BtbN changed the topic of #ffmpeg to: Welcome to the FFmpeg USER support channel | Development channel: #ffmpeg-devel | Bug reports: https://ffmpeg.org/bugreports.html | Wiki: https://trac.ffmpeg.org/ | This channel is publically logged | FFmpeg 7.0 is released
nigetilly has joined #ffmpeg
lavaball has quit [Remote host closed the connection]
nigetilly has quit [Quit: Konversation terminated!]
theobjectivedad has quit [Ping timeout: 260 seconds]
theobjectivedad has joined #ffmpeg
e^pi-1 has quit [Quit: WeeChat 4.2.2]
e^pi-1 has joined #ffmpeg
iive has quit [Quit: They came for me...]
lexano has quit [Ping timeout: 252 seconds]
zsoltiv_ has quit [Quit: Left]
zsoltiv_ has joined #ffmpeg
lucasta has joined #ffmpeg
NotFaely has quit [Remote host closed the connection]
ivanich has quit [Remote host closed the connection]
<bpmedley>
noobaroo: Are you playing a file and getting the error message?
troyt has joined #ffmpeg
Tinos has quit [Ping timeout: 250 seconds]
Haxxa has quit [Quit: Haxxa flies away.]
Haxxa has joined #ffmpeg
zenstoic has joined #ffmpeg
<noobaroo>
bpmedley thanks for your reply. Nope, I'm trying to use -profile:a aac_he_v2 or aac_he
hightower2 has quit [Ping timeout: 260 seconds]
<noobaroo>
LC works fine, it's just the HE stuff that won't work at all
<bpmedley>
So, you are transcoding from an input format to HE-AAC v2?
<noobaroo>
Specifically I was trying to convert 6 channel EAC3 to 2 channel libfdk_aac HE , wouldnt work and I also tried to convert 2 channel FLAC and that wouldnt work either, so I dont think its the source
<noobaroo>
Yeah exactly
<bpmedley>
Can we get an example command line with the issue?
<bpmedley>
noobaroo: What does ffprobe say your input file is? I just tried your command, and "Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 48000 Hz, 2 channels, s16, 1536 kb/s" worked for me.
trillion_exabyte has quit [Read error: Connection reset by peer]
trillion_exabyte has joined #ffmpeg
Jhonny2x4 has quit [Quit: Jhonny2x4]
Jhonny2x4 has joined #ffmpeg
Jhonny2x4 has quit [Client Quit]
Jhonny2x4 has joined #ffmpeg
zenstoic has quit []
superkuh has quit [Read error: Connection reset by peer]
lucasta has joined #ffmpeg
hightower2 has joined #ffmpeg
<noobaroo>
bpmedley, it used to work for me too
<bpmedley>
What FFmpeg version are you using? And/or what changed?
hightower2 has quit [Ping timeout: 260 seconds]
relue has joined #ffmpeg
hightower2 has joined #ffmpeg
<tm512>
so I gave in and am doing a test stream using OBS with the same sort of bitrate that was consistently causing ffmpeg to drop frames, and OBS is just not dropping frames
ferreum has quit [Quit: bye]
<tm512>
feeling kinda vindicated because I felt like it made no sense for me to have so many frame drops at 5Mbit when every speed test I do shows upload speeds around 20Mbit or higher
<tm512>
but it has me wondering what exactly is wrong with ffmpeg
ferreum has joined #ffmpeg
<tm512>
I'm guessing the frame drops are somewhere else along the pipeline from x11grab to the RTMP connection
<tm512>
could having -thread_queue_size or -rtbufsize too small cause frame drops? and if so, could "too small" be dependent on the exact connection conditions, such as latency/jitter?
hightower2 has quit [Remote host closed the connection]
<furq>
yes
<tm512>
would it do this silently, without printing out any kind of messages about buffer issues?
<furq>
i would expect a warning but i've never tried with x11grab
e^pi-1 has joined #ffmpeg
<tm512>
I think suggestions to increase thread_queue_size get printed with my streaming script where I've got multiple input streams, but I'll double check here in a bit (I don't really want to interfere with this OBS test)
<tm512>
in the past, just ignoring the thread_queue_size stuff always seemed harmless
<tm512>
hmm, not currently running into the recommendation that I increase thread_queue_size
<tm512>
x11grab does give me "Stream #0: not enough frames to estimate rate; consider increasing probesize"
lavaball has quit [Remote host closed the connection]
SuicideShow has quit [Ping timeout: 260 seconds]
SuicideShow has joined #ffmpeg
<random83>
Is DASH/HLS going to give any benefits if I'm only serving a single quality +faststart file? While the video starts playing pretty quickly, I'm having some abysmal seeking performance, routinely taking 20 seconds or more. Maybe switching from simple <video> to something like JW player would help?
<random83>
The files were encoded using nothing more than -crf and -movflags +faststart, so there's probably a lot of room for improvement.
Narrat has quit [Quit: They say a little knowledge is a dangerous thing, but it's not one half so bad as a lot of ignorance.]
<bpmedley>
random83: Would you playing the HLS video live or in vod mode?
<bpmedley>
Said another way, would you have a directory of segments and .m3u8 manifest(s)?
<tm512>
trying to figure out what the default rtbufsize is. setting it to 256M does not seem to prevent the frame drops, nor does setting thread_queue_size to 1024. dunno if I need to go larger, or if I'm not on the right track here
<tm512>
not sure how I can figure out which part of the pipeline is choking
Vonter_ has quit [Ping timeout: 260 seconds]
<tm512>
maybe 256M isn't enough. it appears that the RTMP client buffer defaults to 3000ms, but a 256MB buffer is only enough to store a little over a second of a 720p capture from x11grab, which according to ffmpeg is about 216MiB/s
Vonter has joined #ffmpeg
<tm512>
that evidently isn't it either. still get frame drops even setting rtbufsize to 768M, which is more than enough buffer to store all of the input frames that a 3 second RTMP cache would hold
<tm512>
and ffmpeg isn't even coming close to actually using 768MB of memory
<tm512>
I'd prefer not to be required to resort to OBS. it's relatively bloated compared to ffmpeg, and it's kinda unwieldy on a single-monitor streaming setup like mine
<tm512>
but it can stream to RTMP properly, whereas I can't seem to get ffmpeg to behave. frustrating situation.
xx has quit [Remote host closed the connection]
Kroko has joined #ffmpeg
Kroko has quit [Remote host closed the connection]
Kroko has joined #ffmpeg
<random83>
bpmedley: the videos are all pre-recorded, so I'm guessing yes?
stolen has quit [Quit: Connection closed for inactivity]
relue has quit [Read error: Connection reset by peer]
<bpmedley>
random83: I've had good luck with a proof-of-concept, where I used ffmpeg to create the segments and manifest and then view via HLS.js. I had no issues seeking or with playback. However, I only used this setup with a proof-of-concept.