal_1707> hmm, i still can not get msvc to build, getting missing libCMT.lib [20:51:04 CEST] <vel0city> @durandal_1707: either check errors if that didn't build successfully, or if it's not assigned as a dependency you will need to go to the Solution Explorer, find CMT or whatever it will be called, and right click -> Build to build it manually [20:53:02 CEST] <vel0city> also afaik you don't need to build it yourself via VS if you have an .exe and it has debug symbols [20:53:28 CEST] <vel0city> if that's an option for your case (ie. you haven't changed any code) [21:05:33 CEST] <thardin> oh boy, dolby-e [21:09:15 CEST] <cehoyos> We already auto-detect ac3 and dts, so auto-detection in wav should be possible. [21:10:55 CEST] <JEEB> yea, there are some capture formats where I've also thought if there could be a flag to note that "btw try detecting if this is a coded bit stream within a PCM container" [21:11:02 CEST] <JEEB> in my case mostly AAC [21:18:21 CEST] <kierank> cehoyos: what about dolbye inside 302m inside ts [21:18:28 CEST] <kierank> where the dolby e can span over 302m packets [21:19:01 CEST] <kierank> JEEB: there is a flag in SDI but nobody uses it [21:19:40 CEST] <JEEB> :) [21:19:43 CEST] <JEEB> business as usual [21:28:00 CEST] <durandal_1707> michaelni: if 2x slowdown slice thread decoding does not happen because of calling overhead what is next possible source of problem? [21:31:28 CEST] <cehoyos> kierank: I always wanted to look into it, is it possible that only a few lines of code are needed? [21:31:53 CEST] <kierank> cehoyos: for the simple case yes but the complex case is when the dolby e spans a packet [21:31:58 CEST] <kierank> (i.e is misaligned to the video) [21:32:12 CEST] <cehoyos> Needs a parser? [21:32:34 CEST] <durandal_1707> parser does not modify data [21:36:02 CEST] <durandal_1707> i gave up builing msvc toolchain it havent worked and i downloaded GBs of data [21:49:51 CEST] <durandal_1707> i just added execute2 to flacdec for fun, and it causes also slowdown, but even with very small packet size, slowdown was much lower [21:50:15 CEST] <durandal_1707> execute2 was doing nothing, returning immediately [22:06:33 CEST] <durandal_1707> so it is not overhead, it must be something else [22:47:42 CEST] <cehoyos> jamrial: 4/7 has a typo in the commit message [22:48:19 CEST] <jamrial> cehoyos: ah, thanks for noticing. will fix it [00:00:00 CEST] --- Wed Jul 31 2019

1 0

ffmpeg-devel.log.20190815
by burek 16 Aug '19

16 Aug '19

[01:39:21 CEST] <cone-191> ffmpeg 03Michael Niedermayer 07master:c4a88fb546b6: avcodec/ffwavesynth: Fix integer overflow for some corner case values [01:39:22 CEST] <cone-191> ffmpeg 03Michael Niedermayer 07master:cdea0206efec: avcodec/ffwavesynth: Fixes invalid shift with pink noise seeking [01:39:23 CEST] <cone-191> ffmpeg 03Andreas Rheinhardt 07master:f2d522f2a888: fftools: Use right function signature and pointers [01:39:24 CEST] <cone-191> ffmpeg 03Michael Niedermayer 07master:df7e79a5c31e: tools/target_dec_fuzzer: adjust pixel threshold for MSRLE, as it allows coding gigantic images on tiny input [01:39:25 CEST] <cone-191> ffmpeg 03Michael Niedermayer 07master:c812db814ebd: avcodec/loco: Check left column value [01:39:26 CEST] <cone-191> ffmpeg 03Michael Niedermayer 07master:e301736862f1: avcodec/flicvideo: Optimize and Simplify FLI_COPY in flic_decode_frame_24BPP() by using bytestream2_get_buffer() [01:39:27 CEST] <cone-191> ffmpeg 03Andreas Rheinhardt 07master:114775955241: avformat/mux: Use const AVPacket * in compare functions [01:39:28 CEST] <cone-191> ffmpeg 03Andreas Rheinhardt 07master:789dcfaf58db: avformat/mux: Use av_packet_rescale_ts [01:39:29 CEST] <cone-191> ffmpeg 03Shiyou Yin 07master:e1039b09c470: avutil/mips: remove redundant code in TRANSPOSE16x8_UB_UB. [03:03:27 CEST] <cone-191> ffmpeg 03James Almer 07master:03ba38683301: avcodec/libdav1d: allow setting frame size limit in pixels [04:03:31 CEST] <Lynne> is there any way for a checkasm test function to always return a value 0.12 smaller than it should be? [04:04:24 CEST] <Lynne> I can add the offset/negative offset in both the c and asm functions but it makes no difference, its always 0.12 smaller [04:06:42 CEST] <Lynne> wtf, I recompiled the init file and now its fine? [04:43:57 CEST] <cone-191> ffmpeg 03tomajsjiang 07master:3d1506c630eb: lavf/avio: add a ffio_realloc_buf API for AVIO buffer realloc [04:43:58 CEST] <cone-191> ffmpeg 03tomajsjiang 07master:9c3adb7ce235: lavf/utils: fix error like "offset 0x1f85: partial file" [04:43:59 CEST] <cone-191> ffmpeg 03Jun Zhao 07master:3aeb681f074a: libavdevice: Update the class name as uniform style [11:43:41 CEST] <durandal_1707> Lynne: sampling index for our file is 0, which must not be [11:44:10 CEST] <durandal_1707> it must be either 8 or 11 [12:17:07 CEST] <j-b> durandal_1707: I do. [12:17:39 CEST] <durandal_1707> wait, it might not be aac file at all [12:18:08 CEST] <durandal_1707> cehoyos is probably terribly mistaken as usual [12:20:42 CEST] <j-b> Errare humanum est. [12:41:05 CEST] <Lynne> aac celp inside adts [12:52:06 CEST] <durandal_1707> whats difference? [12:52:22 CEST] <durandal_1707> this one uses avi demuxer [13:01:50 CEST] <durandal_1707> this is extradata: 0x401F0000 [13:05:55 CEST] <Lynne> it'll be funny if it turns out its a weird mux and its actually standard aac-lc [13:07:24 CEST] <durandal_1707> their program claims its MPEG2-AAC [13:07:35 CEST] <durandal_1707> 4 channels [13:10:52 CEST] <Lynne> AAC CELP is mpeg4, so maybe it actually is a weird mux [13:38:04 CEST] <nevcairiel> aac celp is the same concept opus uses, seperate coding for voice? [13:40:41 CEST] <Lynne> yes, but its much more limited, for example you can't use celp for certain bands, only for the entire frame [13:41:52 CEST] <Lynne> the spec also seems underspecified as it doesn't say anything about how to handle transitions (celp -> long window/short window and vice versa) [13:52:48 CEST] <Lynne> kinda weird it ended up being a part of mpeg4, same as aac-ltp, since they both fill the same role [14:30:39 CEST] <cone-634> ffmpeg 03Lynne 07master:aac382e9e5f1: aarch64/opusdsp: do not clobber register v8 [14:54:18 CEST] <thardin> is there a test source with cuts? [14:54:30 CEST] <thardin> or sample I guess [15:02:01 CEST] <thardin> found something that works [15:27:37 CEST] <thardin> wait a sec.. one has to clear AV_PKT_FLAG_KEY explicitly? [15:34:16 CEST] <thardin> nm, just something to do with probing growing .avi files [16:57:19 CEST] <J_Darnley> The workstation and the monitor both seem to work just fine. [17:06:27 CEST] <kierank> J_Darnley: wrong channel [17:06:39 CEST] <J_Darnley> my bad [19:12:10 CEST] <durandal_1707> Lynne: the spec you linked just seems outdated compared to what reference code does [19:40:41 CEST] <durandal_1707> yes, with latest spec the file is indeed celp. the only issue is that channels are coded as independent packets with same pts [19:41:26 CEST] <durandal_1707> so 4ch file have 4 packets of same pts one after another [19:41:40 CEST] <durandal_1707> this is insane [19:43:51 CEST] <juandl> Hello, just pinged my patch again, I'm here if anyone wants to talk about it. [19:56:33 CEST] <thardin> guess I should ask here as well: why is gop_size < keyint_min by default? [19:56:39 CEST] <Illya> anyone has a setup for HDR on linux? trying to fix some video but I can't see it :| [19:59:48 CEST] <cone-008> ffmpeg 03Guo, Yejun 07master:2c01434d60db: convert_from_tensorflow.py: add option to dump graph for visualization in tensorboard [19:59:48 CEST] <cone-008> ffmpeg 03Guo, Yejun 07master:ddd92ba2c6c6: convert_from_tensorflow.py: support conv2d with dilation [20:13:03 CEST] <thardin> ooh it's possible to put in defaults on a per-codec basis now? neato [20:20:46 CEST] <jamrial> thardin: yeah, that's what i meant when i said the libx264 wrapper changed them [20:27:15 CEST] <thardin> hum, but in this case keyint_min makes no sense [20:27:30 CEST] <thardin> maybe it makes for smaller index tables? [20:29:00 CEST] <thardin> on the other hand, if the user really wants that who am I to argue? [20:33:47 CEST] <thardin> there we go, successfully prevented a keyframe (even though it codes as one) [20:34:21 CEST] <doughy1> why does doing this? ffmpeg -i input1.mp4 -c copy -bsf:v h264_mp4toannexb -f mpegts output.ts create 4gig file to 6 gig [20:38:54 CEST] <mkver> I see three possibilities: 1. Transport streams generally have a higher overhead. 2. h264_mp4toannexb adds in-band extradata. 3. Your mp4 uses edit lists which specify that some part of the video get looped. In this case it will be several times in the output, but only once in the input. [20:40:38 CEST] <doughy1> i ran that command in the first place because it's 2 hour input.mp4 video and but video show duration as 3 hours and this command has fixed this problem but with 2 more gigs [20:41:03 CEST] <doughy1> now output.ts shows correction video duration [20:41:46 CEST] <doughy1> now output.ts shows correct video duration [20:44:29 CEST] <doughy1> mkver that also means if convert output.ts to output.mp4 again, then it should shrink to 4 gigs but it does not [21:08:11 CEST] <mkver> A size increase by 2. and 3. will not be reverted when remuxing back to mp4. [21:08:30 CEST] <mkver> And anyway, you are on the wrong channel. [21:30:33 CEST] <Lynne> durandal_1707: well, I don't think they're coupled channels like stereo but rather mono streams, one for each mic [21:30:53 CEST] <Lynne> so a demuxer could just say they're streams [21:32:55 CEST] <durandal_1707> yea, will see [22:05:33 CEST] <jamrial> ffplay doesn't seem to like truehd streams [22:08:22 CEST] <mkver> What error messages do you get? [22:08:45 CEST] <mkver> Is it TrueHD in Matroska and you get the "keyframes not correctly marked" error? [22:46:18 CEST] <Lynne> irc logs off? irc logs off. [22:46:25 CEST] <Lynne> carl not here? carl not here. [22:47:31 CEST] <Lynne> nicolas is an awful person who disagrees with everything and does no work like ever, yet hangs around the ml to be obnoxious [22:52:10 CEST] <Lynne> his opinions on asserts should disqualify him from working on any library ever [22:52:32 CEST] <durandal_1707> add reimar to that list [22:55:14 CEST] <Lynne> I'll remove that assert if I push that patch, just because maybe he'll fuck off then [22:56:19 CEST] <durandal_1707> they only work toward covering michael [22:57:29 CEST] <Lynne> reimar does, nickolas is just there to misunderstand and be annoying and demanding [23:07:14 CEST] <durandal_1707> mkver: there is bug in mkv? [23:54:46 CEST] <Compnn> Lynne, who are you ? [23:57:59 CEST] <jamrial> mkver: no, no error. audio playback is wrong, regardless of container [00:00:00 CEST] --- Fri Aug 16 2019

1 0

ffmpeg.log.20190815
by burek 16 Aug '19

16 Aug '19

[00:01:01 CEST] <Classsic> I read this https://trac.ffmpeg.org/wiki/Hardware/QuickSync but canŽt find any example like this "ffmpeg -hwaccel vaapi -hwaccel_output_format vaapi ... -i interlaced_input.mp4 -vf 'deinterlace_vaapi,scale_vaapi=w=1280:h=720,hwdownload,format=nv12' -c:v libx264 ... progressive_output.mp4 [00:38:43 CEST] <petecouture> BtbN It worked! thanks again mate. You have a patron account? [00:38:52 CEST] <BtbN> no [00:39:11 CEST] <petecouture> Accept donations? [02:24:38 CEST] <AiNA_TE> hello [02:24:56 CEST] <AiNA_TE> i am trying to encode on the fly from a directshow device [02:25:13 CEST] <AiNA_TE> ffmpe is giving me "Interger division by zero." errors [02:31:11 CEST] <AiNA_TE> https://pastebin.com/raw/iMnFLp1v [02:32:56 CEST] <AiNA_TE> https://i.imgur.com/H8eUtiK.png [02:33:09 CEST] <AiNA_TE> this is the crash error [02:33:15 CEST] <AiNA_TE> anyone know whats up [02:56:59 CEST] <durandal_1707> the image is blurer [20:30:59 CEST] <doughy1> why does doing this? ffmpeg -i input1.mp4 -c copy -bsf:v h264_mp4toannexb -f mpegts output.ts create 4gig file to 6 gig [20:34:57 CEST] <furq> what does the last line of the ffmpeg output say under "muxing overhead" [20:35:52 CEST] <doughy1> furq are you talking to me? [20:36:11 CEST] <furq> yes [20:36:31 CEST] <doughy1> i closed the box [20:36:41 CEST] <doughy1> so i missed it [20:36:56 CEST] <pink_mist> then reopen it, rerun the command, and check it [20:37:07 CEST] <doughy1> i deleted input1.mp4 [20:37:14 CEST] <doughy1> i still haved output.ts [20:37:33 CEST] Action: zeromind sighs [20:38:02 CEST] <furq> well yeah mpegts has a lot of overhead compared to mp4 [20:38:10 CEST] <furq> 2GB on a 4GB file sounds like too much but shrug [20:38:17 CEST] <furq> you'd normally expect ~10% overhead or so [20:38:39 CEST] <doughy1> i ran that command in the first place because it's 2 hour input.mp4 video and but video show duration has 3 hours and output.ts has fixed this duration [20:39:34 CEST] <doughy1> but that also means if i convert output.ts to ouput.mp4 it should go back to 4 gigs but it does not [20:41:43 CEST] <pink_mist> presumably whatever your -bsf:v argument did is a cause then [20:41:50 CEST] <pink_mist> since that's the only other variable you have [20:42:06 CEST] <doughy1> what does bsf-v do [20:42:20 CEST] <furq> it isn't [20:42:23 CEST] <pink_mist> no idea, why did you put it in your commandline? [20:42:36 CEST] <furq> that bsf gets autoinserted for mp4 to ts in recent ffmpeg [20:42:53 CEST] <pink_mist> huh, ok [20:42:56 CEST] <doughy1> because this help page says so: https://trac.ffmpeg.org/wiki/Concatenate#differentcodec [20:44:13 CEST] <furq> that doesn't add a significant amount of data and it'll be reversed when going back to mp4 anyway [20:45:27 CEST] <furq> annex b should be slightly smaller than avcc if anything [20:45:41 CEST] <furq> but very slightly [20:47:37 CEST] <pink_mist> doughy1: well, without the original file, it would be my guess that it's impossible to figure this out [22:57:07 CEST] <firelegend> Hi all. I would like to ask about dav files, whether the format has been documented somewhere? [22:57:50 CEST] <durandal_1707> firelegend: be more specific [22:58:12 CEST] <durandal_1707> do you mean dahua cctv stuff? [22:58:49 CEST] <firelegend> Yes [22:59:01 CEST] <firelegend> I see that ffmpeg supports decoding the format [22:59:26 CEST] <durandal_1707> everything is documented in dhav demuxer [23:00:02 CEST] <durandal_1707> just because it may have also audio [23:00:21 CEST] <firelegend> Relevant part? https://github.com/FFmpeg/FFmpeg/blob/master/libavformat/dhav.c [23:00:28 CEST] <durandal_1707> yes [23:01:00 CEST] <durandal_1707> why are you interested in that format? [23:01:18 CEST] <firelegend> Because it's obscure [23:01:46 CEST] <firelegend> I even busted out ghidra and x64dbg ready to dig into the proprietary code [23:02:02 CEST] <firelegend> but if it's been implemented it will help me [23:02:21 CEST] <durandal_1707> feel free to improve it more [23:02:35 CEST] <durandal_1707> it was also REd [23:02:57 CEST] <firelegend> I'd imagine so, RE is a fun subject. [23:03:07 CEST] <firelegend> But I am glad someone has already done the heavy lifting :) [23:04:37 CEST] <durandal_1707> it was done with limited number of samples, format appears to support more obscure codecs [23:05:06 CEST] <firelegend> Obscure as in fully in-house [23:05:11 CEST] <firelegend> or modified existing codecs. [23:05:43 CEST] <durandal_1707> all codecs it use are normal existing codecs [23:06:32 CEST] <firelegend> So they just insert more metadata and different header\? [23:07:36 CEST] <durandal_1707> appears so [23:10:12 CEST] <firelegend> Is there any reason for them to use modified codecs and not whats publicly available? [23:11:17 CEST] <durandal_1707> it use unmodified codecs in own clumsy container [00:00:00 CEST] --- Fri Aug 16 2019

1 0

ffmpeg.log.20190726
by burek 15 Aug '19

15 Aug '19

[00:12:39 CEST] <MomokoGC> [mpegts @ 0000000000660580] Could not find codec parameters for stream 0 (Video: hevc (HEVC / 0x43564548), none): unspecified size [00:12:39 CEST] <MomokoGC> Consider increasing the value for the 'analyzeduration' and 'probesize' options [00:12:49 CEST] <MomokoGC> how do i increase these values? [00:13:15 CEST] <another> -analyzeduration 10M -probesize 10M [00:13:24 CEST] <MomokoGC> ty [00:36:53 CEST] <MomokoGC> another that worked but my output has no audio track [00:37:18 CEST] <MomokoGC> i did "-c copy" but only video is present in output [00:37:23 CEST] <MomokoGC> you got any ideas? [02:27:23 CEST] <grkblood13> why is this failing to record the stream coming out of the specified input device? I'm using the azap program to fetch an OTA signal using an ATSC tuner card. https://pastebin.com/m7xgrnjy [02:42:11 CEST] <another> move the -s option before the -i [02:52:52 CEST] <grkblood13> Option video_size not found. [02:54:45 CEST] <Beam_Inn> https://ffmpeg.org/ffmpeg.html [02:54:56 CEST] <Beam_Inn> can someone tell me in the manual here where I can figure out how to just trim a video? [02:56:17 CEST] <Beam_Inn> ffmpeg -i movie.mp4 -ss 00:00:03 -t 00:00:08 -async 1 cut.mp4 maybe [02:56:26 CEST] <Beam_Inn> try it [03:06:37 CEST] <nine_milli> blb [04:17:00 CEST] Last message repeated 1 time(s). [06:28:45 CEST] <brx_> Hi everyone. My command mixes an mp4 with an mp3 keeping the audio stream from the mp4 using 'amix' filter. [06:29:26 CEST] <brx_> The command works fine on ubuntu but fails on ANdroid when using a different version of ffmpeg, here is my pastebin, it shows the 2 versions of ffmpeg that I am using and the error that I get on android when using the same command to mix the files [06:29:28 CEST] <brx_> https://pastebin.com/muSjkQsq [06:35:45 CEST] <tdr> disect the line and see what part fails. start with ffmpeg -hide_banner -filters | grep amix [06:39:10 CEST] <furq> that's not the error you'd get if amix was missing [06:39:38 CEST] <furq> at a guess that's some kind of weird shell issue [06:41:35 CEST] <brx_> i think amix is present, this is the output of tdr's command https://pastebin.com/523U5dY6 [06:41:55 CEST] <brx_> it only gave me 1 page of filters but i see amix in there [06:42:08 CEST] <furq> yeah if the filter was missing it wouldn't put the labels in the error message [06:42:15 CEST] <brx_> ok i see [06:42:15 CEST] <furq> maybe try using doublequotes [06:42:39 CEST] <brx_> you mean for this portion: '[0:a][1:a]amix[aout]' [06:42:42 CEST] <furq> yeah [06:42:45 CEST] <brx_> ok [06:46:13 CEST] <brx_> ok i had to escape the double quotes with the \ operator, so my filter was like: -filter_complex \"[0:a][1:a]amix[aout]\" [06:46:22 CEST] <brx_> and i got the error: [AVFilterGraph @ 0xf196f040] No such filter: '"' [06:46:42 CEST] <furq> uhh [06:46:51 CEST] <furq> i don't know what shell this is but it sounds really broken [06:47:41 CEST] <furq> maybe try dropping the quotes entirely [06:47:45 CEST] <brx_> ok [06:48:30 CEST] <brx_> oooo its rendering a file, looking good... [06:48:39 CEST] <brx_> i had to drop the single quotes from the next portion as well [06:48:49 CEST] <brx_> i.e: -filter_complex [0:a][1:a]amix[aout] -map [aout] [06:49:00 CEST] <brx_> checking the resultant file 1 sec [06:49:57 CEST] <brx_> ok the resultant output file is good! [06:50:11 CEST] <brx_> thanks [07:23:37 CEST] <brx_> building on that previous command, I am trying to 'itsoffset' the mp4 file by a given value. The command runs but the output file doesnt offset the mp4 as I was expecting. Here is the command I am using: https://pastebin.com/ [07:37:07 CEST] <brx_> i just realized i was doing "-itsoffset -00:00:01", the second hyphen shouldnt be there. I removed the second hyphen and the output is the same. I dont get the offsetted mp4 [08:04:57 CEST] <brx_> sorry i posted an empty pastebin, here is the actual pastebin with my command to 'itsoffset' the mp4... https://pastebin.com/jBNpvKuY [09:49:35 CEST] <Beam_Inn> see the issue i keep having with my command is that it basically always seems to give me an 8-second video from the start [09:49:45 CEST] <Beam_Inn> what I'm trying to do is set a start and end time [09:49:51 CEST] <Beam_Inn> ffmpeg -i movie.mp4 -ss 00:00:03 -t 00:00:08 -async 1 cut.mp4 [09:51:12 CEST] <Beam_Inn> is there only a duration tag and no end-time tag? [09:54:24 CEST] <Beam_Inn> ffmpeg -i test.mp4 -c copy -ss 00:09:23 -to 00:25:33 cut.mp4 this seems to work well [11:05:58 CEST] <user_oreloznog> Beam_Inn: Worked fine for me ;) [13:36:40 CEST] <nine_milli> blb [14:50:36 CEST] <JEEB> TIL -channel_layout in ffmpeg.c [14:51:02 CEST] <JEEB> seems to be more useful than -ac in the sense that you are specifying the exact channel layout [14:51:16 CEST] <JEEB> (which then also defines the amount of channels) [19:29:16 CEST] <grkblood13> why is this failing to record the stream coming out of the specified input device? I'm using the azap program to fetch an OTA signal using an ATSC tuner card. https://pastebin.com/qjSajw2f [20:22:57 CEST] <nine_milli> blb [22:02:00 CEST] Last message repeated 1 time(s). [22:02:15 CEST] <another> can someone kick nine_milli please? [22:04:42 CEST] <DHE> seconded [22:05:22 CEST] <another> durandal_1707: ^^ [22:06:33 CEST] <durandal_1707> another: why? [22:07:30 CEST] <another> spamming the same line every couple hours over many many days [22:08:20 CEST] <DHE> suspiciously it matches a user's name, presumably to constantly highlight them [22:29:47 CEST] <furq> he actually had a question about libvpx on the 10th [22:30:34 CEST] <furq> and then since then he's just highlighted that guy 82 times [22:45:30 CEST] <durandal_1707> furq: did that guy "helped" him somehow? [22:45:40 CEST] <furq> not as far as i can see [22:45:47 CEST] <furq> i guess in another channel maybe [00:00:00 CEST] --- Sat Jul 27 2019

1 0

ffmpeg.log.20190729
by burek 15 Aug '19

15 Aug '19

[01:31:59 CEST] <Henry151> heya #ffmpeg ... can you inspect this file: https://romp.network/nextcloud/index.php/s/9tYC5zn5PZsCwbr for me, i am getting these errors: https://bpaste.net/show/oVj5 when using this bash script: https://termbin.com/gglw to try and convert the aiff to flac. [01:32:18 CEST] <Henry151> I am getting the same error on dozens of files, so I suspect it's something other than damaged files. [01:33:06 CEST] <Henry151> specifically, all of the files in this folder: https://romp.network/nextcloud/index.php/s/5ja7cHLXTCKijtF and a bunch of others that are in similarly-titled folders [01:39:24 CEST] <Henry151> it's saying "invalid data found when processing input" ... I don't understand what that might mean [01:39:48 CEST] <Henry151> well, i mean, it sounds like "the file is no good" but there's a whole bunch of them, so i'm really hoping that it's something else [01:41:16 CEST] <furq> i'm guessing that those files aren't supposed to all be 187KB [01:44:46 CEST] <Henry151> ah, definitely they are not [01:45:03 CEST] <furq> 12-Things Are Going My Way.aiff: MacOS Alias file [01:45:14 CEST] <furq> there's your problem [01:45:28 CEST] <Henry151> oh, i get it now [01:45:31 CEST] <Henry151> reading here: https://www.lifewire.com/fpbf-file-2621399 [01:45:55 CEST] <Henry151> i understand now -- and it's OK, because I have the actual files as well for each of these "burn folders" [01:46:04 CEST] <Henry151> thank you [02:57:29 CEST] <^Neo> hello friends, I'm trying to convert a 29.97i video with frame repeats into a 59.94p video with frame repeats using FFmpeg [02:57:41 CEST] <^Neo> usling libx264 [05:09:01 CEST] <nine_milli> blb [06:35:05 CEST] <koz_> I'm trying to use my AMD graphics card to VAAPI code H264. Here is my command and result: http://paste.debian.net/1093430/ What am I missing? [06:35:32 CEST] <koz_> (my device is correct, I have a second GPU in /dev/dri/renderD129) [08:01:18 CEST] <nine_milli> blb [08:07:16 CEST] <Saccarab> would you be able to demux concat up to 100 files with different specific inpoints [08:07:35 CEST] <Saccarab> or is amix a better solution [08:14:20 CEST] <furq> concat should work fine [08:22:19 CEST] <Saccarab> whats the proper way to give inpoint and outpoint by miliseconds [08:22:43 CEST] <Saccarab> I get `Estimating duration from bitrate, this may be inaccurate` error whatever I've tried out, and output ends up being garbage [15:52:35 CEST] <nine_milli> blb [15:55:04 CEST] <DHE> durandal_1707: I think he needs to be kicked harder [15:55:16 CEST] <durandal_1707> lol [15:55:56 CEST] <durandal_1707> did he ever asked why he got kicked? [15:57:00 CEST] <DHE> I have not seem him say any other thing since June 18th (which is also his first appearance in my logs) [16:40:25 CEST] <Adcock> would I feel the effects of DD 5.1 if using earphones? [16:42:38 CEST] <durandal_1707> Adcock: with sofalizer filter and correct parameters to it, maybe [16:53:35 CEST] <lagzilla> Is it possible for ffmpeg to generate a sine wave with multiple frequencies so far I've only found examples like ```ffmpeg -f lavfi -i "sine=frequency=1000:sample_rate=48000:duration=21600" -c:a libmp3lame -q:a 6 test.mp3``` [16:55:17 CEST] <durandal_1707> lagzilla: you can generate multiple frequencies with aevalsrc filter and custom expressions [17:10:12 CEST] <Adcock> What? [17:11:16 CEST] <durandal_1707> Adcock: what info you could not understand? [17:16:34 CEST] <Adcock> sofalizer filter? [17:17:21 CEST] <durandal_1707> Adcock: yes, look at ffmpeg online documentation of all filters [19:01:23 CEST] <Gopa> I have a video of 60 sec, [19:01:23 CEST] <Gopa> I need a video from `10 sec to 20 sec` and from `50 sec to 60 sec`, [19:01:23 CEST] <Gopa> Also I need to rotate the video to `90 degree clock wise`, [19:01:23 CEST] <Gopa> I also need to crop the video as `40:20:100:100` [19:01:23 CEST] <Gopa> I also reffered this https://stackoverflow.com/questions/47054881/ffmpeg-remove-multiple-segment… but I couldnt understand, can any one help me with the command I have to write to get this output ? [19:02:47 CEST] <nine_milli> blb [19:07:38 CEST] <Gopa> I'm building the video-cut-tool, An online opensource tool to cut/trim videos using ffmpeg, So can any one guide me with the above case, to move smoothly with the project :) [19:24:17 CEST] <kepstin> Gopa: to do cuts of video like that, best option is probably to use multiple inputs each with the -ss and -to input option, and the concat filter to join them [19:25:09 CEST] <Gopa> kepstin: yeah I was doing that currently, Is there any new method to do that? [19:25:27 CEST] <kepstin> Gopa: everything else you mentioned is easily done with transpose and crop filters [19:26:53 CEST] <kepstin> Gopa: if you want frame-accurate seeking, that's the best method [19:27:09 CEST] <kepstin> all of these steps can be done together in a single ffmpeg command, of course. [19:27:21 CEST] <Gopa> Can you please help me with the example command? [19:30:34 CEST] <kepstin> `ffmpeg -ss 10 -to 20 -i video.mp4 -ss 50 -to 60 -i video.mp4 -filter_complex 'concat=n=2,transpose=clock,crop=40:20:100:100' output.mp4` [19:31:22 CEST] <kepstin> you'd also want to add options to select video / audio codecs and settings for the output. [19:31:56 CEST] <Gopa> codec ? [19:33:10 CEST] <Gopa> kepstin: Is there any way to see the progress of the ffmpeg computation ? [19:33:29 CEST] <kepstin> Gopa: it prints a status line with progress information on the terminal by default [19:34:44 CEST] <kepstin> if you're doing an api thing... I think there's some way for it to output progress reports in a computer readable format. haven't used it myself [19:35:25 CEST] <Gopa> I'm using ffmpeg version 4.1.3-0york1~16.04 [19:35:25 CEST] <Gopa> I can only see all the computation output , I'm not seeing any progress bar in my terminal [19:36:28 CEST] <kepstin> it's not a progress bar (ffmpeg can't do that do to some technicalities about not knowing the output file length because of how filters work), but it doesn't print the current position in the output [19:36:42 CEST] <kepstin> e.g. "frame= 480 fps=120 q=28.0 Lsize= 507kB time=00:00:19.87 bitrate= 209.0kbits/s speed=4.96x" [19:37:06 CEST] <kepstin> that time=00:00:19.87 bit means that encoded 19.87 seconds worth of output so far [19:37:26 CEST] <kepstin> but it does print* [19:37:58 CEST] <Gopa> Oo :) [19:38:46 CEST] <kepstin> if you know how long the output is supposed to be, then you can calculate the percentage yourself :) [19:39:59 CEST] <Gopa> Is 19.87 the total time to encode the video? [19:40:12 CEST] <kepstin> no, that's the current position in the output video [19:40:53 CEST] <Gopa> so how can I caluclate the percentage with out knowing, how much time it takes to encode the video? [19:41:08 CEST] <kepstin> it has nothing to do with the amount of time that the encode will take [19:41:25 CEST] <kepstin> it's the position in the output file, so it depends on how long the output file will be [19:41:52 CEST] <Gopa> little confused, Thinking about it ;) [19:44:00 CEST] <Gopa> what do you ment by position in the output file? [19:45:15 CEST] <kepstin> if you stop ffmpeg at the moment that it prints time=00:00:19.87, then the file output.mp4 will be 19.87 seconds long when you play it. [19:46:46 CEST] <Gopa> Ahh got it :) [19:46:53 CEST] <Gopa> Thank you kepstin :D [19:49:01 CEST] <Gopa> kepstin: One doubt for consedring the sec [19:49:01 CEST] <Gopa> Is there any specifix fixed formate `hh:mm:ss` or can I directly use `ss`? [19:49:54 CEST] <kepstin> the time format accepted in the -ss and -to options is documented here: https://www.ffmpeg.org/ffmpeg-utils.html#Time-duration [19:50:34 CEST] Action: Gopa going through documentation [20:00:01 CEST] <Gopa> kepstin: Yeah I got it regarding the time, Is there any way extract only the time with out frames, fps, q, Lsize ? [20:00:54 CEST] <kepstin> use a regular expression? Otherwise for an api sort of thing, you might try the -progress option [20:02:40 CEST] <Gopa> Oohkay :) [21:02:51 CEST] <quesker> I ripped a dvd with ffmpeg using -acodec libmp3lame and the audio doesn't play when you play the video with quicktime on a mac. I also tried mp2. same thing [21:03:21 CEST] <quesker> the audio works in vlc though [21:03:42 CEST] <quesker> I want to use qt so I can easily trim the video [21:04:39 CEST] <kepstin> quicktime might have issues with audio codecs other than aac :/ [21:05:13 CEST] <quesker> so try -acodec libaac? I think I have that [21:05:44 CEST] <durandal_1707> only in historic version of ffmpeg for museum [21:05:44 CEST] <kepstin> use `-c:a aac` to use the builtin encoder, or `-c:a libfdk_aac` if you're using a build with that enabled [21:05:45 CEST] <quesker> Unknown encoder 'libaac' or not :) [21:05:53 CEST] <quesker> ok [21:06:39 CEST] <kepstin> and you might want to do your trimming with ffmpeg to skip an extra lossy encoding step :/ [21:06:47 CEST] <quesker> what should I use for vcodec? [21:06:57 CEST] <quesker> ok it is just hard to know where to trim from command line [21:08:00 CEST] <kepstin> libx264 is a reasonable video codec to use if you don't have any strong opinions on the matter. it's the default for mp4 in most ffmpeg builds. [21:08:23 CEST] <quesker> ok thanks [21:08:43 CEST] <quesker> looks like I can use quicktime to see where to trim and then pass that time to ffpmeg [21:11:36 CEST] <kepstin> note that if you want help picking video encoding settings for libx264 in ffmpeg, https://trac.ffmpeg.org/wiki/Encode/H.264 goes over it. [21:11:44 CEST] <quesker> thanks [21:28:56 CEST] <Abbott> I was recording my screen with gdigrab and stopped recording with ^C instead of q [21:29:21 CEST] <Abbott> I did ffmpeg -i output.mkv output1.mkv to get ffmpeg to fix the improperly closed file, but will it re-encode the file or know to just fix the stream? [21:29:43 CEST] <kepstin> i dunno how ctrl-c behaves on windows, but on linux ctrl-c properly stops ffmpeg and closes the file :/ [21:30:11 CEST] <Abbott> I was unable to seek in output.mkv so I figured I messed something up [21:30:13 CEST] <kepstin> Abbott: ffmpeg transcodes by default, use "-c copy" to copy the already encoded stuff [21:30:29 CEST] <Abbott> ok I'll try that [21:31:12 CEST] <kepstin> at least you didn't use mp4, those are broken and difficult to recover if only partly written :) [21:37:57 CEST] <Abbott> okay, so I got the malformed mkv working, now I'm trying to combine the stream with an audio stream and I used: `$ ./ffmpeg -i output.mkv -ss 00:00:15 -to 00:00:25 -i audio.wav -ss 00:00:00.5 -c:v copy -c:a aac -shortest final.mkv` [21:38:18 CEST] <Abbott> so I want to encode a test segment of 10 seconds to see if the audio matches up correctly [21:38:33 CEST] <Abbott> because there's some stuttering at the beginning of the audio and some trash at the beginning of the mkv [21:38:54 CEST] <Abbott> but that gave me a single frame of the video and the sound started not 0.5s in, but much later [21:38:58 CEST] <Abbott> did I do this wrong? [21:40:08 CEST] <kepstin> the -ss and -to options should go *before* the input they apply to [21:41:02 CEST] <kepstin> if you want to encode a short 10s test, I recommand using only the -ss options before the inputs, and then "-t 10" before the output [21:41:20 CEST] <quesker> well that was weird. after ripping and trimming, I tried to upload to youtube. after a few minutes of uploading it said "Invalid Request. This video was rejected" [21:42:59 CEST] <quesker> this is video I shot on mini-DV back in 1999, and there is no music. I just recorded to DVD-R and then ripped that with cat /mnt/dvd/VIDEO_TS/VTS_01_*VOB | ffmpeg -i - -deinterlace -c:a libfdk_aac foo.mp4 [21:43:48 CEST] <quesker> oh. too long [21:44:18 CEST] <quesker> after writing a long story about it which is now apparently lost. :( [21:50:28 CEST] <kepstin> the "-deinterlace" options isn't the recommended way to deinterlace stuff nowadays, if you're gonna do that again I'd recommend explicitly specifying a filter - probably the "bwdif" filter is the best deinterlacer in ffmpeg right now [21:51:58 CEST] <quesker> ok [21:53:06 CEST] <kepstin> (-deinterlace currently inserts a "yadif" filter, which isn't terrible, but isn't the best) [21:54:23 CEST] <Abbott> okay! I think I'm getting closer to what I wanted. Last thing is that my monitor is 1440p I want to scale the video to 1080p, but with -vf scale=1920:1080p, the quality seems to scale, but the video still has a 1440p resolution. Is this what the setsar option is for? [21:55:58 CEST] <kepstin> the scale filter doesn't take a "p", it's just "scale=1920:1080" [21:56:29 CEST] <kepstin> I'm not sure if that's what caused the problem? it might not have parsed the height correctly [22:15:55 CEST] <Abbott> I think that was a problem just with me typing the question [22:16:07 CEST] <Abbott> I think me thinking the video resolution was wrong was actually a problem with my player [22:16:10 CEST] <Abbott> I think scale worked fine [22:16:33 CEST] <Abbott> most answers on the internet always seem to include the setsar=1:1 option with scale, though. Why is that important? [22:18:48 CEST] <kepstin> in some cases you can get rounding errors where the sar gets set to something that's slightly off from 1:1, so the filter fixes that [22:33:07 CEST] <Abbott> thanks for all the help kepstin [00:00:00 CEST] --- Tue Jul 30 2019

1 0