michaelni changed the topic of #ffmpeg-devel to: Welcome to the FFmpeg development channel | Questions about using FFmpeg or developing with libav* libs should be asked in #ffmpeg | This channel is publicly logged | FFmpeg 6.1.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
Marth64 has joined #ffmpeg-devel
<Marth64>
good evening
rvalue has quit [Ping timeout: 252 seconds]
rvalue has joined #ffmpeg-devel
mkver has quit [Ping timeout: 268 seconds]
<compn>
Lynne, my brainnnnn
iive has quit [Quit: They came for me...]
Marth64 has quit [Ping timeout: 264 seconds]
Gramner has quit [Ping timeout: 264 seconds]
Gramner has joined #ffmpeg-devel
kurosu has quit [Quit: Connection closed for inactivity]
Marth64 has joined #ffmpeg-devel
thilo has quit [Ping timeout: 272 seconds]
thilo has joined #ffmpeg-devel
tufei has quit [Remote host closed the connection]
tufei has joined #ffmpeg-devel
thilo has quit [Ping timeout: 268 seconds]
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
thilo has joined #ffmpeg-devel
Marth64 has quit [Remote host closed the connection]
jamrial has quit []
System_Error has quit [Ping timeout: 260 seconds]
System_Error has joined #ffmpeg-devel
markh has quit [Remote host closed the connection]
Sean_McG has quit [Quit: leaving]
markh has joined #ffmpeg-devel
Martchus_ has joined #ffmpeg-devel
Martchus has quit [Ping timeout: 260 seconds]
AbleBacon has quit [Read error: Connection reset by peer]
<cone-428>
ffmpeg Marth64 master:8991cedd6aa2: avcodec/ccaption_dec: clarify log message when out-of-display columns are ignored
<cone-428>
ffmpeg Marth64 master:590b2d6517ee: avcodec/ccaption_dec: use consistent naming convention of Closed Captions
<sdc>
question re: how check_func works (in x86 asm if that matters)
<sdc>
I'm a bit confused on how to use it specifically with how call_ref and call_func are defined
* sdc
I see that check_func( func_new =, ... ) but not sure how func_ref = check_func( ... ) works in terms for getting the symbol for the reference function
<jdek>
Works based on the setup function
<jdek>
The setup function is called with no cpuflags enabled such that it always resolves the default C functions
<Gramner>
sdc: first all c functions are called and the function pointers are stored in a tree structure, then later when the asm functions are tested it looks up that function in the tree to obtain a pointer to the reference function
<Gramner>
the tree lookup is by name, i.e. the string passed to check_func()
s55 has joined #ffmpeg-devel
<Gramner>
it's implemented as a self-balancing LLRB binary search tree so lookups are very fast even with a huge number of functions
<cone-428>
ffmpeg Andreas Rheinhardt master:cc95fd4531fc: doc/bitstream_filters.texi: Document types used by filter_units
<wbs>
mkver: ok, I'll test it out within a few days, hopefully sooner than that
<sdc>
ah cool thank you jdek, Gramner
<mkver>
wbs: Thanks. Take your time.
pzy has quit [Ping timeout: 268 seconds]
rvalue has quit [Ping timeout: 264 seconds]
rvalue has joined #ffmpeg-devel
hamzah has joined #ffmpeg-devel
cubicibo has quit [Ping timeout: 250 seconds]
cubicibo has joined #ffmpeg-devel
TheSashm_ has quit [Remote host closed the connection]
TheSashmo has joined #ffmpeg-devel
TheSashmo has quit [Client Quit]
cubicibo has quit [Quit: Client closed]
TheSashmo has joined #ffmpeg-devel
rodgort has quit [Quit: Leaving]
zsoltiv_ has joined #ffmpeg-devel
rodgort has joined #ffmpeg-devel
<Marth64>
ok so mpeg12video closed caption decoding has a conceptual bug
<Marth64>
if source has two types of closed captions (e.g. A53CC and SCTE-20), they will write over eachother and you will get weird output like THANK YOU --> THTHANANK K YOYOUU
<Marth64>
i am working on a fix
<Marth64>
there is real-world sample material with both
<Marth64>
fix is to pick the first one and stick with it (done), and allow user to choose forcefully as an option if that doesn't work (in progress)
<cone-428>
ffmpeg Michael Niedermayer master:5db09574dfd4: avcodec/8bps: Consider width in the minimal size check
<cone-428>
ffmpeg Michael Niedermayer master:b2d7cbc378fa: avformat/concatdec: Check in and outpoints to be to produce a positive representable duration
<cone-428>
ffmpeg Michael Niedermayer master:dc89cf804a81: avcodec/vorbisdec: Check remaining data in vorbis_residue_decode_internal()
<cone-428>
ffmpeg Michael Niedermayer master:e0e30e07a175: avcodec/proresenc_kostya: Remove bug similarity text
<Marth64>
pick the first one seen with actual cc byte data*
<Marth64>
does anyone knows if MPEG-1 allows closed caption?
<Marth64>
looks like no :)
<Marth64>
according to search online
cubicibo has joined #ffmpeg-devel
Traneptora has quit [Quit: Quit]
BtbN has quit [Remote host closed the connection]
BtbN has joined #ffmpeg-devel
<kierank>
Marth64: maybe in vcd?
<Marth64>
kierank: yeah good call. I have exactly 1, gonna try it :)
<Marth64>
looks like somebody tried to fix this once, but the patch set fell off (I can assume because feedback wasn't addressed)
<Marth64>
Honestly ffmpeg's CC decoder is really good. it just needs some little fixes
<Marth64>
the assa output is great
<Marth64>
roll-up captions are cursed to begin with, so I'll leave those alone
<kierank>
legally required in the usa...
<Marth64>
i can't think of any other format that does roll-up mode either so i have nothing to cleanly archive them to
<Marth64>
at least with pop-on ffmpeg can render them to ass nicely and that serves a spiritual backup of what they should look like
BtbN has quit [Remote host closed the connection]
BtbN has joined #ffmpeg-devel
<kierank>
is cea608 -> ass -> cea608 bitexact?
<kierank>
it's one of the reasons I don't use ffmpeg cea decoder as I don't trust ass
<Marth64>
it is not, but it is close in both positioning and formatting for the average pop-on captions. it effectively renders the canvas and positions text literally x/y. it does a decent job. however, the limits to the approach are that ASS doesn't have justification with x/y positioning, only with the preset positions, so text supposed to be center justified looks awkward.
<Marth64>
so for an average sample it will produce a pretty good synthetic rendering of the cc.
hamzah has quit [Ping timeout: 250 seconds]
<Marth64>
a good monospace font is a must
<Oneric>
> ASS doesn't have justification with x/y positioning, only with the preset positions, so text supposed to be center justified looks awkward
<Oneric>
you can very much combine \an and \pos to get each line center-justified and the event at a fixed position
<Oneric>
or am i misunderstanding what you want to do?
<Oneric>
what you can’t do is using \an for implicit positioning but using different justification settings
<Marth64>
it's the former. i didn't think it was possible, i will try it out :D
<Marth64>
if that's the case, maybe the decoder can be improved to make center justified text not look awkward
microchip_ has quit [Quit: There is no spoon!]
microchip_ has joined #ffmpeg-devel
<Lynne>
how did you end up converting the bitmap subs to ass?
<Marth64>
CC is not bitmap, they're coded bytes representing characters/formatting basically
<Marth64>
using lavfi input one can get the eia608 stream and encode it to ass
<Marth64>
there can be multiple languages, afaik one limitation now is that method only allows the first language (or "service 1")
<Oneric>
as a general warning though, ffmpeg’s existing converter(s) to ASS have issues with it’s output not conforming to standard ASS
<Oneric>
some easy ones of which would be fixed by patches i’ve first sent two years ago, but no one (besides one off-topic troll for v1) ever looked at
<Marth64>
Oneric: I did notice a behavior in the CC encoder where it prints multiple \pos in one cue. e.g. "\pos(x,y)Line1\N\pos(x,y)Line2" . Is that allowed? I couldn't get that to work without using seperate cues for each position.
<Marth64>
s/encoder/decoder
<Oneric>
Nope, only one \pos per event takes effect (and that’s not one of the things fixed by my patches)
<cone-428>
ffmpeg Martin Storsjö master:cd420c294941: makefile: Clean up missed object files with "make clean"
<cone-428>
ffmpeg Martin Storsjö master:c31d23265600: libavcodec: Don't include libavcodec/x86/vvc/Makefile on any architecture
<Marth64>
I am curious to apply/test them if they're up online somewhere. I do think CC->ASS is a promising fit with the right improvements
<Marth64>
The text justification being my highest priority to try, that would instantly make a bunch of CCs look great