cancername has quit [Remote host closed the connection]
mishehu has joined #ffmpeg
iive has quit [Quit: They came for me...]
<noobaroo>
I have a really slow encode going that's going to last for the next 12 hours probably, but I want to encode a quick 2 min long video right now. Is it possible to pause the first one while I do the second? Sometimes I get crashes from running too many things at once
<noobaroo>
If not, how bad do you guys think it is to run 2 operations at once on your own PCs? I never run 2 operations simultaneously but mine crashes sometimes with just one operation, and I am approx 8 hours into this encode process so I would really hate for it to crash and to have to start all over
<linext>
what's the fastest way to resize video on a fast computer?
<Marth64>
noobaroo: you should be able to use ctrl+z to suspend the process if on linux . then `fg` to bring it back. this is a posix signaling concept guides can be found online
<Marth64>
linext: you said fastest that would be using a GPU which will have some degree quality loss (may be minimal degree)
<linext>
right now it's 37x
<Marth64>
that is pretty good with just cpu
<noobaroo>
Marth64 have you tested this yourself with ffmpeg? I just want to be sure I won't have to start over, this took forever
<evilscreww>
Error opening input: Invalid data found when processing input
Hackerpcs has quit [Quit: Hackerpcs]
<iconoclast_hero>
i need to recompile, don't i?
<iconoclast_hero>
thanks to this handy dandy compile script i have...
Gaboradon_ has joined #ffmpeg
l4yer has joined #ffmpeg
Gaboradon has quit [Ping timeout: 264 seconds]
chitchor has joined #ffmpeg
chitchor has left #ffmpeg [WeeChat 3.8]
rv1sr has quit []
l4yer has quit [Ping timeout: 268 seconds]
<iconoclast_hero>
with the pulse capture... it says it's getting data since it has a size and a bit rate. where is this being saved if the output file is still 0 bytes?
lexano has joined #ffmpeg
ivanich has joined #ffmpeg
SystemError has quit [Remote host closed the connection]
chiselfuse has quit [Read error: Connection reset by peer]
xx has quit [Read error: Connection reset by peer]
SystemError has joined #ffmpeg
xx has joined #ffmpeg
chiselfuse has joined #ffmpeg
yans has joined #ffmpeg
SystemError has quit [Remote host closed the connection]
SystemError has joined #ffmpeg
YuGiOhJCJ has quit [Quit: YuGiOhJCJ]
<jafa>
neither valgrind nor heaptrack show a memory leak... that suggests the leak has to be mmap or similar non-heap allocator right?
yans has quit [Remote host closed the connection]
<JEEB>
jafa: I recommend specifically valgrind's memory allocation profiler, I remembered its name earlier but right now it escapes my mind but if you scroll back there you should have me mentioning it :P
<JEEB>
massif, right
<JEEB>
that should show you what keeps allocating memory even if it is all freed at the end
yans has joined #ffmpeg
<JEEB>
ans Massif Visualizer that's a Qt app is quite great at visualizing a massif dump as well
jemius has joined #ffmpeg
fossdd has joined #ffmpeg
junaid_ has joined #ffmpeg
e^pi-1 has quit [Quit: WeeChat 4.2.1]
emanuele6 has quit [Ping timeout: 255 seconds]
<jafa>
does valgrind track mmap and similar apis or only malloc?
fossdd has quit [Ping timeout: 256 seconds]
<JEEB>
did you actually try massif or just basic valgrind? since valgrind only tells you about possible invalid memory access or leaks, not just rising allocations (but everything getting nicely cleaned in the end)
SystemError has quit [Remote host closed the connection]
SystemError has joined #ffmpeg
<jafa>
ffmpeg 6.1.1 won't compile for me with that configure option
<JEEB>
compile option?
<jafa>
I will need to check - I tried a week or 2 ago
<JEEB>
you don't need to set any compile options for valgrind or massif
l4yer has joined #ffmpeg
<JEEB>
setting it might make FATE tests be run under valgrind but will not affect your builds
<iconoclast_hero>
Specifically, does this mean that ffmpeg will not take sink inputs: https://ffmpeg.org/ffmpeg-devices.html#pulse: "To list the PulseAudio source devices and their properties you can invoke the command pactl list sources."
<BtbN>
What do you mean by "sink inputs"?
<BtbN>
You can only use inputs as inputs.
<BtbN>
If your sink has a monitor-source, of course you can use that. But you can't straight up use a sink as input.
fossdd has joined #ffmpeg
<iconoclast_hero>
i'm looking at pavucontrol and I see "Monitor of audio" with a signal (i.e., the bar's moving) and that is the thing I want to grab.
<iconoclast_hero>
when I made it I tried to include a space in the name: `pactl load-module module-null-sink sink_name=monitor sink_properties=device.description='audio hole'`
emanuele6 has joined #ffmpeg
<iconoclast_hero>
when I look in pactl list|grep audio\ hole I do see this: Argument: sink_name=monitor sink_properties=device.description=audio hole
<iconoclast_hero>
so i'm looking for ` -i monitor `?
<evilscreww>
aaabbb: concat demuxer works WAY better than concat protocol for my purposes
<evilscreww>
i dont even need to convert to ts
chiselfuse has quit [Remote host closed the connection]
darkapex has joined #ffmpeg
chiselfuse has joined #ffmpeg
JanC has quit [Ping timeout: 260 seconds]
JanC_ has joined #ffmpeg
JanC_ is now known as JanC
Gaboradon_ has quit [Quit: Quitting]
junaid_ has quit [Remote host closed the connection]
chiselfuse has quit [Remote host closed the connection]
chiselfuse has joined #ffmpeg
fossdd has joined #ffmpeg
<noobaroo>
Got it thanks ayecee
<noobaroo>
Oops wrong chat
evilscreww has quit [Remote host closed the connection]
evilscreww has joined #ffmpeg
rv1sr has joined #ffmpeg
waleee has quit [Ping timeout: 260 seconds]
Marth64 has quit [Ping timeout: 252 seconds]
Marth64 has joined #ffmpeg
Tano has joined #ffmpeg
fossdd has quit [Ping timeout: 256 seconds]
evilscreww has quit [Remote host closed the connection]
fossdd has joined #ffmpeg
<iconoclast_hero>
what should an SDP file include for an RTP audiostream created via $ pactl load-module module-null-sink sink_name=rtp channels=2 rate=48000; pactl load-module module-rtp-send source=rtp.monitor destination=192.168.1.10 port=4010 mtu=320
<iconoclast_hero>
i see people getting the output of ffmpeg and using that for the parameters but i just get an error
<jafa>
testing vaapi h264 encode on a Ryzen 7000 series iGPU... it is messing up the video
<Marth64>
What is the error iconoclast_hero
<iconoclast_hero>
I might have gotten it...but I'm not getting any audio
<iconoclast_hero>
i'm trying to get this RDP stream working on android and i figured it would be better if i can get it working in linux first.
<iconoclast_hero>
hence ffplay
<iconoclast_hero>
however, it's no longer playing on the pulseaudio rtp receiver app on the phone so i'm giving up for ow i think
<Marth64>
Does android even have pulse stack?
<Marth64>
or its custom app
<iconoclast_hero>
it's from fdroid
<iconoclast_hero>
that's its name
<jafa>
vaapi - solved - need to explicitly specify nv12
<iconoclast_hero>
i could get sound out of it but it sounded like it was coming in at 200% volume or something
<iconoclast_hero>
ok, as for ffplay. i checked and had the wrong ip address when i created the rtp send so i fixed that and changed the port
noobaroo has quit [Remote host closed the connection]
noobaroo has joined #ffmpeg
<noobaroo>
I have a video that I recorded from a friend's brand new phone, the file is like 80Mbps 4K video and it's fine on the phone but in my PC the color is really washed out and there is a weird rainbow effect
<dv_>
hm, I guess you can do intra refresh in VP8 by not generating I frames (except for the first one) and manually force intra macroblocks instead, but decoders have no idea about this, and thus don't know when to start producing decoded pictures when they join mid-stream
<JEEB>
yea, H.264 was also quite hacky regarding that
<noobaroo>
My mom's red car looks grey but it changes to red instantly like from 0 to 1 in a single frame at a certain proximity
<dv_>
so, in other words, intra refresh in VP8 can't work in general - the decoder must be specifically modified to handle it
<dv_>
VP9 can do it though because its frames can be marked appropriately
<dv_>
via the error_resilient_mode flag
<JEEB>
I'd say H.264 is quite similar since in most cases you have no idea that within 10 frames or 50 frames or whatever you would have all the necessary refs
<JEEB>
sometimes in the container you may have this flagged but I think almost nothing supports this in mp4 even though the flags are possible
<dv_>
h264 does have some intra refresh metadata tho, hang on
<Marth64>
noobaroo: sounds like colorspace issue possibly?
<dv_>
ah, not exactly. the SEI messages can contain recovery point information.
<JEEB>
yea that is used for open gop
<JEEB>
so you have that SEI plus a normal I frame
<JEEB>
but that is different from informing the decoder that in X frames it can start showing
<JEEB>
(since in X frames it should have all the necessary refs)
<dv_>
so why does intra refresh still work mid-stream?
<dv_>
do decoders just keep track of the intra macroblocks that arrived thus far?
<dv_>
at least it seems to work with ffmpeg's h264 decoder
<JEEB>
well yes, that is not surprising :D
<JEEB>
since x264 was one of the first major things I know of that supported intra refresh
<JEEB>
for game streaming in like... 2009-2010 or so?
<dv_>
sure, but what does that decoder use to know when the frames can be shown if intra refresh is in use?
<JEEB>
so avcodec was made to support that
<dv_>
does it use the SEI?
<JEEB>
dv_: since intra refresh is not flagged in any form as far as I can tell, you just wait until you seem to have a fully decode'able frame
<JEEB>
and until then you decode and store refs
<JEEB>
I have not looked into h264dec code but that's how I'd expect it to work
<dv_>
and this waiting is done by the decoder by keeping track of what macroblocks got intra-refreshed, and once all were refreshed that way -> output?
<JEEB>
what slices were full and thus the reference frame is available I'd say
gchound has joined #ffmpeg
<dv_>
because perhaps something similar could be used for VP8
<JEEB>
usually intra refresh is done with low latency and you can't have frame threading there, so you just have X slices
<JEEB>
and at least how x264 intra refresh works is it codes one of those slices as intra in a pattern that after X frames you have all of the slices decoded and all refs available
Marth64 has quit [Remote host closed the connection]
<dv_>
right
lucasta has joined #ffmpeg
zsoltiv_ has joined #ffmpeg
fossdd has quit [Ping timeout: 256 seconds]
lavaball has quit [Quit: lavaball]
<noobaroo>
Marth64 It looks like it plays fine when I use software decoding but my computer fan starts going crazy
rue_mohr has joined #ffmpeg
<rue_mohr>
anyone know how I can take a web stream camera and loop it back as a v4l device?
jemius has quit [Quit: Leaving]
thomas_D8 has quit [Read error: Connection reset by peer]
thomas_D8 has joined #ffmpeg
SnakesAndStuff has joined #ffmpeg
KombuchaKip has joined #ffmpeg
dreamon has joined #ffmpeg
Haxxa has quit [Quit: Haxxa flies away.]
fossdd has joined #ffmpeg
Somebody3456 is now known as SystemError
Haxxa has joined #ffmpeg
lucasta has quit [Quit: Leaving]
Tinos has joined #ffmpeg
fossdd has quit [Ping timeout: 255 seconds]
averne has quit [Quit: quit]
AbleBacon has joined #ffmpeg
averne has joined #ffmpeg
dreamon has quit [Ping timeout: 240 seconds]
theobjectivedad has quit [Ping timeout: 268 seconds]
sonicrules1234 has quit [Remote host closed the connection]
theobjectivedad has joined #ffmpeg
fossdd has joined #ffmpeg
MrZeus_ has joined #ffmpeg
MrZeus_ has quit [Read error: Connection reset by peer]