<cone-032>
ffmpeg Paul B Mahol master:49719d3cb58c: avfilter/avf_showcwt: add fm frequency scaler
ubitux has quit [Ping timeout: 240 seconds]
ubitux has joined #ffmpeg-devel
dellas8386 has quit [Remote host closed the connection]
<cone-032>
ffmpeg Paul B Mahol n6.1:HEAD: avfilter/avf_showcwt: add fm frequency scaler
Chagall has quit [Server closed connection]
Chagall has joined #ffmpeg-devel
<michaelni>
Lynne, ive pushed 6.1 tag, download & CVE updates, if you want to push anything you can do it anytime thats convenient
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 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
<Lynne>
michaelni: thanks, website update pushed
<kierank>
Lynne: posted on X and linkedin
<Lynne>
thanks, for once we're ahead of phoronix :)
thilo has quit [Ping timeout: 240 seconds]
thilo has joined #ffmpeg-devel
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
dellas has joined #ffmpeg-devel
dellas has quit [Remote host closed the connection]
lemourin has quit [Read error: Connection reset by peer]
lemourin has joined #ffmpeg-devel
jamrial has quit []
durandal_1707 has quit [Ping timeout: 258 seconds]
durandal_1707 has joined #ffmpeg-devel
cone-032 has quit [Quit: transmission timeout]
<kurosu>
So I imagine you want to keep the extra voters discussion public, at least for accountability. But it feels like a bad showing for the project, for every side of the arguments. I unfortunately understand what is happening here, and there's no cooling it off :-(
<aaabbb>
can someone give a tl;dr of what this vote and the drama is about?
<Lynne>
people who aren't coding are arguing
<aaabbb>
about what?
<aaabbb>
i hope this isn't going to be as damaging as the whole libav/avconv crap
<kurosu>
People able to decide on the directions of the project, I'd try to summarize
<aaabbb>
what direction are people pushing to go?
<Lynne>
none at all, this is just arguing about who should have very little voting rights in case there is a community-wide voting
<aaabbb>
oh
<kurosu>
I better not try to summarize that part. But at this stage, it feels a bit like that time, yes. Or at least, I wonder if it's all worth it. But easy for me to say just that
<Lynne>
no community-wide voting has even happened in the 3+ years we've had this scheme, though!
<Lynne>
surely everyone else can see there's as little worth giving someone inactive in 3 years voting rights when those on the general assembly are changed every 6 months
<aaabbb>
so the question is whether inactive contributors should have equal voting rights as active ones?
<Lynne>
s/equal//
<Lynne>
what's worse is that we'd repeat this every 6 months at this rate
<aaabbb>
what percentage of contributors are inactive that would have voting rights?
<Lynne>
in the grand scheme of things, very little
<Lynne>
but for some reason some are bringing a shovel and digging up corpses to give voting rights to
<kurosu>
I don't think that way of presenting things is helpful. I understand you feel it is important. But it showcases the level of the discussion, and I really feel that question is not worth all this.
<Lynne>
I don't have any interest into that discussion, I do wish everyone would just drop it
<another|>
I'd like to give voting rights to Joseph Fourier and Cornelius Lanczos.
<another|>
/s
HarshK23 has joined #ffmpeg-devel
<aaabbb>
what contributors would get voting rights? all, even people who only contributed a single patch on the ml once?
invertedguy0485 has quit [Server closed connection]
invertedguy0485 has joined #ffmpeg-devel
derpydoo has joined #ffmpeg-devel
AbleBacon has quit [Read error: Connection reset by peer]
dellas has joined #ffmpeg-devel
TD-Linux has quit [Server closed connection]
dellas has quit [Remote host closed the connection]
TD-Linux has joined #ffmpeg-devel
<Swedaniel>
Im voting same as Lynne!
<Swedaniel>
what is the voting about? :D
<Lynne>
not important
<Lynne>
what is important is that avio is inadequate!
<Lynne>
why can't it accept open FDs and sockets?
markh has quit [Read error: Connection reset by peer]
mark4o has joined #ffmpeg-devel
Mikhail_AMD has joined #ffmpeg-devel
mark4o is now known as markh
MikhailAMD has quit [Ping timeout: 255 seconds]
<Swedaniel>
ok:D what is avio? i dont know what FDs is either and ive seen some electrical sockets to connect an lightbulb with :D
<Lynne>
libavformat/avio.h
elastic_dog has quit [Ping timeout: 246 seconds]
<Swedaniel>
ooh lol, you had be thinking withint Sony Vaio laptops there :D
elastic_dog has joined #ffmpeg-devel
<Lynne>
I agree with elenril, let's not repeat this same vote a third time
jarthur has quit [Quit: jarthur]
<Swedaniel>
I must go chck what Inadequtae is allso! 3-4 things in one sentence that must be googled and ive got masters degree in english:D haha
<elenril>
who are you anyway
<elenril>
Lynne: pls say so on the ML
rvalue has quit [Read error: Connection reset by peer]
<Lynne>
wow, zhao mentioned the same thing I did
rvalue has joined #ffmpeg-devel
<elenril>
it's ridiculous
<elenril>
a tempest in a teapot is the english idiom I think
<Lynne>
close, a storm in a teapot
<j-b>
The best way to block all decisions is to block votings
<jamrial>
ideally everyone would do that, take a breath, and we can start fresh on monday
<durandal_1707>
no, take breath from your drama queens, exp. those with 0 recent commits
<elenril>
(like durandal_1707)
<jamrial>
durandal_1707: things are pretty fucking bad right now without you providing us with more trolling, so could you please refrain from doing so at least for a couple weeks?
<durandal_1707>
no
<durandal_1707>
the people with little or no recent involment in project take over FFmpeg already
<courmisch>
well if you'd rather long-time devs like NG take over the project
ccawley2011 has quit [Ping timeout: 252 seconds]
<courmisch>
but if even jamrial is on edge, maybe you could exceptionally tone it down for just a bit?
* elenril
wonders how evil is goto'ing into an if(0) {} block
<courmisch>
in a macro, it might be okay
<courmisch>
in normal code, that just sounds unnecessarily ugly and confusing
<jamrial>
and the compiler will probably hate you
<courmisch>
of course, you should switch to Rust for all zero-cost abstractions, rather than use C macros
<elenril>
yeah, guess I'll just put it at the end
<Traneptora>
I'm having a hard time seeing the benefit of that over gotoing to the end past a return statement
<elenril>
there are two cleanup blocks
<elenril>
one to be executed on failure, followed by one to be executed always
<elenril>
guess i'll have the former goto into the latter
<elenril>
more gotos for everyone
<courmisch>
impl Drop for ElenrilHack { ...
Krowl has quit [Read error: Connection reset by peer]
dellas has quit [Remote host closed the connection]
Swedaniel has joined #ffmpeg-devel
<durandal_1707>
courmisch: why you think NG is long-time dev, his recent work is exactly 0.
lemourin has joined #ffmpeg-devel
<courmisch>
durandal_1707: I don't "think" it; I just trust derek's word
NuoMi has joined #ffmpeg-devel
Nuo has joined #ffmpeg-devel
cone-020 has joined #ffmpeg-devel
<cone-020>
ffmpeg Nuo Mi master:09f783692e21: avcodec/cbs_h266: H266RawSliceHeader, expose curr_subpic_idx
NuoMi has quit [Ping timeout: 260 seconds]
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
dellas has joined #ffmpeg-devel
Nuo has quit [Quit: Leaving]
derpydoo has quit [Ping timeout: 246 seconds]
<Traneptora>
durandal_1707: 1097 Nicolas George
<Traneptora>
commit count
<Traneptora>
that's why
<durandal_1707>
since when?
<Traneptora>
since "initial commit"
<Traneptora>
he's been a dev for a long-time
<Traneptora>
hence, a long-time dev
<durandal_1707>
he is not LEADER or DICTATOR to have rights to vote/troll/bikeshed/nak/review patches if no longer really active
<Traneptora>
that's not the same thing as being a long-time dev
<courmisch>
Lynne: so exrdsp.predictor... I got the RVV within 10% of C... but then I managed to improve the C by 30%... so RVV is hopeless
<courmisch>
now to see how that works out on opusdsp.deemphasis but I'm not too optimistic
<durandal_1707>
courmisch: you improved C by 30% how?
<durandal_1707>
are you wizard?
<BBB>
courmisch the magician
<BBB>
courmisch for CC, will improve CC by 30%
<setnan>
hi, i guess that's not the good time to ask a question but i try :) in theory does anyone know if AAC packet decoding can be done in parallel ? not sequencially (i think that each decoding step depends on previous pkt but not 100% sure)
<courmisch>
durandal_1707: patch is on the ML, pretty trivial really
zane has quit [Quit: WeeChat 4.1.1]
Krowl has joined #ffmpeg-devel
<durandal_1707>
courmisch: there is x86 SIMD for this function?
<courmisch>
durandal_1707: i think so, lets
<courmisch>
see
Traneptora has quit [Quit: Quit]
<courmisch>
durandal_1707: there's ssse3, avx and avx2
<durandal_1707>
and new C version from compiler does not beat those x86 SIMD ?
<courmisch>
if you are afraid your hand-written x86 is being deprecated, no. The x86 SIMD is still much faster than C
<courmisch>
well it was written by Martin Vignali anyway
<courmisch>
at least on my AuthenticAMorDor Ryzen 5
<durandal_1707>
if AMD is Mordor what is Nvidia?
<courmisch>
durandal_1707: you don't want to know
<courmisch>
I cannot confirm that I am under an NDA not to talk about my former employment at NVIDIA
<courmisch>
And this is totally not a suspiciously specific denial.
<microchip_>
Nvidia = Eisengard
<durandal_1707>
3dfx = Shire
<courmisch>
3dfx is technically Nvidia now, so err
<JEEB>
RIP 3dfx voodoo
auri has quit []
<another|>
They took the Hobbits to Isengard
<jamrial>
another|: they are taking
<another|>
3dfx bankruptcy was over 20 years ago. I think it's safe to say they arrived
<j-b>
Ridiculous
<j-b>
We discuss how to do the bootstrapping vote, with 25+ people during VDD, while we have people online too. We implement that vote, but it's not good, because it's missing an option that, is supposedly important. The vote is redone and the result is very clear with a huge margin (and the option added, is, of course, completely burried) and then people want again to nullify the vote...
<j-b>
All for a bootstrapping vote...
<courmisch>
it's not ridiculous. it's discrediting the decision-making process 101
<durandal_1707>
25+ people - 255+ opinions
<j-b>
Vote durandal_1707 as President of the World!
<durandal_1707>
first decision as new LEADER : rm -rf ffmpeg/
<elenril>
setnan: never heard of frame threading for mdct-based codecs
<durandal_1707>
perhaps it can be delay until final overlap?
<elenril>
i guess it's not entirely out of question, otherwise seeking wouldn't be possible
auri has joined #ffmpeg-devel
<j-b>
jamrial: where we do stand on EVC? I'm kinda lost.
<jamrial>
j-b: i'm checking the decoder/encoder wrapper patches and pushing them today or tomorrow if there's nothing wrong
<j-b>
yay
<jamrial>
as for VVC, nuomi said he'd send a new version of his native decoder soon
<jamrial>
so we're getting and the VCs in one form or another :p
<j-b>
We NNEEED more
<j-b>
EVC, EVC_base, VVC, LC_EVC
<j-b>
VV-EVC
<j-b>
ZV-EVC
<JEEB>
with LC-EVC I have no idea why some people thin it's more performant for single variant encoding (than just normally encoding that variant)
<JEEB>
and no-one can of course provide any other numbers than those provided by its author company
<j-b>
MPEG politics are weird, I don't get how they can have all of those.
<cone-020>
ffmpeg Paul B Mahol master:fa4c2884dd56: avfilter/avf_showcwt: do not return initial black frames with negative timestamps
<JEEB>
j-b: LC-EVC at least is separate from the other coding things since it's basically a "codec indifferent" base layer + enhancement layer thing. so eh, if you wanted that now that's specified
<j-b>
Oh, I know.
<JEEB>
EVC I have no idea... is it their Nth attempt for non-paid licensing thing?
<j-b>
it's just they all get out at the same time, but with different names, but all named *VC
<j-b>
EVC vs LC-EVC, come on
<durandal_1707>
AV1VC
<durandal_1707>
AVC1
darkapex has quit [Ping timeout: 246 seconds]
mkver has joined #ffmpeg-devel
<j-b>
AV2-VC
<setnan>
elenril: iiuc, seeking is ok as long as next frames are read in sequence using the same decoder, i'm doing some experimentation (for fun and to learn, rainy saturday afternoon at my end, you see…), and trying to use several decoders in a concurrency context (in my test, one by processor's core / thread) gives incorrect decoded data (not far but not exact)
<setnan>
i'm not surprised but i search comprehensible infos sources about this, if any
<jkqxz>
The LCEVC people might run out of money at some point, and then it will disappear without a trace.
<nevcairiel>
the world never needed EVC or LCEVC, so that would not be so bad
<durandal_1707>
the world never needed FFmpeg or LibAV, so that would not be so bad.
<durandal_1707>
the world never needed nevcairiel or lavfilters, so that would not be so bad.
<jkqxz>
Hmm, their next results might be published soon. (Previous year they had 220k revenue and 8.8m expenses; at some point the dead horse must bleed out completely.)
<durandal_1707>
they can go open-source, like undead horse FFmpeg
<BtbN>
hm, configure seems to get the order wrong when a pkg-config file decides to do this: Libs: -L${libdir} -laribcaption /opt/ct-ng/x86_64-w64-mingw32/lib/libstdc++.a
<BtbN>
It then does this: x86_64-w64-mingw32-gcc [......] /opt/ct-ng/x86_64-w64-mingw32/lib/libstdc++.a -o /tmp/ffconf.keCmW6aC/test.exe /tmp/ffconf.keCmW6aC/test.o -laribcaption [.....]
<durandal_1707>
how much i need to wait for review?
<durandal_1707>
you should not be with people who do not value you or your work.
<courmisch>
Lynne: RVV opusdeemp is 3% faster than C
<durandal_1707>
not possible
<durandal_1707>
lets fork from this FFmpeg
<courmisch>
that worked out so well the last time
Swedaniel has quit [Quit: Connection closed]
<durandal_1707>
pross: there are other formats of lead codec, see trac
<j-b>
JEEB: I have had competent people who told me they got positives with LC-EVC
Krowl has quit [Read error: Connection reset by peer]
<pross>
durandal_1707: that other lead codec is a bit 'tuff' to re. i might try again someday. it almost as difficult as the RED codecs...
<JEEB>
I like how simple RED used to be
<JEEB>
J2K with bayer
<durandal_1707>
Lynne: local patch (adaptive threshold picked by wiener entropy estimation score + log domain gain instead of linear in current master) with 7 adynamicequalizer filters cascaded together (uses showcwt to display spectrogram)
haasn has quit [Server closed connection]
haasn has joined #ffmpeg-devel
HarshK23 has quit [Quit: Connection closed for inactivity]
<Lynne>
how do you compute the entropy?
<Lynne>
it should correspond inversely to tonality
<durandal_1707>
Lynne: wiener entropy formula
<durandal_1707>
spectral flatnes
<Lynne>
do you have any weight adjustments to handle high frequencies in speech, so you don't get rid of them?
<durandal_1707>
nope, this formula works on any band
kurosu has quit [Quit: Connection closed for inactivity]
ccawley2011 has quit [Read error: Connection reset by peer]
cone-424 has joined #ffmpeg-devel
<cone-424>
ffmpeg Paul B Mahol master:553b31da6840: avfilter/avf_showcwt: fix invalid write for full bargraph
derpydoo has joined #ffmpeg-devel
Swedaniel has joined #ffmpeg-devel
<Sebastinas>
Lynne: dfcd4bbf5118b9fa601ba36e3c5984db61478d5f introduces a segfault in av_fft_end since ctx2 is not initialized. Should av_fft_init call av_mallocz instead of av_malloc?