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 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
thilo has quit [Ping timeout: 256 seconds]
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
thilo has joined #ffmpeg-devel
aaabbb- is now known as aaabbb
aaabbb has quit [Changing host]
aaabbb has joined #ffmpeg-devel
AbleBacon has quit [Read error: Connection reset by peer]
IndecisiveTurtle has quit [Ping timeout: 240 seconds]
<BtbN>
That patch for aacenc makes me think that this code was intended to do something else, and maybe should be looked at first before just removing it, hm
<courmisch>
plausible enough
<ramiro>
mkver: thanks. there's another occurrence of "pic->reference = 3;" in load_input_picture(). I'm trying to download your entire patchset to see if it's dealt with in another patch, but I'm still struggling with patchwork.
<mkver>
ramiro: That has been removed in an earlier patch, namely 54/71.
<ramiro>
mkver: great! thanks
ccawley2011 has joined #ffmpeg-devel
jamrial has joined #ffmpeg-devel
HarshK23 has joined #ffmpeg-devel
___nick___ has joined #ffmpeg-devel
___nick___ has quit [Client Quit]
SystemError has quit [Remote host closed the connection]
___nick___ has joined #ffmpeg-devel
SystemError has joined #ffmpeg-devel
PasswordsInPlain has joined #ffmpeg-devel
PasswordsInPlain has quit [Remote host closed the connection]
sadome has joined #ffmpeg-devel
sadome has quit [Changing host]
sadome has joined #ffmpeg-devel
rodgort has quit [Quit: Leaving]
sadome has quit [Remote host closed the connection]
Krowl has joined #ffmpeg-devel
rodgort has joined #ffmpeg-devel
sadome has joined #ffmpeg-devel
sadome has quit [Remote host closed the connection]
sadome has joined #ffmpeg-devel
sadome has quit [Remote host closed the connection]
sadome has joined #ffmpeg-devel
sadome has quit [Changing host]
sadome has joined #ffmpeg-devel
sadome has quit [Remote host closed the connection]
<rcombs>
(fixes an issue where we fail to demux the first packets in some files with DVHS/BD timestamps that contain an 0x47)
<rcombs>
somebody just pinged me about it
Krowl has joined #ffmpeg-devel
<JEEB>
rcombs: can't directly say much, but I see it moves from using the raw_packet_size to just removing the static sizes from pos instead. and then adds specific skipping logic for the DVHS/BDMV stuff, which is not clear is it related to the first diff or not (as in, is this one or two changes)
<rcombs>
it really needs a commit message, but the point is that the existing code assumes that any extra data beyond the 188 bytes is at the end of the packet, when in reality with DVHS/BD timecodes it's at the start
<rcombs>
or, well, the code was inconsistent about this
<JEEB>
yea, that was a TIL for me too, I never really looked at the spec too much
<JEEB>
so timecodes in the beginning and the FEC stuff at the end
<rcombs>
the avio_skip happened at the end, but the packet pos was set as if it was at the start
<rcombs>
so this makes the code consistent both internally and with the actual correct location of the data
<rcombs>
i.e. timecode is skipped at the beginning and FEC at the end, and the pos is set to before the timecode on streams with that, but after the previous packet's FEC on streams with that
<JEEB>
also this means that raw_packet_size was at various points not correct since it's not being utilized for the pes_cb any more
<JEEB>
my guess without looking at things and the 0x47 in timestamp comment is that this was some really small size or something?
<rcombs>
setting pos correctly fixes seeking, and skipping the timecode at the start fixes demuxing from the start on files with 0x47 in timecodes
<rcombs>
not specific to small sizes, just, we'd start at the beginning of the file and syncword-seek, and find the 0x47 in the timecode instead of the actual start of the MPEGTS packet
<JEEB>
although if it was small and then you give it 188 backwards on a packet that was in the beginning then not sure how that'd work (negative pos?)
<JEEB>
eh, need to I guess generate a sample for that or something
<rcombs>
I'll be boarding a flight soon so might vanish for a few hours
<rcombs>
in theory we should probably improve the robustness of the syncword search to these kinds of cases, but I think this change would be worthwhile either way (esp since it fixes packet pos)
Livio has quit [Quit: leaving]
<JEEB>
yea I'll try to look at this patch at some point and to see if there's no negatives to ignoring/overriding raw_packet_size in the pes callback
<JEEB>
thanks for the sample, grabbed it
<JEEB>
right, so raw_packet_size is supposed to be one of 188, 192 and 204 always
<JEEB>
at least if get_packet_size is called, since the returns that are positive are within that range
Livio has joined #ffmpeg-devel
<rcombs>
yeah
<rcombs>
note this doesn't actually remove use of raw_packet_size, it just special-cases the DVHS size in particular
<rcombs>
any other size should still work (well, ≥188)
<JEEB>
so that expects that where that logic stops is not the actual end of the packet, but just before the possible FEC
<JEEB>
which is how 188 and 192 are the only values you need to poke at
<JEEB>
so that one actually fixes the FEC case, and not the DVHS case, since DVHS case should already be OK with raw_packet_size being 192
<JEEB>
so that would probably be if FEC, remove the FEC amount from raw_packet_size, kind of thing? or some other way that makes it clear that you're not really adjusting the DVHS case (as in that case the raw_packet_size should already be TS_DVHS_PACKET_SIZE), you're adjusting the FEC case.
<JEEB>
since it just now took me a while to grasp that from the logic due to the amount of comments that specifically were talking about DVHS :P
unlord has joined #ffmpeg-devel
___nick___ has quit [Ping timeout: 252 seconds]
Krowl has quit [Read error: Connection reset by peer]
Livio has quit [Ping timeout: 268 seconds]
<jamrial>
courmisch: just sent another dsp function
<jamrial>
for this one i guess i have no other choice than using avx512, because vpsraq
cone-578 has joined #ffmpeg-devel
<cone-578>
ffmpeg James Almer master:467d84a06d1b: checkasm/flacdsp: run lpc benchmarks with an unmodified buffer
<cone-578>
ffmpeg James Almer master:c00c8679ed35: x86/flacdsp: add a SSE4 version of lpc16
<cone-578>
ffmpeg James Almer master:479d26cea29e: checkasm/flacdsp: sanitize lpc arguments
<cone-578>
ffmpeg James Almer master:5acec189afa3: checkasm/flacdsp: add a test for wasted32
<cone-578>
ffmpeg James Almer master:1ba528769702: avcodec/flacdsp: split off wasted bit handling into dsp functions
<cone-578>
ffmpeg James Almer master:0664cbd732fd: x86/flacdsp: add a SSE2 version of wasted32
<llyyr>
could someone take a look at my patch that fixes 10585? it's linked in the trac issue
ccawley2011 has quit [Read error: Connection reset by peer]
IndecisiveTurtle has joined #ffmpeg-devel
Traneptora has joined #ffmpeg-devel
dasome has joined #ffmpeg-devel
dasome has quit [Changing host]
dasome has joined #ffmpeg-devel
sadome has quit [Remote host closed the connection]
dasome has quit [Remote host closed the connection]
sadome has joined #ffmpeg-devel
sadome has joined #ffmpeg-devel
sadome has quit [Changing host]
MisterMinister has joined #ffmpeg-devel
Mister_D has joined #ffmpeg-devel
mkver has quit [Ping timeout: 240 seconds]
MisterMinister has quit [Ping timeout: 272 seconds]
cone-578 has quit [Quit: transmission timeout]
sadome has quit [Remote host closed the connection]
sadome has joined #ffmpeg-devel
sadome has quit [Remote host closed the connection]
sadome has joined #ffmpeg-devel
sadome has joined #ffmpeg-devel
sadome has quit [Changing host]
IndecisiveTurtle has quit [Ping timeout: 256 seconds]