[Ffmpeg-devel-irc] ffmpeg.log.20160312

burek burek021 at gmail.com
Sun Mar 13 02:05:01 CET 2016


[00:00:06 CET] <limbo_> c_14: when you say "throws away", I don't know what you mean.
[00:00:21 CET] <limbo_> I know data is lost, but not how.
[00:00:26 CET] <kepstin> limbo_: multiple things - lowpass filter, perceptual masking, etc.
[00:00:39 CET] <kepstin> limbo_: you really have to get deep into theory (more than I know) to understand it all
[00:01:18 CET] <c_14> psychoacoustic optimizations as well
[00:01:35 CET] <furq> is the max -q for jpeg2000 documented anywhere
[00:01:46 CET] <furq> oh.
[00:02:02 CET] <furq> Stream #0:0: Video: mjpeg, yuvj444p(pc), 1280x720, q=2-31, 200 kb/s, 25 fps, 25 tbn, 25 tbc
[00:02:05 CET] <furq> that was fast
[00:02:53 CET] <kepstin> furq: iirc, that shows the acceptible values for '-qp', which aren't the same as the range for '-q' in all codecs
[00:03:26 CET] <TD-Linux> limbo_, quantization is the primary way to lose information
[00:03:47 CET] Action: kepstin still doesn't know why the crf option for x264 wasn't mapped to -q in ffmpeg, but too late to change that now.
[00:03:52 CET] <growery> llogan thanks a lot
[00:04:31 CET] <llogan> growery: unfortunately it does not yet have much info on videotoolbox
[00:04:49 CET] <limbo_> so, given that. how do I get ffmpeg to encode these files as best I can? I can't listen to all of them 100% after encoding to check for slight lfutrther loss.
[00:04:58 CET] <furq> limbo_: what was your actual problem again
[00:05:15 CET] <furq> i remember it had something to do with a badly stitched together mp3 file which is unplayable somehow
[00:05:52 CET] <kepstin> limbo_: if you need to re-encode the files, and you don't want noticable quality loss, then just set the bitrate higher than you would otherwise need. and avoid re-encoding it multiple times.
[00:06:04 CET] <furq> nvm it's still in my buffer
[00:06:27 CET] <limbo_> I have 7.1 GB of fucked up (probably) 30kbps mp3 files that I should probably re-encode, since more than on of them may be incorrectly made.
[00:06:55 CET] <kepstin> ew, 30kbps. I assume they're all something like mono 16khz voice?
[00:06:57 CET] <limbo_> the only file I tried listening to had seeking issues on an ipod.
[00:07:07 CET] <furq> have you tried remuxing it?
[00:07:19 CET] <furq> failing that, if you need it to work on an ipod then i guess your best bet is reencoding to fdk-aac
[00:07:29 CET] <limbo_> kepstin: I don't know about the frequency, but it's all radio recordings.
[00:07:49 CET] <furq> i would suggest opus or speex for speech but it's not going to work on an ipod
[00:07:59 CET] <limbo_> furq: remuxing how? I tried ffmpeg -codec copy and that was more obviously wrong.
[00:08:06 CET] <furq> that's how i would have suggested
[00:08:11 CET] <limbo_> since it reported a shorter time.
[00:08:11 CET] <kepstin> if they're actually music, than it's almost certain that the bitrate is being shown incorrectly by your tool due to probably missing vbr headers.
[00:08:25 CET] <kepstin> particularly if they're stereo and actually sound ok
[00:08:41 CET] <furq> can ffmpeg actually decode the whole thing
[00:09:21 CET] <limbo_> kepstin: I'm not getting the bitrate from a program, it wasin the description where I got this.
[00:09:30 CET] <kepstin> would be a good thing to check first - just decode some samples to a wav and see if they sound ok
[00:10:15 CET] <furq> if it is mono and low sample rate then 32kbps isn't so bad
[00:11:10 CET] <kepstin> yeah, mono 16kHz would probably be listenable at that. Maybe ok for AM radio recordings?
[00:11:15 CET] <limbo_> furq: I'm gett errors like "[mp3 @ 0xd031e0] overread, skip -5 enddists: -1 -1" occasionally
[00:11:25 CET] <furq> that sounds fun
[00:12:11 CET] <limbo_> How would I quickly check if an encode worked right? since my PC could play them back just fine.
[00:12:25 CET] <furq> fwiw you should probably decode the whole thing to wav anyway, unless your ffmpeg was compiled with fdk-aac
[00:12:47 CET] <furq> doing that and using the standalone fdk encoder will be much less hassle than recompiling ffmpeg
[00:14:09 CET] <limbo_> I'm not touching aac.
[00:14:52 CET] <limbo_> I have 7.1 GB of these things, and I'm not looking to increase that much.
[00:15:53 CET] <furq> aac is more efficient than mp3, especially at low bitrate
[00:17:24 CET] <furq> i wouldn't normally recommend it but it's probably your best bet here
[00:22:21 CET] <growery> llogan i am not getting it do i have to compile or do i have to ask for the videotoolbox hardware acceleration on the encoding command?
[00:35:35 CET] <limbo_> is there any other way to get he smallest mp3 possible without losing a notible amount of quality? not supplying any options leads to a 47MB file increasing to 57MB.
[00:35:57 CET] <TD-Linux> decrease the bitrate or quality setting until it sounds bad
[00:37:21 CET] <pzich> yeah, the problem is "notable" is totally relative
[00:40:26 CET] <limbo_> TD-Linux: the test file I used is 04:04:58 long.
[00:40:44 CET] <TD-Linux> listen to a segment
[00:40:51 CET] <limbo_> file says it's 56kbps though.
[00:56:40 CET] <growery> llogan i see! i am compiling it
[01:23:48 CET] <llogan> growery: once you get it compiled you can encode with "-c:v h264_videotoolbox". i assume you're wanting to encode H.264 and not decode
[01:24:07 CET] <llogan> you can look at the commits that contain the word for some info: http://git.videolan.org/?p=ffmpeg.git&a=search&h=HEAD&st=commit&s=videotoolbox
[01:34:17 CET] <somethingelse> Is there any way to decode frames in file order
[01:34:21 CET] <somethingelse> As opposed to display order?
[01:34:41 CET] <somethingelse> (Without doing a second manual buffering and reordering of the frames)
[01:35:35 CET] <J_Darnley> I know of no option but look at the help, possibly for instances of dts
[01:35:57 CET] <J_Darnley> (not to be confused with the DTS audio format)
[01:37:16 CET] <Lucifer333> hi
[01:37:32 CET] <Lucifer333> i am looking for the (unofficial) discription of dvd-video file layout
[01:37:53 CET] <Lucifer333> does anyone have a link?
[03:46:04 CET] <jookiyaya> i heard that aac encoder that comes ffmpeg sucks, is this true?
[03:50:33 CET] <TD-Linux> it's better than it used to be
[03:50:56 CET] <jookiyaya> which is
[03:50:56 CET] <TD-Linux> the correct solution is to not use aac
[03:51:02 CET] <jookiyaya> then use what
[03:51:04 CET] <kepstin> and it's definitely better than vo-aacenc
[03:51:04 CET] <TD-Linux> but it is still behind qaac
[03:51:21 CET] <jookiyaya> what's qaac ?
[03:51:28 CET] <kepstin> quicktime aac encoder
[03:51:32 CET] <jookiyaya> oh apple one
[03:51:33 CET] <kepstin> (apple itunes encoder)
[03:51:51 CET] <TD-Linux> jookiyaya, opus is currently the best
[03:51:58 CET] <jookiyaya> why can't ffmpeg include qaac
[03:52:04 CET] <TD-Linux> it's proprietary software
[03:52:40 CET] <jookiyaya> opus is better than qaac ?
[03:52:43 CET] <TD-Linux> yes
[03:52:49 CET] <jookiyaya> says who?
[03:52:55 CET] <TD-Linux> listening tests
[03:53:00 CET] <jookiyaya> fair enough
[03:53:11 CET] <jookiyaya> does opus do 5.1 ?
[03:53:19 CET] <TD-Linux> yes
[03:53:25 CET] <furq> jookiyaya: http://listening-test.coresv.net/s/scores_by_tracks_en.png
[03:53:30 CET] <jookiyaya> and can you use it with mp4
[03:53:41 CET] <furq> no
[03:53:54 CET] <TD-Linux> there are now patches for opus in mp4 but I would consider it experimental at best
[03:53:59 CET] <TD-Linux> mkv is your best bet right now
[03:54:03 CET] <furq> wtf is opus 0.1.9
[03:54:11 CET] <jookiyaya> mp3 is better than faac ?
[03:54:11 CET] <TD-Linux> furq, opus-tools
[03:54:17 CET] <furq> ah
[03:54:37 CET] <TD-Linux> jookiyaya, note that mp3 is run at 128kbps in that test
[03:54:39 CET] <furq> i guess that's opus 1.0.3 or one of the 1.1 release candidates then
[03:54:48 CET] <furq> so not too old
[03:55:07 CET] <jookiyaya> TD-Linux true, that's not a fair comparison then
[03:55:43 CET] <TD-Linux> jookiyaya, well it was done because otherwise mp3 would be so far away from the others that it would be meaningless
[03:56:14 CET] <TD-Linux> note that faac is not ffmpeg's aac encoder
[03:56:32 CET] <furq> that test was done before the ffaac overhaul anyway
[03:57:06 CET] <jookiyaya> then where is ffmpeg's aac encoder test?
[03:57:12 CET] <furq> there isn't one
[03:57:18 CET] <furq> perhaps you'd like to organise one
[03:57:52 CET] <TD-Linux> there are some on hydrogenaud.io already
[03:58:02 CET] <furq> the only one i found was one guy comparing fdk and ffaac
[04:45:57 CET] <jookiyaya> furq i don't see fdk one in that graph
[04:50:38 CET] <jookiyaya> <jookiyaya> [18:53] <furq> jookiyaya: http://listening-test.coresv.net/s/scores_by_tracks_en.png
[04:50:38 CET] <jookiyaya> [19:49] <+j45> http://lmgtfy.com/?q=the+cake+is+a+lie
[04:50:38 CET] <jookiyaya> [19:50] <+j45> the listening tests are rigged
[04:50:38 CET] <jookiyaya> [19:50] <jookiyaya> really?
[04:55:18 CET] <TD-Linux> that was a portal joke
[04:56:03 CET] <jookiyaya> <jookiyaya> i heard opus is better than fdk-aac? why doesn't handbrake support that
[04:56:03 CET] <jookiyaya> [19:46] <+j45> because the cake is a lie
[04:56:03 CET] <jookiyaya> [19:46] <jookiyaya> huh
[04:56:03 CET] <jookiyaya> [19:47] <jookiyaya> cake?
[04:56:24 CET] <TD-Linux> you're being trolled
[04:56:58 CET] <jookiyaya> i am not
[04:59:43 CET] <jookiyaya> these are serious people
[05:00:54 CET] <jookiyaya> what aac encoder does google use now?
[05:01:18 CET] <jookiyaya>  vo-aacenc?
[05:16:34 CET] <jookiyaya> [20:07] <bogo_lode> afaict, opus isn't really as good as aac or mp3 in most situations where you'd want to use those.
[05:16:34 CET] <jookiyaya> [20:08] <bogo_lode> the big uptake for opus is voip, because it's designed to work well for speech and has low latency options
[05:16:34 CET] <jookiyaya> [20:08] <+j45> yes, that's what opus was designed for
[05:16:34 CET] <jookiyaya> [20:09] <jookiyaya> why is google starting use opus then
[05:16:34 CET] <jookiyaya> [20:09] <bogo_lode> because google uses tons of stupid shit?
[05:35:19 CET] <TD-Linux> jookiyaya, yeah that's not true, it is true that it was originally designed for low latency
[05:35:26 CET] <TD-Linux> but it ended up outperforming anything else
[05:36:35 CET] <TD-Linux> all the data for the picture posted earlier is at http://listening-test.coresv.net/results.htm
[05:36:58 CET] <jookiyaya> td-linux can you join #handbrake and tell the people that
[05:52:44 CET] <mr_lou> Hello
[05:53:13 CET] <durandal_1707> Hello
[05:54:01 CET] <mr_lou> I'm creating a home-video-album for my family, on a blu-ray disc. Using ffmpeg to encode video-recordings into a compatible h264 video stream and ac3 audio stream, then muxing them together with tsMuxer.
[05:54:28 CET] <mr_lou> At first I used Linear PCM for audio, but since that takes up way too much space compared ot ac3, I switched.
[05:55:16 CET] <mr_lou> But after switching to ac3, my blu-ray players won't play the audio. So I've been searching and searching, trying to figure out whether it's an encoding problem by ffmpeg, or something else.
[05:57:47 CET] <mr_lou> This is the command line: ffmpeg -i Video.mp4 -acodec ac3 -ab 128k -ar 44100 -vn Audio.ac3
[05:58:35 CET] <mr_lou> Spot anything wrong with that?
[06:00:24 CET] <durandal_1707> I dunno, but aren't there special programs for that
[06:03:17 CET] <mr_lou> Well there are certainly other programs yes, like Mencoder and Aften to encode ac3. But since I'm already using ffmpeg for the other steps in the process.....
[06:50:55 CET] <durandal_1707> mr_lou: perhaps they already have working command combinations
[07:02:33 CET] <mr_lou> durandal_1707, They might. I'll try Aften.
[08:19:14 CET] <mr_lou> durandal_1707, Well, that didn't exactly improve things. Aften's ac3 audio results in noise....  not sure which I prefer then. ;-)
[08:22:45 CET] <durandal_1707> mr_lou: handbrake?
[08:28:20 CET] <mr_lou> Trying mencoder now.
[08:36:05 CET] <lukeshu> I'm probably just being stupid, but: with ffmpeg 3.0, I'm using ffmpeg to convert rgb24 streams into multipart jpeg/mjpg, but the color is... off.  If I encode to something else, it seems fine.
[08:37:13 CET] <lukeshu> The command I'm using is: ffmpeg -pixel_format rgb24 -s 640x480 -f rawvideo -i video.rgb24 -q:v 1 -f mpjpg - > video.mjpg
[08:37:53 CET] <lukeshu> If I change it to: ffmpeg -pixel_format rgb24 -s 640x480 -f rawvideo -i video.rgb24 -f video.avi
[08:37:56 CET] <lukeshu> then it works fine
[08:38:40 CET] <lukeshu> Also, with mpjpeg, it prints a warning from swscalar: "deprecated pixel format used, make sure you set range correctly"
[08:39:08 CET] <lukeshu> I'm purplexed, any help would be greatly appreciated.
[08:41:56 CET] <lukeshu> Nevermind, I've figured it out.
[08:42:21 CET] <lukeshu> Apparently ffmpeg's mpjpeg doesn't play nice with stdout instead of a real file.
[08:46:39 CET] <mr_lou> durandal_1707, Mencoder seems to work.
[08:51:46 CET] <lukeshu> I take it back, it seems to actually be a client issue.
[08:54:21 CET] <lukeshu> Any idea why javax.imageio.ImageIO would screw up the colors for an mpjpeg stream comming from ffmpeg, but not other sources.
[08:54:22 CET] <lukeshu> ?
[11:46:57 CET] <edakiri> Where does the information for preset Ultrafast come from?  I find no file with that name.
[11:53:25 CET] <jkqxz> It depends on the encoder.  (For libx264, see <http://git.videolan.org/?p=x264.git;a=blob;f=common/common.c#l190>.)
[13:34:39 CET] <Wader8> hi
[13:36:08 CET] <Wader8> Is there any intermediate guide for converting, or just increasing bitrare, I have FRAPS video, want to put it on youtube, but I need  to make a good intermediate, in order to avoid further degradation when youtube reconverts it, thus I cannot use the settings youtube uses
[13:36:36 CET] <Mavrik> Then upload it as it is?
[13:36:38 CET] <Wader8> I always do it like this with video software, but I have to use ffmpeg for Fraps
[13:36:49 CET] <Mavrik> No conversion means no quality loss :)
[13:37:15 CET] <Wader8> 3 GB will take a long time to upload for 3 minutes of footage
[13:37:20 CET] <Mavrik> Mhm.
[13:37:32 CET] <Mavrik> Then you want to decrease bitrate I think? (You wrote increase)
[13:37:56 CET] <Mavrik> I'd say do losless x264: https://trac.ffmpeg.org/wiki/Encode/H.264
[13:37:59 CET] <Wader8> I don't have that kind of time, I just put it on youtube, embedd into a thread, them im off to other things besides PCs
[13:38:21 CET] <Wader8> Fraps is uncompressed - just using MPEG2 would lower size by 100x
[13:38:45 CET] <Wader8> well that was a off-the-head number
[13:38:56 CET] <Wader8> basically, any codec will decrease size significantly
[13:39:34 CET] <Wader8> what about the B frames thing ?
[13:40:07 CET] <Wader8> I forgot a bit, which ones are which, I-frames are the ones that are full right ?
[13:42:25 CET] <Wader8> Isn't the -preset intended for lowering side ? or does it also do both image quality enhancement ?
[13:42:30 CET] <Wader8> size *
[13:43:45 CET] <Mavrik> Losless :)
[13:43:55 CET] <Mavrik> Do you understand that means that image quality does not change? :)
[13:44:06 CET] <Mavrik> preset just controls size vs. encoding time in this cas
[13:44:07 CET] <Mavrik> *case
[13:44:15 CET] <Mavrik> Also don't forget to encode audio as well
[13:50:58 CET] <Wader8> Well the size thing isn't that needed now, it's not going to make that much of a difference, 500MB or 300MB, the time it takes , the intermediate is going to be deleted after upload, so only use slowest for archiving when size is also important to save space
[13:51:20 CET] <Wader8> So if I use medium, I would still get as good of a quality as in slow ?
[13:51:57 CET] <Wader8> Because I know they do a lot of stuff behind scenes when presets are changed, but I do not know that deep to know if all those things only affect size
[13:56:14 CET] <Wader8> Well actually it's going better than I expected,
[13:56:48 CET] <Wader8> for 1 minute of footage some 60FPS 1080p shots are 5GB large, but on slow it's going 18 FPS so it shouldn't take long at all
[13:57:00 CET] <Wader8> Other videos are in 30 and 15 FPS
[14:21:55 CET] <Mavrik> Wader8, why do you keep asking about quality
[14:22:01 CET] <Mavrik> When we're talking about LOSLESS.
[14:22:09 CET] <Mavrik> No losses = same output as it's the input :P
[14:22:23 CET] <Mavrik> Quality doesn't change :)
[14:22:58 CET] <Wader8> Not sure, I used CRF=16 as I did with some other project when I was archiving old DVD into X265
[14:23:24 CET] <Wader8> I do not believe X264 and X265 have true lossless
[14:23:41 CET] <Wader8> nor I do not need it, never said I wanted lossless, I wanted a good intermediate
[14:24:16 CET] <Wader8> I heard some I-Frame only MPEG2 seems to be looking good, but this was some time ago
[14:28:02 CET] <Wader8> Here, here's the output it made, from 5GB Fraps, to 137 MB x264 at 14K Mbps variabled http://pastebin.com/H28h92mk
[14:28:06 CET] <Mavrik> Wader8, they do have true losless.
[14:28:16 CET] <Mavrik> Which is described in the wiki you posted.
[14:30:03 CET] <Wader8> That's just increasing bitrate I believe, not sure, I didn't heard from community, X264 wasn't their pick, I always heard I-Frame only MPEG2 to be the choice for actual practical use, ignoring intermediate-only codecs
[14:31:08 CET] <Wader8> I think it's something more to it than just increasing bitrate, but heck I'm not even that of an expert on ffmpeg so anyway, I already almost finished with the 4 videos, goign to use this on a monthly basis to come, I just needed some refreshment
[14:31:39 CET] <Wader8> 14000 Mbps probably enough, it's not real 60FPS, the FPS ingame keeps fluctuating
[14:32:36 CET] <Wader8> Well I used that for like 30FPS, if I was doing some serious HD footage obviously I would double that if it was 60FPS, but this is a Simulator game which has simple graphics so no need
[14:33:10 CET] <Wader8> But if I used the super-compression-agressive youtube settings I would get the text all blurred, so that's why I need a good one so the text is clear
[14:33:43 CET] <furq> what youtube settings
[14:34:30 CET] <Mavrik> Jesus I remember you now.
[14:34:41 CET] <Mavrik> You're that moron that already once came here and kept ignoring everything everyone said.
[14:34:44 CET] <Mavrik> Like now.
[14:34:52 CET] Action: Mavrik moves out of the debate.
[14:36:07 CET] <Wader8> furq, the stuff Youtube has officialy on their site, and all the other mainstream people which have videos about "perfect youtube settings for upload", but youtube re-converts that, so you always have to upload a much better version, not give it the crappy version so it makes an even crappier one once it's uploaded, general stupidity out there
[14:37:09 CET] <furq> people on youtube, stupid? never
[14:37:12 CET] <Wader8> Mavrik, yeah, I don't really follow the system that much, and, I've indicated like 2 times I already figured out what I needed, I posted the pastebin of the results, I'm already uploading, not sure what's your problem
[14:37:39 CET] <furq> i assume it's the fact that you inexplicably don't believe that x264 has a lossless mode
[14:40:01 CET] <Wader8> furq, I never wanted lossless, I said this 2 times now, if I'm wrong on X264 then I apologize
[14:59:48 CET] <jokoon> I want to add text on a small video
[14:59:57 CET] <jokoon> is there so free software to do that
[15:00:02 CET] <J_Darnley> ffmpeg?
[15:00:04 CET] <jokoon> I mean a lightweight thing
[15:00:08 CET] <J_Darnley> How small?  Where?
[15:00:09 CET] <jokoon> I mean with an UI
[15:00:21 CET] <J_Darnley> Then no, nothing "light weight"
[15:00:45 CET] <J_Darnley> All will probably use ffmpeg behind the scenes anyway
[15:00:47 CET] <jokoon> can avidemux do it ?
[15:00:56 CET] <J_Darnley> I don't know
[15:01:00 CET] <J_Darnley> Read their manual
[15:23:50 CET] <great> hi!
[15:23:56 CET] <Mavrik> ohai
[15:24:06 CET] <great> i have some questions
[15:24:11 CET] <great> could you help me
[15:24:25 CET] <durandal_1707> nope
[15:25:18 CET] <great> we use ffmpeg to convert our videos but this process is so slow
[15:25:38 CET] <great> we use ssd servers 12 cores etc.
[15:26:21 CET] <great> there is a app called total video converter
[15:26:41 CET] <great> and it converts same video with same settings faster than us
[15:27:08 CET] <BtbN> probably not the same settings.
[15:27:24 CET] <great> yep commands are diffrent
[15:27:34 CET] <great> but output is same
[15:27:52 CET] <ritsuka> define same
[15:28:20 CET] <great> same bitrate and encoder
[15:28:28 CET] <J_Darnley> I wonder whether you disabled assembly or threads when compiling ffmpeg (or other libraries)
[15:29:16 CET] <BtbN> same bitrate doesn't mean anything
[15:29:25 CET] <Mavrik> Yp, same bitrate doesn't mean much.
[15:29:33 CET] <great> i complied it with this https://github.com/rdp/ffmpeg-windows-build-helpers
[15:29:34 CET] <Mavrik> great, easiest way to tell would be to compare ffmpeg parameters :/
[15:29:48 CET] <Mavrik> great, but if you can at least provide your command line and it's output we can help with optimization
[15:29:51 CET] <BtbN> if it uses ultrafast with cbr you get the same bitrate with a notably reduced encoding load, but also much worse quality.
[15:30:01 CET] <great> i can send commands wait
[15:30:52 CET] <J_Darnley> It takes that script an hour to build libx264, fdkaac and ffmpeg?
[15:30:55 CET] <J_Darnley> ONE HOUR?
[15:31:10 CET] <J_Darnley> Goog god
[15:31:58 CET] <great> http://pastebin.com/UhsWiNpY
[15:32:08 CET] <great> it takes more :(
[15:32:11 CET] <BtbN> so probably the author never heard of parallel make.
[15:32:34 CET] <Mavrik> great, hmm, that's some old ffmpeg commandline there in total converter :)
[15:33:04 CET] <Mavrik> great, the biggest hurt I see for you is that you don't pass a "-preset" parameter which controls CPU usage and encoding time
[15:33:16 CET] <Mavrik> that's a "general quality vs. CPU used" setting that implies bunch of other
[15:33:31 CET] <Mavrik> default is -preset "medium", but it ranges from "ultrafast" to "veryslow"
[15:33:48 CET] <great> i guess aisee using ultrafast
[15:34:15 CET] <great> is there a quality diffrence
[15:34:55 CET] <furq> great: run mediainfo on both files and paste the encoding settings entry
[15:35:15 CET] <Mavrik> great, yes, there's a quality difference
[15:35:21 CET] <Mavrik> they pass "-vpre ..." and path to some file
[15:35:25 CET] <Mavrik> which can be whatever :)
[15:35:30 CET] <BtbN> ultrafast degrades h264 to mpeg2 level
[15:35:48 CET] <furq> at the same bitrate then faster presets will obviously look worse
[15:35:55 CET] <furq> otherwise there wouldn't be much point having slower presets
[15:36:05 CET] <Mavrik> Yes.
[15:36:12 CET] <Mavrik> Use "fast" or "faster" perhaps
[15:36:15 CET] <Mavrik> Wouldn't go beyond that
[15:36:45 CET] <Mavrik> But using mediainfo helps too yeah
[15:36:48 CET] <great> in -vpre file
[15:36:51 CET] <great> preset=faster vprofile=baseline
[15:36:57 CET] <furq> well there's your answer
[15:37:04 CET] <Mavrik> mhm
[15:37:12 CET] <Mavrik> Baseline is fast to encode, preset is set to faster :)
[15:37:23 CET] <Mavrik> I strongly advise against encoding to baseline though, looks like arse.
[15:37:32 CET] <Mavrik> Just set preset to faster and you should be set
[15:38:15 CET] <great> yeah thanks
[15:38:32 CET] <great> i will send mediainfo after converts finished
[15:38:39 CET] <furq> there's not really any need now
[15:39:00 CET] <Mavrik> mhm, we got the reason :)
[15:39:48 CET] <great> ok thanks :)
[15:40:21 CET] <furq> i should really clean up my ffmpeg mingw build script
[15:40:28 CET] <furq> since it takes much less than an hour
[16:13:00 CET] <Wader8> why does -i "concat: ||" not work with mp4 anymore
[16:13:09 CET] <Wader8> nor with mpegts
[16:13:15 CET] <Wader8> worked for vob files
[16:13:19 CET] <Wader8> not for AVC/MP4
[16:13:25 CET] <Wader8> i have the same command
[16:14:03 CET] <Wader8> maybe i need to add map 0:0, that's the only difference I have in the command, but the source files don't contain any other streams than 1 video and 1 audio
[16:15:54 CET] <furq> the concat protocol has never worked for mp4
[16:15:57 CET] <furq> it should work for ts though
[16:16:35 CET] <Wader8> the files aren't ts
[16:16:44 CET] <Wader8> im not even sure if AVC can be in TS
[16:16:59 CET] <furq> sure it can
[16:17:16 CET] <Wader8> this just always gets more frustrating by the minute
[16:17:19 CET] <furq> https://trac.ffmpeg.org/wiki/Concatenate#demuxer
[16:17:29 CET] <furq> that's what you want
[16:17:42 CET] <Wader8> i should have finished this 1 and half hours ago to go do other stuff but no im stuck here
[16:21:52 CET] <Wader8> could not write header for output file 0# incorrect codec parameters, invalid argument
[16:21:59 CET] <Wader8> when i use MYLIST option
[16:24:27 CET] <Wader8> anyone please ?
[16:25:11 CET] <Wader8> I don't understand how this doesn't work, what kind of nonsense is this that such a basic thing doesn't work
[16:28:09 CET] <Wader8_> anyway screw this im going to recovert it later buch of nonsense really, concat doesn't work on MP4, i mean, that's like saying you can't drive up a hill with a car, i better get the heck away from the PC before I blow up
[16:29:37 CET] <furq> what a nice young man
[16:33:04 CET] <kepstin> "concat doesn't work on MP4" is more like saying "you can't drive through a tree with a car"
[16:33:49 CET] <kepstin> (concat protocol, at least)
[16:34:08 CET] <furq> what do you mean you can't weld four cars together and go four times as fast
[16:34:18 CET] <Mavrik> Well, he apparently read some forums from 1999
[16:34:26 CET] <Mavrik> And doesn't know the heck he's doing.
[16:43:13 CET] <JEEB> well to be 100% honest concatenating ISOBMFF is less than simple, which is why L-SMASH doesn't yet do it, and why GPAC's mp4box often does ISOBMFF concat completely wrong
[18:02:12 CET] <kubast2> Hey ,how to create .mkv file with empty video track[as long as audio track I'll be adding to the empty file] ,with audio track from .flac file ?
[18:04:45 CET] <J_Darnley> "empty" means what exactly?
[18:04:54 CET] <J_Darnley> You can create a black video
[18:05:05 CET] <J_Darnley> but that is still a video
[18:41:01 CET] <dsl42> i guess he meant ffmpeg -i input.flac -c copy out.mkv .. and of course i love dudes who asked questions and leave a minute later ;)
[19:02:43 CET] <chungy> heh yeah... and the appropriate extension there would really be .mka :P
[19:39:38 CET] <wellen> hello, i have just used ffmpeg on windows cli to get a 720p video to output as just audio m4a file, it took a while eg 1hr as it was 2hrs7mins long... is it not possible to just run a line that quickly outputs only the audio from the video file?
[19:40:26 CET] <c_14> ffmpeg -i video -c copy -map 0:a out.m4a
[19:41:48 CET] <wellen> hey thanks for quick response - i only had the ffmpeg -h to refer to ... is there a good online manual for ffmpeg i can plow through?
[19:42:16 CET] <Mavrik> wellen, wiki tends to describe common operations: https://trac.ffmpeg.org/wiki
[19:42:32 CET] <Mavrik> wellen, and online documentation is more useful in my opnion: https://www.ffmpeg.org/documentation.html
[19:42:33 CET] <J_Darnley> and http://ffmpeg.org/documentation.html
[19:43:09 CET] <wellen> thanks all, sorry for being lazy and just coing here to ask - after watch the cli for over an hour i need to try and make some time back up ! cheers
[19:45:43 CET] <wellen> one last question before i go... what exactly did i get ffmpeg to do for an hr with this command... ffmpeg -i e:\files\wipez400\cfiles\videos\prs2016\720p\720pav.mp4 720pav.m4a
[19:46:19 CET] <wellen> i mean to ask, what exactly was it doing that too 1hr+
[19:48:06 CET] <J_Darnley> Encoding the video
[19:49:08 CET] <wellen> that is what i thought but the end result file 720pav.m4a i would have expected to just be audio ...it is actually the audio and video
[19:49:37 CET] <J_Darnley> Of course, m4a is just mp4 with a different extension.
[19:50:27 CET] <wellen> ah my stupid assumption was that m4a was just audio file format extension am i very wrong on that ?
[19:51:26 CET] <J_Darnley> Sure, apple uses it like that but it is no different to mp4
[19:52:02 CET] <wellen> good to know thanks (and even more reason i should get stuck into the ffmpeg manual no doubt!
[19:52:36 CET] <J_Darnley> If you think ffmpeg is doing something unexpected then please come ask us.
[19:56:03 CET] <wellen> im normally the person who everyone else comes to for tech answers so i try not to ask others and research first, but this time im breaking my own rules.. sorry (as for ffmpeg doing something unexpected, in my case it was pebcak for sure)
[19:59:35 CET] <wellen> found what i need https://trac.ffmpeg.org/wiki/Encode/HighQualityAudio
[20:00:00 CET] <wellen> many thanks all for the help bye
[20:03:38 CET] <wellen> is there a specific reason that the ffmpeg project does not have an official GUI?
[20:04:08 CET] <sfan5> can you imagine the complexity of such a gui?
[20:04:21 CET] <sfan5> there are waaay to many options to be able to pack them into a gui
[20:05:09 CET] <wellen> i can sfan5..i can also imagine a way through decent ui/ux to actually cover it in a decent way
[20:05:39 CET] <sfan5> then go write one
[20:05:43 CET] <dsl42> :)
[20:05:44 CET] <sfan5> maybe it'll become the ffmpeg official gui
[20:06:46 CET] <wellen> has it not been done as it is an intimidating task to everyone? i figured everyone would love a gui but i wondered if it has neveer been done as the original project devs had a reason for it not to be exisitng eg i wondered if they were against ffmpeg project having a gui?
[20:08:21 CET] <dsl42> i dont care about a gui
[20:08:24 CET] <sfan5> ^ same
[20:10:17 CET] <J_Darnley> What toolkit aregoing to to use that will work across all platforms?
[20:10:24 CET] <J_Darnley> *are you going
[20:11:08 CET] <wellen> i am impressed at the collective effort and talents that have gone (and continue to go into) the ffmpeg project and i have nothing against cli tools either, but i bet ffmpeg would be even more widely adopted by users if the project had a gui as well..i wonder if the collective would be interested in persuing a gui effort at some point
[20:11:31 CET] <J_Darnley> Who doesn't use it?
[20:11:56 CET] <J_Darnley> It is the backing to every other a/v program out there
[20:12:41 CET] <wellen> i have noticed various services using ffmpeg and even making their own basic, limited windows applications to carry out single task functions via ffmpeg
[20:16:58 CET] <wellen> so is it accurate to say that an official ffmpeg gui does not exist due to there not being a real need for its own gui?
[20:18:48 CET] <J_Darnley> That is certainly a large part
[20:20:16 CET] <wellen> also, does it not slightly annoy the ffmpeg developers that others just grab ffmpeg and use it tucked away under their programs as though it is their own work?
[20:20:54 CET] <wellen> or are they just happy to provide ffmpeg to anyone and everyones use
[20:20:56 CET] <J_Darnley> As long as they follow the license we don't really mind
[20:20:57 CET] <furq> only if they're violating the license
[20:21:47 CET] <wellen> is that in the legal section?
[20:26:13 CET] <wellen> i have read the legal section on the project website... it just seems the usual standard resonable things and of note is the mini section on patents etc and licensing
[20:30:38 CET] <wellen> would an official gui be a waste of time?
[20:38:59 CET] <wellen> sorry j darnley just noticed your question further up -re toolkit i dont know without giving it serious consideration but im sure there are various options - my first question would be is it worth doing, then the hows
[20:43:02 CET] <wellen> gtk+ possibly although this doesnt bode well... "Dirk Hohndel, co-developer of Subsurface and member of Intel's Open-Source Technology Center, criticized the GTK+ developers for being abrasive and ignoring most community requests"
[20:44:25 CET] <J_Darnley> If you're seriously considering writing an official gui then definitely email ffmpeg-devel
[20:44:34 CET] <J_Darnley> but I don't reckon it would be well recieved
[20:44:50 CET] <J_Darnley> It would be a large maintainace burden
[20:45:45 CET] <J_Darnley> And I would expect people to complain about portability, libraries, static builds...
[20:46:48 CET] <J_Darnley> Since we already link with gdi and other libs on Windows, a win32 specific gui might be an option
[20:46:55 CET] <wellen> i understand why it wouldnt be well received, tbh i'd be in that same camp...but i'm not 'every user' and everyone has different needs
[20:47:04 CET] <J_Darnley> but I think it would be shot down as "not portable"
[20:47:38 CET] <wellen> why would it be not portable? can you elaborate slightly?
[20:48:05 CET] <J_Darnley> How can you run the win32 gui on Linux, Mac, Android?
[20:49:12 CET] <J_Darnley> I mean using native components and native functions
[20:51:12 CET] <wellen> could it not be worked around at idea stage eg wrappers - i would have figured there are various ways/means/methods/workarounds eg if considered at the very early stages
[20:51:48 CET] <J_Darnley> Sure: a cross platform toolkit like gtx or qt
[20:52:26 CET] <wellen> call me simplistic but also if it was kept lean enough, what is stopping there being 1 download containing version of the gui for each platform? or even cross platform consideration at the code stage of the gui?
[20:53:02 CET] <wellen> or is my naiveity showing?
[20:55:28 CET] <J_Darnley> Size would be the main reason, especially if you want a statically linked binary
[20:56:33 CET] <wellen> what if it was all handcoded in c/c++/c# ? wouldnt that free up most of the concerns as then the only cost would be time and ability of the creator right?
[20:59:02 CET] <c_14> I think the biggest problem is that nobody is going to want to maintain it.
[21:01:56 CET] <wellen> while i cant blame anyone for feeling that way - it is funny when you think about it... eg the amount of collaborative effort time and talent that has gone into the cli ffmpeg - kinda shows what can result from continued joint effort by many
[21:02:43 CET] <J_Darnley> I now think you have no idea how complicated a gui really is.
[21:02:45 CET] <ethe> c_14: no one wants to maintain Qt code
[21:02:58 CET] <wellen> are you all user of ffmpeg or do any of you also help out on the project as well?
[21:03:08 CET] <ethe> or code for *every* os either.
[21:03:12 CET] <J_Darnley> Yes, but I'm a shit programmer
[21:03:53 CET] <ethe> wellen: I *try* to, but I have the same problem as J_Darnley
[21:04:09 CET] <wellen> i call bs i bet you are all just being modest haha
[21:04:52 CET] <wellen> as for having no idea how complicated j darnley... most things are... but some like the challenge ;o)
[21:08:23 CET] <wellen> also why does it just have to be 'one person'?
[21:10:42 CET] <ethe> who else is going to make a gui? It doesn't have to be one person. If you want it to happen, then make it happen
[21:11:53 CET] <wellen> for example what if there was a decent effort made to visualise/rough design an official gui whereby everyone was agreed on a few main critieria eg clarity and ease of use
[21:12:20 CET] <ethe> There's no point saying "what if"... there isn't.
[21:12:32 CET] <wellen> i know often its a case of 'too many cooks' and peoples egos can cause extra grief in group projects but everyone can agree on they just want things to be usable right?
[21:15:26 CET] <sfan5> wellen: >clarity and ease of use
[21:15:32 CET] <sfan5> that is a wonderful example
[21:16:22 CET] <wellen> it is...?
[21:16:30 CET] <sfan5> if you want to make it really easy to use you might not want to confuse the user with what a libvpx-vp9 might be and just say VP9 or even just WebM
[21:16:37 CET] <sfan5> but that would go again clarity, wouldn't it?
[21:16:52 CET] <sfan5> s/again/against/
[21:17:09 CET] <wellen> nothing stopping the gui having 2 tabbed pages eg basic and advanced....
[21:17:23 CET] <furq> i think you're going to need more than two tabs
[21:17:24 CET] <sfan5> just deciding on a few main criteria is not going to get a good result
[21:17:30 CET] <sfan5> you need to think about every detail
[21:17:47 CET] <ethe> wellen: why dont you design a draft for the whole gui?
[21:18:01 CET] <furq> `ffmpeg -h full` will give you an idea of the scale of this endeavour
[21:18:11 CET] <wellen> most common choices on basic tab (also default) and all the power user settings on the advanced tab..also can have a main options menu to specify by default if gui loads up iin basic or advanced displaying on startup.. right?
[21:18:12 CET] Action: dsl42 grabs some popcorn
[21:18:23 CET] <sfan5> % ffmpeg -h full 2>/dev/null | wc -l
[21:18:23 CET] <sfan5> 8210
[21:18:33 CET] <sfan5> so we need about 821 tabs?
[21:18:33 CET] <furq> i got 6400 but this is an old version
[21:19:00 CET] <wellen> (butterkist toffee please dsl42)
[21:19:10 CET] <furq> i think it'd be charitable to say that 1000 of those are headers and empty lines
[21:19:45 CET] <sfan5> adding sort -u gives 3987
[21:20:20 CET] <furq> a lot of those will be mapped to multiple codecs with different value ranges though
[21:20:33 CET] <wellen> without being too much of a killjoy - i would be thinking not just the 2 tab approach but also pages eg next and back at bottom right of gui for a wizard type approach
[21:20:42 CET] <sfan5> btw
[21:20:46 CET] <sfan5> have you looked at handbrake?
[21:20:56 CET] <wellen> how come you thought all the options would just be o 2 static pages?
[21:20:57 CET] <sfan5> https://handbrake.fr/img/slides/slide2_lin.jpg
[21:21:06 CET] <sfan5> it's about as good as it gets
[21:21:13 CET] <wellen> yeah i have seen handbrake a while ago
[21:21:20 CET] <sfan5> you can't fit every ffmpeg option into a gui
[21:21:48 CET] <furq> you can, but it'll be some kind of nightmarish vision of a future gone wrong
[21:22:17 CET] <sfan5> like this? https://pbs.twimg.com/media/BylctRxCEAEv-i2.png:orig
[21:22:21 CET] <ethe> wellen: handbrake is great, for users. If a user *really* needs 100 obscure flags, then they can just use the cli (the docs really aren't that hard to understand)
[21:22:26 CET] <furq> just the codec options needs to cover >400 codecs
[21:22:36 CET] <dsl42> :)
[21:22:44 CET] <wellen> just seen that handbrake gui and - it could be made simpler (from a user epxerience pov)
[21:23:09 CET] <sfan5> huh
[21:23:34 CET] <sfan5> if you want to make it that simple you'll need 3 tabs
[21:23:35 CET] <dsl42> handbrake only encodes to x264, afaik
[21:23:43 CET] <furq> it does x265 and mpeg2 as well
[21:23:49 CET] <dsl42> ok
[21:23:54 CET] <furq> and maybe xvid
[21:24:11 CET] <dsl42> not anymore i guess
[21:24:17 CET] <sfan5> vp8 too it seems https://trac.handbrake.fr/wiki/Encoders
[21:24:43 CET] <wellen> what would make it simpler is a wizard screens approach (as well as the already mentioned 2 page basic/advance tab)
[21:24:45 CET] <dsl42> oh ok, i did not use it for a long time
[21:24:47 CET] <furq> mine doesn't do vp8 ;_;
[21:25:16 CET] <ethe> does ffmpeg have CoreAudio's encoders yet? (I guess it would if handbrake has them)
[21:25:35 CET] <furq> handbrake uses some non-libav* stuff
[21:25:43 CET] <sfan5> does ffmpeg have intel quicksync?
[21:25:49 CET] <furq> most notably dvd ifo parsing
[21:25:59 CET] <furq> which is the only reason i have handbrake installed in the first place
[21:26:20 CET] <wellen> btw let me clarify i am thinking usable by users (not power users)
[21:26:22 CET] <c_14> sfan5: yes
[21:26:27 CET] <c_14> But it's a pain.
[21:28:29 CET] <sfan5> fortunately nvenc isn't a pain
[21:29:00 CET] <wellen> power users just 'get' the concept that all options are usable and that is what the advanced tab would be for .... the basic gui tab along with wizards would literally guide average users through parameters getting narrower and more defined as each wizard screen is completed.. this would not only allow average users to get things done easier but more importantly would free up the power users and the project devs from being asked for s
[21:29:04 CET] <sfan5> wellen: you'll not even be able to mention codec names if you want it that easy
[21:29:10 CET] <sfan5> it will just be
[21:29:16 CET] <sfan5> input file [....]
[21:29:21 CET] <sfan5> output file [....]
[21:29:24 CET] <sfan5> format <selectbox>
[21:29:57 CET] <wellen> sfan5 - give users some credit..... most know about common codecs, containers etc
[21:30:14 CET] <dsl42> most do not, that is my experience
[21:30:18 CET] <wellen> in fact sfan i owuld agree on your last thought
[21:30:31 CET] <sfan5> i wouldn't expect "users" to know what a container is
[21:30:36 CET] <dsl42> right
[21:30:52 CET] <sfan5> most people know the "formats" by file extension
[21:31:09 CET] <wellen> the user shouldnt need ot get 'stuck in' if all they want to do is alter the resolution etc as the wizard task would not call for a conversion so the codec would remian the same for the usage
[21:31:28 CET] <sfan5> i don't know what an user is too you
[21:31:40 CET] <sfan5> but i rarely see someone i would call "user" wanting to scale a video
[21:32:05 CET] <c_14> What you need is a natural language processing neural network that tries guessing what the user wants and then just does it.
[21:32:20 CET] <sfan5> like cdecl.org for ffmpeg
[21:32:22 CET] <sfan5> but more advanced :D
[21:33:13 CET] <wellen> c_14 ... we need this guy then.... Derek Smart
[21:35:10 CET] <furq> derek smart is famous for his intuitive guis
[21:38:06 CET] <wellen> intuitive rhymes with guis.... lies?
[22:54:12 CET] <skokkk> I want to convert my full video library to .mp4 with minimal quality loss. My library has avi's, flv's, and a few more..
[22:54:23 CET] <skokkk> What is the recommended way to do this?
[22:54:36 CET] <skokkk> This is for a few thousand videos
[22:55:00 CET] <c_14> You should be able to remux most of them, you'll probably have to reencode the flvs
[22:55:14 CET] <skokkk> remux?
[22:55:23 CET] <J_Darnley> use -c copy where possible
[22:56:15 CET] <skokkk> I'm very new to ffmpeg. What does -c mean?
[22:56:48 CET] <furq> codec
[22:57:24 CET] <BtbN> why convert it? mp4 isn't exactly a good format, and those formats should play almost everywhere.
[22:58:13 CET] <skokkk> not for html5 web on my phone's opera..
[23:55:15 CET] <giusef> Hi all
[23:56:14 CET] <giusef> I want ffserver to stream to clients without re-encoding since the feed is already in the right format (to not overload the server). Is that possible
[23:56:15 CET] <giusef> ?
[23:59:54 CET] <giusef> done :)
[00:00:00 CET] --- Sun Mar 13 2016



More information about the Ffmpeg-devel-irc mailing list