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
rv1sr has quit []
<JesseTG>
klaxa, I'm not sure that's what I'm looking for (that's for file I/O, not my issue here), but I appreciate the effort anyway.
DauntlessOne47 has joined #ffmpeg
Kei_N has quit [Read error: Connection reset by peer]
DauntlessOne4 has quit [Ping timeout: 276 seconds]
DauntlessOne47 is now known as DauntlessOne4
Kei_N has joined #ffmpeg
olndrxyz has quit [Ping timeout: 252 seconds]
DauntlessOne4 has quit [Ping timeout: 244 seconds]
olndrxyz has joined #ffmpeg
cryptic has joined #ffmpeg
EmleyMoor has quit [Ping timeout: 265 seconds]
EmleyMoor has joined #ffmpeg
olndrxyz has quit [Read error: Connection reset by peer]
olndrxyz has joined #ffmpeg
ttys000 has quit [Read error: Connection reset by peer]
ttys000 has joined #ffmpeg
xx has quit [Ping timeout: 264 seconds]
olndrxyz has quit [Read error: Connection reset by peer]
olndrxyz has joined #ffmpeg
squeaktoy has joined #ffmpeg
rvalue has joined #ffmpeg
vincejv has quit [Ping timeout: 244 seconds]
olndrxyz has quit [Read error: Connection reset by peer]
olndrxyz has joined #ffmpeg
olndrxyz has quit [Remote host closed the connection]
olndrxyz has joined #ffmpeg
olndrxyz has quit [Read error: Connection reset by peer]
olndrxyz has joined #ffmpeg
olndrxyz has quit [Read error: Connection reset by peer]
olndrxyz has joined #ffmpeg
olndrxyz has quit [Remote host closed the connection]
minimal has quit [Quit: Leaving]
JesseTG has quit [Ping timeout: 252 seconds]
Vonter has quit [Read error: Connection reset by peer]
Sketch has quit [Remote host closed the connection]
Sketch has joined #ffmpeg
wziko has joined #ffmpeg
wziko has quit [Ping timeout: 265 seconds]
Blacker47 has joined #ffmpeg
j45 has quit [Ping timeout: 276 seconds]
j45 has joined #ffmpeg
JanC has joined #ffmpeg
JanC is now known as Guest9592
Guest9592 has quit [Killed (platinum.libera.chat (Nickname regained by services))]
rsx has joined #ffmpeg
wziko has joined #ffmpeg
wziko has quit [Ping timeout: 248 seconds]
deetwelve has quit [Quit: null]
alexherbo2 has quit [Remote host closed the connection]
alexherbo2 has joined #ffmpeg
JesseTG has joined #ffmpeg
deetwelve has joined #ffmpeg
lavaball has joined #ffmpeg
Thurmond has joined #ffmpeg
<Thurmond>
i have an annoying problem with a set of OGM files and my hope is that someone here can help me find a solution
<Thurmond>
the original files are .ogms all with the same general layout: stream 0 (video) = ogg/mpeg. stream 1 (audio) = aac (English). Stream 2 (audio) = aac (Japanese). Stream 3 (subs) = English.
<Thurmond>
these files play very smoothly in general, but have a couple of weaknesses, the most obvious being that the subs don't appear at the right times, if at all, when played.
<Thurmond>
using ffmpeg to convert to mkv shows some of the underlying issues. no timestamps in the video stream (probably the cause of the subs not playing) and the audio streams are marked as Vorbis though ffmpeg says they are aac (and during the conversion it throws errors "not a vorbis packet")
<Thurmond>
at any rate, using ffmpeg to convert from .ogm to .mkv is successful, and playback of the resulting mkv works in most respects (audio is fine, timing/content of subs are fine, video quality is good) EXCEPT...
Vonter has quit [Ping timeout: 244 seconds]
<Thurmond>
in "panning" shots - say a slow upward crawl - the pan is choppy. it will jump and skip. it doesn't lose frames or do anything bad to the overall timing. it's just no longer a smooth pan.
<Thurmond>
at first, I thought this was a weakness in VLC, so I got a copy of IINA and tried that, but the playback issues are exactly the same.
<Thurmond>
SO...
rvalue has quit [Ping timeout: 265 seconds]
<Thurmond>
suggestions on how to convert these files to another format (to resolve the timing issues) but without the jumpy playback?
trillion_exabyte has quit [Ping timeout: 246 seconds]
trillion_exabyte has joined #ffmpeg
rv1sr has quit []
Arokh has quit [Ping timeout: 252 seconds]
sihloo has joined #ffmpeg
l4yer has joined #ffmpeg
StephenLynx has joined #ffmpeg
lucasta has quit [Quit: Leaving]
alexherbo2 has quit [Remote host closed the connection]
Arokh has joined #ffmpeg
arkanoid has joined #ffmpeg
<arkanoid>
Hello! I'm trying to give rtsp input to ffmpeg using "ffmpeg -i "rtsp://foo:bar@192.168.1.1/stream" -f null" but all I get is "[rtsp @ 0x76bd7820] Could not find codec parameters for stream 0 (Video: hevc, none): unspecified size"
<arkanoid>
if I use "-c copy" I get "[null @ 0x7e8090] dimensions not set \n Could not write header for output file #0 (incorrect codec parameters ?): Invalid argument"
<arkanoid>
I don't understand what the error is saying to me. The same rtsp stream plays correctly in vlc
<arkanoid>
what is the simplest ffmpeg line I can use to get rtsp stream and save to file, no transcoding, no, conversion
Arokh has quit [Quit: //ThisShouldNeverHappen]
nasso has joined #ffmpeg
l4yer has quit [Ping timeout: 248 seconds]
Arokh has joined #ffmpeg
rv1sr has joined #ffmpeg
Arokh has quit [Ping timeout: 244 seconds]
Arokh has joined #ffmpeg
<arkanoid>
I don't understand why ffmpeg is looking for a decoder device if my pipeline wants copy
Unit640 has joined #ffmpeg
<Unit640>
I tried to make a slideshow with ffmpeg but it ended up overwriting all my input images with the same image instead. The one time I didn't have backups of the data... hundreds of carefully sorted images. I didn't think that the `-i` flag would ever output to those images -- only use them as input... :/
<Unit640>
(Slideshow exported as a video file.)
Arokh has quit [Quit: //ThisShouldNeverHappen]
Arokh has joined #ffmpeg
minimal has joined #ffmpeg
Arokh has quit [Ping timeout: 252 seconds]
Arokh has joined #ffmpeg
Arokh has quit [Ping timeout: 248 seconds]
Arokh has joined #ffmpeg
manwithluck has quit [Ping timeout: 252 seconds]
manwithluck has joined #ffmpeg
JanC has joined #ffmpeg
JanC is now known as Guest8798
Guest8798 has quit [Killed (mercury.libera.chat (Nickname regained by services))]
<arkanoid>
can't succeed in the most trivial ops, moving to gstreamer
<arkanoid>
what happened to ffmpeg? It used to be a reliable multimedia swiss army knife
<ops>
excuse me?
<arkanoid>
haha
<arkanoid>
I expect "haha" user to show up now
<Thulinma>
arkanoid: I'm not an ffmpeg dev, but AFAIK that means the RTSP description doesn't include the codec init data, which quite a few formats expect to find in the headers. RTSP does allow sending it as part of the video data as well, but perhaps it's not getting far enough for that. Does that help at all?
<ops>
init data? is that like the moov atom?
<arkanoid>
"ffmpeg -rtsp_transport tcp -i "$RTSP_URL" -c copy -f mpegts -y foo.ts" -> [h264_v4l2m2m @ 0x77f2c530] Could not find a valid device [h264_v4l2m2m @ 0x77f2c530] can't configure decoder [mpegts @ 0x7c0e40] dimensions not set Could not write header for output file #0 (incorrect codec parameters ?): Invalid argument
<arkanoid>
why it wants a decoder, when I don't want to decode, just store
<Thulinma>
ops: I like calling it "init data", but it's also often referred to as "codec private data". Some codecs need out-of-band information to initialize themselves for decoding. The moov atom does contain this stuff as well for MP4, but it's just a small part of it.
<ops>
oh ok
Arokh has quit [Ping timeout: 265 seconds]
Everything has quit [Ping timeout: 272 seconds]
<Thulinma>
arkanoid: I'm not sure. That's strange indeed - it's a weird decoder it's trying to open to, not the default on most systems as far as I'm aware.
<Thulinma>
too*
l4yer has joined #ffmpeg
JanC has joined #ffmpeg
JanC has quit [Killed (erbium.libera.chat (Nickname regained by services))]
Arokh has joined #ffmpeg
wziko has joined #ffmpeg
Arokh has quit [Ping timeout: 265 seconds]
Blacker47 has quit [Quit: Life is short. Get a V.90 modem fast!]
wziko has quit [Ping timeout: 265 seconds]
Blacker47 has joined #ffmpeg
rizino has quit [Quit: ZNC 1.8.2+deb3.1+deb12u1 - https://znc.in]
rizino has joined #ffmpeg
jannau has joined #ffmpeg
<jannau>
hej, do I misremember that it was documented that frame buffers for get_buffer*() should be over-allocated like update_frame_pool() does with 'size[i] + 16 + STRIDE_ALIGN - 1'?
Arokh has joined #ffmpeg
<jannau>
looking at firefox VP9 crashes on arm64 caused by over reads in put_8tap_16h_34 and co when the frame size is a multiple of the page size like 1280x640
robobub has quit [Quit: Connection closed for inactivity]
lavaball has quit [Remote host closed the connection]
<JEEB>
jannau: ah, a familiar name :) (just noticed your nick on the fedora asahi linux article)
Arokh has quit [Ping timeout: 265 seconds]
<jannau>
JEEB: yes, that's taking most of my free time these days
<JEEB>
:)
<JEEB>
anyways, I recall overallocation was documented somewhere, but things like av_frame_get_buffer mostly talk of alignment
<JEEB>
right, linesize
<JEEB>
that would enable overallocation, but it does indeed take us back to the point of whether SIMD is supposed to be overreading
<JEEB>
(once again, it was mentioned somewhere but on which level I'd have to check)
Arokh has joined #ffmpeg
<JEEB>
could be packet buffers as opposed to frame buffers
<jannau>
might be related to edge emulation but I forgot all details
<jannau>
yes, it could be input what I remember. there FF_INPUT_BUFFER_PADDING_SIZE is clearly documented
<JEEB>
is this btw repro'able with fftools?
<JEEB>
or only in ffvp9 ripped into Firefox?
dv_ has quit [Quit: WeeChat 3.8]
rsx has quit [Quit: rsx]
<jannau>
it repos with valgrind / ffmpeg if I remove the overallocation from update_frame_pool()
<jannau>
the '+ 16' has been there since a long time. my git annotate is currently in 2005
<jannau>
it shouldn't be needed and it's and issue in either vp9 which switches not soon enough to edge emulation or in the neon asm which over reads too much
JanC_ has joined #ffmpeg
JanC has quit [Killed (calcium.libera.chat (Nickname regained by services))]
JanC_ is now known as JanC
sentriz has quit [Ping timeout: 252 seconds]
Arokh has joined #ffmpeg
sentriz has joined #ffmpeg
Arokh has quit [Ping timeout: 265 seconds]
minimal has quit [Quit: Leaving]
Arokh has joined #ffmpeg
wziko has joined #ffmpeg
wziko has quit [Max SendQ exceeded]
wziko has joined #ffmpeg
Everything has joined #ffmpeg
wziko has quit [Ping timeout: 260 seconds]
System_Error has quit [Remote host closed the connection]
Arokh has quit [Ping timeout: 252 seconds]
wziko has joined #ffmpeg
Arokh has joined #ffmpeg
JanC_ has joined #ffmpeg
JanC is now known as Guest1677
Guest1677 has quit [Killed (molybdenum.libera.chat (Nickname regained by services))]
Blacker47 has quit [Quit: Life is short. Get a V.90 modem fast!]
wziko has joined #ffmpeg
wziko has quit [Ping timeout: 248 seconds]
wziko has joined #ffmpeg
wziko has quit [Ping timeout: 264 seconds]
wziko has joined #ffmpeg
Marth64 has quit [Quit: Leaving]
wziko has quit [Ping timeout: 265 seconds]
<bcheng>
Is there anyway to skip the av1_metadata bsf while doing an av1 encode? I just want the raw bitstream that came from the encoder, no need for any parsing/muxing
Marth64 has joined #ffmpeg
wziko has joined #ffmpeg
<BtbN>
Pretty sure without it the resulting raw bitstream is unparseable, that's why the raw "encoder" inserts it
<bcheng>
I'm trying to debug an encoder that's generating bad headers, which makes the av1_metadata bsf fail, so I can't get any output out of ffmpeg :/
wziko has joined #ffmpeg
<BtbN>
Just comment it out when you're debugging anyway
<bcheng>
sure... was hoping to not have to build it
<bcheng>
I'm dealing with a cursed VAAPI on windows stack
<JesseTG>
Is it possible to list the possible frame sizes of a video camera with the C interface? I'm implementing an API that allows clients to request webcam input of a specific size, but I want to pick something close enough if it's not supported.
CruxOfTheB has quit [Ping timeout: 252 seconds]
CruxOfTheB has joined #ffmpeg
<BtbN>
Can ffmpeg list that at all?
bitbinge has quit [Remote host closed the connection]
Everything has quit [Quit: leaving]
mw1144 has joined #ffmpeg
<ePirat>
JesseTG, BtbN, I am relatively sure avdevice has no way to do that.
<Unit640>
I won't blame the ffmpeg developers for my lack of backups, but why would using anything as -i cause it to write to those under any circumstances?
<Unit640>
I thought it was a safe command because the output was unrelated to the -i dir/files.
<JesseTG>
ePirat, dang. Oh well, I guess I'll just have to fall back to whatever libavdevice says is the default and hope for the best
<JesseTG>
Thanks anyway
<BtbN>
Nothing should (and does, to my knowledge) ever write to the input file
<BtbN>
If you did not quote that *.png, that will expand to a lot more options, and all but the first one will be an output
<Unit640>
I may have tried "/home/a/Desktop/tmp/nostalgia/"*.png before/after it when it didn't seem to work. Unsure when exactly the destruction happened.
<Unit640>
(Of course, I have the photos still on many backups, but not carefully picked out and sorted like I had them in this dir.)
<BtbN>
Well, shell expansion there just makes no sense, and _will_ cause all but the first file it expands to to be overwritten
JesseTG has quit [Quit: Leaving]
<BtbN>
FFmpeg will just see ffmpeg -framerate 1/3 -i 1.png 2.png 3.png 4.png ...
<BtbN>
so all but 1.png are outputs
<Unit640>
In my mind, it meant "include all the PNG files from this dir".
<BtbN>
Your shell does not agree with your mind I'm afraid
<Unit640>
I hate the "shell expansion" stuff, frankly. It's caused me other similar issues recently.
<BtbN>
FFmpeg has its own syntax if you want to turn a sequence of images into a video
<BtbN>
Even if all those png images had an -i in front of them, it would not result in them turning into a video
<BtbN>
Also, quoting it like you did here would just result in it not finding the literal "*.png" file
<Unit640>
I think it still wouldn't have looked good even if it worked. I kind of want a smooth and very slightly "zoom-in" effect for the duration of each picture in the slideshow.
<Unit640>
But I've searched in vain for a working and good video editor, although it seems like this should be scripted somehow.
<ePirat>
also never use -y when tinkering with ffmpeg commands
<Unit640>
Yeah, that -y did worry me a bit.
<BtbN>
Yeah, without -y it'd just have asked you if you really want to overwrite your "input" pictures
<BtbN>
Why did you put it then?
coldfeet has joined #ffmpeg
bitbinge has joined #ffmpeg
<Unit640>
ChatGPT added it, but I noticed it and didn't remove it. (Of course I don't blindly run anything from it without carefully reading through it, but here I missed the shell expansion stuff.)
<Unit640>
At worst I thought it'd give some syntax error.
<BtbN>
It's possible to craft what you want with an ffmpeg cli, but it'd be a surprisingly huge commandline, with a huge filter_complex
LainExperiments has joined #ffmpeg
LainExperiments has quit [Client Quit]
mw1144 has quit [Quit: Client closed]
LainExperiments4 has joined #ffmpeg
SuicideShow has quit [Ping timeout: 260 seconds]
SuicideShow has joined #ffmpeg
Mister_Magister_ has joined #ffmpeg
Mister_Magister has quit [Ping timeout: 265 seconds]
Mister_Magister_ is now known as Mister_Magister
Traneptora has quit [Quit: Quit]
<ePirat>
BtbN, even then zoompan can not do properly what it says on the tin
<BtbN>
I'd probably use Davinci Resolve if I wanted a video like that. Though it seems insanely overkill
<ePirat>
I think shotcut should be able to do it
lavaball has quit [Remote host closed the connection]