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
rvalue has quit [Ping timeout: 252 seconds]
rvalue has joined #ffmpeg-devel
lexano has quit [Ping timeout: 246 seconds]
mkver has quit [Ping timeout: 246 seconds]
dellas has quit [Remote host closed the connection]
navi has quit [Quit: WeeChat 4.1.2]
thilo has quit [Ping timeout: 240 seconds]
taniey has joined #ffmpeg-devel
thilo has joined #ffmpeg-devel
sepro has quit [Quit: Bye!]
sepro has joined #ffmpeg-devel
MrZeus has quit [Ping timeout: 255 seconds]
kurosu has joined #ffmpeg-devel
lemourin has joined #ffmpeg-devel
lemourin has quit [Killed (zirconium.libera.chat (Nickname regained by services))]
orthoplex64 has joined #ffmpeg-devel
BradleyS has quit [Quit: quit]
<BBB> cworley: that could use an example of impacted result, y'know
<BBB> like a before/after on a real file, or something
BradleyS has joined #ffmpeg-devel
jamrial_ has quit []
<cworley> Sure. I'll show with a dds file or something
<cworley> As-is, color comes out extra dark when an already premultiplied file is premultiplied again during decoding
Marth64 has joined #ffmpeg-devel
Martchus has joined #ffmpeg-devel
Martchus_ has quit [Ping timeout: 264 seconds]
rooisnoek has joined #ffmpeg-devel
kurosu has quit [Quit: Connection closed for inactivity]
<Traneptora> I just discovered the hell that is MakerTag in EXIF
<Traneptora> how do we wish to handle this?
<Traneptora> MakerNote is an EXIF tag with id 0x927c, and by the EXIF spec it's an opaque binary blob
<Traneptora> the reality is that it's often actually an IFD itself, which means in TIFF files it contains offsets relative to the head of the file, not to the head of the ExifTag
<Traneptora> and whether or not it's actually an IFD itself depends entirely on Manufacturer. for example, Canon cameras use MakerNote to just be an IFD. Samsung also uses an empty IFD but its offsets are relative to the start of the MakerNote
<Traneptora> copying the binary blob as-is will corrupt it if it's an IFD, because all the offsets will be incorrect
<Traneptora> this is only an issue for TIFF files, but how do we wish to handle this?
taniey has quit [Ping timeout: 264 seconds]
qeed_ has quit [Quit: qeed_]
AbleBacon has quit [Read error: Connection reset by peer]
qeed has joined #ffmpeg-devel
rooisnoek has quit [Quit: Leaving]
Marth64 has quit [Ping timeout: 268 seconds]
Marth64 has joined #ffmpeg-devel
epony has joined #ffmpeg-devel
epony has quit [Read error: Connection reset by peer]
Teukka` has joined #ffmpeg-devel
Teukka has quit [Ping timeout: 252 seconds]
epony has joined #ffmpeg-devel
kurosu has joined #ffmpeg-devel
Marth64 has quit [Ping timeout: 268 seconds]
Krowl has joined #ffmpeg-devel
ngaullier has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
Raz- has quit [Quit: Leaving]
Rathann has joined #ffmpeg-devel
<Rathann> hello
<Rathann> How does one request a particular commit to be back ported to older branch?
<Traneptora> try mentioning the commit
<Traneptora> and which branch
Krowl has joined #ffmpeg-devel
mkver has joined #ffmpeg-devel
<jdek> Anyone around in brussels already?
<JEEB> Rathann: you just request a backport of hash x to older branches on the ML. you can usually check if it's zero effort by doing a `git cherry-pick -x HASH` in your local git clone when on that old branch
<Traneptora> tbh you can also ask here, you just need to actually ask
<Daemon404> jdek, koda is
<Daemon404> but he is not in this cgannel
<Daemon404> channel even
* Daemon404 currently on train
<JEEB> trains are good civilization
<haasn> great
<haasn> my train got cancelled due to vandalism
<Daemon404> thats a new one
<Rathann> JEEB, Traneptora: thanks
<Rathann> I'd like to request backporting a0a0a804998da8d1a397479c9bafeb000e6be088 to 6.0.x
<Rathann> it fixes a segfault
<Rathann> reproducer: ffmpeg -y -f rawvideo -s 1x1 -i /dev/zero -vf palettegen=doesnt_exist=64 dummy.png
<Rathann> reported in RPM Fusion bugzilla: https://bugzilla.rpmfusion.org/show_bug.cgi?id=6773 originally
<Rathann> I haven't checked earlier branches
<cworley> >Sure. I'll show with a dds file or something
<cworley> ah, message held due to size
<JEEB> Rathann: ah, uninitialized variables. classic
<haasn> Part of the train tracks were removed / stolen
<haasn> Hilarious
<JEEB> o_O
<ePirat> haasn, germany?
<haasn> ePirat: yep
<nevcairiel> well thats a sort of vandalism that explains why the train doesnt go anymore
dellas has joined #ffmpeg-devel
epony has quit [Remote host closed the connection]
<another|> haasn: I guess I'm lucky my train is still on
ccawley2011 has joined #ffmpeg-devel
<another|> btw: https://bahn.expert/ is a great resource
<haasn> Seems only trains via frankfurt -> köln are affected
<another|> yeah, vandalism between frankfurt - cologne
<another|> am currently on this leg. train is taking a >2h detour
<another|> oh, well. at least I'l get money back
<another|> lol. just got an email that my "arrival in Brussels will be belated by 5 min"
<elenril> haasn: I'd expect this kind of thing further east
<haasn> I have the vague suspicion it’s caused by protesting farmers
<haasn> They also block roads
<elenril> outsource farming to poland
<elenril> no farmers=no problem
dellas has quit [Remote host closed the connection]
<Daemon404> didnt the farmers specifically say they aimed to cause train issuws
epony has joined #ffmpeg-devel
<nevcairiel> i really fail to see the point in such actions, it will just totally kill any goodwill in the population, and you want to keep that if you want the government to be the bad guy, not yourself
jamrial has joined #ffmpeg-devel
<haasn> Daemon404: source? lol if true
dellas has joined #ffmpeg-devel
dellas has quit [Remote host closed the connection]
<JEEB> elenril: I think this time poland is also gonna protest since черноземь is OP with multiple harvests per year
<JEEB> just like spain/france/etc protested in the turn of the century
<JEEB> when poland etc entered the EU
kurosu has quit [Quit: Connection closed for inactivity]
BradleyS has quit [Quit: quit]
BradleyS has joined #ffmpeg-devel
<galad> about one third of the European Union budget is dedicated to agricolture, and Poland will be equally affected by same reworking of EU agricultural subsidies
<galad> but some protests are about local laws, so I don't know if there will be some in Poland too
Krowl has quit [Read error: Connection reset by peer]
TheElixZammuto has quit [Remote host closed the connection]
cone-053 has joined #ffmpeg-devel
<cone-053> ffmpeg Niklas Haas master:dcc7263b0e79: fftools/ffplay: add missing YUV metadata to buffersrc
<cone-053> ffmpeg Niklas Haas master:f7b1017d7378: fftools/ffmpeg_filter: simplify buffersrc arg printing
novaphoenix has quit [Quit: i quit]
novaphoenix has joined #ffmpeg-devel
lexano has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
navi has joined #ffmpeg-devel
navi has quit [Ping timeout: 264 seconds]
navi has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
Krowl has joined #ffmpeg-devel
Rathann has quit [Ping timeout: 264 seconds]
rvalue has quit [Quit: ZNC - https://znc.in]
rvalue has joined #ffmpeg-devel
epony has quit [Remote host closed the connection]
psykose has quit [Remote host closed the connection]
<elenril> JEEB: should grow potatoes then, like the czechs do
<elenril> never enough potatoes
psykose has joined #ffmpeg-devel
<cone-053> ffmpeg Anton Khirnov master:70ecb8238cff: lavd: deprecate the bktr device
<cone-053> ffmpeg Anton Khirnov master:aa3cfd4b5a04: lavc/bsf: add a showinfo filter
dellas has joined #ffmpeg-devel
<thardin> gonna grow the hell out of some 'tatoes this year
<thardin> got a friend of mine excited to plant some ares
<elenril> excited?
<JEEB> :)
psykose has quit [Remote host closed the connection]
Marth64 has joined #ffmpeg-devel
AbleBacon has joined #ffmpeg-devel
<kasper93> >filter context - w: 1920 h: 1080 fmt: 99 csp: unknown range: pc
<kasper93> this doesn't seem to be readable
<kasper93> I look and all I see is "unknown range"
<kasper93> some separators would be nice there
<haasn> Oh god, now we can’t continue because people are standing on the tracks
<haasn> Next time I will drive to Belgium, I swear
Krowl has quit [Read error: Connection reset by peer]
<courmisch> won't help if tractors stand on the highway
<kasper93> need a bigger tractor or snowplow
<courmisch> doubt that would work
<another|> I finally arrived and checked in
<courmisch> in the nineties, France used army tanks to "move" trucks blocking the roads
<courmisch> I am boarded on the first of my rerouting flights
<courmisch> in theory I should arrive in BRU in 6 hours
<elenril> courmisch: of how many?
<courmisch> elenril: HEL-BRU/AY turned into HEL-FRA-BRU/LH-SN, departint 1h late, arriving 4h late
<elenril> not so horrible
* elenril stood in the bus queue for about an hour
<elenril> not much fun with the wind
<courmisch> I had to take a taxi since no metro, no train
<courmisch> 80€
APic has quit [Ping timeout: 268 seconds]
<elenril> SPI pays for it surely
<kierank> lol
dellas has quit [Remote host closed the connection]
<kierank> just mark a random coverity issue as a false positive
derpydoo has joined #ffmpeg-devel
georgecox has quit [Read error: Connection reset by peer]
APic has joined #ffmpeg-devel
<Marth64> which link to stream talks remotely? https://fosdem.org/2024/schedule/streaming/
kurosu has joined #ffmpeg-devel
georgecox has joined #ffmpeg-devel
georgecox has quit []
epony has joined #ffmpeg-devel
<elenril> >If I followed correctly, Android AOSP had started migrating from Make to "Kati" and "Soong", and then partway through that migration they started a big migration to Bazel.
<Marth64> i thought they were namde after desserts
<JEEB> those are make-like systems
<Marth64> ahh
<JEEB> so them build system migrations 8)
dellas has joined #ffmpeg-devel
<haasn> elenril: how does gradle fit into this ecosystem?
<elenril> ask someone who actually uses or develops for android
<elenril> I'm just giggling from the sidelines
<Daemon404> bazel is a sin
<Daemon404> fun fact: anything built with gcc in bazel gets a bunch of CFLAGS added by default
<Marth64> elenril: a good read
ccawley2011 has quit [Read error: Connection reset by peer]
jarthur has joined #ffmpeg-devel
cone-053 has quit [Quit: transmission timeout]
philipl has quit [Remote host closed the connection]
philipl has joined #ffmpeg-devel
lns has joined #ffmpeg-devel
<BBB> haasn: http://trac.ffmpeg.org/ticket/10842#ticket I think that's your changes (but not sure)
<elenril> i feel targeted
rvalue has quit [Ping timeout: 252 seconds]
<BBB> by me?
<haasn> BBB: thanks, I'll have a look at it (probably tomorrow, as I'm about to arrive)
<haasn> BBB: is this a regression caused by f7b1017d7378e44daa7971fa2bf79738964d2002?
<haasn> it seems that on previous versions it should have tried to format NULL into %s
<haasn> which I can't imagine being particularly healthy
<BBB> I think that works on most systems
<BBB> testing... (takes a while)
<jamrial> haasn: it printed "Undefined constant or missing '(' in 'null)'" before that commit
<jamrial> but it's not the one that broke it
<haasn> Okay
<BBB> happens both before and after that commit for me
<BBB> plus before some extra errors indeed
rvalue has joined #ffmpeg-devel
<elenril> clearly must be an av1 bug
<BBB> ¯\_(ツ)_/¯
<elenril> av1 is the buggiest codec on earth
kurosu has quit [Quit: Connection closed for inactivity]
ccawley2011 has joined #ffmpeg-devel
<haasn> I wonder why DB employees, instead of going on strike, don’t just publicly announce that tickets won’t be inspected on day XYZ
<haasn> More effective way to financially harm your superiors AND maintain public goodwill
<JEEB> that sounds like the bus strike that was in japan last yea
<JEEB> the drivers still went to work, but covered the payment terminals
<BtbN> Cause the drivers aren't the ones checking the tickets
<elenril> wouldn't that be illegal?
<BtbN> No, that happens every year, when the ticket check personell goes on strike
<BtbN> they don't announce it loudly though, but it's effectively what happens
<elenril> they go on strike regularly every year?
<elenril> is that like a teambuilding thing?
MrZeus has joined #ffmpeg-devel
<BtbN> it's a money thing :D
<BtbN> well, realistically it's every two years, that's the usual duration of the contracts these days
ngaullier has quit [Ping timeout: 246 seconds]
<Traneptora> haasn: isn't there a train strike in germany
<Traneptora> in either case, how should we handle MakerNote inside TIFF files when attempting to export EXIF?
<elenril> does it follow a spec?
<JEEB> apparently the spec is a "this is a blob"
<elenril> there are many like it, but this one is mine?
cosimone has joined #ffmpeg-devel
<rodeo> French language video warning: https://youtu.be/eFCn2Fq3vB0 (about farmer discontent)
<Traneptora> elenril: the spec says that it's a binary blob
<Traneptora> the practice is that it's often an IFD with offsets relative to the head of the file in which it occurs
<Traneptora> so extracting it from the TIFF file will corrupt it
<Traneptora> whether or not it's an IFD depends on the manufacturer
<elenril> can we meaningfully parse it ourselves?
<Traneptora> elenril: yes, if we know what format it takes. the problem is the format depends highly on the manufacturer
<elenril> so ideally we'd parse it ourselves
<Traneptora> some manufacturers can be relocated as a blob either because it's not an IFD or because it is an IFD but relative to the start of the file
<elenril> for the rest, export the blob and the offset it was stored at?
<Traneptora> for the others we have to parse it, which will result in a big switch statement based on the initial bytes
<Traneptora> which we *can* do but it seems really ugly
Marth64 has quit [Ping timeout: 260 seconds]
<elenril> is there any realistic alternative?
Marth64 has joined #ffmpeg-devel
<Traneptora> if we want to move it to a non-TIFF format (e.g. PNG) then it will be corrupted unless we parse it
<Traneptora> atm we don't write MakerNote at all
<Traneptora> are we willing to corrupt it when camera manufacturers don't follow spec or are we not I guess is the question
<elenril> could be an option
<elenril> e.g. you could have multiple side data types
<elenril> one for parsed, one for the blob
<elenril> the muxer would require an option to be set when writing the blob
<Traneptora> another thought is to add a metadata option to AVFrameSideData which is "exiftag's original offset in the original file"
<Traneptora> but that doesn't fix the writing-corrupt problem
* elenril stabs Traneptora
<Traneptora> like, I saw your suggestion but I don't think it fixes the problem
<Traneptora> since in order to adjust the IFD offsets, you'll have to parse the IFD itself
<elenril> someone has to, no?
<Traneptora> yea, but if we're going to do that why don't we just fix them when we do
<elenril> the suggestion is: when we can parse the blob, we do so
<Traneptora> the biggest issue is that whether or not it's an IFD requires a big lookup table based on manufacturer, and is that something we are willing to do?
<elenril> when we can't, we give the blob to the caller with all the information they need to parse it
<elenril> I have no fundamental problem with a big lookup table when there's no better way
<Traneptora> if you think that's fine then sure
<elenril> big lookup tables were just unjustly maligned
<elenril> they never actually murdered anyone and danced on the corpse
<Traneptora> it just feels kinda gross to have a manufacturer-based LUT to parse something. like we're back in the DVD era
<Marth64> lol dvd
<elenril> don't try to parse exif then
<elenril> "doctor it hurts when I stab myself"
<Traneptora> doctor it hurts when I work on FFmpeg
<elenril> and now you see why it's only crazy people here
<BBB> thanks!
iive has joined #ffmpeg-devel
Marth64 has quit [Remote host closed the connection]
<elenril> wtf, V4L2 Stateless Video Encoding is in automotive and embedded?
<JEEB> embedded sounds like it
ngaullier has joined #ffmpeg-devel
<JEEB> ARM stuff
<elenril> I guess hoping they'd design it for everybody, not just embedded, is naive?
deus0ww has quit [Ping timeout: 268 seconds]
deus0ww has joined #ffmpeg-devel
ngaullier has quit [Ping timeout: 255 seconds]
Marth64 has joined #ffmpeg-devel
dellas has quit [Remote host closed the connection]
dellas has joined #ffmpeg-devel
Marth64 has quit [Remote host closed the connection]
<BtbN> last time I looked at it, it didn't look like it was designed at all
AbleBacon has quit [Read error: Connection reset by peer]
AbleBacon has joined #ffmpeg-devel
<elenril> who volunteers to go there and complain?
cosimone has quit [Remote host closed the connection]
cosimone has joined #ffmpeg-devel
cosimone has quit [Remote host closed the connection]
rodgort has quit [Quit: Leaving]
<courmisch> elenril: because that's who pays for linux-media devs
rodgort has joined #ffmpeg-devel
HarshK23 has quit [Quit: Connection closed for inactivity]
cosimone has joined #ffmpeg-devel
cosimone has quit [Remote host closed the connection]
cosimone has joined #ffmpeg-devel
cosimone has quit [Remote host closed the connection]
cosimone has joined #ffmpeg-devel
dellas has quit [Remote host closed the connection]
derpydoo has quit [Ping timeout: 268 seconds]
kurosu has joined #ffmpeg-devel
cosimone has quit [Remote host closed the connection]
cosimone has joined #ffmpeg-devel
cosimone has quit [Remote host closed the connection]
cone-475 has joined #ffmpeg-devel
<cone-475> ffmpeg Andreas Rheinhardt master:b9596daafbc7: avformat/iamf_writer: Avoid using dynamic buffer
<cone-475> ffmpeg Andreas Rheinhardt master:8d4b22edc7b5: avformat/iamf_writer, iamfenc: Avoid allocations when using dyn buffers
<cone-475> ffmpeg Andreas Rheinhardt master:9d364fbdb05b: avcodec/vlc: Remove unused macros
<cone-475> ffmpeg Andreas Rheinhardt master:648df1c25046: avcodec/leaddec: Remove unnecessary VLC structures
<cone-475> ffmpeg Andreas Rheinhardt master:d50a24668775: avformat/iamf*: Improve included headers
ccawley2011 has quit [Read error: Connection reset by peer]