[Ffmpeg-devel-irc] ffmpeg.log.20191113

[burek]

[02:24:25 CET] <Sazpaimon> so I ended up automating the drawbox filter by calculating the width of my text in inkscape on the command line, seems to work pretty well
[02:24:46 CET] <Sazpaimon> took me less time than i assume learning ass subtitles would take
[02:34:40 CET] <ThePendulum> ass subtitles?
[02:40:10 CET] <nicolas17> Advanced SubStation Alpha
[02:40:16 CET] <nicolas17> https://en.wikipedia.org/wiki/SubStation_Alpha
[03:34:33 CET] <Atlenohen> Hello
[03:36:39 CET] <Atlenohen> So I'm working on an older program which ffmpeg API, and it seems to have this distinction between Encoder and Codec, but this "Codec" is only used once while the "encoder" var is mostly used, it uses diferent code but I think it's the same thing
[03:37:11 CET] <Atlenohen> Because it's the encoder ID that ffmpeg, or codec id that needs to be specified right? But not both?
[03:38:28 CET] <Atlenohen> Or, well, I guess H264 codec could be done by multiple encoders, x264, x264_rgb, nvenc_h264, like that's the idea right?
[03:39:26 CET] <Atlenohen> Well the program uses the codec as if it was the encoder or the other way around, it's actually the encoder I need to specify as I want to be specific
[03:40:17 CET] <Atlenohen> It's a bit confusing with the terminology, avcodec->id, av_codec_descriptor_get_by_name etc ... in ffmpeg command line -vcodec is actually encoder name right.
[03:40:27 CET] <Atlenohen> So more like -vencoder
[03:51:24 CET] <Atlenohen> there's just codec_ids in the public headers hmm..
[03:57:17 CET] <Atlenohen> Meh, IDK, I'll have to look better at API, but this is werid, there is no encoder selections just codecs, what is this?!?!
[03:59:33 CET] <Atlenohen>  * The functions avcodec_find_decoder_by_name(), avcodec_find_encoder_by_name(),
[03:59:33 CET] <Atlenohen>  * avcodec_find_decoder() and avcodec_find_encoder() provide an easy way for
[03:59:33 CET] <Atlenohen>  * retrieving a codec.
[04:00:33 CET] <Atlenohen> Yeah I guess I can skip the whole codec id stuff and just go to the encoder then or what, except i have to fill out the codec for the avformat context just for the internals
[04:01:25 CET] <Atlenohen> Sorry for my talk, the point is, whether a user needs to be able to enter both, whether it's redundant, we can always work the codec out of the encoder, but not the other way around is what I wanted to make sure.
[04:02:01 CET] <Atlenohen> So I don't need both options, specify encoder, code will figure out the rest if necessary.
[06:01:34 CET] <kepstin> if you have a user-provided field, you want to use the find_encoder_by_name function, yeah, that'll return the specific encoder implementation (or if they gave a codec name without a builtin encoder, it'll pick a random encoder for that codec)
[14:14:16 CET] <Atlenohen> Hey guys
[14:21:09 CET] <alex```> yo
[14:22:01 CET] <DHE> 'sup
[14:43:53 CET] <Atlenohen> pixfmt changes must go to swsscale?
[14:44:16 CET] <Atlenohen> I'm wondering how much processing power that uses
[14:45:03 CET] <Atlenohen> it goes from BGRA to BGR0 in one case, another case is YUV420
[14:45:52 CET] <furq> bgra and bgr0 are identical
[14:45:58 CET] <furq> bgr0 just ignores the last eight bits
[14:46:11 CET] <furq> idk if swscale even zeroes them out
[14:46:35 CET] <DHE> well it would have to set them to 0xff if going the other way around right?
[14:46:39 CET] <furq> yeah
[14:50:01 CET] <Atlenohen> Currently the program uses source and destination pix_fmt, and then it puts it into sws_context to convert, so the actual conversion happens there then?
[14:50:34 CET] <Atlenohen> However if BGR0 is identical as you said, and just ignores the last part, could it skip the conversion?
[14:52:34 CET] <Atlenohen> Sorry, the source is RGBA and destination is BGR0
[14:53:11 CET] <furq> no because then it has to change the order
[14:53:29 CET] <furq> it's still a quick conversion though, it's just shuffling bytes around
[14:53:41 CET] <furq> rgb to yuv actually involves maths
[14:54:14 CET] <furq> if the source and destination are the same then it just skips the conversion entirely
[14:55:48 CET] <Atlenohen> Oh ok, trying to optimize it as much as possible. I'm wondering if it's possible to throttle the encoding speed, like a limit of how much FPS it can encode at?
[14:56:25 CET] <Atlenohen> So when the queue builds up it would store in memory or even filesystem perhaps?
[15:00:34 CET] <Atlenohen> or at least to assign CPU affinity to not take up all cores,
[15:06:09 CET] <Rhada> Hello,
[15:06:22 CET] <Rhada> I'm facing a issue trying to deliver a dash stream from an hls live input. our partner ask us to not use timed manifest so, i use -use_timeline 0. Unfortunately, with this option, the stream is not playable through any player (bitmovin, akamai, dash.js). After the first segments, the player start to ask for segments too early and after a while there is a huge gap between the actual segment and the one requested resulting in only 404 errors. Am i missing some
[15:06:22 CET] <Rhada> option ? If i set -use_timeline 1 evrything works fine
[15:06:56 CET] <Rhada> Here is the command i use : ffmpeg -re -i playlist.m3u8 -c copy -vtag avc1 -atag mp4a -bsf:a aac_adtstoasc -map 0:p:5 -map 0:p:4 -map 0:p:3 -map 0:p:2 -b:v:0 2500k -b:v:1 1500k -b:v:2 1000k -b:v:3 500k -use_template 1 -use_timeline 0 -utc_timing_url 'https://time.akamai.com/?iso' -seg_duration 2 -window_size 30 -extra_window_size 30 -media_seg_name 'chunk-stream$RepresentationID$-$Number$.$ext$' -remove_at_exit 1 -adaptation_sets 'id=0,streams=v
[15:06:56 CET] <Rhada> id=1,streams=a' -f dash index.mpd
[15:24:11 CET] <Atlenohen> Does FFV1 even need BGR ? could it just be RGB0 ?
[15:30:53 CET] <Atlenohen> Hey guys, ffmpeg does not have RGBA8 pixel format, is this some kind of a equivalent name, like RGBA can be substitute?
[15:32:04 CET] <JEEB> packed RGBA is just RGBA yea
[15:33:27 CET] <Atlenohen> Okay so the number afterward doesn't matter, as long as the letters are the same?
[15:33:51 CET] <JEEB> if there are no numbers it's 8bit generally
[15:34:22 CET] <JEEB> if you don't expect to utilize alpha, or mean that it doesn't matter there's 0RGB and RGB0
[15:34:53 CET] <Atlenohen> Oh, heh that feels kind of a coincidence, it does work as of right now, but I'm reimplementing it from scratch and figuring out whether anything's wrong.
[15:35:53 CET] <Atlenohen> Well the source format needs to be pixed AFAIK first, what the program outputs, so the output is captured, and then the destination fmt for whatever the codecs are used.
[15:44:08 CET] <pabs3> are there subtitle formats that list the audio language and subtitle language for each subtitle item? so that for a film with mostly English audio with some parts Spanish, the video player could show only the subtitles for the Spanish audio?
[15:46:20 CET] <furq> no
[15:46:25 CET] <furq> you would normally just use two sub tracks for that
[15:50:38 CET] <pabs3> why two sub tracks? in my scenario there are only English subtitles but some of the English subtitles don't need to be shown for people who understand English but not Spanish
[15:51:30 CET] <JEEB> some formats have "forced" subtitles for that, but that generally speaking is ignored by open source stuff
[15:52:03 CET] <JEEB> and for different languages/configurations otherwise separate subtitle tracks with different metadata is utilized
[15:53:57 CET] <pabs3> so one sub track would be "English throughout" and another would be "English for the non-English parts", with the latter duplicated within the former? hmm
[15:54:42 CET] <JEEB> f.ex., and the default flag being on one of them depending on the container
[15:57:37 CET] <pabs3> both forced subtitles & dual sub tracks seem a bit hacky :(
[15:57:49 CET] <pabs3> but thanks for the information :)
[15:57:57 CET] <JEEB> forced subs aren't really a thing in open source
[15:58:02 CET] <JEEB> (un)fortunately
[15:58:18 CET] <JEEB> I meant the default flag in matroska for example, which IIRC is actaully supported
[15:58:39 CET] <pabs3> how do they work where they are implemented?
[15:58:53 CET] <JEEB> matroska for some reason also has a forced flag on the stream level, which is something that nothing else does wrt "forced" subs
[15:59:11 CET] <JEEB> pabs3: in DVD and BD forced subpictures are shown whether or not the stream is selected
[15:59:21 CET] <JEEB> open source players generally don't show anything until the stream is selected
[16:01:18 CET] <pabs3> hmm, ok
[16:26:45 CET] <Hello71> okular: Enable DRM
[16:48:46 CET] <Atlenohen> Hey
[16:49:51 CET] <Atlenohen> How does ffmpeg deal with mistmatches in source versus what's initialized. For example if pixel format is initialized in AV Codec Context as RGBA and let's say 10 frames are ok, but suddenly one frame comes in YUV420P, how does FFmpeg deal with such an encounter?
[16:51:11 CET] <JEEB> avframes will.have what
[16:51:18 CET] <JEEB> comes out of decoder
[16:51:36 CET] <JEEB> normal stuff with f.ex. h.264
[16:51:51 CET] <JEEB> resolutipn changes, pixel format...
[16:52:20 CET] <JEEB> you should not expect decoders' context values to be anything but initial vslues
[16:52:43 CET] <Atlenohen> I'm not dealing with decoders
[16:52:56 CET] <Atlenohen> Frames are created by a game program
[16:54:09 CET] <Atlenohen> Sorry, I should have mentioned, this is dealing with ffmpeg's API, programming. Not the command line.
[16:54:13 CET] <DHE> how are you getting them though? does the game directly make use of libav ?
[16:54:35 CET] <Atlenohen> Yeah, the ffmpeg external is integrated into it.
[16:55:16 CET] <Atlenohen> It works like a dumper, capturing the render output, I checked with some who are familiar, they told me while there's
[16:55:32 CET] <JEEB> Atlenohen: AVFrames are what the decoding api pushes out
[16:55:38 CET] <Atlenohen> one single format that is finally output, read from the GPU, that's RGBA8
[16:55:39 CET] <JEEB> so not cli
[16:56:02 CET] <JEEB> and what the encosing api takes in
[16:56:12 CET] <JEEB> so not sure what your comment was about
[16:56:48 CET] <JEEB> i was just explaining that with inputs you should not expect things to.be static and according to.decoder AVCodecContext
[17:00:50 CET] <bzb> when selecting x265 for 2 pass transcoding of a videofile, what is the commandline switch to specify the name of the stats file to use?
[17:00:57 CET] <Atlenohen> I don't see AVFrame in the current implementaiton, it's just av_codec_context->pix_fmt
[17:01:24 CET] <JEEB> the decoder and encoder interfaces give and tkae.in.avframes
[17:01:27 CET] <bzb> I tried using -passlogfile and it doesn't do anything
[17:01:39 CET] <JEEB> if you do decoding or encoding with ffmpeg it is there
[17:01:43 CET] <JEEB> no buts
[17:02:19 CET] <Atlenohen> Well, resolution changes are clear, they mismatch, so it requires stopping the video output and creating a new file, reinitalizing everything with a new resolution.
[17:02:36 CET] <Atlenohen> So I was wondering if pixel format changes would require the same.
[17:03:17 CET] <Atlenohen> Logic tells me pixel format changed while the dump is being actively written/encoded, is a bad idea.
[17:04:02 CET] <JEEB> encoding wise while there could be wncoders that can do that, it is generally probablly a good idea to reinit. also whether that can.nicely happen depends on the formats at play etc
[17:04:45 CET] <JEEB> usually the output encoder format and resolution is the static thing and filters are put into the workflow should cpnversion or scaling etc be needed
[17:05:37 CET] <Atlenohen> Right, thanks.
[17:18:21 CET] <void09> does ffmpeg have some functionality to detect the main dvd video track files ? I want to take screenshots from the whole movie, at once, at different timestamps. How to do this without passing a single .vob file to ffmpeg ?
[17:26:55 CET] <void09> for bluray it's easy as it's one single m2ts file
[17:32:36 CET] <brimestone> Hey guys,  I have a odd requirement.. I need to detect where the first human voice in a video.. is this something ffmpeg can handle?
[17:33:05 CET] <JEEB> void09: it's not always a single
[17:33:22 CET] <JEEB> void09: you need dvdnav/dvdread support and for bluray libbluray
[17:33:30 CET] <JEEB> I think we have something for libbluray but it isn't perfect
[17:33:40 CET] <void09> JEEB: oh really ? I haven't seen blurays with split video files
[17:33:46 CET] <void09> for the main movie, not extras and stuff, that is
[17:34:06 CET] <void09> JEEB: I think I have ffmpeg compiled with everything.. but how would I use it ?
[17:34:24 CET] <JEEB> void09: I have quite a few where the main movie or main playlist consists of multiple m2ts
[17:34:49 CET] <JEEB> void09: there is a libbluray protocol I think
[17:35:04 CET] <JEEB> see `ffmpeg -protocols`
[17:35:07 CET] <JEEB> should have bluray
[17:35:32 CET] <void09> it does
[17:36:36 CET] <JEEB> ffmpeg -v verbose -i "bluray:D:\dvdbdiso\heavens_feel_pt2"
[17:36:40 CET] <JEEB> this worked for me
[17:36:48 CET] <JEEB> (it's not ISO but that's beyond the point)
[17:36:51 CET] <JEEB> it's the root directory
[17:37:13 CET] <JEEB> in which you have the BDMV dir
[17:41:23 CET] <kepstin> but yeah, since ffmpeg doesn't have native dvd support, you'd want to use other tools to extract a single continuous vob for a particular title+angle to feed to ffmpeg.
[17:41:58 CET] <kepstin> (ffmpeg will decode corrupt output if given a vob with multiple angles, that's fun)
[17:45:01 CET] <void09> sounds too complicated :(
[17:47:52 CET] <void09> i just want something that abstracts the different .vob files into one continous video. just interested in the main (biggest) one
[17:52:44 CET] <another> tracks can begin in the middle of a vob and end in the next one
[17:53:44 CET] <kepstin> dvds don't have tracks, they have titlesets, titles, and chapters
[17:53:56 CET] <kepstin> titles and chapters can begin in the middle of a vob, yeah
[17:54:52 CET] <furq> that's very common
[17:55:17 CET] <furq> even with films you'll often have the preroll stuff and the start of the movie in the same vob
[17:55:26 CET] <furq> for tv shows it's more common than not
[17:55:50 CET] <kepstin> really, a titleset can be thought of as a continuous vob file, it's just split into chunks by filesize
[17:56:21 CET] <furq> you also have multiangle stuff to consider which absolutely requires special handling
[17:57:12 CET] <kepstin> anyways, I normally use the 'dvdbackup' backup tool, select a title, chapter range, angle, and then it outputs a single vob file that ffmpeg can read happily.
[17:57:13 CET] <furq> since all the angles are interleaved so you can jump between them instantly
[17:57:19 CET] <kepstin> that's not the only way to do it tho
[17:57:29 CET] <furq> will dvdbackup output to stdout
[17:58:15 CET] <kepstin> not as far as i know
[17:58:36 CET] <furq> can you even make it output a single oversized vob
[17:58:40 CET] <furq> i remember that being the reason i don't use it
[17:59:10 CET] <kepstin> if you select a title or chapter range it does do a single large vob, yeah
[17:59:22 CET] <furq> fun
[17:59:31 CET] <furq> i should mess around with it now that transcode is out of all the good repos
[18:00:01 CET] <kepstin> also, the fact that angles are interleaved is one of the reasons dvds are awful
[18:00:17 CET] <kepstin> because both angles count towards the bitrate limit, so you have to encode each at 1/2 bitrate :)
[18:00:54 CET] <furq> lol really
[18:01:01 CET] <furq> that doesn't make any sense at all
[18:01:05 CET] <furq> so it must be true
[18:01:11 CET] <kepstin> the bitrate limit is based on drive read speed
[18:01:22 CET] <furq> oh i guess that works then
[18:01:44 CET] <furq> i only have one multiangle disc and it's used for something entirely worthless
[18:01:58 CET] <furq> nice to have for testing stuff though
[18:02:09 CET] <kepstin> i have an anime dvd that uses angles to switch between english and japanese text in the opening
[18:02:18 CET] <kepstin> the opening is typically the part that requires the most bitrate
[18:02:23 CET] <kepstin> the result looks *terrible*
[18:03:58 CET] <another> kepstin: i'll look into dvdbackup, thx
[18:04:52 CET] <kepstin> dvdbackup also has a mode to automatically guess the main title, you could give that a try.
[18:11:55 CET] <kepstin> (amusingly, most modern dvd players don't really care about the bitrate/vbv limits since they have lots of ram and fast drives)
[18:17:31 CET] <JEEB> yea, at this point the 1990s limits aren't as much a thing I bet
[18:17:44 CET] <JEEB> maxrate 8mbps and bufsize 1.5 megabit
[18:17:46 CET] <JEEB> ay lmao
[18:17:56 CET] <JEEB> every meg of buffer was precious :D
[18:18:50 CET] <kepstin> and the keyframe interval was about 1/3 second, iirc?
[18:19:12 CET] <kepstin> so the bufsize was basically always just full of the next keyframe
[18:19:37 CET] <JEEB> pretty much
[18:19:51 CET] <JEEB> with the SD resolutions & frame rates it was /really/ easy to go full
[18:19:57 CET] <JEEB> and hit the VBV limits
[18:37:14 CET] <familiyaF> Hey, I have an video 29.97i(59.94 fields), if I check with mediainfo it displays framerate as 29.97 fps and ffprobe displays it as 59.94 fps. Can someone help me understand why the difference.
[18:38:11 CET] <kepstin> which ffprobe field are you looking at?
[18:38:27 CET] <familiyaF>  Video: h264 (Main), yuv420p(tv, top first), 1920x1080 [SAR 1:1 DAR 16:9], 59.94 fps, 59.94 tbr, 1k tbn, 59.94 tbc (default)
[18:39:27 CET] <kepstin> huh. can you pastebin the complete output of ffprobe -show-format -show-streams on that file?
[18:39:39 CET] <familiyaF> sure
[18:40:10 CET] <kepstin> can't say more without knowing about the container, but the "fps" is usually a guess because it's not stored anywhere in many formats
[18:40:33 CET] <furq> -show_format -show_streams
[18:40:36 CET] <kepstin> (also, the values you're looking at are rounded decimals, the exact field rate is 60/1.001)
[18:43:57 CET] <familiyaF> https://pastebin.com/2uWBprWN
[18:48:35 CET] <familiyaF> Media Info https://pastebin.com/hCzgm31M
[18:49:47 CET] <kepstin> hmm, i guess it just hit some quirk of ffmpeg's frame rate guessing. increasing the probe size / duration might help.
[18:50:27 CET] <kepstin> media info is lying to you unless it checked every single frame in the file - mkv doesn't really do "constant" framerate :)
[18:50:50 CET] <kepstin> i guess it's heuristics for guessing the probably framerate work a bit better on this file.
[18:56:43 CET] <familiyaF> Thanks
[18:57:37 CET] <another> kepstin: hmm.. how do i get dvdbackup to output one file?
[18:57:38 CET] <familiyaF> Is it possible for you point me to a good source for understanding the constant and variable frame rates. Also different interlacing mode
[18:58:28 CET] <kepstin> another: it should do that if you specify the -t option to pick a title, i think. It'll still put it in folders that look like a dvd structure, tho :/
[19:02:12 CET] <another> nope
[19:02:29 CET] <another> splits at 1 GiB
[19:02:53 CET] <furq> another: you can use tccat from transcode if your distro still packages it
[19:03:09 CET] <furq> you can pipe that straight into ffmpeg which is nice
[19:03:19 CET] <furq> although you'll need to mux in chapter markers etc separately anyway
[19:06:59 CET] <another> hmm.. i think i can just concat the vob files
[19:07:07 CET] <another> lost chapters though
[20:23:14 CET] <MarchHare> Question: I want to use ffmeg to handle audio from a FIFO. I've managed to start this. Is there an option to tell ffmpeg not to terminate when the FIFO is empty (simply wait for more)? I want to basically have an encoding ffmpeg dumping to the fifo, and a second ffmpeg sending to an icecast server. But I also want to be able to kill the encoding
[20:23:15 CET] <MarchHare> process (and so the slave ffmpeg just sends silence), and to be able to simply dump an audio file into the encoding ffmpeg to go into the fifo
[20:24:35 CET] <MarchHare> Essentially an audio pipeline (1) ffmpeg to mp3 -> FIFO -> (2) ffmpeg sending to icecast.
[20:24:50 CET] <MarchHare> I want to be able to kill and replace 1 as needed while keeping 2 live.
[20:26:25 CET] <MarchHare> There might be easier ways to do it, but I never found them, and this is for a radio station live stream that I set up a machine to encode sound from the line-in and send out to a server
[20:27:24 CET] <kepstin> ffmpeg will wait block when reading froma  fifo that just goes empty, the problem is when the app writing to the fifo closes it, it'll get an eof and exit.
[20:28:15 CET] <kepstin> i don't know any workaround for that.
[20:28:21 CET] <MarchHare> :(
[20:28:43 CET] <MarchHare> Damn. I was hoping to be able to leverage ffmpeg into something bigger and better for this
[20:29:28 CET] <MarchHare> (Actually, just alternate audio for blackouts, I want to do those from our end, rather than scheduling them on the server, since their interface really sucks and it's prone to me making mistakes)
[20:41:42 CET] <kepstin> MarchHare: hmm. if you're using a named fifo, you might actually be able to hack it with the "-stream_loop" input option - that way when ffmpeg sees the input close, it'll reopen it.
[20:42:05 CET] <kepstin> well, i forget if it re-opens or just seeks
[20:42:10 CET] <kepstin> either way, worth a try
[20:43:17 CET] <MarchHare> Let me attempt that. Basically, I want (2) to stay open no matter what.
[20:44:29 CET] <kepstin> ffmpeg doesn't close the output when using the -stream_loop input option.
[20:44:49 CET] <MarchHare> So should I be using that on (1) then?
[20:45:11 CET] <kepstin> on (2)
[20:45:40 CET] <MarchHare> Let me test it.
[20:45:59 CET] <kepstin> what you want is when something stops writing to (closes) the fifo, then the ffmpeg trying to read from the fifo should instead of seeing "hey, an eof, exiting", it should go "hey, an eof, i'm gonna re-read that input"
[20:46:23 CET] <MarchHare> Yeah, that was my first impression of what you meant. too
[20:46:29 CET] <kepstin> with any luck (i haven't tried it), "-stream_loop -1" might do it.
[22:54:39 CET] <jemius> Can I try to repair broken frames somehow?
[23:48:53 CET] <Bugz000> Hey guys i'm a complete newb at ffmpeg, i'm toying around making a little timelapse utility, now i understand ffmpeg has it's own ways of working out "which image comes first", my plan was to work around that (and probably develop a headache from all the possible formats the images could be in), but i just seen that you can pass in a text file with the paths
[23:49:09 CET] <Bugz000> (ctrl + F "imagepaths.txt") https://stackoverflow.com/questions/24961127/how-to-create-a-video-from-images-with-ffmpeg
[23:50:12 CET] <Bugz000> can it accept a complete "hardcoded" file list generated by my utility and use them in the order that they are within the file?
[23:51:07 CET] <Bugz000> so rather than %3d.jpg, it has each unique file name, say 2000 of them
[23:51:37 CET] <Bugz000> if so my life will have just got a hella lot easier :p but i don't want to get carried away just yet!
[23:52:33 CET] <DHE> ffmpeg -h demuxer=image2   # in particular -pattern_type glob
[23:53:15 CET] <DHE> so you can just do "*.jpg" and images will be eaten in asciibetical order
[23:53:35 CET] <DHE> is this workable?
[23:53:59 CET] <Bugz000> https://i.imgur.com/vyMYuWE.png
[23:54:03 CET] <Bugz000> idk you tell me
[23:54:26 CET] <DHE> that's actually reassuring. ISO standard naming conventions are nice because they sort naturally
[23:54:27 CET] <Bugz000> hahaha, if its any consolation this is the order that windows shows the images, and the order that my program picks them up (i don't actively sort them this way)
[23:55:13 CET] <DHE> so... this will probably work just fine
[23:57:08 CET] <Bugz000> could i be a super annoying newb and ask you to throw a more complete example command at me because i am lost here
[23:57:45 CET] <DHE> ffmpeg -pattern_type glob -i "*.jpg" -c:v libx264 -crf 30 output.mp4
[23:57:55 CET] <DHE> quotes around the *.jpg are mandatory
[23:58:14 CET] <Bugz000> ffmpeg -h demuxer=image2 -pattern_type glob -i 'C:/path/to/images/**.png' \ -c:v libx264 -pix_fmt yuv420p out.mp4
[23:58:35 CET] <DHE> I don't know where that backslash came from but it doesn't belong...
[23:58:43 CET] <furq> or the second *
[23:58:54 CET] <DHE> also -h is the help parameter and doesn't belong here
[23:59:05 CET] <Bugz000> rofl SEE WHAT HAPPENS WHEN I TRY TO DO THIS MYSELF
[23:59:10 CET] <Bugz000> okayokay
[23:59:34 CET] <Bugz000> ffmpeg demuxer=image2 -pattern_type glob -i 'C:/path/to/images/*.png' -c:v libx264 -pix_fmt yuv420p out.mp4
[23:59:59 CET] <DHE> ffmpeg -pattern_type glob -i 'C:/path/to/images/*.png' -c:v libx264 -pix_fmt yuv420p out.mp4
[00:00:00 CET] --- Thu Nov 14 2019