Not sure if this has been answered before or if I should post to devel
instead, but here goes.
I have a vfr video
I reencode the vfr in cfr
ffmpeg -i INPUT -codec:a aac -b:a 128k -codec:v libx264 -crf 23 -vf
"pad=width=3840:height=2160:x=0:y=312:color=black,scale=-1:720,fps=30"
-preset fast -pix_fm
Is it possible to start a live capture at the end of an existing video in
the hls format?
Say you have a captured m3u8, something happens and you want to restart
ffmpeg but continue capturing at the end of the previous m3u8?
I guess concat can't do it since it has to be done on complete videos.
> What about the option "-hls_flags append_list"
Well what do you know, change request from you on Aug. 11, 2016, what a
terrific idea.
My version doesn't have it, I'll compile a new one and try it as soon as I
can.
Thanks a lot Liu! That sounds exactly like what I need.
Louis
___
Just a question before I start diving in the code, why doesn't fast
seeking work on live hls feeds/streams/files?
Context:
If you create a dummy video:
mkdir -p a;ffmpeg -y -s 640x480 -f rawvideo -pix_fmt rgb24 -r 25 -i
/dev/zero -vf
"drawtext=fontfile=/usr/share/fonts/truetype/freefont/FreeMono.
so it's been in the 3.2 branch
for awhile now.
Louis
On Fri, May 19, 2017 at 5:49 PM, Louis Letourneau wrote:
> Just a question before I start diving in the code, why doesn't fast
> seeking work on live hls feeds/streams/files?
>
> Context:
> If you create a dummy video:
> Does anybody know of any hardware that would work with my MacBook Pro and
speed up conversion to H265 (HEVC) ?
AJA Corvid HEVC
https://www.aja.com/products/developer/corvid-hevc
I've used it extensively under Linux and they do support MAC
This being said, I never tried it under MAC.
(none of
I seem to be having issues seeking accurately in HLS videos, but it
works fine in other containers (mp4, mkv, etc)
tried with ffmpeg version 3.4.4 and master
If I create an HLS video like this
mkdir a;ffmpeg -y -s 640x480 -f rawvideo -pix_fmt rgb24 -r 25 -i
/dev/zero -vf
"drawtext=fontfile=/usr/
> > > I seem to be having issues seeking accurately in HLS videos, but it
> > > works fine in other containers (mp4, mkv, etc)
> >
> >
> > Hello,
> >
> > Refer to this issue which sounds the same/related/similar:
> >
> > https://trac.ffmpeg.org/ticket/7359
> >
> > I fixed this in my own ffmpeg buil
Not the same problem and patch doesn't fix the issue at all.
Louis
On Wed, Oct 10, 2018 at 5:11 AM Nick Ryan wrote:
>
> > I seem to be having issues seeking accurately in HLS videos, but it
> > works fine in other containers (mp4, mkv, etc)
>
>
> Hello,
>
> Refer to this issue which sounds the sa
On Thu, Oct 25, 2018 at 1:39 PM Louis Letourneau wrote:
>
> > However, I have now submitted a new patch which correctly fixes the issue I
> > was solving.
> >
> > I took the time to reproduce your issue outlined in your original post. I
> > managed to reproduce
> However, I have now submitted a new patch which correctly fixes the issue I
> was solving.
>
> I took the time to reproduce your issue outlined in your original post. I
> managed to reproduce it and then resolve it with this new patch.
>
> The patch I submit has changes to his.c (relevant to yo
I was wondering if anyone had experience streaming 720p ~5000kbit/s
video overseas successfully? If so which protocol did you use?
I'm trying to stream between Stockholm sweden to toronto canada.
Using mpegts over tcp, I sometimes get so many retries that the
bandwith falls below 5MBit/s for many
> Try rtmp
Please don't top post, thanks
I just tried same issue as mpegts over tcp. doesn't handle packet
loss/retries well, tcp buffer exhausts itself sometimes.
Louis
___
ffmpeg-user mailing list
ffmpeg-user@ffmpeg.org
http://ffmpeg.org/mailman/list
> You can do it by 2 ways
> 1. Push it as RTMP and client can then convert RTMP to HLS
>
> 2. However if you still see packet loss then dump stream to ts packets on
> S3 or any SSD storage and serve it through CDN HLS
Please don't top post thanks.
Yeah I don't see any other way than to find a clo
> Hello Louis,
>
> Have you take a look at SRT protocol ?
>
> Source code : https://github.com/Haivision/srt
>
> Latest FFmpeg handle this protocol (if i am not wrong).
>
I didn't know about it. I will try it as soon as i can. It seems
interesting.
Louis
>
__
Has anyone managed to capture a feed using the "USB Capture SDI 4k"
from Magewell with ffmpeg?
It runs for awhile but then a 0 byte packet comes along and ffmped just stops.
It doesn't quit, but the frame number just stops and it doesn't do
anything after rhis.
I saw 2 bugs from 2014 and both see
Hello,
I pretty much have the same problem listed in this post in Sept 2015 which
got no answer for.
http://ffmpeg.org/pipermail/ffmpeg-user/2015-September/028281.html
When cutting a video, even when transcoding, I often have a small delay
between the video and audio.
When looking at the video wi
On Jul 7, 2016 13:08, "Christian Ebert" wrote:
>
> * Louis Letourneau on Thursday, July 07, 2016 at 12:38:26 -0400
> > When cutting a video, even when transcoding, I often have a small delay
> > between the video and audio.
> >
> > When looking at the vi
>
> > When cutting a video, even when transcoding, I often have a small delay
> > between the video and audio.
>
> After many problems, I started converting my audio to uncompressed
> PCM to do cutting, then re-encoding it later (of course this works
> for me because I only have stereo audio :-).
If you encode your mp4 with '*-movflags faststart' it will put the moov
atom at the beginning of the movie. This way ffplay can play it BUT it
won't play it until the end.*
*You'll just see the state of the video when you started ffplay.*
*To see the encoding live until it ends, you need segmented
I'm a bit confused on what is expected when seeking with ffmpeg on m3u8
files.
I was trying to get frame 43 from a m3u8. I converted the frame to time in
seconds and when I extract that frame, the frame that I get is WAY off.
about 10 seconds off.
I did some tests and:
FFMPEG 3.1.1 with ss 0
ffmp
ways, this seeking bug is the opposite, fine on 3.1.3 but broken on
3.1.1 and HEAD.
Louis
On Tue, Sep 13, 2016 at 12:44 PM, Carl Eugen Hoyos
wrote:
> 2016-09-13 18:15 GMT+02:00 Louis Letourneau :
>
> > FFMPEG 3.1.3 with ss 0
> > ffmpeg -y -loglevel error -ss 00:00:00 -vsync 0 -i 2708
> > Anyways, this seeking bug is the opposite, fine on 3.1.3 but broken on
> > 3.1.1 and HEAD.
>
> You already said so.
I had mentioned it, but I just wanted to show the numbers in case it helped
> Please do not top-post here, Carl Eugen
Yeah, thanks for the reminder (darn gmail)
I ran bisect
> On Thu, Sep 15, 2016 at 6:27 AM, Carl Eugen Hoyos
wrote:
> Is the issue reproducible with current FFmpeg git head?
Yes
> Do I understand correctly that the issue is only reproducible
> with FFprobe but not FFmpeg?
No, I have 2 symptoms
1- Frame counts with ffprobe is too high
2- seeking to th
> On Thu, Sep 15, 2016 at 8:45 AM, Moritz Barsnick wrote:
>
> The commit on master corresponding to your bisected commit on 3.1 is
> 9884f17e343b37aef442fafa05bd0113cdf47087, according to the commit
> message.
I found that commit but it didn't fix or cause the problem on master?!
So on branch 3.
>On Thu, Sep 15, 2016 at 10:59 AM, Moritz Barsnick wrote:
>
> "broken seeking" = "incorrect frame"? (You use both wordings, which is
> fine - I just want to be sure.)
Yes exactly. I've been running a bunch of checkouts, build, test since
yesterday, it might explain my wording lapses.
> > I've se
> That would mean that it was only ever "fixed" (even if by accident) on
> branch 3.1, while the issue remains on master?
>
> Then the actual causes for the issues need to be understood, not just
> the commits. ;-) (I'm not volunteering, I don't understand the code.)
>
> Moritz
Should I open a bu
> I forgot to ask: Do you have samples, i.e. an m3u8 and its contained
> segments, which allow to reproduce the issue?
Yup I posted that earlier in the thread.
https://s3.amazonaws.com/slffmpegdebug/seekProb/seekProb.m3u8
It's publicly accessible
---> From that earlier post
I shared the video on
I'm having trouble understanding if I'm doing something wrong or if I'm
missing a flag and I hope you guys can help
If I create an m3u8 with the following command (version 3.1.4)
mkdir a
./ffmpeg -y -s 640x480 -f rawvideo -pix_fmt rgb24 -r 25 -i /dev/zero -an
-vcodec libx264 -preset medium -tune s
>
> "ffmpeg -loglevel verbose" at least shows you which segments are opened.
>
Well what do know, I should have tried that. Very useful on m3u8s.
> Why do you need a release, anyway? Do you rely on getting your ffmpeg
> from a distribution?
>
For development it doesn't really matter. For our pr
> I don't use releases myself, but I always advise AV
> archivists, who implement FFmpeg in their workflow, to use
> the last release of an actively maintained branch. Why?
> Because for the people living outside our "gang" it is much
> easier to keep the software updated this way
So do you sugges
I'm capturing a video using:
ffmpeg -y -f decklink -i "DeckLink Mini Recorder (2)@14" -pix_fmt yuv420p
-vf "fps=29.97" -force_key_frames "expr:gte(t,n_forced*5)" -codec:a
libfdk_aac -b:a 128k -codec:v libx264 -crf 24 -preset fast -bf 0 -hls_time
5 -hls_list_size 0 -hls_wrap 0 -hls_allow_cache 1 -h
>
> I do realise that a wrapping HLS stream could be problematic, but still I
> was wondering if a parameter existed to force it to try anyways.
>
> Sorry missed it in the doc -live_start_index does exactly this.
If it can help someone else:
ffmpeg -y -live_start_index 0 -ss 00:10:00 -i segment.m3
I'm not sure if this belongs in user or devel, but here goes.
I'm working with m3u8 files. They can sometimes be live, sometimes VOD.
For live, I need to start the processing from the start. using the command
line I would
ffmpeg --live_start_index 0 -i file.m3u8 [...]
How do I reproduce this in
>
> Is priv_data the right parameter? I would have expected a FormatContext.
>
> There is a dedicated mailing list for using the FFmpeg libraries, see:
> https://ffmpeg.org/contact.html
I didn't see the libav-user mailing list when I first looked. I'll know
next time.
I finally solved the proble
I seem to be having issues seeking accurately in HLS videos.
ffmpeg version 3.2 and master N-82664-g801b5c1
If I create an HLS video like this
mkdir a;ffmpeg -y -s 640x480 -f rawvideo -pix_fmt rgb24 -r 25 -i /dev/zero
-vf "drawtext=fontfile=/usr/share/fonts/truetype/freefont/FreeMono.ttf:
text=%{
>
> It's missing feature iirc, if this is static file you could create
> index once and use it for precise seeking after.
>
Didn't quite get that.
Are you saying to re-encode using 'single_file' to use EXT-X-BYTERANGE
elements in the m3u8?
I tried and it does seem to work
mkdir a;ffmpeg -y -s 64
I'm trying to concatenate multiple videos using filter_complex.
Problem is, the movie filter always removes the last frame.
The more movie filters there are, the more frames are removed
here are example commands to reproduce the problem
ffmpeg -y -s 640x480 -f rawvideo -pix_fmt rgb24 -r 60 -i /de
38 matches
Mail list logo