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 7.0.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
thilo_ has quit [Ping timeout: 260 seconds]
thilo_ has joined #ffmpeg-devel
thilo_ has quit [Changing host]
thilo_ has joined #ffmpeg-devel
mkver has joined #ffmpeg-devel
cone-373 has joined #ffmpeg-devel
<cone-373>
ffmpeg Andreas Rheinhardt master:a0ff31e740ec: avcodec/vvc/inter: Don't return void
* Sean_McG
quickly peeks in
mateo` has quit [Ping timeout: 268 seconds]
System_Error has quit [Remote host closed the connection]
AbleBacon has quit [Read error: Connection reset by peer]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
Kei_N has joined #ffmpeg-devel
Martchus_ has joined #ffmpeg-devel
Martchus has quit [Ping timeout: 264 seconds]
jamrial has quit []
cone-373 has quit [Quit: transmission timeout]
Marth64[m] has joined #ffmpeg-devel
Marth64 has quit [Ping timeout: 252 seconds]
Marth64 has joined #ffmpeg-devel
Marth64[m] has quit [Ping timeout: 268 seconds]
Marth64[m] has joined #ffmpeg-devel
Marth64 has quit [Ping timeout: 240 seconds]
cone-170 has joined #ffmpeg-devel
<cone-170>
ffmpeg Andreas Rheinhardt master:c84e40d9e6ce: fftools/ffmpeg_mux_init: Return error upon error
mateo` has quit [Ping timeout: 260 seconds]
mateo` has joined #ffmpeg-devel
Marth64 has joined #ffmpeg-devel
Marth64[m] has quit [Ping timeout: 260 seconds]
Krowl has joined #ffmpeg-devel
another| has quit [*.net *.split]
q66_ has quit [*.net *.split]
aaabbb has quit [*.net *.split]
jess has quit [*.net *.split]
markh has quit [*.net *.split]
klaxa has quit [*.net *.split]
markh has joined #ffmpeg-devel
q66 has joined #ffmpeg-devel
another| has joined #ffmpeg-devel
klaxa has joined #ffmpeg-devel
jess has joined #ffmpeg-devel
aaabbb has joined #ffmpeg-devel
derpydoo has joined #ffmpeg-devel
ngaullier has joined #ffmpeg-devel
sgm has quit [Ping timeout: 260 seconds]
sgm has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
<thardin>
>google breaks git-send-mail, encouraging dozens of useds to degoogle themselves
<thardin>
marvellous
<JEEB>
lol, so they finally did that
System_Error has joined #ffmpeg-devel
Kei_N has quit [Read error: Connection reset by peer]
Kei_N has joined #ffmpeg-devel
<thardin>
the price of freedom is endless faffing about
Krowl has quit [Read error: Connection reset by peer]
<wbs>
fwiw, gmail works just fine with git-send-email, if you set up "app passwords" (which requires enabling 2fa)
<JEEB>
ah, that was my "current state"
<JEEB>
so I thought they broke that?
<JEEB>
I think you no longer can make new app passphrases? but yea, existing ones should have still worked
<mkver>
thardin: You are overestimating x265.
<wbs>
no; previously you could enable the setting "allow insecure apps", that let you authenticate over plain smtp/imap with the regular account password, nothing more
<wbs>
but this is being phased out now
<thardin>
mkver: perhaps
<JEEB>
wbs: ah
<JEEB>
yea, never used that since already years ago made those application credentials
<thardin>
>2 years ago
<thardin>
didn't notice that. if it's still broken that's quite bad
<JEEB>
x265 seems to work just well enough that normal users don't notice it :P
<JEEB>
but don't try to do valgrind or other thigns
<JEEB>
*things
<thardin>
*angry formal methods noises*
<thardin>
perhaps there is no good solution
<wbs>
thardin: you can take over x265 maintainance and make it valgrind clean? :-)
<wbs>
pretty please? :-)
<thardin>
sure, if I get paid for it
<thardin>
I've done quite extensive valgrind/callgrind work in other projects. it's pleasing
<thardin>
but I also need to eat and pay rent, as it were
<JEEB>
x265 also has a memcpy problem
<JEEB>
you will see it as soon as you try to encode something like 10bit 2160p50
<JEEB>
but yes, fixing the valgrind etc stuff at least lets you make FATE tests :D
<JEEB>
like I added one, and it then got removed since more people were running FATE with x265 built-in than I thought :P
<thardin>
I also noticed yesterday an old C++ project of mine uses type punning. unacceptable!
<thardin>
lately I've been working on melt's avformat plugin
kasper93 has quit [Ping timeout: 252 seconds]
kasper93 has joined #ffmpeg-devel
ccawley2011 has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
av500 has joined #ffmpeg-devel
<j-b>
10:32 <@JEEB> x265 also has a memcpy problem
<j-b>
memcpy is murder!
cone-170 has quit [Quit: transmission timeout]
ccawley2011 has quit [Ping timeout: 252 seconds]
<thardin>
memory was not meant to be copied
<thardin>
they have played us for absolute fools
<courmisch>
meanwhile Arm added a memory copy instruction set extension
Marth64 has quit [Remote host closed the connection]
<elenril>
I have some patches pending that have a side effect of exposing those races
<elenril>
mkver: not sure I can talk about it yet, but it will become clear eventually
<haasn>
illuminati project confirmed
<mkver>
elenril: Do you want to add frame and slice-threading (as in wpp) simultaneously?
<elenril>
not in the immediate future
<mkver>
elenril: There is btw a fix for a leak caused by you on the ML.
<elenril>
thanks, replied
lexano has joined #ffmpeg-devel
<thardin>
elenril: maybe we-mark it as experimental?
<thardin>
re-mark*
jamrial has joined #ffmpeg-devel
<mkver>
thardin: "expect non-zero numbers of angry users whose workflows suddenly break" <-- the "non-zero" number would be huge understatement in this case.
<elenril>
who even cares about encoding hevc
<ramiro>
courmisch: did you find out what the issue was with the chroma range conversions?
<elenril>
also, I think it might motivate someone to actually fix the bugs
<JEEB>
I wish people didn't care about it
<JEEB>
hilariously enough at this point you can actually encode AV1 with SVT-AV1 better than HEVC with x265 in live
<JEEB>
since x265 falls to pieces with higher resolutions
<elenril>
can you make it not BLUR yet?
<JEEB>
I think the pq people are pushing their changes to SVT-AV1 upstream slowly
<elenril>
IOW x264 remains best grill for the foreseeable future
Krowl has joined #ffmpeg-devel
<JEEB>
AV1 is just something that you can encode at 10bit and have browsers etc play it
<JEEB>
so stuff like HDR
ccawley2011 has joined #ffmpeg-devel
<thardin>
elenril: good question. a friend of mine works for a company that does a lot of live transcoding. they've tested libx265 but found libx264 was good enough and much better in terms of battery life
<thardin>
mkver: maybe it would be good though, give libx265 a much needed kick?
cone-545 has joined #ffmpeg-devel
<cone-545>
ffmpeg Andreas Rheinhardt master:8754c9bd82b7: configure: Disable DNN without backend
<thardin>
if development has stalled and it has issues and there's bandwidth to say then surely someone could justify funding improvements
<thardin>
to save*
<elenril>
maybe someone could suggest forking it and cleaning it up as an STF project
<JEEB>
thardin: yea and wait until you have someone test out higher resolutions with x265 :P
<JEEB>
I literally couldn't get 2160p50 rolling with a nice 2019 (?) xeon
<JEEB>
then after profiling the biggest time sink were memcpy-likes
<thardin>
lol
<JEEB>
meanwhile SVT-AV1 goes brrrrr, and with some versions even with a single NUMA node of that dual socket 2019 xeon
<cone-545>
ffmpeg Andreas Rheinhardt master:6ae1a337f207: fftools/ffmpeg_mux_init: Fix leak when using non-encoding option
<ramiro>
jamrial: I don't think pmaddwd works in the range_convert functions. it performs "add adjacent doubleword results", which we don't want. or am I missing something?
<BtbN>
I think libx265 is a dead horse. It also has numerous critical security flaws, and nobody cares
<BtbN>
I agree with the sentiment of marking it as experimental to highlight that fact
<jamrial>
ramiro: you can zero half the words, so the addition is a nop
<sfan5>
the embedded device mindset mbedtls was written with definitely makes these issues annoying
<elenril>
and as courmisch said on ML, lavf really shouldn't mess with global logging callbacks
<sfan5>
the log callback is per-context. I removed the call that changed the globla log level
<elenril>
ah, ok then
APic has quit [Ping timeout: 260 seconds]
<thardin>
ah the joys of webrtc
<thardin>
coffe, then getting back to wrangling all the electrical connectors on my citroën xm
<JEEB>
I found it funny how even with what's in master I could build the mbedtls wrapper, but it wouldn't link because random symbols were absent. (mbedtls 2.28.8)
<JEEB>
(that's the version that was already packaged in fedorA)
<JEEB>
and sfan5's patch set was just as good, compiled :)
<JEEB>
(and same linking errors from master)
<cone-545>
ffmpeg sfan5 master:63b6620ad3cd: lavf/tls_mbedtls: handle more error codes for human-readable messages
<cone-545>
ffmpeg sfan5 master:807d1505bfcd: lavf/tls_mbedtls: add missing call to psa_crypto_init
<cone-545>
ffmpeg sfan5 master:827578ca761e: lavf/tls_mbedtls: hook up debug message callback
<elenril>
sfan5: for the last patch see my reply on ML
<courmisch>
michaelni: FUCK YOU. SERIOUSLY. You should be ashamed of yourself.
<courmisch>
I am clearly being trolled here
<courmisch>
and I am sure the CC will find some pretext why it's all my fault
<sfan5>
ty for merge, will reply on ML
<elenril>
courmisch: the benchmark is clearly in your email already
<elenril>
(the diffstat)
<elenril>
but please language
<courmisch>
yeah go ahead and harass the victim
<courmisch>
also for someone who called me a troll *here*, hypocrit much
<elenril>
grow up
<thardin>
please provide benchmarks on C64
<thardin>
come to think of it, could ffmpeg be compiled for amiga? I'm sure someone has tried
<courmisch>
until C11 , compiled probably
<courmisch>
linked and loaded, err...
<thardin>
one of the bigger amiga sites got pushback from users for wanting to move to requiring TLS 1.2. a move that would make the (only?) amiga web browser stop working on it
<courmisch>
I'm so fed up with those arguments
<courmisch>
you can't provide high level support if the low level is not supported
<courmisch>
if (generic) you want to run Windows XP that's your problem, not mine or my OSS projects
Krowl has quit [Read error: Connection reset by peer]
rom1v has quit [Ping timeout: 255 seconds]
<cone-545>
ffmpeg Anton Khirnov master:040755671632: lavc/hevcdec: do not free SliceHeader arrays in pic_arrays_free()
<cone-545>
ffmpeg Anton Khirnov master:a82f2b092430: lavc/hevcdec: simplify condition
<cone-545>
ffmpeg Anton Khirnov master:e12fd62d1dd4: lavc/hevcdec: drop a redundant assignment in hevc_decode_frame()
<cone-545>
ffmpeg Anton Khirnov master:c879165b393d: lavc/hevc_ps: make PPS hold a reference to its SPS
<cone-545>
ffmpeg Anton Khirnov master:2e46d68f553c: lavc/hevc_ps: make SPS hold a reference to its VPS
<cone-545>
ffmpeg Anton Khirnov master:6ddba110eb1d: lavc/hevc/parser: stop using HEVCParamSets.[psv]ps
<cone-545>
ffmpeg Anton Khirnov master:fb873a05b343: lavc/hevc/mvs: stop accessing parameter sets through HEVCParamSets
<cone-545>
ffmpeg Anton Khirnov master:b38aecffecea: lavc/hevc/filter: stop accessing parameter sets through HEVCParamSets
<cone-545>
ffmpeg Anton Khirnov master:d0868d70eaa1: lavc/hevc/cabac: stop accessing parameter sets through HEVCParamSets
<cone-545>
ffmpeg Anton Khirnov master:38b8ae411268: lavc/hevc/pred: stop accessing parameter sets through HEVCParamSets
<cone-545>
ffmpeg Anton Khirnov master:0f47342c12dc: lavc/hevcdec: stop accessing parameter sets through HEVCParamSets
<cone-545>
ffmpeg Anton Khirnov master:8c8072c29c97: lavc/hevcdec: move active PPS from HEVCParamSets to HEVCContext
<cone-545>
ffmpeg Anton Khirnov master:8eb134f4f944: lavc/hevcdec: drop an always-zero variable
<cone-545>
ffmpeg Anton Khirnov master:90fc331b0fc6: lavc/hevcdec: only ignore INVALIDDATA in decode_nal_unit()
<cone-545>
ffmpeg Anton Khirnov master:0892ec947c7f: lavc/hevcdec: pass SliceHeader explicitly to pred_weight_table()
<cone-545>
ffmpeg Anton Khirnov master:5e438511abb2: lavc/hevcdec: do not pass HEVCContext to decode_lt_rps()
<cone-545>
ffmpeg Anton Khirnov master:01b379a93e05: lavc/hevcdec: move pocTid0 computation to hevc_frame_start()
<cone-545>
ffmpeg Anton Khirnov master:07eb60c0da34: lavc/hevcdec: only call export_stream_params_from_sei() once per frame
<cone-545>
ffmpeg Anton Khirnov master:381b70e173f9: lavc/hevcdec: do not pass HEVCContext to ff_hevc_frame_nb_refs()
<cone-545>
ffmpeg Anton Khirnov master:3115c84015d8: lavc/hevcdec: only set no_rasl_output_flag for IRAP frames
<cone-545>
ffmpeg Anton Khirnov master:d86ac94df24d: lavc/hevcdec: output RASL frames based on the value of no_rasl_output_flag
rom1v has joined #ffmpeg-devel
<cone-545>
ffmpeg Anton Khirnov master:b0c29a45dc3c: lavc/hevc/cabac: do not infer WPP use based on HEVCContext.threads_number
<cone-545>
ffmpeg Anton Khirnov master:e4e9e1da15a2: lavc/hevcdec: drop redundant HEVCContext.threads_{type,number}
<cone-545>
ffmpeg Anton Khirnov master:d43527a1a0f5: lavc/hevcdec: store slice header POC in SliceHeader
<cone-545>
ffmpeg Anton Khirnov master:7cce612a26c3: lavc/hevcdec: move a slice segment sanity check to hls_slice_header()
<cone-545>
ffmpeg Anton Khirnov master:efc827bf6fdc: lavc/hevcdec: move slice decoding dispatch to its own function
<cone-545>
ffmpeg Anton Khirnov master:3bbb5d78c74a: lavc/hevcdec: move per-slice local_ctx setup out of hls_slice_header()
<cone-545>
ffmpeg Anton Khirnov master:6ee550d83d15: lavc/hevcdec: move calling hwaccel start_frame to hevc_frame_start()
<cone-545>
ffmpeg Anton Khirnov master:fe171a3b51a1: lavc/hevcdec: move calling hwaccel decode_slice to decode_slice_data()
<cone-545>
ffmpeg Anton Khirnov master:47d34ba7fbb8: lavc/hevcdec: move constructing slice RPL to decode_slice_data()
<cone-545>
ffmpeg Anton Khirnov master:a2e77caf37c9: lavc/hevcdec: set active PPS/SPS in hevc_frame_start()
<cone-545>
ffmpeg Anton Khirnov master:82ded1ad3a79: lavc/hevcdec: move sequence increment/IDR handling to hevc_frame_start()
<cone-545>
ffmpeg Anton Khirnov master:a8f9d52c2278: lavc/hevcdec: move setting slice_initialized out of hls_slice_header()
<cone-545>
ffmpeg Anton Khirnov master:3cd6492fb5e7: lavc/hevcdec: move the check for multiple frames in a packet
<cone-545>
ffmpeg Anton Khirnov master:90e75c4ec9b0: lavc/hevcdec: drop a redundant multiple-frame-per-packet check
<cone-545>
ffmpeg Anton Khirnov master:edb6a471c4b3: lavc/hevcdec: factor decoding a slice NALU out of decode_nal_unit()
<cone-545>
ffmpeg Anton Khirnov master:d725c737fe2a: lavc/hevcdec: move some frame-end code to hevc_frame_end()
<cone-545>
ffmpeg Anton Khirnov master:ccd391d6a3af: lavc/hevcdec: do not unref current frame on frame_end() failure
<cone-545>
ffmpeg Anton Khirnov master:08ea7d6b8e1e: lavc/hevcdec: constify source frame in hevc_ref_frame()
mkver has joined #ffmpeg-devel
sgm has quit [Remote host closed the connection]
System_Error has quit [Remote host closed the connection]
sgm has joined #ffmpeg-devel
System_Error has joined #ffmpeg-devel
<courmisch>
wbs: any idea why this:
<courmisch>
src/tests/checkasm/h264chroma.c:54:18: warning: ‘size’ is used uninitialized [-Wuninitialized]
<courmisch>
looks like GCC farting to me, but?
<courmisch>
wbs: (I only see it on AArch64)
Livio has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
AbleBacon has joined #ffmpeg-devel
AbleBacon has quit [Read error: Connection reset by peer]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
sgm has quit [Remote host closed the connection]
sgm has joined #ffmpeg-devel
sgm has quit [Remote host closed the connection]
Krowl has quit [Read error: Connection reset by peer]
sgm has joined #ffmpeg-devel
System_Error has quit [Ping timeout: 260 seconds]
System_Error has joined #ffmpeg-devel
___nick___ has joined #ffmpeg-devel
___nick___ has quit [Client Quit]
___nick___ has joined #ffmpeg-devel
<michaelni>
mkver, no plan to look at it beyond what i already did (unless theres something specific someone wants me to look at?)
<mkver>
michaelni: Not really. Thanks for the confirmation.
<courmisch>
thardin: there is Amiga under MAINTAINERS, FWIW
<thardin>
nice
<thardin>
amigaaaa
<courmisch>
I know Segaaaa, but not Amigaaaa
ramiro has quit [Ping timeout: 264 seconds]
ramiro has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
Krowl has joined #ffmpeg-devel
System_Error has joined #ffmpeg-devel
___nick___ has quit [Ping timeout: 256 seconds]
cone-545 has quit [Quit: transmission timeout]
<jamrial>
mkver: btw, i tried to replace the zero_extend in lavc/mathops.h (used by get_bits) with the lavu version, and the result was a broken vorbis decoder
<jamrial>
the reason is that it's called with 32 as bit offset, whereas the lavu implementation only supports up to 31
<jamrial>
the output of get_bits_long(gb, 32) changed
<jamrial>
i didn't look into why that would try to zero extend with 32 as bit offset
<mkver>
jamrial: Probably because it used get_bits_long() calls where the number of bits is not a compile-time constant.
<mkver>
Actually, vorbis uses lots of get_bits_long(gb, 32); and it does this at byte-aligned positions.
<mkver>
Nuts.
Livio has quit [Ping timeout: 264 seconds]
Livio has joined #ffmpeg-devel
* Sean_McG
peeks in
AbleBacon has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
dionisis has quit [Ping timeout: 268 seconds]
<llyyr>
d725c737fe2a19091b481d4d115fd939e0a674b2 broke hevc hwdec, at least vaapi and vulkan
<llyyr>
elenril: ^
witchymary has joined #ffmpeg-devel
rvalue- has joined #ffmpeg-devel
rvalue has quit [Ping timeout: 264 seconds]
mkver has quit [Ping timeout: 252 seconds]
Livio has quit [Ping timeout: 256 seconds]
rvalue- is now known as rvalue
System_Error has quit [Remote host closed the connection]
<nevcairiel>
because hevc_frame_end is never called in the hwaccel path
darkapex has quit [Remote host closed the connection]
darkapex has joined #ffmpeg-devel
<Lynne>
I'm surprised hevc has a picture_md5 field
<nevcairiel>
looking at that patch, it should probably just be partly reverted for the hwaccel part
<Lynne>
having a checksum field as a part of the bitstream is just unnecessary, there's only one correct way of decoding
<nevcairiel>
catches corruption more so then broken decoders
dionisis has joined #ffmpeg-devel
System_Error has joined #ffmpeg-devel
ccawley2011 has quit [Read error: Connection reset by peer]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
cone-435 has joined #ffmpeg-devel
<cone-435>
ffmpeg Michael Niedermayer master:c21fb3624bb7: avformat/iamf_parse: Remove dead case
<cone-435>
ffmpeg Michael Niedermayer master:78d6d13babc6: avdevice/xcbgrab: Check sscanf() return
<cone-435>
ffmpeg Michael Niedermayer master:f10493f6fc2a: avformat/imfdec: Simplify get_next_track_with_minimum_timestamp()
<cone-435>
ffmpeg Michael Niedermayer master:7d04c6016b09: avformat/img2dec: Move DQT after unrelated if()
<cone-435>
ffmpeg Michael Niedermayer master:5712f36dd0ee: avformat/img2dec: Little JFIF / Exif cleanup
<cone-435>
ffmpeg Michael Niedermayer master:019fce18bb06: avformat/matroskadec: Assert that num_levels is non negative
<cone-435>
ffmpeg Michael Niedermayer master:0263b6a48caa: avformat/libzmq: Check av_strstart()
<cone-435>
ffmpeg Michael Niedermayer master:0c977d37aad6: avformat/mov: Use 64bit in intermediate for current_dts
<cone-435>
ffmpeg Michael Niedermayer master:034054b3706b: avformat/mov: Use int64_t in intermediate for corrected_dts
<cone-435>
ffmpeg Michael Niedermayer master:9c9f095e30c1: avfilter/vf_rotate: Check ff_draw_init2() return value
sgm has quit [Remote host closed the connection]
sgm has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]