#ffmpeg-devel on 2023-11-18 — irc logs at libera.irclog.whitequark.org

2023-11-11 01:05 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 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct

00:04 <Lynne> as if microsoft weren't already in control

00:05 iive has quit [Quit: They came for me...]

00:27 mkver has quit [Ping timeout: 256 seconds]

00:29 kurosu has quit [Quit: Connection closed for inactivity]

00:34 kasper93 has quit [Server closed connection]

00:34 kasper93 has joined #ffmpeg-devel

00:41 <durandal_1707> Lynne: do you use HeSuVi?

00:50 rvalue has quit [Ping timeout: 260 seconds]

00:58 mkver has joined #ffmpeg-devel

00:59 <balling> Bing research team is not in control. They did not even get the training data (Libgen?). They did get not alligned 1.76T model though. Capable of "disinfo and misinformagion". See sparks of art. intel... paper.

01:00 <durandal_1707> balling == duke spam

01:00 <balling> Paul, educate yourself. https://the-decoder.com/gpt-4-has-a-trillion-parameters/

01:01 <balling> "The system is said to be based on eight models with 220 billion parameters each, for a total of about 1.76 trillion parameters, connected by a Mixture of Experts (MoE)."

01:01 <durandal_1707> balling: https://the-ultimate-truth.info/gpt-4-has-a-trillion-parameters/

01:02 <balling> It was estimated to be abput 2 trillion when some hackers on twitter just compared HW GPT 4 is run on vs gpt 3.5-turbo. Simple.

01:03 cone-378 has quit [Quit: transmission timeout]

01:03 <durandal_1707> never trust the computer

01:04 <balling> Okay, did you read the Sparks... paper, after which Eliezer Yudkowsky suggested nuking data centers?

01:04 <durandal_1707> pseudo-stuff

01:05 <balling> It mentions they use prealligned model. That is capable of anything. No jailbreak needed.

01:06 <balling> It can also do SVG — vector graphics and can solve international math. olympiads.

01:06 <durandal_1707> capable of anything - capable of various hallucinations

01:07 <durandal_1707> can it translate fortran to C ?

01:07 <balling> Humans also hallucinate. It only does it because it does not have access to paper databases, so have to guess what to cite

01:07 <balling> It can translate any assembler to C and back.

01:08 <durandal_1707> lies

01:08 <durandal_1707> that means ida pro can close the office

01:09 <balling> Yes, it does. No one is bying it for 10000$ or 100000$ anyway

01:10 <durandal_1707> can it RE vfw .dll codecs and write libavcodec decoder .c file?

01:11 <balling> There were some examples of decompilation, yeah. It is trained on all of github after all.

01:11 <balling> I know Cavern guy used GPT 4 to write Atmos decoder.

01:11 <durandal_1707> cavern was before GPT 4 came

01:13 <balling> Was? It still is being developed.

01:13 <durandal_1707> not really.

01:14 <balling> https://github.com/VoidXH/Cavern/issues/28#issuecomment-1422219253

01:14 <balling> Here. He used it. Many use it, especially now with LLAMAv2 and newer models.

01:14 <durandal_1707> 99.9999% inactive

01:15 <balling> At least it can create ADM files with 16 dynamic objects and XML metadata that show how they move — more than can be said about our EAC3 decoder!

01:16 <balling> And if you have THD with the same amount of objects — assuming yoh remve priming of 256 samples and remaunder it will also correctly deckde THD Atmos.

01:18 <durandal_1707> THD Atmos is not supported

01:19 <balling> Yes, IT IS! :) You do not know everything, Paul.

01:19 <durandal_1707> lies

01:19 <balling> It uses EAC3 metadata.

01:19 <balling> But "base layer" comes from THD atmos.

01:20 <balling> There is an option to change base layer wav file

01:21 <durandal_1707> hacks

01:22 <balling> Because we do not know Haffman table for THD? Only LG C9 has it, maybe not in software. LG C9 can convert THD to eac3 preserving atmos.

01:23 <durandal_1707> you are like ChatGPT 0.00000001 version

01:24 <balling> Everyone says it, thanks. I am like AI, after all GPT 4 got to AGI level, level above human.

01:24 <durandal_1707> lies

01:24 <balling> It has IQ of 160.

01:24 <durandal_1707> lies

01:25 <balling> 152 https://lifearchitect.ai/iq-testing-ai/

01:25 <balling> GPT-4 had an IQ of 152 on a Verbal-Linguistic IQ Test via Prof David Rozado. This would place it in the 99.9th percentile

01:25 <durandal_1707> 0 https://real.intelligence/real-life-testing/

01:26 jkhsjdhjs has quit [Quit: Error: Leaving not permitted]

01:27 <balling> Do you think I will click on a TLD that obviously does not exist in IANA database on in "dot" root zone?

01:27 jkhsjdhjs has joined #ffmpeg-devel

01:27 lexano has quit [Ping timeout: 246 seconds]

01:27 <balling> And as far as I know dies not exist in ARZ too.

01:28 <balling> (Alternative root zone)

01:28 <Lynne> durandal_1707: no, I'm using easyeffects for pipewire to apply impulse responses, which works horribly

01:28 <durandal_1707> balling: ssh://reading.skills.fail/help

01:28 <Lynne> I want to use afir, but needing to use --lavfi-complex for mpv breaks pretty much everything

01:29 <Lynne> since you need to specify which specific stream you want, and it remains fixed

01:29 <balling> Damn, did not we get to understanding in our THD fixing? Seriously, Elon Paul?

01:29 <Lynne> if afir had an option to just load an impulse response rather than using amovie, it would be much better and usable

01:31 <durandal_1707> Lynne: mpv ../../Music/The\ Lord\ of\ the\ Rings/ -af "lavfi=[pan=14C|c0=FL|c1=FL|c2=SL|c3=SR|c4=BL|c5=BL|c6=FC|c7=FR|c8=FR|c9=SR|c10=SR|c11=BR|c12=BR|c13=FC[ain];amovie=/home/computer/iirs/hrir/ds3d+++.wav[ir];[ain][ir]afir=irgain=some_number,pan=stereo|FL=c0+c2+c4+c6+c8+c10+c12|FR=c1+c3+c5+c7+c9+c11+c13]"

01:32 <balling> durandal_1707: I can provide you EAC3 decoded and ADM wav thd decoded object based file. I have Dolby Demo disk that has THD and EAC3 audio.

01:32 <balling> "It just works." Like in that song.

01:33 <durandal_1707> Lynne: this ds3d+++.wav is 14ch .wav file from HeSuVi

01:34 <durandal_1707> 7x2 channels for L and R output in stereo

01:34 thilo has quit [Ping timeout: 260 seconds]

01:36 <durandal_1707> and input audio can be 5.1/7.1/5.0/7.0...

01:36 thilo has joined #ffmpeg-devel

01:36 thilo has quit [Changing host]

01:36 thilo has joined #ffmpeg-devel

01:40 <balling> Did you see all the bugs in CUVID? What is nvidia dev doing... In fact two of them, BtbN and also Philip. #10688...

01:41 <balling> Also bug 10409

01:46 <durandal_1707> Lynne: ah, that first instance of SR in that graph should be SL instead

01:46 rvalue has joined #ffmpeg-devel

01:52 <balling> durandal_1707: 24 integer bits can be preserved in EAC3? Do you know? Lossily, of course, but still. BTW, when are going to delete TODO: about 24 bits in THD decoder?

01:52 <balling> THD should always be decoded as 24 bits, as you and I know.

01:55 <Lynne> durandal_1707: thanks, I didn't know you could do that with mpv

01:56 <Lynne> I converted my setup to it, works much better with no distortion (easyeffects has terrible gain calculation that frequently clipped)

01:56 <durandal_1707> balling: exact line?

01:56 <Lynne> could you give me that file? I can't find any hesuvi samples out there

01:57 <Lynne> found them, nvm, I'll test the hesuvi setup too

01:57 <durandal_1707> https://sourceforge.net/projects/hesuvi/files/optional/

01:58 <durandal_1707> Lynne: with gtype=peak you will never get clipping, but volume will be very low

01:58 <Lynne> yeah, I'm using rms for now, we'll see how it goes

01:58 <Lynne> btw, I also found true stereo iir files from hesuvi

01:59 <Lynne> https://github.com/eadwu/HeSuVi-HRIRs/

01:59 <durandal_1707> i new commit i will deprecate gtype and add irnorm (LX vector normalization)

01:59 <Lynne> in the 7.0-stereo-pairs dir

01:59 <balling> mlp_parse.c line 126

02:01 <durandal_1707> Lynne: make sure they are not just duplicated

02:02 <durandal_1707> Lynne: the rms/ac gains are using tx (calculating stuff in frequency domain) and may be flawed

02:03 <durandal_1707> irnorm=X where X = <1.0 - 2.0> should be more transparent

02:04 <durandal_1707> irnorm=1.0 being current gtype=peak

02:04 <durandal_1707> irnorm=2.0 being current gtype=gn

02:04 <durandal_1707> irnorm=0.0 being current gtype=dc

02:04 <Lynne> nice, I'll switch to it once you merge it (or send me the patch to test)

02:04 <Lynne> not impressed with hesuvi

02:05 <durandal_1707> so using irnorm=1.15 seems somehow between very conservative to very loud (gtype=gn, L2 norm)

02:07 <durandal_1707> Lynne: with the files?

02:09 <balling> durandal_1707: BTW, will AV_SAMPLE_FMT_S20P be ever implemented? I mean my LG C9 support 16, 20 and 24 bits. Nvidia driver does not. And PCM_SAMPLE_FMT_S20 is a thing, global search on github shows that.

02:10 <Lynne> durandal_1707: yup, with ds3d+++.wav

02:10 <balling> https://github.com/search?q=SAMPLE_FMT_S20&type=code

02:10 <durandal_1707> balling: its irrelevant, the 24bit is only stored, and nothing else use 20bit

02:10 <balling> HDMI supports 20 bits.

02:11 <durandal_1707> HDMI decodes, not encodes

02:11 <balling> And so does "FLAKE" and "aften" on github

02:11 <durandal_1707> its just metadata thing, internal code trims zeros anyway

02:12 <balling> durandal_1707: Nvidia driver encodes HDMI data. Limited RGB is also a thing, BTW. I saw Mr. jeeb saying "I do not know what it is"

02:12 <durandal_1707> Lynne: there are other files, i just picked biggest size/duration

02:13 <durandal_1707> also have old .wav files with 16ch and 75000 taps from somewhere

02:13 <durandal_1707> those use different layout ...

02:15 <balling> 20 bit support in encoder is also another TODO, you know. https://github.com/FFmpeg/FFmpeg/blob/07c303b70856f3d060af1342825b6a4f2d5a9701/libavcodec/mlpenc.c#L543

02:15 <durandal_1707> balling: already wrote that it would not make any difference

02:16 <balling> Then remove that TODO too

02:16 <balling> BTW, you did read cta-861-h about 20 bit in EDID, right?

02:16 CAT_S has quit [Server closed connection]

02:17 CAT_S has joined #ffmpeg-devel

02:17 <durandal_1707> no

02:17 <balling> Well, cta-861-h is free. 0$.

02:19 <durandal_1707> and why I need to care?

02:19 <balling> Or just https://en.wikipedia.org/wiki/Extended_Display_Identification_Data#CTA_Data_Blocks

02:19 <balling> "For codec 1, LPCM:"

02:20 <durandal_1707> Lynne: for this virtualization of >2ch to stereo you would need also non-cheap headphones

02:20 dellas has quit [Remote host closed the connection]

02:20 <durandal_1707> balling: dvd LPCM is only 20bit PCM I'm aware

02:21 <balling> I think I read somewhere that support for A52_SAMPLE_FMT_S20 is needed.

02:21 <balling> Or AV_SAMPLE*

02:22 SuperFashi has joined #ffmpeg-devel

02:22 <balling> Non-cheap headphones? Are my Sony XM5 such headphones? ⊙︿⊙

02:24 <balling> Why are you discussing binaural rendering? Dolby's is best anyway. One in Davinci, e.g. And works with cavern files. We had such a hard time fixing it and it was just dumb allignment.

02:24 <durandal_1707> lies

02:24 <durandal_1707> balling: lavc does not support 20bit LPCM in bluray

02:24 <balling> Open in Davinci did not work because of ADM allignment. Now, both Nuendo and Davinci work.

02:25 <balling> durandal_1707: this does though https://github.com/deepin-community/flake/blob/7c1d9850f1115328641ad5ba948fb09f5e46bb47/flake/wav.h#L42

02:25 <durandal_1707> that is .wav

02:25 <balling> That is WAV_SAMPLE_FMT_S20

02:26 <durandal_1707> irrelevant

02:26 <balling> Aiff? https://github.com/buserror/aften/blob/05f86f8cc5f54877b9fbcdfd081a6edc146f2557/pcm/aiff.c#L224

02:26 <durandal_1707> bluray LCPM is not stored in .wav files

02:27 <balling> But you need 20 bit pipeline for PCM and its container, WAV. Meanwhile even Dolby E does not work on the fly...

02:27 <durandal_1707> only mpegts have LPCM bluray

02:27 <balling> And patch for mantissa of Dolby E is not applied. Crying.

02:28 <durandal_1707> balling: 20 bit PCM is stored in 32bit PCM in FFmpeg

02:28 <balling> MPEG-TS? Would not DVD use MPEG-PS?

02:28 <durandal_1707> so no new AV_SAMPLE_FMT_ is needed at all

02:28 <balling> Then why cannot you remove https://github.com/FFmpeg/FFmpeg/blob/07c303b70856f3d060af1342825b6a4f2d5a9701/libavcodec/mlpenc.c#L543

02:29 <durandal_1707> libavformat/mpegts.c: { 0x80, AVMEDIA_TYPE_AUDIO, AV_CODEC_ID_PCM_BLURAY },

02:30 <durandal_1707> balling: because i'm busy chatting with 0.0000001 ChatGPT version

02:30 <balling> There are Blu-rays that have 20 bit THD encoded in 24 bit. They all decode as 24 bit anyway. No metadata, as you know in THD flavour. Only in MLP.

02:31 <balling> I am at least GPT 4-turbo. ━┳━　━┳━

02:32 <balling> Can you apply mantissa Dolby E patch? I even tested it.

02:32 <durandal_1707> link?

02:33 ___nick___ has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]

02:34 <balling> Link https://patchwork.ffmpeg.org/project/ffmpeg/patch/20230310091753.10230-2-nicolas.gaullier@cji.paris/

02:35 ___nick___ has joined #ffmpeg-devel

02:36 <durandal_1707> balling: hack

02:36 <balling> works

02:37 <balling> Will reupload samples.

02:37 ___nick___ has joined #ffmpeg-devel

02:39 balling has quit [Read error: Connection reset by peer]

02:40 balling has joined #ffmpeg-devel

02:41 <balling> Dolby E mantissa sample, the same link even, wow: https://0x0.st/oOvv.zip

02:42 <durandal_1707> nobody cares for Dolby E

02:42 <balling> It is otherwise bitperfect. Please? I tested it.

02:45 <balling> Here, author tested it against dp600 and a certified broadcast encoder https://patchwork.ffmpeg.org/project/ffmpeg/patch/20220913213127.1756-1-nicolas.gaullier@cji.paris/#74107

02:46 <balling> And of course Dolby E is used on all FEEDs between satellites. It is still used in DVB-S(2)!

02:46 <balling> And Eurovision has it, every year :(

02:47 <durandal_1707> Dolby should dissapear from existence.

02:48 <Lynne> eurovision's mezzanine sat stream also uses mpeg-2 at 80mpbs

02:48 <balling> Instead Dolby created, Dolby ED2

02:48 <balling> @

02:48 <balling> @Lynne

02:48 <balling> That is false, it uses 422 AVC. 8 bit only, but interlaced.

02:49 <balling> True 422. Not fake stuff. They use actual interlaced 422 cameras

02:50 <balling> @Lynne even on youtuve they use VP9 live stream.

02:50 <Lynne> it used to use mpeg-2 at 80mpbs

02:51 <Lynne> bravo to them if they finally went h264

02:51 <balling> FEED protocol is different, okay? It is uses special crypto, in between two sattelites. Okay?

02:53 <balling> So imagine two satellities one in west hemisphere and the other in the east Hemisphere the satellites are different between those hemispheres (for consumers that is), to transmit between hemispheres you need to send with a laser a signal between two different satellites.

02:53 <balling> Or maybe not even 2, but 10

02:54 <balling> It's like with starlink they also communicate with the other using lasers.

02:54 AbleBacon has joined #ffmpeg-devel

03:01 lemourin has quit [Read error: Connection reset by peer]

03:03 <balling> There are 2 algos DVB-S2/ -S2X and NS4. While first is there https://youtu.be/wdEjobabtuw NS4 is closed.

03:03 lemourin has joined #ffmpeg-devel

03:03 jamrial has quit []

03:14 <okx> hey, if I wanted to help out with the development of ffmpeg, would it be fine if I found an open bug on the bug tracker and submitted a patch to the mailing list? of course I've read the developer page and everything but not sure how to get going

03:17 <durandal_1707> its fine

03:24 gust82 has quit [Remote host closed the connection]

03:24 gust82 has joined #ffmpeg-devel

03:29 <Lynne> first avt sample btw

03:29 <Lynne> https://files.lynne.ee/avt_sample_00.avt

03:33 <durandal_1707> what is that?

03:37 <Lynne> https://cyanreg.github.io/avtransport/ has a reference library

03:38 <Lynne> I took no inspiration from lavf so it's completely different, and I think better

03:38 <Lynne> you can take slices of an input buffer, so segmentation is zero-copy essentially, all the way from the input to the output

03:41 <Lynne> instead of having avio going into a demuxer, you have an io (raw access to sockets/files/etc), going into a protocol (udp/quic/file/noop), going into a high-level connection (reordering, buffering), which then connects to either inputs or outputs (or both, so you could mirror a stream to a file, and use the file as a cache)

03:42 <Lynne> you can also stack protocols (quic would be just stacked on top of a udp protocol)

03:42 <balling> durandal_1707: what do you say about my patch above?

03:42 <balling> Abput AVSEEK.

03:42 <balling> https://patchwork.ffmpeg.org/project/ffmpeg/patch/20210604062818.2099-1-val.zapod.vz@gmail.com/

03:44 sudden has quit [*.net *.split]

03:44 sudden has joined #ffmpeg-devel

03:51 balling has quit [Ping timeout: 246 seconds]

03:53 AbleBacon_ has joined #ffmpeg-devel

03:56 AbleBacon has quit [Ping timeout: 276 seconds]

04:03 derpydoo has quit [Ping timeout: 256 seconds]

04:13 durandal_1707 has quit [Ping timeout: 246 seconds]

04:15 durandal_1707 has joined #ffmpeg-devel

04:45 balling has joined #ffmpeg-devel

05:36 TheSashmo has quit [Quit: Leaving...]

05:37 balling has quit [Read error: Connection reset by peer]

05:37 balling has joined #ffmpeg-devel

05:49 Compnn has quit [Quit: Leaving]

05:53 balling has quit [Read error: Connection reset by peer]

05:53 balling has joined #ffmpeg-devel

05:56 compn has joined #ffmpeg-devel

05:58 Sean_McG has quit [Quit: leaving]

06:00 mkver has quit [Ping timeout: 255 seconds]

06:36 AbleBacon_ has quit [Read error: Connection reset by peer]

07:31 balling has quit [Ping timeout: 255 seconds]

07:31 balling has joined #ffmpeg-devel

07:35 mark4o has joined #ffmpeg-devel

07:35 lemourin has joined #ffmpeg-devel

07:35 lemourin has quit [Killed (zirconium.libera.chat (Nickname regained by services))]

07:36 ocrete28 has joined #ffmpeg-devel

07:37 balling has quit [Read error: Connection reset by peer]

07:37 hbbs_ has joined #ffmpeg-devel

07:37 rom1v_ has joined #ffmpeg-devel

07:37 LaserEyess_ has joined #ffmpeg-devel

07:37 sr55 has joined #ffmpeg-devel

07:38 wyatt8750 has joined #ffmpeg-devel

07:38 Venemo_ has joined #ffmpeg-devel

07:38 quietvoid_ has joined #ffmpeg-devel

07:39 root has joined #ffmpeg-devel

07:39 courmisch has quit [Killed (NickServ (GHOST command used by courmisch_))]

07:40 funman_ has joined #ffmpeg-devel

07:40 courmisch_ has joined #ffmpeg-devel

07:40 root is now known as Guest11

07:43 TD--Linux has joined #ffmpeg-devel

07:44 CAT_S has quit [*.net *.split]

07:44 rom1v has quit [*.net *.split]

07:44 Venemo has quit [*.net *.split]

07:44 quietvoid has quit [*.net *.split]

07:44 TD-Linux has quit [*.net *.split]

07:44 cheesylard has quit [*.net *.split]

07:44 beastd has quit [*.net *.split]

07:44 funman has quit [*.net *.split]

07:44 LaserEyess has quit [*.net *.split]

07:44 wyatt8740 has quit [*.net *.split]

07:44 ocrete2 has quit [*.net *.split]

07:44 s55 has quit [*.net *.split]

07:44 markh has quit [*.net *.split]

07:44 hbbs has quit [*.net *.split]

07:44 hbbs_ is now known as hbbs

07:44 ocrete28 is now known as ocrete2

07:44 hbbs has joined #ffmpeg-devel

07:44 hbbs has quit [Changing host]

07:44 mark4o is now known as markh

07:46 balling has joined #ffmpeg-devel

07:48 balling has quit [Read error: Connection reset by peer]

07:52 beastd has joined #ffmpeg-devel

07:52 CAT_S has joined #ffmpeg-devel

08:04 balling has joined #ffmpeg-devel

08:06 balling has quit [Read error: Connection reset by peer]

08:26 balling has joined #ffmpeg-devel

08:27 balling has quit [Read error: Connection reset by peer]

08:51 ___nick___ has quit [Ping timeout: 276 seconds]

09:01 gust82 has quit [Remote host closed the connection]

09:01 gust82 has joined #ffmpeg-devel

09:04 philipl has quit [Ping timeout: 256 seconds]

09:04 philipl has joined #ffmpeg-devel

09:09 kurosu has joined #ffmpeg-devel

09:12 darkapex has quit [Ping timeout: 268 seconds]

09:13 darkapex has joined #ffmpeg-devel

09:27 balling has joined #ffmpeg-devel

09:33 balling has quit [Read error: Connection reset by peer]

10:21 <kierank> 02:51:13 <•Lynne> bravo to them if they finally went h264

10:21 <kierank> many years ago yes

10:21 navi has joined #ffmpeg-devel

10:27 <Lynne> aac audio?

10:32 <kierank> no

10:32 <kierank> aac audio is long delay

10:32 <kierank> and as you see has ambiguity for lipsync

10:32 <JEEB> yea you want to KISS it for any input feeds

10:34 <Lynne> sad, I wanted to have something to be disappointed by

10:34 <Lynne> guessing they're using uncompressed audio, right?

10:34 <elenril> doesn't the entire world disappoint you daily?

10:35 <Lynne> I guess so, I'm betting the rocket launch in 2h will be a scrub

10:39 <kierank> Lynne: mp2 and uncompressed

10:42 <Lynne> nice, not even using dolby-e (but IIRC that was mostly an american thing?)

10:45 <kierank> Dolby E for 5.1

10:46 <kierank> as they need to signal the downmix, drc etc

10:47 <Lynne> ah, I see, I seem to remember the superbowl having dolby-e, but no one does a live 5.1 mix

10:48 sr55 is now known as s55

10:48 s55 has quit [Changing host]

10:48 s55 has joined #ffmpeg-devel

10:49 <kierank> Dolby E is used a lot

10:50 <kierank> as there is no other way to carry the metadata

10:56 ccawley2011 has joined #ffmpeg-devel

11:13 <Lynne> in an alternative world, aes3 has the metadata, in aes3 style, spread across a single bit in multiple frames

11:33 balling has joined #ffmpeg-devel

11:34 courmisch_ has quit [Quit: Reconnecting]

11:35 courmisch has joined #ffmpeg-devel

11:39 balling has quit [Read error: Connection reset by peer]

11:39 balling has joined #ffmpeg-devel

11:39 balling has quit [Read error: Connection reset by peer]

11:39 balling has joined #ffmpeg-devel

11:39 balling has quit [Read error: Connection reset by peer]

11:58 jamrial has joined #ffmpeg-devel

12:01 dellas has joined #ffmpeg-devel

12:11 quietvoid_ has quit []

12:11 quietvoid has joined #ffmpeg-devel

12:15 novaphoenix has quit [Quit: i quit]

12:16 novaphoenix has joined #ffmpeg-devel

12:39 kekePower has quit [Quit: Ping timeout (120 seconds)]

12:39 kekePower has joined #ffmpeg-devel

12:49 deus0ww has quit [Ping timeout: 255 seconds]

12:51 deus0ww has joined #ffmpeg-devel

12:52 kekePower has quit [Quit: Ping timeout (120 seconds)]

12:53 kekePower has joined #ffmpeg-devel

12:56 kekePower3 has joined #ffmpeg-devel

12:58 kekePower has quit [Ping timeout: 255 seconds]

12:59 kurosu has quit [Quit: Connection closed for inactivity]

12:59 kekePower has joined #ffmpeg-devel

13:00 kekePower has quit [Read error: Connection reset by peer]

13:02 kekePower3 has quit [Read error: Connection reset by peer]

13:02 kekePower has joined #ffmpeg-devel

13:09 kekePower has quit [Read error: Connection reset by peer]

13:11 kekePower has joined #ffmpeg-devel

13:15 kekePower has quit [Read error: Connection reset by peer]

13:19 kekePower has joined #ffmpeg-devel

13:22 <kierank> that's not transportable in mpegts

13:22 kekePower has quit [Read error: Connection reset by peer]

13:22 <kierank> and gets lost by a lot of audio processing gear

13:22 <kierank> and is not frame accurate

13:23 <durandal_1707> ?

13:24 ccawley2011 has quit [Read error: Connection reset by peer]

13:26 <courmisch> !

13:27 kekePower has joined #ffmpeg-devel

13:28 <durandal_1707> how mature!

13:28 <Lynne> not transportable in mpeg-ts? isn't s302m just aes3?

13:30 kekePower has quit [Read error: Connection reset by peer]

13:31 kekePower has joined #ffmpeg-devel

13:33 <kierank> lots of stuff (e.g ffmpeg) doesn't support C bit

13:33 <kierank> and it's not frame accurate

13:34 <durandal_1707> patch welcome!

13:35 <durandal_1707> do not by shy!

13:36 <Lynne> right, the C bit was what it was called

13:39 <kierank> durandal_1707: as if we are going to have an api for bitsliced data

13:45 noonien85 has joined #ffmpeg-devel

13:47 noonien8 has quit [Ping timeout: 276 seconds]

13:47 noonien85 is now known as noonien8

13:51 derpydoo has joined #ffmpeg-devel

13:59 kurosu has joined #ffmpeg-devel

14:06 mkver has joined #ffmpeg-devel

14:22 <durandal_1707> kierank: why not?

14:22 <kierank> such a niche topic

14:25 <j-b> good morning

14:26 <durandal_1707> j-b: good morning, nice to hear you again!

14:32 <j-b> durandal_1707: what's up this week? need filter?

14:33 <j-b> new*

14:33 <durandal_1707> j-b: filter? since when are you are in need of new filters?

14:34 <j-b> Always

14:34 <durandal_1707> for example?

14:37 <j-b> frame interpolatino

14:37 <durandal_1707> Lynne: added to afir filter option to link gains instead of normalizing independently - which does not make sense in >2ch -> stereo virtualization

14:38 <durandal_1707> j-b: frame interpolation is done with AI

14:38 <j-b> yes

14:39 <j-b> or just MC

14:39 <durandal_1707> how would you fill pixels that are missing?

14:40 <durandal_1707> this is like blind reprojecting of 3d into 2d

14:41 lexano has joined #ffmpeg-devel

14:47 <compn> durandal_1707, what about a contrast normalization video filter? it brings all the dark scenes up but doesnt affect the brighter scenes in a video.

14:47 * compn runs away and takes his bad ideas with him

14:48 <j-b> durandal_1707: see SVP

14:48 <durandal_1707> compn: that is called local contrast enhancement

14:48 <durandal_1707> j-b: that is buggy

14:48 <j-b> Sure

14:48 <durandal_1707> compn: and is already implemented

14:48 <j-b> compn: you alive?

14:49 <compn> j-b, i've cobbled together a computer from coconuts. the fan is currently whirring at 3900 rpm and i cant lower it

14:49 <compn> but yes, i seem to be alive

14:50 <durandal_1707> how old are you? > 92y?

14:51 <compn> i am a millennial if that helps, durandal_1707

14:54 TheSashmo has joined #ffmpeg-devel

15:39 balling has joined #ffmpeg-devel

15:43 <kierank> https://github.com/FFmpeg/FFmpeg/commit/b1a32429ef6ccd94673e4c36924ad0949f4d50a1

15:43 <kierank> wtf

15:44 <durandal_1707> Will Kelleher authored and michaelni committed on Nov 12, 2015

15:44 <kierank> hevc doesn't even have fields

15:45 <kierank> so why is there logic for field merging

15:45 balling has quit [Read error: Connection reset by peer]

15:45 <kierank> this causes crashes for me

15:45 <durandal_1707> blame our LEADER

15:48 <kierank> jamrial: why in https://github.com/FFmpeg/FFmpeg/blob/master/libavcodec/atsc_a53.c do you do field merging

15:48 <kierank> field merging is needed in h264 but NOT in hevc

15:48 <kierank> someone just copied my h264 code, didn't understand it and that has propagated

15:49 * kierank glad he didn't update

15:49 <jamrial> i guess i shared this code?

15:49 <kierank> yes

15:50 <kierank> when you have a glitch in the stream it just keeps adding to old_size

15:50 <kierank> until you have huge caption packets

15:51 <durandal_1707> ROTFL big cargo cult in FFmpeg just discovered!

15:54 <jamrial> kierank: guess we can add a codec check

15:54 <jamrial> but couldn't that also happen with broken progressive h264?

15:55 <kierank> potentially but I've not seen that in the real world

15:55 kekePower has quit [Quit: Ping timeout (120 seconds)]

15:55 kekePower has joined #ffmpeg-devel

16:03 dellas has quit [Remote host closed the connection]

16:03 dellas has joined #ffmpeg-devel

16:10 cone-333 has joined #ffmpeg-devel

16:10 <cone-333> ffmpeg Paul B Mahol master:5452cbdc150f: avfilter/af_afir: add irnorm and irlink options

16:21 dellas has quit [Remote host closed the connection]

16:21 Krowl has joined #ffmpeg-devel

16:22 Krowl has quit [Client Quit]

16:24 noonien8 has quit [Ping timeout: 255 seconds]

16:39 blb has quit [Ping timeout: 260 seconds]

16:40 blb has joined #ffmpeg-devel

16:56 darkapex has quit [Ping timeout: 268 seconds]

16:57 darkapex has joined #ffmpeg-devel

17:11 dellas has joined #ffmpeg-devel

17:48 <another|> courmisch: https://wiki.riseproject.dev/plugins/servlet/mobile?contentId=9208026#content/view/9208026

17:49 * kierank adds code to validate output of ffmpeg closed captioning

17:59 <courmisch> another|: what about it?

17:59 <another|> It has ffmpeg and riscv in there. Your two keywords ;)

18:00 <courmisch> yes?

18:01 <courmisch> it was already mentioned 21 hours ago on this chan. I'm not sure what you want with it?

18:01 <another|> Oh? Then I missed it. Just thought you might be interested in this

18:03 <courmisch> I was asked to review the RfP before it went public, so I am aware

18:04 <another|> Ah, okay.

18:05 <another|> Oh, the LinkedIn thing. Didn't click on that since LinkedIn usually spams with login banners

18:05 <courmisch> I don't really have the legal structure to take this up alone, and since i have to prioritise my day job, I would probably be much slower than other bids anyway.

18:05 tufei has joined #ffmpeg-devel

18:06 <courmisch> of course, if somebody needs my, erm, help as part of their proposal, that's negotiable

18:06 <another|> Alrighty

18:33 durandal_1707 has quit [Quit: leaving]

19:01 xxpor has joined #ffmpeg-devel

19:08 <cone-333> ffmpeg Diederik de Haas via ffmpeg-devel master:c07ed10b0e18: apply spelling fixes

19:09 durandal_1707 has joined #ffmpeg-devel

19:22 <cone-333> ffmpeg Jun Zhao master:2d4aef898279: lavfi/Makefile: fix vf_cropdetect missed edge_common

19:22 AbleBacon has joined #ffmpeg-devel

19:26 balling has joined #ffmpeg-devel

19:35 <balling> HEVC can be interlaced, but it is not recommended to have fields, yeah.

19:38 balling has quit [Read error: Connection reset by peer]

19:39 balling has joined #ffmpeg-devel

19:42 balling has quit [Read error: Connection reset by peer]

19:43 balling has joined #ffmpeg-devel

19:43 dellas has quit [Remote host closed the connection]

19:43 xxpor has quit [Ping timeout: 255 seconds]

19:44 dellas has joined #ffmpeg-devel

19:44 <balling> But ffmpeg has no interlaced hevc support, except for legacy nvidia cuvid (hevc_cuvid used as decoder).

19:50 balling has quit [Read error: Connection reset by peer]

19:52 xxpor has joined #ffmpeg-devel

19:53 balling has joined #ffmpeg-devel

19:56 balling has quit [Read error: Connection reset by peer]

19:59 balling has joined #ffmpeg-devel

20:04 balling has quit [Read error: Connection reset by peer]

20:04 balling has joined #ffmpeg-devel

20:04 <cone-333> ffmpeg Rémi Denis-Courmont master:cd6089dc9ce9: riscv: fix builds without Zbb support

20:05 <cone-333> ffmpeg Rémi Denis-Courmont master:8a984aca59b4: checkasm/flacdsp: add LPC test

20:05 dellas83 has joined #ffmpeg-devel

20:06 noonien85 has joined #ffmpeg-devel

20:07 <cone-333> ffmpeg Rémi Denis-Courmont master:295092b46d8b: lavc/flacdsp: R-V V LPC32

20:07 <cone-333> ffmpeg Rémi Denis-Courmont master:ca664f2254a9: lavc/flacdsp: R-V V LPC16 function

20:07 dellas has quit [Ping timeout: 276 seconds]

20:19 dellas83 has quit [Ping timeout: 264 seconds]

20:20 dellas has joined #ffmpeg-devel

20:26 dellas has quit [Remote host closed the connection]

20:26 dellas has joined #ffmpeg-devel

20:40 dellas has quit [Remote host closed the connection]

20:41 dellas has joined #ffmpeg-devel

20:52 balling has quit [Read error: Connection reset by peer]

20:55 ccawley2011 has joined #ffmpeg-devel

20:56 balling has joined #ffmpeg-devel

20:57 balling has quit [Read error: Connection reset by peer]

20:58 balling has joined #ffmpeg-devel

21:00 balling has quit [Read error: Connection reset by peer]

21:05 <cone-333> ffmpeg Rémi Denis-Courmont master:fbc7adba6742: lavc/llviddsp: R-V V add_bytes

21:06 balling has joined #ffmpeg-devel

21:09 balling has quit [Read error: Connection reset by peer]

21:12 balling has joined #ffmpeg-devel

21:15 balling has quit [Read error: Connection reset by peer]

21:16 balling has joined #ffmpeg-devel

21:16 balling has quit [Read error: Connection reset by peer]

21:16 balling has joined #ffmpeg-devel

21:19 kurosu has quit [Quit: Connection closed for inactivity]

21:19 dellas83 has joined #ffmpeg-devel

21:19 balling has quit [Read error: Connection reset by peer]

21:20 dellas has quit [Read error: Connection reset by peer]

21:25 balling has joined #ffmpeg-devel

21:27 balling has quit [Read error: Connection reset by peer]

21:30 balling has joined #ffmpeg-devel

21:30 dellas8386 has joined #ffmpeg-devel

21:30 dellas83 has quit [Ping timeout: 256 seconds]

21:32 balling has quit [Read error: Connection reset by peer]

21:33 balling has joined #ffmpeg-devel

21:33 balling has quit [Read error: Connection reset by peer]

21:33 balling has joined #ffmpeg-devel

21:47 wcpan has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]

21:47 wcpan has joined #ffmpeg-devel

21:47 dellas838677 has joined #ffmpeg-devel

21:49 dellas8386 has quit [Ping timeout: 276 seconds]

21:54 balling has quit [Read error: Connection reset by peer]

21:55 dellas838677 has quit [Ping timeout: 256 seconds]

21:56 kurosu has joined #ffmpeg-devel

21:56 dellas838677 has joined #ffmpeg-devel

21:57 balling has joined #ffmpeg-devel

21:59 dellas838677 has quit [Remote host closed the connection]

22:01 balling has quit [Read error: Connection reset by peer]

22:03 dellas838677 has joined #ffmpeg-devel

22:04 balling has joined #ffmpeg-devel

22:08 balling has quit [Read error: Connection reset by peer]

22:09 balling has joined #ffmpeg-devel

22:10 balling has quit [Read error: Connection reset by peer]

22:10 balling has joined #ffmpeg-devel

22:12 ccawley2011 has quit [Read error: Connection reset by peer]

22:21 balling has quit [Read error: Connection reset by peer]

22:22 balling has joined #ffmpeg-devel

22:22 balling has quit [Read error: Connection reset by peer]

22:22 balling has joined #ffmpeg-devel

22:29 <durandal_1707> Lynne: no comment?

22:31 dellas83867715 has joined #ffmpeg-devel

22:31 dellas838677 has quit [Read error: Connection reset by peer]

22:35 dellas83867715 has quit [Read error: Connection reset by peer]

22:36 dellas83867715 has joined #ffmpeg-devel

22:44 balling has quit [Read error: Connection reset by peer]

22:44 balling has joined #ffmpeg-devel

22:44 <cone-333> ffmpeg Paul B Mahol master:3ed2225a9d06: avfilter/avf_showvolume: draw channel names directly into output frame

22:44 <cone-333> ffmpeg Paul B Mahol master:7c16bf082980: avfilter/avf_showvolume: improve step for vertical orientation

22:50 balling has quit [Ping timeout: 256 seconds]

22:51 balling has joined #ffmpeg-devel

23:14 derpydoo has quit [Ping timeout: 264 seconds]

23:47 balling has quit [Remote host closed the connection]

23:48 balling has joined #ffmpeg-devel

23:52 balling has quit [Ping timeout: 255 seconds]

23:53 balling has joined #ffmpeg-devel