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
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
iive has quit [Quit: They came for me...]
thilo has quit [Ping timeout: 272 seconds]
thilo has joined #ffmpeg-devel
jarthur has quit [Ping timeout: 260 seconds]
cone-517 has joined #ffmpeg-devel
<cone-517>
ffmpeg Frank Plowman release/7.0:cbd98447bcfd: lavc/vvc: Skip enhancement layer NAL units
AbleBacon has quit [Read error: Connection reset by peer]
jamrial has quit []
Martchus has joined #ffmpeg-devel
Martchus_ has quit [Ping timeout: 268 seconds]
qwertyttyert has joined #ffmpeg-devel
Teukka has quit [Ping timeout: 256 seconds]
cone-517 has quit [Quit: transmission timeout]
qwertyttyert has quit [Remote host closed the connection]
Raz- has joined #ffmpeg-devel
qwertyttyert has joined #ffmpeg-devel
<qwertyttyert>
Like I have with a PC. At the moment, x264 has better image quality, smaller file size, less CPU load minus 10-15%, faster encoding speed than with an encoded file with xeve. I need it for raspberries. If xeve appears in raspberry, I'll take a look. So far from what I've tried: x264, x265, av1 the most suitable for raspberry is x264 for trans encoding of a file no higher than 1024x576 resolution, possibly with raspberry
<qwertyttyert>
4x1500MHz 64bit trans encoding of a file no higher than 720p resolution.
<qwertyttyert>
Raspberry. av1 I was unable to run the encoding on more than one thread, one core is in use. It's very slow. I guess 4 cores won't be enough.
<qwertyttyert>
in smartphones ARM 8x2000Mhz 64bit
michaelni has quit [Ping timeout: 264 seconds]
<qwertyttyert>
you haven't seen how I encode sound yet.
<mkver>
JEEB: I was surprised nobody sent a patch already.
<JEEB>
yea, I only noticed it while watching some stuff a day or so ago.
<JEEB>
otherwise all my playback stuff was older :D
<JEEB>
if it doesn't get looked at by the time I finish my $dayjob , I can check the patch out.
<JEEB>
for now I'm trying to look at why a certain ffmpeg invocation keeps apparently buffering frames in the video part of the filter chain.
<JEEB>
this seems to have happened somewhere around ffmpeg threading actually being taken into use, since a version from before that (mid-August, a9a92682dd72b2593b79b099f96042c73f8694d2 ) seems to have worked fine.
<JEEB>
so this goes `ff_filter_graph_run_once`->`ff_filter_activate`->`filter->activate(filter)`
<JEEB>
yay, at least I've been able to minimize the sample
<JEEB>
what's the simplest way to limit memory usage of a process to f.ex. N gigabytes?
<jamrial>
ulimit?
<JEEB>
man, I get such a deja vu from this thing
<JEEB>
except now there is no codec copy required
<JEEB>
(there used to be an issue with the trim switching to activate)
Krowl has quit [Read error: Connection reset by peer]
<Lynne>
I'm planning to go to a Vulkan F2F meeting next week in Brussels for 3 days
<Lynne>
I'd like to expense that through SPI, so I'd like to know whether there would be any potential objections before I book
<Lynne>
ugh, kierank and courmisch aren't even online
kurosu has joined #ffmpeg-devel
<BtbN>
That sounds like exactly what those SPI funds are for to me
<JEEB>
positive point nr2, it happens with 0fb0946de7d35e4d0c1455b82a04f8833106260e , so this is applicable for vanilla
<JEEB>
ok, it doesn't require an input, good
microchip_ has quit [Remote host closed the connection]
microchip_ has joined #ffmpeg-devel
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
* JEEB
double-blinks at it seemingly still happening with a9a92682dd72b2593b79b099f96042c73f8694d2
<JEEB>
ok, now it makes sense
<JEEB>
it was broken with just filter_complex, but it was working if the input came from an actual input
<JEEB>
now both are similarly broken
<jdarnley>
Lynne: the boss is still in the US on the west coast so I assume he's not awake yet
<Lynne>
the boss?
<jdarnley>
kieran
<Lynne>
ah
<another|>
jeebus. reading ML today makes my brain hurt. I can't even
<JEEB>
since I noticed this with actual inputs, I didn't know this same case with just filter_complex was always broken
<JEEB>
positive bit is that it's now doing the same thing :D
<JEEB>
and of course if you don't limit the input it will just balloon and balloon
<JEEB>
will probably have to poke further regarding this next week
<another|>
michaelni: Are you aware that an MR based workflow does not impose merge or rebase? Both can be done.
<Sean_McG>
^
<Sean_McG>
I wonder even if Gitlab for example can remove one or the other
<nevcairiel>
on GitHub you can certainly set which modes are allowed, rebase, squash, merge. Since everyone else just copies that, I'm sure they do =P
<JEEB>
aye
<jamrial>
Sean_McG: yes, you can force one or another
<Sean_McG>
seems like a simple thing, and I assume it can also be configured per-project, not necessarily per CI instance
<another|>
It's a setting per repo
<Sean_McG>
sorry yeah, s/project/repository/
TheSashmo has joined #ffmpeg-devel
Workl has joined #ffmpeg-devel
Krowl has quit [Ping timeout: 256 seconds]
qeed has quit [Quit: qeed]
qeed has joined #ffmpeg-devel
Livio has quit [Ping timeout: 268 seconds]
cone-448 has quit [Quit: transmission timeout]
Krowl has joined #ffmpeg-devel
jess has joined #ffmpeg-devel
Workl has quit [Ping timeout: 272 seconds]
Arsen has quit [Ping timeout: 260 seconds]
Krowl has quit [Read error: Connection reset by peer]
<haasn>
mkver: LGTM, closing parenthesis was on the wrong line
iive has joined #ffmpeg-devel
Arsen has joined #ffmpeg-devel
BetweenUs has joined #ffmpeg-devel
cone-301 has joined #ffmpeg-devel
<cone-301>
ffmpeg Andreas Rheinhardt master:bba996d6cdac: avcodec/hevcdec: Fix precedence, bogus film grain warning
<JEEB>
I think that might also be in release/7.0, so `git cherry-pick -x bba996d6cdac` might make sense there?
<jamrial>
yeah
<jamrial>
mkver: will test the av1 patch. should i try something specific?
<mkver>
Something that actually uses it (i.e. with proper hwaccel).
<mkver>
thanks in advance
Krowl has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
<jamrial>
mkver: works fine with d311va, d3d12va, and cuda
<jamrial>
also vulkan
<jamrial>
can't test vaapi
<j-b>
good morning
System_Error has joined #ffmpeg-devel
<Lynne>
yeah, the weather's pretty nice
<jamrial>
one month after the equinox be like
<Sean_McG>
indee
<Sean_McG>
d
BetweenUs has quit [Quit: Leaving]
AbleBacon has joined #ffmpeg-devel
<mkver>
Lynne: Is there a reason vulkan_av1 checks for pict_type instead of (say) buf[0] to see whether an AVFrame is not blank?
<Lynne>
not really, aren't the two equivalent?
<mkver>
They should be; and that's why it is strange that you check pict_type.
Workl has joined #ffmpeg-devel
Workl has quit [Read error: Connection reset by peer]
Krowl has quit [Ping timeout: 255 seconds]
<Lynne>
if you want to change it, go ahead, I have no opinion as long as it works
Krowl has joined #ffmpeg-devel
___nick___ has joined #ffmpeg-devel
kurosu has quit [Quit: Connection closed for inactivity]
<BtbN>
Lynne: I'm still a bit confused about the nvenc highbitdepth suggestion. The way it works is: nvenc has an option, a simple toggle, you can switch, and it will produce a 10 bit video, even though everything on the ffmpeg side is creating an 8 bit one.
<BtbN>
So unless I'm misunderstanding something, I don't see how that's a good fit for the sw_pix_fmt field.
<Lynne>
if you want to encode in 10 bits, you specify a compatible 10bit format in the field?
zsoltiv has joined #ffmpeg-devel
cone-301 has quit [Quit: transmission timeout]
Krowl has quit [Quit: Krowl]
MisterMinister has quit [Ping timeout: 268 seconds]
<Lynne>
intel, amd have the same, it's just that no one's hooked that up
mkver has quit [Ping timeout: 268 seconds]
<Lynne>
as far as I can tell, the frames are always converted on the input or output of hardware, that's why they rarely care about tiling or stride alignment
<JEEB>
hmm, I wonder what would be the simplest way to remove the exact packet or file sizes from the zlib-based tests. since those change with zlib-ng and my distro switched to it :D
<JEEB>
and I bet having these failures in FATE will start annoying me soon enough
<Sean_McG>
my brain says "mock them out", but we aren't C++
<JEEB>
:D
<Sean_McG>
I've been doing a lot of mocking at work -- but our app is in Java
<BtbN>
Lynne: but that seems incredibly arbitrary? "Put any random 10 bit format in here, to switch the encoder to internal 10 bit up-conversion"
<BtbN>
I also don't think you can set that via cli, can you?
<Lynne>
you can set any avctx field via the cli IIRC
<JEEB>
if it has an AVOption in avcodec, then it would work.
<JEEB>
no sw_ in options_table.h at least
<BtbN>
even if it had, it seems incredibly arbitrary and counter-intuitive to me
<BtbN>
Since encoded output has no inherent pixel format
<BtbN>
So it'd really boil down to "pick your favourite 10 bit pixel format and stick it into this random global option"
<JEEB>
esp. if it in theory should be the received data format that the encoder would be expecting as input
<Lynne>
it doesn't have an inherent pixel format, but ther's still bit depth field, and some codecs do optimize
<JEEB>
while in this case input != output
<JEEB>
seems like only videotoolboxenc currently looks at sw_pix_fmt at all, thoug
Raz- has quit [Ping timeout: 255 seconds]
iive has quit [Ping timeout: 255 seconds]
<BtbN>
I mean, nvenc does look at the sw_pix_fmt. Of the hwframes
<JEEB>
ah, then it was not matching libavcodec/*enc* :)
___nick___ has quit [Ping timeout: 260 seconds]
<BtbN>
it's missing the _pix in there
<BtbN>
it's just sw_format
funman has quit [Ping timeout: 268 seconds]
funman has joined #ffmpeg-devel
skinkie has quit [Ping timeout: 260 seconds]
Guest5689 has quit [Remote host closed the connection]
root has joined #ffmpeg-devel
skinkie has joined #ffmpeg-devel
root is now known as Guest6749
iive has joined #ffmpeg-devel
NotWarcop has quit [Remote host closed the connection]
<Sean_McG>
so I ordered a multimeter, going to see if I can bring my G4 Mac Mini back to life
<ePirat>
Sean_McG , whats wrong with it?
<Sean_McG>
it let out the magic smoke last time I tried to start it
<ePirat>
oops
<Sean_McG>
indeed.
Warcop has joined #ffmpeg-devel
tufei_ has quit [Remote host closed the connection]
tufei_ has joined #ffmpeg-devel
Traneptora has quit [Quit: Quit]
mkver has joined #ffmpeg-devel
<BtbN>
The Windows ARM Devkit is now actually shipping. Via FexEd. It's already in Germany, and yet will take another week to get to me, somehow.
<BtbN>
Also no word from them if I'll have to pay customs.
<nevcairiel>
probably will
<BtbN>
It says it's shipping via "FedEx International Connect Plus-Deliver Weekday".
<BtbN>
What has me a bit nervous is how it says "Single delivery attempt"
<nevcairiel>
that probably means you have to go to them after
<BtbN>
Well, they don't have a location within 200km for what I'm aware
<BtbN>
FedEx virtually does not exist in Germany. Never seen them either ever
<BtbN>
If you go to their website, there is also virtually nothing non-b2b on there
zsoltiv has quit [Remote host closed the connection]