<noobaroo>
I'm trying libsvtav1 now and it seems to be working okay so far, I'm 8 minutes into the video but that's farther than libx265 was getting me
Livio has quit [Ping timeout: 252 seconds]
yawkat has quit [Ping timeout: 240 seconds]
MrZeus has quit [Ping timeout: 252 seconds]
<noobaroo>
Well it looks like the libsvtav1 process failed too, this time the whole terminal window is gone
Tinos has quit [Ping timeout: 250 seconds]
lexano has quit [Ping timeout: 252 seconds]
waleee has quit [Ping timeout: 255 seconds]
hamzah has quit [Quit: Client closed]
<aaabbb>
noobaroo: that's interesting. have you tried reproucing it on the latest BtbN build?
<aaabbb>
it should never segfault, that means there's a bug
yans has quit [Read error: Connection reset by peer]
<iive>
noobaroo, it would help if you can run ffmpeg from gdb and provide backtrace.
<aaabbb>
that would be ideal, but it might be a bug fixed long ago. it would suck to waste time trying to track something down that has been fixed
<iive>
if you can't... then maybe try to isolate what system crashes. e.g. try to disable sound entirely (no decoding, no encoding).
<iive>
you can always compile a recent one.
<iive>
then you could try to figure out if it is in the decoder or encoder, e.g. you can decode to raw image/pcm (and pipe it to /dev/null)
<iive>
if you want a meaningful backtrace, you'd need to enable debug info and disable binary stripping
<iive>
oh, i see subtitles too, try disabling the subtitle processing too.
<aaabbb>
there's no need to compile one yourself, the BtbN builds are available
<iive>
well, the builds are for red hat and win7+
<iive>
i don't know if they have debug info in them...
Keshl_ has quit [Read error: Connection reset by peer]
<znf>
I had something much more complicated in my head lol
<znf>
colorchannelmixer and overlay
<znf>
and changing the pixel format
treefrob has quit [Ping timeout: 255 seconds]
e^pi-1 has joined #ffmpeg
treefrob has joined #ffmpeg
KombuchaKip has quit [Quit: Leaving.]
KombuchaKip has joined #ffmpeg
BUSY has quit [Read error: Connection reset by peer]
AbleBacon has quit [Read error: Connection reset by peer]
<ShaedS>
furq: that easy?
<noobaroo>
iive This is ffmpeg 6.2 from git and aaabbb the BtbN build doesn't have libfdkaac so I don't use those
<ShaedS>
furq: is it easier to replace the audio afterwards, or is it not a big deal to online it
<ShaedS>
*oneline
<noobaroo>
How can I decode to raw image/pcm?
<noobaroo>
I may not be able to backtrace but I think it's a good idea to try to decode to raw image/pcm and see if that works
minimal has quit [Quit: Leaving]
<ShaedS>
I know ffmpeg has silent detect, is there a video equivalent? My use case: i'm standing in a black frame, street light in the back, minimal but some light present. I then go turn on TWO UV lights at the same time. I'm hoping I can use ffmpeg to mark the time in miliseconds or frame# of the light turning on, I do this 3 or 4 times in succession, as in practice when trimming i get black frames and am n ot
<ShaedS>
sure if was synced correctly.
<ShaedS>
This is my firs ttime doing it more than once so I'll see if my cuts are dead-nuts accurate or just impreceptably accurate
<ShaedS>
if i could do that then i could automate a good chunk of my workflow
bitoff has quit [Read error: Connection reset by peer]
yawkat has joined #ffmpeg
turlando has quit [Quit: No Ping reply in 180 seconds.]
turlando has joined #ffmpeg
zsoltiv_ has quit [Ping timeout: 255 seconds]
zsoltiv_ has joined #ffmpeg
darkstardevx has joined #ffmpeg
mrelcee has quit [Ping timeout: 255 seconds]
mrelcee has joined #ffmpeg
evilscreww has joined #ffmpeg
<evilscreww>
so theres concat demux and concat protocol?
<evilscreww>
which one suits which suituation best?
gvg_ has quit [Quit: Reconnecting]
gvg_ has joined #ffmpeg
darkstardevx has quit [Quit: Leaving]
Xaldafax has quit [Quit: Bye...]
evilscreww has quit [Remote host closed the connection]
evilscreww has joined #ffmpeg
rvalue has quit [Ping timeout: 255 seconds]
jafa has quit [Remote host closed the connection]
noobaroo has quit [Quit: Konversation terminated!]
<noobaroo>
I have a browser extension that saves videos of HLS stream origin, but the one from Chrome web store saves the web video as .mp4 and the one from Firefox saves as .ts Obviously one of these extensions is converting the videos but which one, does anyone know?
JanC has quit [Ping timeout: 268 seconds]
JanC has joined #ffmpeg
Keshl__ is now known as Keshl
xx has joined #ffmpeg
<ShaedS>
i don't get how this works. I am iusing VDJ to get an exact frame number of when a light turns on, i go one fram back, mark the frame #, convert it to miliseconds, feed it to ffmpeg, the videos don't sync. I have to pixelpush/fudge the miliseconds around one direction or the other to get it to 'be locked'. Is this due to the timecode -> ms conversion tool? what was that other fftool that did timecode?
<ShaedS>
ffmsometning
Ox7C5_ has joined #ffmpeg
FH_thecat has quit [Quit: Leaving]
zsoltiv_ has quit [Quit: Left]
zsoltiv_ has joined #ffmpeg
e^pi-1 has quit [Ping timeout: 268 seconds]
junaid_ has joined #ffmpeg
Sakura`Kinomoto has quit [Remote host closed the connection]
FH_thecat has joined #ffmpeg
evilscreww has quit [Remote host closed the connection]
evilscreww has joined #ffmpeg
<evilscreww>
why do i need to convert to ts before i concatenatye
Sakura`Kinomoto has joined #ffmpeg
rv1sr has joined #ffmpeg
evilscreww has quit [Remote host closed the connection]
MisterMinister has quit [Ping timeout: 252 seconds]
waleee has joined #ffmpeg
evilscreww has joined #ffmpeg
e^pi-1 has joined #ffmpeg
scoobykj has joined #ffmpeg
scoobykj has quit [Quit: Konversation terminated!]
Suchiman has joined #ffmpeg
elastic_dog has quit [Ping timeout: 260 seconds]
scoobykj has joined #ffmpeg
elastic_dog has joined #ffmpeg
scoobykj has quit [Quit: Konversation terminated!]
evilscreww has quit [Remote host closed the connection]
evilscreww has joined #ffmpeg
Sakura`Kinomoto has quit [Remote host closed the connection]
<aaabbb>
>The demuxer is more flexible – it requires the same codecs, but different container formats can be used; and it can be used with any container formats, while the protocol only works with a select few containers.
evilscreww has quit [Remote host closed the connection]
evilscreww has joined #ffmpeg
robobub has quit [Quit: Connection closed for inactivity]
Sakura`Kinomoto has joined #ffmpeg
lavaball has joined #ffmpeg
<evilscreww>
ts being one of them i suppose
e^pi-1 has quit [Ping timeout: 268 seconds]
<ShaedS>
if I wanted to merge two videos, one at 50% opacity, but I wanted the one that's 100% opacity to be 20px lower than the one at 50% (truncating the bottom 20px from the shift); is this possible? how do i do this? do I first crop the one video by adding a black strip up top? how would you guys do this
<klaxa>
thinking about it without looking at the docs or code made me come to the conclusion you need set all the flags and stuff before allocating, but i don't think there is an API for that
<mitigate>
i'll eventually start reading libavformat.h comments myself :)
<mitigate>
i.e.memcpy the result into an allocated struct and change it there before passing it to avformat_alloc_output_context2
berndj has joined #ffmpeg
<klaxa>
that's one way to do it albeit a very hacky one. but like the person on SO suggests, it's either a bug or we're doing something wrong ¯\_(ツ)_/¯
<klaxa>
if i am not mistaken this was simply an oversight when these fields were made const, and it has either to be reverted or a new api created (as opposed to current ones being adapted) to be able to set those fields before allocating
<mitigate>
but i dont think its supposed to be possible. write_header segfaults
<JEEB>
the flags are just supposed to be informational for you? if some internal module is trying to change its behavior by them, that's a bug
<JEEB>
quickly looked through "flags" in the AVI writer and it isn't reading that
<JEEB>
avdevice/xv.c is also just exposing the flags
<JEEB>
it's not reading them or changing its behavior based on them
<JEEB>
so it seems to me that you are really confused regarding something :P
<klaxa>
in my case at least i used the avoutputformat->flags to set my matroska muxers to AV_FMT_NOFILE (or similar?) so that's gone :D
<JEEB>
yea those were not meant to be modified. the muxing general logic does utilize the flag but it really sounds like you instead want to have custom AVIO or something. I'm not even sure what you are trying to achieve there.
System_Error has joined #ffmpeg
<JEEB>
AVFMT_NOFILE is for (de)muxers that do not actually deal with a specific file (often meta stuff like dashenc or hlsenc or tee|fifo
<klaxa>
it *is* using custom avio, but either i copied that part blindly from examples/muxing.c or without setting the flags files were created (or were tried to be created and failed because no filename)
<klaxa>
or something else entirely
<JEEB>
files were not created in 2013 when I did my baby's first custom AVIO context thing
<klaxa>
which seems more likely after your explanation of AVFMT_NOFILE
<JEEB>
or well, technically I was using AVIO for reading but writing as far as I know works the same way if you override the IO callbacks (since you essentially override the possible file protocol that could be utilized)
Marth64 has quit [Ping timeout: 264 seconds]
junaid_ has joined #ffmpeg
<JEEB>
but yea, I'd recommend checking what you thought you were making happen with that stuff
<JEEB>
in the best case that's just not required, in the worst case there might be a use case that could be thought about how to handle it
<mitigate>
not that it matters after all this but i think mkserver_mk2 has a typo where itscalling ofmt->flags &= AVFMT_NOFILE instead of logior
<klaxa>
indeed it does
<klaxa>
yeah those flag assignments seem quite superfluous looking at it now
vlm has quit [Quit: Leaving]
Tinos has joined #ffmpeg
ivanich has joined #ffmpeg
Marth64 has joined #ffmpeg
Tinos has quit [Remote host closed the connection]
Tinos has joined #ffmpeg
five6184803 has quit [Remote host closed the connection]
five6184803 has joined #ffmpeg
cancername has quit [Ping timeout: 264 seconds]
linext has quit [Read error: Connection reset by peer]
Tinos has quit [Remote host closed the connection]
junaid_ has quit [Remote host closed the connection]
<mitigate>
so this particular fix should be to set AVFMT_FLAG_AUTO_BSF on the AVFormatContext flags and drop all changes to AVOutPutContext flags?
billchenchina- has quit [Read error: Connection reset by peer]
billchenchina- has joined #ffmpeg
j45 has joined #ffmpeg
<mitigate>
i still havent figured out how to test both. with input from a file, the http server exits before i can try to access it, and with input from a http url, i cant figure out if ffmpeg can output to that url
<mitigate>
i can send the patches but they are trivial
Narrat has joined #ffmpeg
junaid_ has joined #ffmpeg
MrZeus has joined #ffmpeg
Wenbin_Chen_ has joined #ffmpeg
luser18 has joined #ffmpeg
mven979 has joined #ffmpeg
tofran5 has joined #ffmpeg
<mitigate>
i think i's http problems both with libav listen and with microhttpd
garfvl_ has joined #ffmpeg
jkkm_ has joined #ffmpeg
function1 has joined #ffmpeg
galad_ has joined #ffmpeg
m5zs7k_ has joined #ffmpeg
DarkG_ has joined #ffmpeg
raccct has joined #ffmpeg
llyyrr has joined #ffmpeg
function1_ has quit [Ping timeout: 260 seconds]
theobjectivedad has quit [Ping timeout: 260 seconds]
jkkm has quit [Ping timeout: 260 seconds]
garfvl has quit [Ping timeout: 260 seconds]
upekkha has quit [Remote host closed the connection]
DarkG has quit [Read error: Connection reset by peer]
auri has quit [Remote host closed the connection]
turlando has quit [Quit: No Ping reply in 180 seconds.]
LimeOn has quit [Read error: Connection reset by peer]
luser1 has quit [Read error: Connection reset by peer]
m5zs7k has quit [Quit: m5zs7k]
gvg has quit [Ping timeout: 264 seconds]
bjarnis has quit [Ping timeout: 264 seconds]
jkkm_ is now known as jkkm
luser18 is now known as luser1
gvg has joined #ffmpeg
Wenbin_Chen__ has quit [Ping timeout: 260 seconds]
Mavrik has quit [Ping timeout: 260 seconds]
mven97 has quit [Ping timeout: 260 seconds]
bencoh has quit [Ping timeout: 260 seconds]
mven979 is now known as mven97
tofran has quit [Ping timeout: 260 seconds]
CounterPillow has quit [Ping timeout: 260 seconds]
kmikita has quit [Ping timeout: 260 seconds]
MetaNova has quit [Ping timeout: 260 seconds]
llyyr has quit [Ping timeout: 260 seconds]
monokrome has quit [Ping timeout: 260 seconds]
galad has quit [Ping timeout: 260 seconds]
tofran5 is now known as tofran
CounterPillow has joined #ffmpeg
murb has quit [Ping timeout: 260 seconds]
upekkha has joined #ffmpeg
auri has joined #ffmpeg
theobjectivedad has joined #ffmpeg
MetaNova has joined #ffmpeg
turlando has joined #ffmpeg
murb has joined #ffmpeg
llyyrr is now known as llyyr
MrZeus has quit [Read error: Connection reset by peer]
monokrome has joined #ffmpeg
JanC has quit [Ping timeout: 255 seconds]
bencoh has joined #ffmpeg
MrZeus has joined #ffmpeg
JanC has joined #ffmpeg
m5zs7k_ is now known as m5zs7k
<mitigate>
success. mkvserver works
galad_ is now known as galad
waleee has joined #ffmpeg
darkstardevx has joined #ffmpeg
raccct has quit [Quit: Client closed]
wacko has quit [Remote host closed the connection]
Marth64 has quit [Ping timeout: 255 seconds]
Marth64 has joined #ffmpeg
wacko has joined #ffmpeg
kmikita has joined #ffmpeg
System_Error has quit [Ping timeout: 260 seconds]
<znf>
I must have been confused when I wrote this:
<znf>
-init_hw_device cuda=cuda:1 -filter_hw_device cuda -extra_hw_frames 3 -reinit_filter 1 -hwaccel nvdec -hwaccel_output_format cuda
System_Error has joined #ffmpeg
<znf>
does hwaccel nvdec make sense here? why would I put that instead of cuda?
<znf>
is there any actual difference?
<znf>
(other than the fact that "nvdec" doesn't exist in -hwaccels)
<znf>
I'm trying to figure out why I'm seeing wild differences in CPU usage with different ffmpeg versions
five6184803 has quit [Remote host closed the connection]
five6184803 has joined #ffmpeg
junaid_ has quit [Remote host closed the connection]
Tinos has quit [Ping timeout: 250 seconds]
todi has joined #ffmpeg
lavaball has quit [Remote host closed the connection]
Juest has quit [Ping timeout: 268 seconds]
gchound has joined #ffmpeg
Juest has joined #ffmpeg
j45 has quit [Ping timeout: 268 seconds]
j45 has joined #ffmpeg
gchound has quit [Quit: Leaving]
sgm has left #ffmpeg [#ffmpeg]
ivanich has quit [Remote host closed the connection]
System_Error has quit [Remote host closed the connection]
gchound has joined #ffmpeg
jemius has joined #ffmpeg
elastic_dog has quit [Ping timeout: 246 seconds]
rv1sr has quit []
<znf>
BtbN, in your readme, for `lto` it is mentioned that it *sometimes works for linux*, does this refer only to the building process? ie: if I succesfully build everything with `lto`, can I expect it to run everywhere/everytime? or should I expect the resulting binary to *sometimes not work*?
<BtbN>
your millage may vary. LTO is known to cause issues everywhere.
<znf>
oh ok then, I'll pass
<BtbN>
LTO'ing dozens and dozens of dependencies plus ffmpeg at once very much is uncharted waters
<znf>
indeed it doesn't, but if I can ping from any other machines, but not from this ASN...
<znf>
where could I plug a HTTP_PROXY variable?
<znf>
would it work properly in makeimage.sh ?
<BtbN>
no, you would have to bake it into the image
<znf>
well this is unfortunate
Blacker47 has quit [Quit: Life is short. Get a V.90 modem fast!]
<BtbN>
Just copy the download cache over to that machine?
<znf>
where is that stored?
<BtbN>
.cache
FlorianBad has quit [Quit: Konversation terminated!]
FlorianBad has joined #ffmpeg
<znf>
can I make ./makeimage skip to download part so I can copy it from another machine, but not spend 40 minutes on building everything else before?
<BtbN>
It uses the image it just built to download
<BtbN>
If you already build the two base images, they should not be rebuilt
Wenbin_Chen_ has quit [Remote host closed the connection]
Wenbin_Chen_ has joined #ffmpeg
<znf>
do I need the symlink?
<BtbN>
yes
<znf>
ok, great, let's see
gchound has quit [Quit: Leaving]
<znf>
yay, another one, opencore-amr
<BtbN>
Yeah, your network is screwed
<znf>
it's still from sf.net
<znf>
they're most likely blocking the whole ASN and/or IP ranges of this DC
<znf>
oh, actually they're using cogent's ASN
Ox7C5_ has quit [Quit: Lost terminal]
Tinos has joined #ffmpeg
lavaball has joined #ffmpeg
System_Error has joined #ffmpeg
five6184803 has quit [Remote host closed the connection]
five6184803 has joined #ffmpeg
iive has joined #ffmpeg
theracermaster has quit [Read error: Connection reset by peer]
jemius has quit [Quit: Leaving]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg
Livio has joined #ffmpeg
lavaball has quit [Quit: lavaball]
shibboleth has joined #ffmpeg
Marth64 has quit [Ping timeout: 260 seconds]
Marth64 has joined #ffmpeg
shibboleth has quit [Remote host closed the connection]
shibboleth has joined #ffmpeg
AbleBacon has quit [Read error: Connection reset by peer]
Icedream has quit [Ping timeout: 264 seconds]
Icycle has joined #ffmpeg
Icedream has joined #ffmpeg
Icycle has quit [Ping timeout: 252 seconds]
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.]
<znf>
zzz, building 5.0, 5.1, 6.0, 6.1 images takes forever
<BtbN>
Why bother building them all?
<znf>
trying to figure out where the CPU usage increase I'm seeing has started to appear
<BtbN>
That's what git bisect is for
<BtbN>
Knowing a release version is not very useful for that, due to numerous backports
<znf>
dunno how I'd effectively use bisect, considering that all I've got is N-105688 and N-113745
<znf>
which is 6k+ commits
<BtbN>
That's EXACTLY what bisect is good for
<BtbN>
6000 commits is like what, 8~10 steps?
<znf>
I don't know?
five6184803 has quit [Remote host closed the connection]
five6184803 has joined #ffmpeg
<znf>
I wouldn't even know where to look in anyway
<BtbN>
Knowing the version it "broke" in doesn't help you either way. You'll still need to bisect then
<BtbN>
And memory/cpu use increasing can just be a change of default settings, like the nvenc preset-name-change or default-bframes
<psykose>
that's exactly what bisect is good for yeah, you even have a great starting point
<psykose>
you would have already been done
<znf>
I wouldn't know where to start while building it with your script
<furq>
do you really need his script
<furq>
normally for this you would want to build with as few dependencies as possible
<furq>
otherwise it will take forever
<BtbN>
You even could use it. Just run bisect beside it, and use the commits it wants to test
<BtbN>
But yeah, seems like a waste of time vs. minimal set of deps
<furq>
6k commits is at most 13 iterations
<furq>
compared to not doing it with bisect in which case it's at most 6k iterations
<BtbN>
You could also spawn a shell in the build container, and bisect in there
<BtbN>
Got all the tools and deps you might need
KombuchaKip has quit [Quit: Leaving.]
<znf>
furq, I usually like keep my build machine and the transcoding ones completely separate, and I much prefer to not have to figure out what libs/libs versions I have to install on both ends, hence why I prefer to build it statically with his scripts, it just makes sharing the binary to multiple machines so much better
<furq>
building static is pretty easy
<furq>
especially with few dep
<furq>
s
<znf>
plus, they're not all on the same OS version, hence why a static one is just "better" in the long run for me
<BtbN>
If all you want to test is nvenc, there is nothing to even consider. It's always the same
<furq>
the same os version doesn't really matter
<furq>
unless one of them is on an ancient kernel
<BtbN>
Just bisect inside of the static build image then
<BtbN>
The entire configure incantation if in some env vars, if you want to copy the full build. Or only enable what you need to speed things up
<znf>
well, it's not nvenc causing the CPU usage difference in this case, I'd guess
KombuchaKip has joined #ffmpeg
<BtbN>
It might be, and would be my most likely guess
<BtbN>
at some point its defaults changed quite drastically when nvidia introduced the new presets
<znf>
now that I've done this, at least I'll try to see what happens and keep the binaries around at least
<znf>
but seems that 5.0 doesn't compile
Marth64 has quit [Ping timeout: 260 seconds]
<BtbN>
5.0 hasn't been on CI in ages, so that wouldn't surprise me
Marth64 has joined #ffmpeg
<furq>
i take it the script only uses up-to-date versions/master of the dependencies
<BtbN>
mostly
<furq>
right
<BtbN>
some scripts have special cases in it, to use older versions for older branches
<znf>
some libs have different 5.0/5.1/6.0 etc. settings I've seen
<BtbN>
but most of them just always use the latest versions
<furq>
yeah
shibboleth has quit [Remote host closed the connection]
<furq>
can't imagine there's any desire to test everything for old version compat
<znf>
5.1 seems to work fine, let me try 5.0 again, at least I'll get an error message (as it scrolled out)
<BtbN>
Well, it still builds 4.4, 5.1 and 6.1 daily
<znf>
not that I'm expecting anyone to fix
<BtbN>
so 5.1 working and 5.0 being broken smells like missing backport
shibboleth has joined #ffmpeg
<znf>
ERROR: libvmaf >= 1.5.2 not found using pkg-config