<brianp>
is there any limit on filters? I have about 400 drawtext filters applied, and they all use the same fontfile and then 1/5 times the encoding fails because "Cannot find a valid font for the family Abel". But the font file is present, and encoding work 4/5 times. So there's no problem with the font file, or the path.
<JEEB>
you'd have to trace the process and check which API fails. could be open file handles or something general like that?
<JEEB>
ltrace for example
FH_thecat has joined #ffmpeg
vtorri22 has joined #ffmpeg
<brianp>
Cool, I'll check that out. This was also my guess that maybe the fontfile was being opened/locked on each filter usage. Which could cause some issue. Although I only seem to run into it after a threshold of filter usage. Almost like the more filters the higher the chance of not closing the file.
<aaabbb>
brianp: could you be running into resource limits? try increasing the stack with ulimit (just a guess)
<aaabbb>
idk how ffmpeg filters work inside but maybe some thread is failing because it's running out of reources
Some_Person has joined #ffmpeg
<brianp>
good call- I just checked and it seems resource friendly. The command fails before it attempts processing it seems. So never actually tried to perform the encoding. But it seems each filter may be attempting to call fopen on the file in preparation. So I'm probably running into some kind of allowed handles per file limit.
<JEEB>
> It is connected with C-runtime limitation of max number of opened files.
<aaabbb>
/6
<JEEB>
but do check if that still is the case
MRiddickW has joined #ffmpeg
<Swedaniel>
im still new with theese ffmpeg command lines, but is this the situation when the '2nd pass' function can help?
fart-omelette40 has quit [Ping timeout: 240 seconds]
fart-omelette40 has joined #ffmpeg
<brianp>
could be a solution. Alternatively, there's code managing the font file before providing it to ffmpeg. I could also create multiple font files and distribute the load. A shame it can't be defined a single time per encode, and used in multiple filters. 1 open would be ideal :D
Some_Person has quit [Ping timeout: 255 seconds]
<Swedaniel>
Yeah i think i read something about using 2nd Pass function would basicly let some procedures complete in one order and then launch the '2nd Pass' with commands to spare resources during the first pass.. i might be the only lucky one and is still finding success with most default settings over here....
ZedHedTed has quit [Ping timeout: 255 seconds]
<Swedaniel>
I have seen WinFF v.1.5.5 have an second pass function
<furq>
brianp: could you maybe do this with a subtitle file and the subtitles filter
<furq>
probably easier unless you're doing dynamic positioning based on timestamps or something like that
<brianp>
I don't think, as the example failing right now has 8 bits of text, in different places over video, where the text changes every second. But yeah the multi-position is the problem
ZedHedTed has joined #ffmpeg
Some_Person has joined #ffmpeg
Shuriko has quit [Ping timeout: 272 seconds]
<brianp>
huh. Maybe subtitles could work. I was under a distinct impression only 1 subtitle could be displayed at a time. And it seems like I was wrong about that
<JEEB>
with non-broadcast subtitles multiple can overlap
<JEEB>
you can utilize something like aegisub to create ASS stylized subtitles, and then render them out
<brianp>
yeah cool. This actually seems hopeful. a few more temp files to write / manage but feels like better processing.
<furq>
better to have a 400 line subtitle file than a 400 line filterchain
<Swedaniel>
does the program collect info from an font folder during this operation? ive tried read all messages im not sure if im understanding anything correct here.. but the font folder is an protected folder and windows can have security protocols to an folder during an specific event on the computer... example would be when the codec database is accessed,
<Swedaniel>
the font folder is protected.. or any other folder
<furq>
this isn't on windows
<Swedaniel>
so basicly when you access the codec the font folder is protected and fail?!
<brianp>
I was wrong about 400 filters. It was 728 lines. '=D
Ogobaga has quit [Quit: Konversation terminated!]
Ogobaga has joined #ffmpeg
<brianp>
Swedaniel the issue is that every filter i define `fontfile:/some/path/font.otf` tries to take an open handle to the file. But the OS limits the amount of times a file can be opened. So after 512 filters the font can no longer be opened, and the encoding fails.
<Swedaniel>
ooh, ah yeah that make sense allso..
<brianp>
but this subtitle solution looks promising. I'll try that.
<brianp>
if I wrote any C i'd look at making a PR to try and re-use an open fontfile handle but that's getting out of my depths :D
<Swedaniel>
i guess anyone would regret not becomeing an RAM memory expert to know how to load the font into the RAM Memory to bypass the limit
<Swedaniel>
maybe if you store the font into an RAM Drive you can bypass OS restrictions
<Swedaniel>
or atleast the font would now be accessd very fast :P
<aaabbb>
what should be done about transcoding a video with a lot of artifacts from generation loss with x265? is denoising a good idea?
wangbin has joined #ffmpeg
lavaball has joined #ffmpeg
hjckr has quit [Remote host closed the connection]
lusciouslover has quit [Ping timeout: 258 seconds]
CoreX has quit [Server closed connection]
CoreX has joined #ffmpeg
hjckr has joined #ffmpeg
wangbin has quit [Quit: Connection closed]
vtorri22 has quit [Ping timeout: 246 seconds]
LeoNice1981 has quit [Read error: Connection reset by peer]
synapt has quit [Server closed connection]
synapt has joined #ffmpeg
ecapi_ has quit [Remote host closed the connection]
elastic_dog has quit [Ping timeout: 252 seconds]
ecapi has joined #ffmpeg
ecapi_ has joined #ffmpeg
elastic_dog has joined #ffmpeg
ecapi has quit [Ping timeout: 255 seconds]
jtgd has joined #ffmpeg
Ogobaga has quit [Quit: Konversation terminated!]
Ogobaga has joined #ffmpeg
kron has quit [Read error: Connection reset by peer]
qaph has joined #ffmpeg
kron has joined #ffmpeg
qaph has quit [Ping timeout: 246 seconds]
Vonter has quit [Ping timeout: 246 seconds]
Vonter has joined #ffmpeg
kron has quit [Remote host closed the connection]
qaph has joined #ffmpeg
andrewrk has quit [Server closed connection]
qaph is now known as kron
andrewrk has joined #ffmpeg
elastic_dog has quit [Ping timeout: 258 seconds]
qaph has joined #ffmpeg
kron has quit [Ping timeout: 258 seconds]
kron has joined #ffmpeg
elastic_dog has joined #ffmpeg
qaph has quit [Ping timeout: 255 seconds]
qaph has joined #ffmpeg
kron has quit [Ping timeout: 240 seconds]
ivanich has quit [Remote host closed the connection]
navi has joined #ffmpeg
qaph is now known as kron
kron has quit [Ping timeout: 246 seconds]
gothicserpent has quit [Server closed connection]
zeromind has quit [Server closed connection]
zeromind has joined #ffmpeg
gothicserpent has joined #ffmpeg
<redeeman>
aaabbb: yes, it is full range(pc), but is there any correct way to specify to get rid of the warning?
navi has quit [Quit: WeeChat 4.0.4]
<JEEB>
that's more on us as the FFmpeg project if you are not specifying that pix_fmt yourself
<aaabbb>
redeeman: if the source is yuvj420p then there's no way to get rid of it. it's just telling you that the format of the input is going to be deprecated
<JEEB>
since we deprecated the pixel format but it was really complex to get to its actual removal
<furq>
didn't it just get removed the other day
<redeeman>
but what would replace it then?
<JEEB>
the flag from which (pc) is looked up from :)
<furq>
yuvj is (was) a hack to accommodate yuv with the full range flag set
<aaabbb>
furq: no there's just a plan on the ml
<JEEB>
it would just be yuv420p (pc)
<JEEB>
aaabbb: no there is an actual patch set
<furq>
yeah that's what i saw
<JEEB>
second one actually, there was an earlier one that got stuck
<furq>
did that get merged yet
<aaabbb>
JEEB: oh awesome!
<JEEB>
its prerequisites did
<furq>
nice
<aaabbb>
i'm excited for yuvj's final removal then !
<JEEB>
yea, long overdue
<aaabbb>
now just to wait another 8 years for the update to trickle down to debian's packages lmao
<redeeman>
so is it complaining that the input I have is yuvj420p(pc, bt709) ?
<JEEB>
redeeman: it's complaining that somewhere in the flow yuvj pixel format is being utilized
<furq>
the warning message comes from the library, not from ffmpeg itself
<redeeman>
thats what it says for input
<furq>
it's warning ffmpeg to not use this because it's been deprecated
<furq>
but it has nowhere to do that except in your terminal
<aaabbb>
redeeman: it's totally harmless
<redeeman>
okay so once this gets removed upstream, i shouldnt need to change anything as the input will just be detected as yuv420p instead of yuvj420p
<furq>
yeah you can just ignore it and hopefully before too long ffmpeg will have stopped using it and the warning will go away
<furq>
right
<redeeman>
excellent, thanks
Anth0mk has quit [Server closed connection]
Anth0mk has joined #ffmpeg
Blacker47 has joined #ffmpeg
lexano has joined #ffmpeg
noonien81 has joined #ffmpeg
noonien8 has quit [Ping timeout: 258 seconds]
noonien81 is now known as noonien8
Swedaniel has quit [Quit: Ping timeout (120 seconds)]
<Guest75>
https://github.com/FFmpeg/FFmpeg/archive/refs/tags/$(FFMPEG_HASH).tar.gz works but the shasum is different (as generated by github I guess) and seems to be some slight differences during the build
esc--ape-- has joined #ffmpeg
<JEEB>
if you actually want tagged releases then there were IIRC downloads for those. the snapshots from the gitweb probably were generated by git automagically
<JEEB>
unless they somehow were signed by the signing key
<_zoggy_>
odd, yeah going to the tag then tree then it ends up showing you 6.1 version but if you go to the tag then shortlog then tree it does stay within 6.0 branch. sorry about that
yazooq has joined #ffmpeg
Kruppt has quit [Quit: Leaving]
elastic_dog has quit [Ping timeout: 246 seconds]
Shine_ has joined #ffmpeg
elastic_dog has joined #ffmpeg
bitbinge has quit [Remote host closed the connection]
klotzhauer has quit [Server closed connection]
klotzhauer has joined #ffmpeg
Ogobaga has quit [Ping timeout: 255 seconds]
Shine_ has quit [Read error: Connection reset by peer]
Ogobaga has joined #ffmpeg
vampirefrog has quit [Quit: Leaving]
lavaball has quit [Remote host closed the connection]
rv1sr has quit []
waleee has joined #ffmpeg
Ogobaga has quit [Read error: Connection reset by peer]
Ogobaga has joined #ffmpeg
ivanich_ has joined #ffmpeg
ivanich has quit [Quit: Konversation terminated!]
adendrag33209987 has quit [Quit: adendrag33209987]
<ShaedS>
hey
<ShaedS>
why's this command taking so long? ffmpeg -i IMG_4526-nadio.mov -i wrazdubs-s3e28-vince-grave-cut.m4a -map 0 -map 1:a -c:v copy IMG_4526-playme.mov it's running at 20x i thought i'd be 200x without re encoding
<furq>
you're reencoding the audio
<furq>
i guess you wanted -c copy
JasonSilver has quit [Remote host closed the connection]
JasonSilver has joined #ffmpeg
<LimeOn>
hello, the previous questions makes me wonder the following: which is the proper method to map streams? or whats the difference between: 1:a:0 / 1:a or 1:0 for example
<LimeOn>
question*
<furq>
1:a is all audio streams, 1:a:0 is the first audio stream, 1:0 is the first stream
<ShaedS>
furq: but: a -c:v copy
<furq>
the v is for video
<ShaedS>
where do I put -c copoy for audio?
<furq>
replace -c:v copy with -c copy
<ShaedS>
arg thta com,,mamd dodm
<ShaedS>
t eve copy mmy video it just copied audio
<ShaedS>
ah yeah it's crunchin itg out with -c copy isntead of -c:v copy
<ShaedS>
oh but no video
<ShaedS>
how do I copy the audio and vidoe without re-encode?
<ShaedS>
that syntax i got from a stackexchanging saying that was the answer but it's clearly not
<ShaedS>
ffmpeg -i IMG_4526-nadio.mov -i wrazdubs-s3e28-vince-grave-cut.m4a -c:a copy -c:v copy IMG_4526-playme.mov only copies the aduio there is no video
<ShaedS>
ffmpeg -i IMG_4526-nadio.mov -i wrazdubs-s3e28-vince-grave-cut.m4a -map 0:v? -map 1:a -c copy IMG_4526-playme.mo doesn't copy the video either
lucasta has joined #ffmpeg
ivanich_ has quit [Remote host closed the connection]