00:00 <fflogger> [newticket] jphowey: Ticket #11472 ([ffmpeg] horizontal bar artifacts from near 0 value pixels.) created https://trac.ffmpeg.org/ticket/11472

02:09 <fflogger> [editedticket] stroltz: Ticket #11465 ([undetermined] SRT stops after 10min (to Castr.com)) updated https://trac.ffmpeg.org/ticket/11465#comment:2

03:16 <fflogger> [editedticket] Wallboy: Ticket #11469 ([ffmpeg] ffmpeg_demux: readrate plays "catch up" if output is blocked, then later resumed) updated https://trac.ffmpeg.org/ticket/11469#comment:3

04:03 <cone-789> ffmpeg Frank Plowman master:e417f939da2d: lavc/vvc: Set fc->ref to NULL at top of decode_nal_units

04:18 <fflogger> [newticket] jumperes: Ticket #11473 ([undetermined] Seeking video in mpv started to cause green and blocky glitches with FFmpeg 7.1 (using VAAPI)) created https://trac.ffmpeg.org/ticket/11473

05:55 <fflogger> [editedticket] Balling: Ticket #11473 ([undetermined] Seeking video in mpv started to cause green and blocky glitches with FFmpeg 7.1 (using VAAPI)) updated https://trac.ffmpeg.org/ticket/11473#comment:2

10:41 <fflogger> [editedticket] jumperes: Ticket #11473 ([undetermined] Seeking video in mpv started to cause green and blocky glitches with FFmpeg 7.1 (using VAAPI)) updated https://trac.ffmpeg.org/ticket/11473#comment:3

12:54 <haasn> does ffmpeg have an existing debug option to dump the generated filter graph (after insertion of auto filters)?

12:54 <haasn> considering adding one

12:54 <Lynne> yeah, avfilter_graph_dump

12:55 <Lynne> there's even an unused option argument so you can have a different dump type

12:56 <haasn> does that print in the same format as what -filter_complex accepts?

12:59 <haasn> seems like it's only used for lavfi inputs atm

13:02 <haasn> ah no, it's some custom format

13:05 <haasn> seems like it needs some love

16:05 <haasn> if (!av_strcasecmp(color_string2, "random") || !av_strcasecmp(color_string2, "bikeshed")) {

16:05 <haasn> I had a laugh, thanks whoever put this here

16:22 <llyyr> wow that is old https://github.com/FFmpeg/FFmpeg/commit/4fa61d1e982ea96f5376959c2abcab4eafddf98a

16:32 <haasn> how reasonable would it be to make interlaced vs progressive a property of the (video) link?

16:32 <haasn> resolution, aspect ratio and frame rate are already link properties

16:39 <Lynne> its not reasonable, sadly

16:40 <Lynne> in telecine, you can have one frame flagged as interlaced, and another as progressive

16:41 <Lynne> not to mention you can have stream switches

17:24 <kepstin> NTSC dvd stuff is a mess :/ iirc that "alternating interlaced/progressive" frames thing in soft-telecined content is actually an encoder bug - it's most often seen when all the frames are encoded progressively with alternating repeat_first_field flags. Actually following the progressive frame flags in that case will result in half the frames having artifacts due to incorrectly using interlaced chroma resampling :/

17:28 <gnafu> I've been dreading that kind of thing as I contemplate ripping some favorite TV shows from DVDs.

17:29 <fflogger> [editedticket] MasterQuestionable: Ticket #11473 ([avcodec] [Regression] H.264 VAAPI hardware decoding unusual artifacts) updated https://trac.ffmpeg.org/ticket/11473#comment:4

17:29 <kepstin> iirc the dvd specs require 30fps progressive content to be encoded as interlaced 2:2; and of course a single stream can contain _any combination_ of 30i, 30p, and 24p content with the 24p content being encoded as progressive soft-telecine or interlaced hard-telecine or both...

17:29 <gnafu> I see even my otherwise very nice Sony UHD Blu-ray player struggle when flags presumably change from one scene tk the next on my _Gilmore Girls_ discs.

17:30 <gnafu> _Gilmore Girls_ seems to switch between 24 and 30 fps content, sometimes eithin an episode. I'm interested to see how ffmpeg handles it.

17:31 <gnafu> s/eithin/within/

17:34 <kepstin> worst case I've seen is the opening from the serial experiments lain anime dvds, which contain content animated on 2s telecined to video tape, then edited _on tape_ to cut in some 60i live action video and 30p cg segments; then when they did the dvd encode, the encoder detelecined some of the animated content and encoded it as 24p with soft telecine flags.

17:35 <kepstin> iirc a few of the cuts have mismatched field dominance, so there's at least one place where you only get one of the two fields of an animation frame.

17:35 <fflogger> [editedticket] dts350z: Ticket #9408 ([avdevice] WASAPI Audio Input/Output Support) updated https://trac.ffmpeg.org/ticket/9408#comment:3

17:39 <kepstin> anyways, interlaced 4:2:0 subsampled video was a mistake :(

17:46 <fflogger> [editedticket] bradisbell: Ticket #9408 ([avdevice] WASAPI Audio Input/Output Support) updated https://trac.ffmpeg.org/ticket/9408#comment:4

17:46 <haasn> Is there a way to explicitly refer to output streams in -filter_complex ? sort of an analog to the way you can do [0:0] [0:1] etc to refer to input streams

17:46 <haasn> I'm trying to decide on a good way to print complex filter graphs in such a way that they can be passed to -filter_complex again

17:46 <haasn> the best I can come up with is to just leave the outputs unconnected and hope the user does the right mapping

17:47 <haasn> or alternatively, connecting them to [out0] [out1] etc and forcing the user to specify the desired -map

17:47 <haasn> for input streams we can use [N:M] to get the right stream automatically

18:19 <kepstin> with the input streams, you're still relying on the user specifying the same input streams in the same order for the [N:M] syntax to work

18:20 <kepstin> ideally you should be printing out output pads as specified in the original filter complex graph, so it'll work if the same set of (or lack of) map options is used.

18:20 <Lynne> I like mpv's terms, vid1, aid1, etc.

18:23 <fflogger> [newticket] adobez: Ticket #11474 ([undetermined] FFmpeg encounters an error when streaming FLV to RTMP after approximately 1 minute and 40 seconds. The error message suggests an issue with submitting a packet to the muxer.) created https://trac.ffmpeg.org/ticket/11474

18:35 <BtbN> mentions error message, does not even include it

18:39 <jamrial> haasn: didn't elenril make it possible to pass a filtergraph's output to another filtergraph?

18:40 <haasn> possibly, although that wouldn't conflict with what I'm working on atm

18:41 <jamrial> haasn: and yes, you give them a name. like -filter_complex "[0:0][0:1]amerge[out]" -map [out]

18:41 <haasn> I guess that the best thing to do is to reuse the original input/output pad names directly

18:59 <fflogger> [editedticket] MasterQuestionable: Ticket #11474 ([undetermined] Failed FLV RTMP streaming to remote source) updated https://trac.ffmpeg.org/ticket/11474#comment:1

19:25 <fflogger> [editedticket] MasterQuestionable: Ticket #11472 ([avcodec] ProRes horizontal bar artifacts from near 0 value pixels) updated https://trac.ffmpeg.org/ticket/11472#comment:1