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 7.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
System_Error has joined #ffmpeg-devel
drv has quit [Ping timeout: 248 seconds]
drv has joined #ffmpeg-devel
iive has quit [Quit: They came for me...]
cone-044 has joined #ffmpeg-devel
<cone-044> ffmpeg softworkz master:c79cdae37772: avutil/log: support logging of date and time information
<cone-044> ffmpeg Michael Niedermayer master:1afbc4087506: avformat/wavdec: Fix overflow of intermediate in block_align check
<cone-044> ffmpeg softworkz master:9fd23f712a99: fftools/opt_common: add time and datetime log flags
<cone-044> ffmpeg Michael Niedermayer master:251d43aef0df: avformat/mlvdec: fix size checks
<cone-044> ffmpeg Michael Niedermayer master:d1be369af6a6: avcodec/aac/aacdec: Clear SFO on error
<cone-044> ffmpeg Michael Niedermayer master:41cd72d76e14: avcodec/aac/aacdec_usac: Simplify decode_usac_scale_factors()
<cone-044> ffmpeg Michael Niedermayer master:464fb861b17b: avcodec/aac/aacdec_lpd: Limit get_unary()
<cone-044> ffmpeg Michael Niedermayer master:8668957ef604: avformat/iff: Check that we have a stream in read_dst_frame()
<cone-044> ffmpeg Michael Niedermayer master:01f74bfaa2e0: avcodec/rv60dec: Initialize slice gb with actually allocated size
thilo has quit [Ping timeout: 248 seconds]
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
thilo has joined #ffmpeg-devel
<fflogger> [newticket] 0x20z: Ticket #11475 ([avutil] memory leaks error at libavutil/mem.c:107) created https://trac.ffmpeg.org/ticket/11475
jamrial has quit []
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
^Neo has quit [Ping timeout: 260 seconds]
cone-044 has quit [Quit: transmission timeout]
tanay-man has quit [Remote host closed the connection]
Martchus has joined #ffmpeg-devel
Martchus_ has quit [Ping timeout: 268 seconds]
tanay-man has joined #ffmpeg-devel
Warcop has joined #ffmpeg-devel
tanay-man has quit [Remote host closed the connection]
tanay-man has joined #ffmpeg-devel
kasper93 has quit [Remote host closed the connection]
kasper93 has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
mkver has joined #ffmpeg-devel
tanay-man has quit [Ping timeout: 260 seconds]
Warcop has quit [Remote host closed the connection]
<fflogger> [newticket] Aleksoid1978: Ticket #11476 ([avcodec] AMF Decoder don't fill AVCodecContext width/height/coded_width/coded_height if no width/height set before avcodec_open2) created https://trac.ffmpeg.org/ticket/11476
sm2n has quit [Remote host closed the connection]
OctopusET has quit [Remote host closed the connection]
MyNetAz has quit [Write error: Broken pipe]
sm2n has joined #ffmpeg-devel
OctopusET has joined #ffmpeg-devel
MyNetAz has joined #ffmpeg-devel
cone-425 has joined #ffmpeg-devel
<cone-425> ffmpeg Zhao Zhili master:d3220ed8181d: avcodec/hevc: Rewrite scalability_mask_flag parse in decode_vps_ext
<cone-425> ffmpeg Zhao Zhili master:d367016d3cca: avcodec/hevc/ps: Add basic HEVC_SCALABILITY_AUXILIARY support
<cone-425> ffmpeg Zhao Zhili master:f926b6045522: avcodec/hevc: Add alpha layer support
<cone-425> ffmpeg Zhao Zhili master:88d9ecaa7b79: avutil/frame: Fix direct pointer compare between different array
ngaullier has joined #ffmpeg-devel
^Neo has joined #ffmpeg-devel
^Neo has quit [Changing host]
^Neo has joined #ffmpeg-devel
<cone-425> ffmpeg Krzysztof Pyrkosz master:64107e22f545: swscale/aarch64/rgb24toyv12: skip early right shift by 2
^Neo has quit [Ping timeout: 244 seconds]
ngaullier has quit [Remote host closed the connection]
pross has joined #ffmpeg-devel
<pross> FFmpeg obviously has lots of follows. GSoC hasn't even started yet and i have three mails from ppl wanting more info about vp6 encoder project.
<wbs> lots of VP6 lovers out there? ;-)
<pross> yeah, that is worrying
<pross> its almost 20 year old codec
<JEEB> some seem to apply due to it being a "known entity", at least that one person who popped here on IRC
<JEEB> although I don't think it has a spec, but I guess multimedia.cx and existing decoder help
Sean_McG has joined #ffmpeg-devel
ngaullier has joined #ffmpeg-devel
<cone-425> ffmpeg Krzysztof Pyrkosz master:b92577405b40: swscale/aarch64/rgb2rgb_neon: Implemented {yuyv, uyvy}toyuv{420, 422}
HarshK23 has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
<mindfreeze> I remember reading the adventures from Kotsya on VP6 Encoder implementation which is not so old might be a good reference to know some encoder implementation nuances
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
tufei_ has joined #ffmpeg-devel
tufei has quit [Ping timeout: 264 seconds]
microchip_ has quit [Quit: There is no spoon!]
microchip_ has joined #ffmpeg-devel
tanay-man has joined #ffmpeg-devel
jamrial has joined #ffmpeg-devel
cone-425 has quit [Quit: transmission timeout]
cone-838 has joined #ffmpeg-devel
<cone-838> ffmpeg James Almer master:0e7c2a6287f9: fftools/ffmpeg: use an int for nb_threads
^Neo has joined #ffmpeg-devel
^Neo has joined #ffmpeg-devel
<haasn> How would people feel about changing "Parsed_%s_%zu" in libavfilter/graphparser.c to "%s@%zu" ?
<haasn> This would give them a consistent format with explicitly named filter instances
<haasn> although it introduces the possibility for conflict if I write -vf "null@1, null"
<haasn> both would end up named null@1
<haasn> probably not worth the trouble, this naming scheme is explicitly documented
bhaskar has joined #ffmpeg-devel
<devinheitmuell-1> haasn: FWIW, I did a patch years ago which dumps out the graph in Graphviz (i.e. dot) format. I was pretty happy with the result. https://github.com/LTNGlobal-opensource/FFmpeg-ltn/commit/16497dafb411d1266dfd0f0a7d4a77ed8368e075
<JEEB> I think on ML there was a patch set to utilize the additional formats in the API
<JEEB> I don't think that ever got merged
<haasn> interesting
<devinheitmuell-1> Also, I extended my tree many years ago to propagate interlaced/progressive as well as TFF/BFF through AVFilterLinks. I brought it up on the mailing list at the time, and extending the framework seemed like it would face significant resistance.
<haasn> devinheitmuell-1: well, I have managed to successfully extend the filter metadata to add colorspace properties
<haasn> what type of resistance did you encounter? do you have a link to the ML thread?
<devinheitmuell-1> It was mainly about the fact that AVPackets and AVFrames represent the interlaced/tff/bff info in diffferent formats (and no consensus on which is “better”). Let me find the link…
<haasn> we did migrate from interlaced_frame/top_field_first to AV_FRAME_FLAG_*
<haasn> though the other frame flags don't seem like they would make sense on the link layer
<devinheitmuell-1> TFF/BFF on the link layer makes sense if you need to convert from TFF to BFF or vice versa (and thus during negoiation need to add an additional filter)
<devinheitmuell-1> For example, if the incoming video was MPEG-2 with TFF (common with 480i), I would need to phase shift it to output on decklink (which requires BFF). If I inserted the filter into the graph, the wrong value would still be announced to decklink (since it wasn’t taking into account the filter being present).
<devinheitmuell-1> Of course in addition to having it in the filterlinks, I also had to update various filters which might change the values (e.g. deinterlacing/interlacing/phase, etc)
<devinheitmuell-1> If such a model is upstreamed, I also have those patches to various filters (at least the ones which I cared about).
<Lynne> decklink can only do BFF?
<devinheitmuell-1> Lynne: If the video is 480i, yes.
<devinheitmuell-1> It follows the SDI standard, and 480i video on SDI is always BFF.
<Lynne> ah, I see
<Lynne> that explains a few things I've seen
<devinheitmuell-1> Let’s face it, most video is TFF so few people care about it. But I have to both put out 480i video as BFF (where it’s commonly TFF in the compressed domain), and I need to be able to upconvert 480i to 1080i59, which requires conversion.
<Lynne> though I've never seen SDI being used for pre-HD systems, only analogue over 75ohm cables, which you couldn't convert over to digital
Marth64 has joined #ffmpeg-devel
* kepstin notes that DV, like used in many SD camcorders, is also bottom field first.
<devinheitmuell-1> Yeah, it’s surprisingly common in the United States. In particular because over-the-air stations can carry “subchannels” which are 480i so they can put multiple video feeds into a single RF channel for broadcast.
<devinheitmuell-1> In fact, I have to more downconversion from 1080i59 to 480i59, than I have upconversion, since in general the channels originate as HD but need to put out SD versions for certain takers.
bhaskar has quit [Quit: Konversation terminated!]
lexano has quit [Remote host closed the connection]
<haasn> What is the difference between tinterlace and interlace?
<haasn> they seem to be overlapping heavily
<devinheitmuell-1> IIRC, one is a rewrite that is LGPL rather than GPL.
<devinheitmuell-1> Hmm, perhaps I am mistaken. I thought that was the rationale originally, but I appear to be misremembering.
bhaskar has joined #ffmpeg-devel
abdu has joined #ffmpeg-devel
microlappy has joined #ffmpeg-devel
microlappy has quit [Quit: Konversation terminated!]
microlappy has joined #ffmpeg-devel
<kierank> haasn: iirc tinterlace came from mplayer
<kierank> and had basically zero docs as one would expect
<kierank> and interlace was meant to be a native libavfilter version
<haasn> fair
<Marth64> speaking of ancient filters, are there reasonable use cases for 'movie' filter today besides closed caption extraction?
<Marth64> i.e. does it add any value besides that since one can pull in and filter complex multiple inputs
<devinheitmuell-1> Going back through my emails, it looks like there was some intention/goal to have tinterlace be a GPL replacement for interlace, but seems like they never accomplished that: https://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2018-March/226080.html
<devinheitmuell-1> Pardon, I meant “LGPL replacement"
Marth64[m] has joined #ffmpeg-devel
Marth64 has quit [Ping timeout: 260 seconds]
<haasn> Lynne: how do you use texture() on ivec4 pos; in vulkan filters?
<haasn> did texture() gain some overload that accepts integer coordinates now?
<haasn> ivec2 pos*
<Lynne> I guess it did, didn't know you needed float coords
<Lynne> these days, with most storage image problems solved, and texel cache no longer being a thing, I'd just use storage images for inputs unless I needed floats
<Lynne> IIRC it was not inconsiderably faster when I benchmarked it for ffv1 encode (which does ~9 lookups per each pixel per plane)
microlappy has quit [Quit: Konversation terminated!]
<Marth64[m]> speaking of movie filter, the below command should not be working ... am i crazy?
<Marth64[m]> ffmpeg -protocol_whitelist http -f lavfi -i "movie='https\://ffmpeg.org/img/coverity-lifetime-2024-08.PNG'"
<Marth64[m]> completely ignores the protocol whitelist ?
Marth64[m] is now known as Marth64
microlappy has joined #ffmpeg-devel
<haasn> Lynne: oh, no, it's just silently promoting ivec2 to vec2
microlappy has quit [Ping timeout: 244 seconds]
<Lynne> right
<haasn> is it just me or do these coordinates not work though
<haasn> I copied the fallback code from avgblur_vulkan exactly but I get linear interpolation on my pixel values when sampling from exactly `pos`
<haasn> this filter also redefines vec4 res
<Lynne> what kind of sampler did you enable?
<haasn> yeah, indeed, this filter mangles passthrough planes
<haasn> try with avgblur_vulkan=planes=0, it blurs the entire frame
<haasn> as if bilinear sampling from the wrong coordinate
<haasn> Lynne: didn't modify the code, so linear sampling
<haasn> if i use nearest instead it works
<haasn> I suspect you need to add in the 0.5 pixel center offset
<haasn> this means that your blur kernel is probably also broken
<haasn> actually, so are probably most vulkan filters since afaict they all have the same issue
<Lynne> I don't get what the issue is
<Lynne> texture(0, 0) and texture(1, 1) should access the values at 0, 0 and 1, 1 respectively, right?
cone-838 has quit [Quit: transmission timeout]
<haasn> no
<haasn> if you image a 1x1 image, then the single sample will be at (0.5, 0.5)
<haasn> imagine a*
<JEEB> mid-sample?
<haasn> for a 2x1 image, normally coordinates are normalized so that your samples are at (0.25, 0.5) and (0.75, 0.5)
<haasn> fun fact, avgblur_vulkan also segfaults if the shader fails to compile
<Lynne> whatever
<haasn> ah but you use .unnormalizedCoordinates = unnorm_coords,
<Lynne> yeah, pretty much everywhere
<haasn> right, but that doesn't seem to align it with integer coordinates
<haasn> it just means the range is now (0.5,1.5) instead of (0.25,0.75)
<haasn> so you need to add on a constant 0.5 in any case
<courmisch> annoying how FATE instances refuse to start due to lock if they crash during a run
abdu has quit [Ping timeout: 240 seconds]
HarshK23 has quit [Quit: Connection closed for inactivity]
abdu has joined #ffmpeg-devel
lexano has joined #ffmpeg-devel
tufei__ has joined #ffmpeg-devel
tufei_ has quit [Ping timeout: 264 seconds]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
rvalue- has joined #ffmpeg-devel
rvalue has quit [Ping timeout: 260 seconds]
<JEEB> hopefully that causes some discussion
WinnerLeader has joined #ffmpeg-devel
rvalue- is now known as rvalue
HarshK23 has joined #ffmpeg-devel
Marth64 has quit [Remote host closed the connection]
Shubham has joined #ffmpeg-devel
<fflogger> [newticket] mikey1959515: Ticket #11477 ([undetermined] webvtt is not in allowed_extensions by recent git master versions only) created https://trac.ffmpeg.org/ticket/11477
<WinnerLeader> is makima good sample rate interpolator for audio?
Shubham has quit [Quit: Client closed]
<WinnerLeader> jamrial there leak somewhere in iamf see trac
Shubham has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
<fflogger> [editedticket] MasterQuestionable: Ticket #11477 ([ffmpeg] "webvtt is not in allowed_extensions" by recent git master versions only) updated https://trac.ffmpeg.org/ticket/11477#comment:4
Marth64 has joined #ffmpeg-devel
<Lynne> haasn: do NEAREST samplers also need +0.5?
<WinnerLeader> Lynne: try kriging resampler for audio, its max delay can be just 1 sample! (i have no skills to implement it)
<jamrial> WinnerLeader: already sent a patch
System_Error has joined #ffmpeg-devel
Shubham has quit [Quit: Client closed]
Shubham has joined #ffmpeg-devel
abdu9 has joined #ffmpeg-devel
abdu has quit [Ping timeout: 240 seconds]
ngaullier has quit [Remote host closed the connection]
<Shubham> Hello I am Shubham Nayak, I am interested to be a GSOC contributor in FFMPEG this year. It would be great if someone could point me to beginner friendly issues where I can contribute.

Also, some getting started tasks that will help me understand the code base would help.
<Shubham> Thanks in advance.
<JEEB> start with building the project, then you can attempt to run the FATE test suite :)
<Marth64> Welcome
<JEEB> whoops, was responding further in the wrong lpace
<JEEB> essentially, something like `mkdir -p lunix_build && cd lunix_build && path/to/ffmpeg/configure` and if the configuration succeeds, `make -jN` should work, where N is the amount of things you want to build/link at the same time
<JEEB> (doing out of tree builds means that if you ever need to hard reset you can just remove the build dir)
<JEEB> on x86_64 the requirements are basically basic compiler tools and nasm for the hand-written SIMD
minimal has joined #ffmpeg-devel
<Lynne> ePirat: could you look into fixing xfade_vulkan, specifically texture() calls need +0.5f offset to get the actual sample
<ePirat> Lynne, sure
<Lynne> plus, you use NEAREST sampler, but for WIPETR/WIPEBL/PIXELIZE crossfades, you use a float offset
<Lynne> is this intended? it won't do any interpolation in NEAREST mode
Shubham has quit [Quit: Client closed]
iive has joined #ffmpeg-devel
<ePirat> Lynne, iiuc we dont do any interpolation in the cpu version of the filter, so I guess its fine if we want to match what that does?
<haasn> Lynne: yes, unless you like implementation defined behavior
<ePirat> then again I am not opposed to changing it if it results in better results, it just did not matter for what I used it for
<Lynne> sure, up to you
<ePirat> Lynne, also whats "old Intel hardware"
<Lynne> Skylake
<WinnerLeader> fg
<ePirat> what happens with the change for people running on that?
System_Error has quit [Remote host closed the connection]
<Lynne> a small speedup, unless you're doing 20+ loads per each pixel, in which case approximately 20% slowdown
<BtbN> JEEB: all those encoder upstreams really are horrible lately...
<BtbN> x265 and svt-av1 both can't API...
<BtbN> Or ABI
<JEEB> aye
System_Error has joined #ffmpeg-devel
pross has quit [Remote host closed the connection]
Flat has quit [Ping timeout: 252 seconds]
<Lynne> haasn: also sent patches to fix the crash and the handling of uncompiled shaders
<haasn> thx
<Lynne> were you working or hacking on a filter?
Flat has joined #ffmpeg-devel
<haasn> I sent it to the ML
<haasn> vf_interlace_vulkan
abdu9 has quit [Ping timeout: 240 seconds]
abdu9 has joined #ffmpeg-devel
IndecisiveTurtle has joined #ffmpeg-devel
WinnerLeader has quit [Quit: Client closed]
cone-014 has joined #ffmpeg-devel
<cone-014> ffmpeg Niklas Haas master:4dc2ae69e72b: avfilter/interlace_vulkan: add interlace_vulkan filter
abdu15 has joined #ffmpeg-devel
abdu9 has quit [Ping timeout: 240 seconds]
Marth64 has quit [Quit: Leaving]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
abdu has joined #ffmpeg-devel
abdu15 has quit [Ping timeout: 240 seconds]
c1480 has quit [Read error: Connection reset by peer]
c1480 has joined #ffmpeg-devel
DauntlessOne4 has quit [Read error: Connection reset by peer]
c1480 has quit [Read error: Connection reset by peer]
c1480 has joined #ffmpeg-devel
DauntlessOne4 has joined #ffmpeg-devel
witchymary has quit [Remote host closed the connection]
witchymary has joined #ffmpeg-devel
IndecisiveTurtle has quit [Quit: IndecisiveTurtle]
c1480 has quit [Read error: Connection reset by peer]
c1480 has joined #ffmpeg-devel
<BtbN> The robots.txt seems to have largely solved the issue of exploding repo-archive storage size
s55 has quit [Read error: Connection reset by peer]
s55 has joined #ffmpeg-devel
sr55 has joined #ffmpeg-devel
s55 has quit [Read error: Connection reset by peer]
pross has joined #ffmpeg-devel
c1480 has quit [Read error: Connection reset by peer]
c1480_ has joined #ffmpeg-devel
<Compn> for now
sr55 is now known as s55
s55 has quit [Changing host]
s55 has joined #ffmpeg-devel
mkver has quit [Ping timeout: 268 seconds]
Guest27 has joined #ffmpeg-devel
cone-014 has quit [Quit: transmission timeout]
<srikanth-> Marth64: for the movie filter, I used it until recently to convert the VFR sources to CFR before applying the trim filter. This avoided the issue with using the fast seek option `-ss` with for wildly VFR sources like screen captures where we always get the NEXT frame after the specified timestamps
Guest27 has quit [Quit: Client closed]
<srikanth-> so instead of `ffmpeg -ss 35 -i vfr.mp4`, I used something like `ffmpeg -f lavfi -i movie=sp=35:filename=vfr.mp4,fps=25,tpad=stop=-1:mode=clone,trim=start=35:end=38` and can simulate the behaviour expected when seen on a player
ramiro has quit [Ping timeout: 252 seconds]
Gramner has quit [Ping timeout: 252 seconds]
ramiro has joined #ffmpeg-devel
DauntlessOne4 has quit [Read error: Connection reset by peer]
abdu has quit [Ping timeout: 240 seconds]
iive has quit [Quit: They came for me...]