[Ffmpeg-devel-irc] ffmpeg.log.20160702

burek burek021 at gmail.com
Sun Jul 3 02:05:02 CEST 2016


[00:46:30 CEST] <lerner> cropdetect returns:  x1:0 x2:719 y1:0 y2:575 w:720 h:576 x:0 y:0 pts:1500440 t:1500.440000 crop=720:576:0:0 , but I have no idea how to use this info to create a new command I can use
[00:47:20 CEST] <furq> crop=720:576:0:0
[00:47:26 CEST] <furq> that's the crop command you should use
[00:47:47 CEST] <furq> or that you would normally use, but it looks like cropdetect hasn't detected anything
[00:48:07 CEST] <furq> you probably want to adjust the limit and round values of cropdetect
[00:48:20 CEST] <furq> https://ffmpeg.org/ffmpeg-filters.html#cropdetect
[00:48:36 CEST] <lerner> furq, why didnt it detect anything? if I get that output it means it detected something...
[00:48:58 CEST] <furq> it means it thinks no cropping is needed
[00:49:08 CEST] <furq> if you think it's wrong then adjust the params
[00:50:30 CEST] <furq> e.g. -vf cropdetect=32:2
[01:05:55 CEST] <lerner> I wonder if Im the only non native speaker finding it painfully slow to learn to run ffmpeg just by reading the manpages
[01:08:45 CEST] <DHE> there is a lot of technical information involved...
[01:09:36 CEST] <friki2015> about the topic: i think ffmpeg 3.0.2 in fact is already in debian stable backports and next stable version :)
[01:09:48 CEST] <lerner> ffplay -i YourMovie.mp4 -vf "cropdetect=24:16:0" did play the file as I want it. On the cli I found this info: crop=720:576:0:0. It looks ok, so I proceed to ffmpeg -i YourMovie.mp4 -vf "crop=720:576:0:0" YourCroppedMovie.mp4 , but the output is not cropped...
[01:10:04 CEST] <lerner> from: https://superuser.com/questions/772795/is-it-possible-to-autocrop-black-borders-of-a-video-with-ffmpeg
[01:10:38 CEST] <DHE> you have to seek into the video to where it it can see the black border around the video
[01:10:57 CEST] <lerner> o...
[01:12:04 CEST] <lerner> so ffplay -i YourMovie.mp4 -vf "cropdetect=24:16:2000 (2000 frames = 2000/24=83 seconds
[01:12:09 CEST] <lerner> right?
[01:12:34 CEST] <DHE> I just leave it at defaults, then seek deep into the movie. maybe half way
[01:15:41 CEST] <lerner> on ffplay there is no positon bar, but by randomly klickling on left or right ethe movie advances or goes back ...
[01:15:44 CEST] <lerner> is that normal?
[01:21:54 CEST] <DHE> iirc the whole screen is a seek bar. click the far left to seek to the start, click the far right to seek to the end.
[01:26:02 CEST] <friki2015> i'm sending a video myself via rtmp (from ffmpeg to nginx). There are missing frames at the nginx recordings. A local recording shows no problems (-tee "/local.flv|rtmp://")
[01:26:40 CEST] <friki2015> same problem avoiding "-f tee"
[02:23:59 CEST] <Spring> with the seek timecode syntax if a user enters just '300' for a time will that be interpreted as '300' seconds, seeing as double digits are accepted. Or does it have to be explicitly written as minutes?
[02:24:35 CEST] <lerner> It sucks not being a native speaker, I struggle to make me understood
[02:25:18 CEST] <llogan> Spring: it can accept seconds or 01:23:45.67
[02:26:01 CEST] <Spring> llogan, so an input of 300 would be understood by ffmpeg as 5 minutes?
[02:26:19 CEST] <Spring> good to know
[02:26:35 CEST] <llogan> 300 or 00:05:00 would be the same
[02:27:19 CEST] <Spring> Excellent. Was wondering whether I'd need to do some calculations on it (which I may anyway).
[02:37:33 CEST] <CoJaBo> furq: ..out of curiosity, where is -2 documented? I've seen -1:180 as a possible value, but never -2...
[02:39:56 CEST] <friki2015> lerner: "left/right [keys]: Seek backward/forward 10 seconds."
[02:43:03 CEST] <furq> CoJaBo: https://ffmpeg.org/ffmpeg-filters.html#scale-1
[02:43:19 CEST] <furq> If one of the values is -n with n > 1, the scale filter will also use a value that maintains the aspect ratio of the input image, calculated from the other specified dimension. After that it will, however, make sure that the calculated dimension is divisible by n and adjust the value if necessary.
[02:45:43 CEST] <CoJaBo> furq: Huh. Dunno how I missed that
[02:52:03 CEST] <kingsob> is it possible to select the first frame, and use it as an overlay for the rest of the video?
[03:00:33 CEST] <kingsob> ffmpeg -i in.mp4 -filter_complex "[0] select=1 [1]; [1] scale=320:240 [1]; [0] [1] overlay=W-w-5:50" output.mp4
[03:01:04 CEST] <kingsob> this is almost what I want, except it not just the first frame
[03:14:21 CEST] <friki2015> kingsob: never done that but i gess you can extract the first frame and, as a second step,watermark the video
[04:04:30 CEST] <llogan> would have been -filter_complex "[0:v]select=eq'(n,1)',scale=iw/2:-1[fg];[0:v][fg]overlay
[07:32:42 CEST] <Matt___> Hi trying to convert DTS 5.1 to aac stereo and ac3 5.1.  For some reason the aac keeps coming out as surround 5.1.  The ac3 works fine.  command line here:
[07:33:02 CEST] <Matt___>  ffmpeg -v debug -i Deadpool\ 2016\ 1080p\ BluRay\ x264\ DTS-JYK.mkv -t 00:00:10 -map_metadata 0:g -map 0:0  -map 0:1 -map 0:1 -c:v copy -c:a:0 aac -ac 2 -af "pan=stereo|FL=FC+0.6FL+0.2BL|FR=FC+0.6FR+0.2BR" -c:a:1 ac3 -ac 6 -movflags +faststart deadpool.mp4
[07:33:44 CEST] <furq> Matt___: -ac:0 2 -ac:1 6
[07:33:53 CEST] <furq> or maybe -ac:a:0, i forget
[07:34:22 CEST] <Matt___> cmd output here:https://gist.github.com/snoby/701cceb1eaf531337618f29ba1917103
[07:35:53 CEST] <Matt___> Here is the crazy part. If I leave out the -c:a:1 basically the part of the cmdline that does the ac3 transcode the aac portion works fine
[07:36:14 CEST] <Matt___> I'm using ffmpeg 3.0.2
[07:39:59 CEST] <Matt___> Here it is working with just the ac3 portion removed: http://pastebin.com/8FNzL5Jn
[08:13:33 CEST] <Matt___> ok I just posted the question up to the ffmpeg forum, hopefully i can get some insight into why this is happening
[10:08:40 CEST] <saucecode> How do I enable everything in ./configure, and is doing that a good idea?
[10:27:22 CEST] <bollocks_k> it doesn't look like you can do it with one flag.  run ./configure --help to get a list of things you can enable.  that's a lot of dependencies though
[10:33:50 CEST] <saucecode> Yeah, I skipped a lot of them.
[10:34:22 CEST] <saucecode> I don't think I'll need to convert x265 to VHS today :P
[10:35:13 CEST] <bollocks_k> i came across some weird google video format the other day.  it is no longer supported and i don't know of anything that will convert it
[10:35:52 CEST] <saucecode> Not VP6?
[10:36:37 CEST] <bollocks_k> i don't think so.  i'll try to find the file again
[10:37:02 CEST] <saucecode> Idk if VP6 is a thing - I'm just watching $ make run down.
[10:41:07 CEST] <bollocks_k> there it is
[10:41:09 CEST] <bollocks_k> *.gvi
[10:41:14 CEST] <saucecode> Lol
[10:41:58 CEST] <saucecode> So I'm trying to compile VLC, and its complaining about libswscale.a and libavformat.a.
[10:42:13 CEST] <saucecode> Which I've just compiled/installed with the latest snapshot.
[10:42:29 CEST] <saucecode> /usr/bin/ld: /usr/local/lib/libswscale.a(swscale.o): relocation R_X86_64_32 against `.rodata.str1.1' can not be used when making a shared object; recompile with -fPIC
[10:42:33 CEST] <saucecode> Thoughts?
[10:42:54 CEST] <bollocks_k> make sure you ./configure to make shared libraries
[10:43:07 CEST] <bollocks_k> and ./configure with the pic option
[10:43:34 CEST] <saucecode> ffmpeg's ./configure ?
[10:43:39 CEST] <bollocks_k> --enable-pic --enable-shared
[10:43:40 CEST] <bollocks_k> yeah
[10:43:44 CEST] <saucecode> Oh I see it.
[10:43:47 CEST] <saucecode> Ty
[10:44:52 CEST] <bollocks_k> https://paste.fedoraproject.org/387363/74490421
[10:45:38 CEST] <bollocks_k> i think i got a similar error before i added the --enable-shared flag
[10:47:05 CEST] <bollocks_k> i built ffmpeg with every codec my distro has and built mpv against it and it has no idea what to think about a *.gvi file
[10:47:14 CEST] <saucecode> Lol
[10:47:19 CEST] <saucecode> ffprobe say anything?
[10:47:54 CEST] <bollocks_k> i'll try
[10:48:34 CEST] <bollocks_k> hmm says divx
[10:48:43 CEST] <bollocks_k> i wonder what the confusion is
[10:49:11 CEST] Action: bollocks_k tries to transcode it
[10:50:09 CEST] <bollocks_k> seems to be working
[13:40:16 CEST] <Spring> question about time codes (again): I just tested and surprisingly ffmpeg recognized 2:34 as being 2 minutes and 34 seconds
[13:40:38 CEST] <Spring> the docs don't indicate that this format is acceptable but it is apparently
[13:41:11 CEST] <Spring> what happens when the video is say 3 hours, would it consider it then to be 2 hours and 34 minutes?
[13:47:24 CEST] <Spring> also I don't understand the 'No bitrate set, defaulting to 96kbps' message during encoding. I've set it to -qscale: 2, shouldn't that tell it the desired bitrate?
[13:51:04 CEST] <ubitux> Spring: the parsing of the time is not related to the target duration
[13:51:11 CEST] <ubitux> it's based solely on its format
[13:51:32 CEST] <ubitux> 2:34 will always be 2 minutes 34 sec
[13:52:01 CEST] <ubitux> qscale is a "quality" factor as opposed to a bitrate
[13:52:09 CEST] <ubitux> it generally implies variable bitrate
[13:52:13 CEST] <ubitux> if supported
[13:52:40 CEST] <ubitux> but i feel like this is an audio bitrate, so probably you want qscale:a ?
[13:58:04 CEST] <Spring> ubitux, yeah I actually use qscale:a but forgot the :a when posting here
[13:58:20 CEST] <Spring> thanks for the confirmation about the timecode syntax
[15:26:52 CEST] <UbuntuDude> I'm trying to reduce the size of mp4 video files using the command : ffmpeg -i interface.mp4 -b 1000000 -strict -2 output.mp4 ended up 59MB instead of 36MB originally
[15:27:20 CEST] <UbuntuDude> is there is a fixed formula to reduce size of videos using ffmpeg?
[15:29:53 CEST] <DHE> well, you didn't specify audio settings.. check the output of ffmpeg to see what it actually did to your audio and video
[15:41:23 CEST] <UbuntuDude> DHE: what is the check command?
[16:03:10 CEST] <UbuntuDude> yes checked and ACC codec was not changed ... ffmpeg -i output.avi -acodec mp3 -vcodec copy output2.mp4 ... remains the same
[16:18:10 CEST] <yagiza> Still can't win my fight with iLBC. Any ideas?
[16:33:27 CEST] <IHAVENONICK> Hi. What's the preferred way to convert a video to a fast forward one? I'm going to discard audio anyway, so increasing fps is just fine (after that I'll re-process it and adjust fps to 60 or 30).
[16:33:30 CEST] <IHAVENONICK> Oh.
[16:33:33 CEST] <IHAVENONICK> Dang.-
[16:33:42 CEST] Action: IHAVENONICK jumps.
[16:41:08 CEST] <Zucca> Boom.,
[16:51:24 CEST] <sor_> is there a way to set the window size without and degradation in the file. scale= always lowers the quality, I just want the same file quality but have it open in a smaller window
[16:54:47 CEST] <Zucca> sor_: So you want the video file to tell the player that it should show smaller video?
[16:55:04 CEST] <sor_> Zucca, yes
[16:55:11 CEST] <Zucca> Afaik, you always lose quality when scaling.
[16:55:19 CEST] <Zucca> Even if it's done by the player.
[16:55:45 CEST] <DHE> maybe select a different scaler than the default?
[16:56:42 CEST] <sor_> Zucca, when I resize  with the player I can adjust it and it maintains quality... I just want it to open like that
[16:57:01 CEST] <sor_> DHE what do you mean
[16:57:37 CEST] <Zucca> sor_: Idk, but if there's a way to fake the dpi (tell higher dpi that what it is) that might be the way. BUT will the player actually show the maller window is only up to the player.
[16:58:15 CEST] <Zucca> I remember I once and a image program that read the dpi values from jpeg images and zoomed them to match the actual size.
[16:58:25 CEST] <Zucca> *once had
[16:59:00 CEST] <Zucca> But other image programs only shoved the image in native resolution (pixel for pixel).
[16:59:09 CEST] <sor_> I want to adjust it just like when you do with the player
[16:59:24 CEST] <sor_> metadata perhaps?
[16:59:50 CEST] <Zucca> sor_: You are asking for a feature that not many players (if none at all) can conform.
[17:02:11 CEST] <sor_> Zucca, players? all player I now can resize a window
[17:02:46 CEST] <Zucca> sor_: But can they resize the window _automatically_ based on some metadata on video?
[17:02:59 CEST] <Zucca> I don't think so.
[17:03:24 CEST] <sor_> Zucca, that was just a guess. I don't know how they do it... that is my question
[17:04:21 CEST] <sor_> Zucca, do you know how to change the metadate size? perhaps I shall try
[17:05:01 CEST] <Zucca> sor_: Metadata depend on container format I guess.
[17:05:20 CEST] <Zucca> You maybe should just wait for someone more exprienced to answer.
[17:05:45 CEST] <sor_> Zucca, ok thanks
[17:06:24 CEST] <sor_> -metadat key=option I just don't know what the key is
[18:15:32 CEST] <Polochon_street> Hi! I'm experiencing some crashes using ffmpeg's rdft, and I was wondering if I was doing something wrong. The relevant code is http://pastebin.com/Jx3rJ7G0 here, and the backtrace thre http://pastebin.com/jHMsGck8 . Is there some problem with proper memory allocation?
[18:40:37 CEST] <EugenA> built ffmpeg with decklink support, but still cannot capture from blackmagic intensity pro 4k
[18:40:41 CEST] <EugenA> http://pastebin.com/7JZuLFJc
[18:43:24 CEST] <EugenA> I try to capture: http://pastebin.com/1nVeMs1U
[18:46:35 CEST] <EugenA> oh, decklink device is in the list http://pastebin.com/vJAAFZew
[18:47:29 CEST] <JEEB> well yeah, your error is coming from the decklink module so it kind of must be around
[18:52:28 CEST] <ChazNunz> Hello - wondering if it's possible to create a rectangle (using drawbox) that increases in size during playback.  Trying to hardcode a "progress bar" to a video.  Even though drawbox supports timeline editing, i can't figure out how to change the box shape.  Any hints?
[18:55:04 CEST] <EugenA> JEEB: so what can I do?
[18:55:31 CEST] <EugenA> JEEB: but media express is working and I can capture video with it
[18:55:36 CEST] <JEEB> I have nfi, never used that module nor I know if you have just failed at enabling something in the driver or something like that. the message is all that I know as well :P
[18:56:13 CEST] <JEEB> so you can of course go look at the code if you want to. also testing building current master is a good idea as well (in general if FATE is green master is a'OK to use)
[19:03:07 CEST] <EugenA> I'll try to rebuild
[19:36:52 CEST] <twohot> Hi all.  Trying to rebuild ffmpeg from SRPM.  I get this error: opencv not found using pkg-config .... meanwhile opencv and opencv-devel are installed.  Why is this happening?  Is there a way to resolve this?
[19:41:46 CEST] <JEEB> twohot: see what it checks for and the result @ config.log
[19:50:12 CEST] <twohot> JEEB: https://paste.fedoraproject.org/387434/48178914/
[19:52:25 CEST] <JEEB> twohot: so yeah - opencv is there but not with the features FFmpeg uses
[19:52:32 CEST] <JEEB> so that's rather simple why it fails
[19:52:49 CEST] <twohot> JEEB: looks like it is expecting version 2 whereas I have 3
[19:53:19 CEST] <JEEB> most probably yes
[19:54:12 CEST] <JEEB> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=configure;h=007c953fca6359d9f5ca4f10332d75504926b6dd;hb=HEAD#l1828
[19:54:17 CEST] <JEEB> still requires it with current master
[19:54:58 CEST] <twohot> ouch, running bleeding edge is fun
[21:04:32 CEST] <TAFB> what's the maximum size for rtbufsize?
[21:54:20 CEST] <dm_nan> g
[22:35:33 CEST] <TAFB> it uses 1.5x the bitrate for half the fps :( lol.
[22:36:24 CEST] <TAFB> can anyone help me fix "[dshow @ 0000000002392cc0] real-time buffer [Decklink Video Capture] [video input] too full or near too full (121% of size: 3041280 [rtbufsize parameter])! frame dropped!" ???
[22:52:23 CEST] <DHE> pick faster encoding options. you can't keep up with the live data. or expand the buffer?
[22:53:52 CEST] <TAFB> i'm using the fastest encoding option I think
[22:54:28 CEST] <BtbN> Get a better CPU then.
[22:54:39 CEST] <furq> isn't that the default rtbufsize
[22:54:59 CEST] <TAFB> I've got a pretty good cpu
[22:55:14 CEST] <furq> if that's 3041280 bytes then that's pretty small
[22:55:21 CEST] <TAFB> i've tried tons of different rtbufsizes, nothing seems to change :(
[22:55:28 CEST] <furq> pastebin the command
[22:55:30 CEST] <TAFB> G:\>ffmpeg -y -f dshow -video_size 1280x720 -framerate 60 -i video="Decklink Video Capture":audio="Decklink Audio Capture" -rtbufsize 1500M -c:v libx264 -pix_fmt yuv420p -preset ultrafast -crf 0 -c:a aac -b:a 384k -r 30 -f mpegts udp://192.168.1.97:1234
[22:55:59 CEST] <BtbN> Buffering 1.5GB of data seems like a bad idea to me.
[22:56:13 CEST] <TAFB> 96gb ram, I can spare a bit :D
[22:56:25 CEST] <BtbN> That will easily cause tons of delay.
[22:56:40 CEST] <TAFB> latency you mean? not an issue, anything shorter than 10 mins is fine
[22:57:07 CEST] <BtbN> But anyway, if your CPU can't keep up with ultrafast, it has to be busy doing something else.
[22:57:28 CEST] <TAFB> when I'm running that command line, cpu is under 2.7% load
[23:01:05 CEST] <ChocolateArmpits> TAFB: rtbufsize is an input option it must go before the input, not after
[23:01:18 CEST] <TAFB> ahhhhhhh
[23:01:29 CEST] <TAFB> what would be a good size to try?
[23:02:24 CEST] <ChocolateArmpits> What format is the input ?
[23:02:31 CEST] <ChocolateArmpits> as in rgb ?
[23:02:31 CEST] <TAFB> raw
[23:02:34 CEST] <TAFB> yeah
[23:03:35 CEST] <ChocolateArmpits> ok so a single second takes up 160 megabytes
[23:04:09 CEST] <ChocolateArmpits> idk maybe try multiplying the previous value by ten
[23:04:22 CEST] <ChocolateArmpits> so it would be 30 megabytes
[23:04:23 CEST] <ChocolateArmpits> 30M
[23:04:34 CEST] <TAFB> k. i'll try 30M, thanks!
[23:04:57 CEST] <ChocolateArmpits> I'm just guessing though, don't have experience with devices
[23:05:49 CEST] <TAFB> testing now, looking really good so far
[23:09:16 CEST] <TAFB> looks really really good!! thanks for the help :) http://tafb.xxx/rtbufsize_30M.png
[23:09:53 CEST] <TAFB> i still don't know why it uses 1.5x the bitrate at 30fps instead of 60fps but not really important, lol
[23:10:32 CEST] <furq> are you planning on encoding a million channels or something
[23:10:52 CEST] <furq> because otherwise you shouldn't use ultrafast, it's the lowest quality by some distance
[23:11:50 CEST] <furq> oh nvm this is lossless
[23:11:58 CEST] <TAFB> I was told that with "-crf 0" it doesn't matter what your use "ultrafast" and "veryslow" seem to be identical
[23:12:06 CEST] <furq> yeah i didn't notice that
[23:12:15 CEST] <furq> slower presets should still compress better though
[23:12:25 CEST] <TAFB> i'll give it a try, I got lots of cpu to spare
[23:12:59 CEST] <TAFB> if the input is 60fps but I use -r 30 on the output does it still have to encode all 60fps or no?
[23:13:05 CEST] <furq> no
[23:14:29 CEST] <TAFB> nice. with the preset "slow" it uses 63mbps vs. ultrafast at 87mbps
[23:14:38 CEST] <TAFB> it looks like ffmpeg is only using 8 out of my 24 cores??
[23:14:53 CEST] <c_14> it should use all of them
[23:14:57 CEST] <c_14> try setting -threads 24?
[23:15:01 CEST] <TAFB> k. 1 sec.
[23:15:17 CEST] <BtbN> it defaults to your number of virtual cores, there's no point in doing that.
[23:15:39 CEST] <TAFB> i have 12 real cores (24 including virtual)
[23:15:43 CEST] <TAFB> so use -threads 12 then?
[23:15:56 CEST] <furq> it doesn't matter if it's using all your cores if you're limited by input framerate
[23:16:12 CEST] <furq> also with 24 ht threads it'll default to -threads 36
[23:16:40 CEST] <TAFB> but maybe I can use slower or slowest instead of slow if I can use all 12/24 :)
[23:16:54 CEST] <furq> it should use them if they're needed
[23:17:01 CEST] <BtbN> you can use whatever you want with that many cores.
[23:17:04 CEST] <c_14> TAFB: when you said it's not using all your cores, does it have threads for them and they're not utilized or does it not have threads for them?
[23:17:05 CEST] <furq> try with -preset placebo and see what happens
[23:17:25 CEST] <TAFB> on the cpu graph, only 8 cores are at about 80% utilization, the rest are idle.
[23:17:39 CEST] <furq> yeah that doesn't really mean anything if the output isn't lagging
[23:17:41 CEST] <c_14> TAFB: check something like htop
[23:17:43 CEST] <furq> you're limited by input framerate
[23:18:00 CEST] <TAFB> lots of dropped frames with "placebo" only using 8 cores.
[23:18:01 CEST] <c_14> something that'll list the threads and their individual usage
[23:18:22 CEST] <c_14> you could switch to slice threading (I think frame is default)
[23:18:37 CEST] <TAFB> c_14: I'm on windows :D
[23:19:00 CEST] <furq> oh yeah i guess frame threading with 36 threads isn't great with a realtime source
[23:19:41 CEST] <c_14> Wait, I don't think x264 has slice threading
[23:19:47 CEST] <c_14> just frame and delay (whatever delay does)
[23:20:03 CEST] <furq> it does
[23:20:14 CEST] <furq> -x264opts sliced-threads
[23:20:51 CEST] <furq> it's less efficient though
[23:20:53 CEST] <c_14> Ah, it's just not listed then
[23:21:06 CEST] <c_14> Probably because the wrapper doesn't handle it
[23:21:30 CEST] <TAFB> lots of dropped frames with sliced-threads but it is using all cores :)
[23:22:00 CEST] <furq> i wonder if it would use more cores with a bigger rtbufsize
[23:22:05 CEST] <c_14> Nah, the wrapper seems to support it
[23:22:07 CEST] <furq> with frame threading
[23:25:01 CEST] <TAFB> working nice with the slow preset, without sliced-threads. Guess I'll just have it run on 8 cores if it does it without issue.
[23:26:56 CEST] <TAFB> hmmm, cpu getting a little toasty, 50c... will have to keep an eye on it and see if the water cooling can keep up
[23:29:12 CEST] <BtbN> Sounds like there's something wrong with it.
[23:30:09 CEST] <TAFB> normally sits around 37c on ultrafast
[23:30:30 CEST] <TAFB> seems to be staying pretty constant at 50c, hasn't spuns the fans up yet
[23:31:17 CEST] <furq> fwiw you might be able to get better compression with ffv1
[23:31:32 CEST] <TAFB> never heard of it :)
[23:31:50 CEST] <furq> https://trac.ffmpeg.org/wiki/Encode/FFV1#FFV1version3
[23:31:54 CEST] <BtbN> the cooling of a system should be designed to keep the system from overheating even at permanent maximum load.
[23:32:15 CEST] <furq> 50C isn't really close to overheating
[23:33:00 CEST] <TAFB> these cpu's are super inefficient :D
[23:33:00 CEST] <BtbN> Still pretty bad for water cooling
[23:34:26 CEST] <TAFB> -vcodec   ffv1 -level 3
[23:34:57 CEST] <furq> you'll probably want to tweak -threads and -slices
[23:35:08 CEST] <furq> and -coder 2 gives better compression iirc
[23:35:18 CEST] <TAFB> is -coder 2 still lossless?
[23:35:28 CEST] <furq> ffv1 is always lossless
[23:35:34 CEST] <TAFB> nice
[23:36:59 CEST] <TAFB> ffv1 -level 3 -threads 12 -coder 2 -context 1 -g 1 -slices 32 -slicecrc 1
[23:38:05 CEST] <furq> -slices should be as low as you can make it without getting dropped frames
[23:39:00 CEST] <BtbN> is that a high core, low clock cpu, or why are you having such a hard time encoding a single stream?
[23:40:00 CEST] <TAFB> 32 slices is unsupported :)
[23:43:32 CEST] <TAFB> 24 slices works, but when I open the udp stream in vlc player, I only get sound, no picture
[23:44:07 CEST] <TAFB> vlc codec info shows no video stream, just audio
[23:44:14 CEST] <c_14> probably needs a really recent version of vlc
[23:44:31 CEST] <c_14> Rather, a vlc linked against a recent ffmpeg
[23:46:33 CEST] <TAFB> just updated to 2.4.4 Weatherwax, will test again
[23:46:45 CEST] <TAFB> 2.2.4 sorry :D
[23:50:25 CEST] <TAFB> nope, still doesn't see the video stream
[23:50:33 CEST] <TAFB> does ffv1 support udp streaming?
[23:50:41 CEST] <furq> i'm not sure if mpegts supports ffv1
[23:50:43 CEST] <BtbN> why would that be up to the codec?
[23:51:07 CEST] <TAFB> ahh, what I can use instead of mpegts for udp streaming?
[23:53:44 CEST] <ChocolateArmpits> nut
[23:53:57 CEST] <ChocolateArmpits> if your end point is ffmpeg too
[23:54:08 CEST] <TAFB> it could be I guess, ffplay?
[23:54:16 CEST] <ChocolateArmpits> that works fine
[23:54:18 CEST] <ChocolateArmpits> -f nut
[23:54:27 CEST] <ChocolateArmpits> accepts anything under the sun
[23:59:27 CEST] <TAFB> with -f nut I get tons of errors on ffplay :( 1 sec.
[00:00:00 CEST] --- Sun Jul  3 2016


More information about the Ffmpeg-devel-irc mailing list