Ffmpeg-devel-irc

Download

ffmpeg-devel-irc@ffmpeg.org

September 2019

2 participants
80 discussions

ffmpeg-devel.log.20190924
by burek 25 Sep '19

25 Sep '19

[00:23:17 CEST] <rcombs> BtbN: GPU arrived, thanks [00:23:46 CEST] <BtbN> oh nice, that went faster than I expected. [00:26:20 CEST] <rcombs> think this is just barely gonna fit in my case [00:26:33 CEST] <rcombs> eyeballing it, looks like it'll have maybe a mm or two of clearance [01:54:33 CEST] <Lynne> no way [01:54:50 CEST] <Lynne> people are using the gain field in the opus extradata for replaygain [02:00:19 CEST] <Lynne> I'm okay with it, its actually smart to let decoders do scaling like that since its free [02:01:00 CEST] <Lynne> just didn't expect it to be used at all [02:37:28 CEST] <kepstin> huh, I haven't actually seen any tools write that field, do you have examples in the wild? [02:39:39 CEST] <Lynne> https://github.com/desbma/r128gain/blob/master/r128gain/opusgain.py [02:43:40 CEST] <Lynne> I think they should have specified a different base, pow(10, 0xffff/(20.0*256)) = 6306736518360.512 [02:44:25 CEST] <Lynne> with base 2, pow(2, 0xffff/(20.0*256)) = 7130.584808402239, more reasonable amplification [02:47:13 CEST] <Lynne> and maybe made anything above 0x7fff inverse to reduce the volume [10:27:32 CEST] <JEEB> ok, so in a case where a stream gets removed from a program in MPEG-TS, should: [10:27:40 CEST] <JEEB> 1. framework send End-of-Stream [10:28:16 CEST] <JEEB> 2. API client like ffmpeg.c catch the PMT update (program updates are communicated to the API client), and send an End-of-Stream of its own to whatever it is doing [10:31:31 CEST] <JEEB> I'm just seeing cases of a client like ffmpeg.c buffering its input because it thinks it will still be getting audio from a stream that has since disappeared from the program [10:31:46 CEST] <JEEB> so I'm wondering if this should be fixed on either the framework, or the client side [11:19:59 CEST] <JEEB> also this reminds me [11:20:05 CEST] <JEEB> what was the lavf way of saying end-of-stream? [11:21:02 CEST] <JEEB> avformat.h doesn't seem to document it [11:29:08 CEST] <cone-244> ffmpeg 03Paul B Mahol 07master:a214c17414bd: avfilter/vf_v360: do not use mod where it is not needed [11:47:01 CEST] <durandal_1707> Lynne: can i pass runtime parameters to vulkan shaders, for example to change view coordinates without need to renit whole filter? [12:04:13 CEST] <cone-244> ffmpeg 03Timo Rothenpieler 07master:89cbbe9f70fc: avcodec/nvenc: fix typo in new Windows driver version [12:05:40 CEST] <cone-244> ffmpeg 03Timo Rothenpieler 07release/4.2:0eb1088960a2: avcodec/nvenc: add driver version info for SDK 9.1 [12:08:53 CEST] <cone-244> ffmpeg 03Timo Rothenpieler 07release/4.1:fe1064f77954: avcodec/nvenc: add driver version info for latest SDKs [12:11:07 CEST] <cone-244> ffmpeg 03Timo Rothenpieler 07release/4.0:25d1d5929fdb: avcodec/nvenc: add driver version info for latest SDKs [12:26:16 CEST] <Lynne> durandal_1707: how big is the data do you want to transfer and do you need to download it later? [12:27:02 CEST] <Lynne> if its small and only needs to be read by the GPU, use pushconstants, like chromaber [12:27:29 CEST] <Lynne> otherwise look at how overlay creates a buffer on init [12:30:59 CEST] <durandal_1707> Lynne: it is about changing roll/yaw/pitch and maybe fov dynamically [12:31:42 CEST] <Lynne> yeah, pushconstants are perfect for this [12:57:51 CEST] <durandal_1707> Lynne: what number means in GLSLC(1, ... ? [13:01:20 CEST] <Lynne> indentation [13:52:18 CEST] <Lynne> can bitstream filters modify ctx->par_in->extradata in the filter or init functions or do they have to create a new AV_PKT_DATA_NEW_EXTRADATA side data? [13:53:00 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:74bbf9bc8279: avcodec/adpcm: Check number of channels for MTAF [13:53:01 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:c7c0229bebbb: avcodec/truespeech: Eliminate some left shifts [13:53:02 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:8c7d5fcfc32d: avcodec/dxv: Check op_offset in both directions [14:05:06 CEST] <mkver> Lynne: Why would you want to do that? Isn't modifying ctx->par_out->extradata enough? [14:09:22 CEST] <Lynne> is ctx->par_out->extradata initialized to be equal to ctx->par_in->extradata? if so that would work [14:10:30 CEST] <mkver> Yes, it is. [15:46:29 CEST] <Lynne> so a few new things I've learned about opus: the gain is in fact signed, not unsigned, which is good, so base 10 makes sense [15:46:49 CEST] <Lynne> and our downmix handling has always been broken [15:47:07 CEST] <Lynne> but so is libopus' downmix handling so its fine! [15:48:53 CEST] <VenomFK> Hi, I am trying to build ffmpeg on AWS EC2 instance, I am facing following error [15:48:54 CEST] <VenomFK> undefined reference to `ff_aac_at_encoder' [15:49:35 CEST] <VenomFK> ./configure --extra-cflags="-I$HOME/ffmpeg_build/include" --extra-ldflags="-L$HOME/ffmpeg_build/lib" --extra-libs=-ldl --enable-version3 --enable-libx264 --enable-libmp3lame --enable-libtheora --enable-nonfree --enable-gpl --enable-postproc --enable-libfdk-aac --disable-optimizations --disable-x86asm [15:50:35 CEST] <JEEB> usage discussion to the users' channel [15:50:45 CEST] <JEEB> -> #ffmpeg [15:50:52 CEST] <VenomFK> thanks [16:00:35 CEST] <durandal_1707> VenomFK: why you disable x86 asm? [16:07:51 CEST] <VenomFK> durandal_1707: its not that important for me now [16:11:18 CEST] <Lynne> durandal_1707: postproc is explicitly enabled and you ask that? [16:13:48 CEST] <BBB> sounds like it's for the android sdk [16:14:07 CEST] <BBB> although no --enable-pic [16:14:10 CEST] <BBB> hm... [16:15:34 CEST] <jamrial> aac_at is audiotoolbox, macos stuff. sounds like a dirty build folder [16:38:50 CEST] <durandal_1707> Lynne: but chromaber_vulkan pushes constants during initialization and not during execution [16:39:56 CEST] <Lynne> durandal_1707: it calls ff_vk_update_push_exec on every frame to update it [16:40:23 CEST] <durandal_1707> ah [17:19:22 CEST] <durandal_1707> Lynne: shouldn't your vulkan filters crash with gray pix fmt inputs? [17:21:34 CEST] <Lynne> no, it still works [17:22:27 CEST] <Lynne> you can always sample .rgba, you'll just get 0 in the .gba channels [17:25:52 CEST] <cone-244> ffmpeg 03Andreas Rheinhardt 07master:f83ac5fd793f: avcodec/cbs_h264: Automatically free SEI payload on error [17:27:34 CEST] <Lynne> durandal_1707: btw if you find a good glsl language reference tell me [17:27:51 CEST] <Lynne> because I wasn't able to find one so all I know is from other code I've seen [17:32:39 CEST] <Lynne> btw since you're doing 360 video to "normal", here's something you should know [17:33:08 CEST] <Lynne> you need to work backwards, that is, determine which pixel of the input image maps to the output image [17:34:48 CEST] <Lynne> in each main() invocation, gl_GlobalInvocationID.xy tell you which output pixel you'll need to fill [17:35:38 CEST] <durandal_1707> Lynne: v360 already works like that [17:38:39 CEST] <Lynne> cool [17:55:38 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:5fe6a9db1539: tools/target_dec_fuzzer: Adjust threshold for MSS2 [17:55:39 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:d217691eec56: libavcodec/mpeg12dec: Check input for minimal frame size [17:55:39 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:59163731e992: tools/target_dec_fuzzer: consider potential padding/edge in pixel threshold [17:55:41 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:cede385018f5: avcodec/aacdec_fixed: Add FF_CODEC_CAP_INIT_CLEANUP [17:55:42 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:8e51f35f81c2: avformat/vividas: Check n_sb_blocks against input space [17:55:43 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:27a2f6594810: avformat/vividas: Test size and packet numbers a bit more [17:55:43 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:c7ccbf40edb8: avcodec/ffwavesynth: Fix integer overflow in timestamps [17:55:45 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:72db18e929cf: avformat/utils: Do not assume duration is non negative in compute_pkt_fields() [17:55:46 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:0831cbfe0991: avcodec/alac: fix undefined behavior with INT_MIN in lpc_prediction() [17:55:47 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:b30c07cc2b9e: avcodec/alac: Fix invalid shifts in 20/24 bps [17:55:48 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:033d2c4884ec: avcodec/smacker: Fix integer overflow in signed int multiply in SMK_BLK_FILL [17:55:49 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:340ab13504dd: avcodec/utils: Use av_memcpy_backptr() in ff_color_frame() [17:55:50 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:1e984a69155c: avcodec/h264_slice: clear frame only on gaps when it is not otherwise initilaized [17:55:51 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:3dce4d03d5a5: avcodec/aacdec: Check if we run out of input in read_stream_mux_config() [17:55:52 CEST] <cone-244> ffmpeg 03Michael Niedermayer 07master:95e5396919b1: avcodec/utils: Optimize ff_color_frame() using memcpy() [17:56:51 CEST] <durandal_1707> Lynne: i can not use function pointers in glslang? [18:10:27 CEST] <Lynne> durandal_1707: I don't think so, AFAIK pointers are somewhat restricted, but you shouln't need to [18:11:08 CEST] <Lynne> after all you can rewrite the shader on init, so you can directly substitute the function you need [18:38:06 CEST] <durandal_1707> Lynne: i get segv when shader failed to compile [18:48:43 CEST] <durandal_1707> Lynne: i can not get size of input image [18:54:27 CEST] <Lynne> can you link the filter somewhere? [18:55:02 CEST] <Lynne> imageSize only works for non-sampled images, so the output planes [19:04:43 CEST] <durandal_1707> Lynne: how to use M_PI in shader? [19:11:58 CEST] <durandal_1707> i got basic working now sort of [19:21:29 CEST] <Lynne> durandal_1707: GLSLF(0, #define M_PI (%f) ,M_PI); [19:21:36 CEST] <durandal_1707> and i got same performance as v360 with x86 asm and multithreading [19:22:00 CEST] <durandal_1707> expect this one use 100% cpu instead of almost 400% cpu [19:22:25 CEST] <Lynne> that's good for a starter, I'm sure performance can be improved with maybe some caching [19:25:15 CEST] <Lynne> are you using an intel gpu? how much cpu is used if you hwmap from vaapi instead of hwupload? [19:26:27 CEST] <durandal_1707> Lynne: this is on Intel(R) HD Graphics 5500 (Broadwell GT2) [19:27:56 CEST] <Lynne> you should be able to put -vaapi_device /dev/dri/renderD128 -hwaccel vaapi -hwaccel_output_format vaapi before the input file and change the hwupload to hwmap [19:32:24 CEST] <durandal_1707> https://github.com/richardpl/FFmpeg/tree/vulkan [19:36:56 CEST] <Lynne> that's it? I thought it would be more complex [19:37:29 CEST] <durandal_1707> Lynne: that is just equirectangular to flat, real v360 have more conversions [19:37:45 CEST] <durandal_1707> this is also without rotation of vector v [19:37:55 CEST] <durandal_1707> no yaw/pitch/roll [19:39:32 CEST] <durandal_1707> also i think this one does not handle wrap around for bilinear [19:40:06 CEST] <Lynne> accessing outside of input image's bounds? [19:40:50 CEST] <durandal_1707> Lynne: no, wrapping x in width when interpolating back to 0 [19:41:26 CEST] <durandal_1707> once i implement yaw/pitch/roll will inspect back view [19:44:58 CEST] <durandal_1707> Lynne: what GPU you have and what speed gain you get? [19:49:26 CEST] <durandal_1707> shit, 820M nvidia card is not for gaming, so no vulkan with it even if it worked by miracle [20:22:23 CEST] <Lynne> no idea, haven't tested yet [20:22:53 CEST] <Lynne> but do keep in mind video players will be able to skip downloading and uploading, so its really only a fair comparison if you hwmp and don't hwdownload [20:23:30 CEST] <Lynne> well, maybe not, if hardware decoders can't decode large resolution 360 video [20:28:26 CEST] <durandal_1707> Lynne: please try it [20:28:35 CEST] <durandal_1707> and report speed [20:51:18 CEST] <Lynne> k, will do in a few hours, can you link a sample? [21:26:01 CEST] <durandal_1707> Lynne: sample is irrelevant, just use same resolution when comparing with v360=e:flat:1 [21:59:13 CEST] <durandal_1707> heh, if i use rgba pixel format than v360_vulkan is almost 2x faster than v360 [22:34:07 CEST] <durandal_1707> Lynne: i guess i can cache results into separate image [22:36:26 CEST] <durandal_1707> Lynne: but how to create read+write cache image ? [22:36:54 CEST] <Lynne> durandal_1707: avgblur_vulkan does that [22:37:14 CEST] <durandal_1707> no, avgblur does not do it [22:37:56 CEST] <durandal_1707> i need to cache all vec2 values for all planes [22:38:18 CEST] <Lynne> it creates a temporary image into which the horizontal blur pass is stored, then runs the vertical blur pass on it [22:38:40 CEST] <Lynne> tmp = ff_get_video_buffer(outlink, outlink->w, outlink->h); [22:43:39 CEST] <durandal_1707> Lynne: and it holds values in floats? [22:47:18 CEST] <durandal_1707> and seems planar with alpha formats are not supported at all? [22:52:26 CEST] <Lynne> durandal_1707: it holds values in whatever the pixel format is [22:53:00 CEST] <Lynne> I also wanted to create a different pixel format image for another filter but this can't be done through lavfi, you need to do it through the avhwframescontext [22:53:01 CEST] <durandal_1707> Lynne: but i need float values, because that allows interpolation [22:54:21 CEST] <durandal_1707> so there is no way to allocate 2 float arrays[W*H] [23:02:57 CEST] <Lynne> you can, just use a buffer, like overlay_vulkan does [23:03:25 CEST] <Lynne> wait, you need to interpolate, that won't work [23:04:02 CEST] <Lynne> I guess I'll write a function which creates a custom avhwframescontext for whatever sw_format you need [23:04:10 CEST] <Lynne> but I'll do it tomorrow [23:07:36 CEST] <durandal_1707> i need to keep those 2 arrays [23:07:52 CEST] <durandal_1707> and redo it only if parameters change [00:00:00 CEST] --- Wed Sep 25 2019

1 0

ffmpeg.log.20190924
by burek 25 Sep '19

25 Sep '19

[00:02:03 CEST] <`St0ner> nonworking file's ffprobe @ https://pastebin.com/EvTxYrnW [00:02:13 CEST] <`St0ner> working file's ffprobe @ https://pastebin.com/bLZdWBHk [00:04:29 CEST] <another> wild stab in the dark: NONWORKING is marked as dash [00:05:00 CEST] <`St0ner> yeah that likely has something to do with it. i set -f mp4 and -brand mp42 [00:05:09 CEST] <`St0ner> but it didnt seem to help [00:08:48 CEST] <c_14> it could be fragmented mp4 [00:08:51 CEST] <c_14> not sure how to check for that [00:08:58 CEST] <`St0ner> so i think i need to tell ffmpeg to convert to a compatible brand. the working files i have show compatible_brands: mp42mp41isomavc1 and major_brand:mp42, nonworking files show compatible_brands: iso6avc1mp41 [00:09:25 CEST] <livingbeef> Hey. How do I copy/write only id3v1 and remove other metadata? [00:09:27 CEST] <`St0ner> the nonworking mp4 doesn't seem to be fragmented, plays fine and isn't corrupt at all [00:10:14 CEST] <livingbeef> -map_metadata -1 -write_id3v1 1 does not seem to work [00:10:28 CEST] <c_14> `St0ner: fragmentation in this sense refers to the moov atom. dash requires "fragmented" moov atoms while mp4 by default has one big one at the end of the file [00:10:34 CEST] <livingbeef> that simply removes all metadata and does not write id3v1 [00:10:57 CEST] <kepstin> livingbeef: yeah, you told it to remove all metadata, so there's nothing to write to the id3v1 tag :) [00:11:04 CEST] <`St0ner> i've set -movflags +faststart to move the MOOV atoms to the front [00:11:37 CEST] <kepstin> livingbeef: i think you can set `-id3v2_version 0` to disable writing id3v2 tags [00:11:46 CEST] <kepstin> but i'd have to double-check that [00:11:51 CEST] <livingbeef> I'll try that [00:12:11 CEST] <kepstin> why do you want to do that anyway? id3v1 tags are awful, and v2 is pretty well supported :/ [00:12:47 CEST] <kepstin> (note that ffmpeg defaults to id3v2.4, but some stuff only works properly with id3v2.3 - use `-id3v2_version 3` to get that instead) [00:15:27 CEST] <livingbeef> Because someone at (google?) was too lazy or incompetent to handle tags properly, and randomly throws away some mp3 files instead. [00:15:35 CEST] <livingbeef> (android) [00:16:30 CEST] <livingbeef> Some metadata is probably triggering some error or something, it works if I remove all metadata and manually set title (can be v2) [00:17:14 CEST] <c_14> `St0ner: does setting -brand mp42 not change the listed brand after converting? [00:19:41 CEST] <livingbeef> So if I just omit -id3v2_version alltogether, it just sets id3v1 tags; but it somehow still writes the metadata to the mp3 file [00:20:17 CEST] <c_14> if you omit that option it'll write id3v2.4 tags [00:20:29 CEST] <livingbeef> id3v2 -l <file> reports v1 tags and says v2 tags are not there; but ffprobe lists all the metadata [00:21:21 CEST] <livingbeef> hmm, so it may be that id3v2 program can't ready v2.4? [00:21:32 CEST] <c_14> possible [00:22:13 CEST] <another> i've been having fun with kid3 for all kids of tagging [00:23:04 CEST] <`St0ner> setting -brand mp42 makes the Codec ID = mp42 (isom/iso2/avc1/mp41), which doesn't load into the iPod. the working files have Codec ID = mp42 (mp42/mp41/isom/avc1) and has a tag that doesnt exist in the nonworking file, "Overall bit rate mode", which = Variable [00:26:03 CEST] <c_14> maybe try looking at the files with mp4box or whatever that other program was called [00:27:35 CEST] <another> MP4Box [00:27:53 CEST] <c_14> yeah, but there's the other one as well. With the japanese website [00:27:57 CEST] <furq> l-smash [00:28:01 CEST] <c_14> yeah, that [00:28:05 CEST] <furq> specifically boxdumper [00:28:20 CEST] <furq> or you could just remux it with l-smash and it'll probably work [00:28:44 CEST] <`St0ner> ffmpeg can't do this? [00:28:46 CEST] <another> or you could remux it with the ipod muxer [00:29:28 CEST] <c_14> I guess try that first, yeah. -f ipod [00:29:30 CEST] <c_14> forgot that existed [00:30:18 CEST] <`St0ner> ok will try that [00:30:57 CEST] <kepstin> i'm surprised just remuxing it with the mp4 muxer with no other options didn't work, tbh. [00:33:18 CEST] <`St0ner> still doesn't work, my command line is -c:v libx264 -crf 23 -profile:v high -level 3.1 -preset medium -f ipod -brand mp42 -movflags +faststart [00:33:52 CEST] <another> have you tried just muxing to ipod w/o setting the brand? [00:33:58 CEST] <livingbeef> OK, some progress. I can output id3v2 and v1 tags, use id3v2 -d <file> to remove v2 and then the file gets recognized by the player [00:33:59 CEST] <`St0ner> no, will try that now [00:36:47 CEST] <`St0ner> no go [00:41:18 CEST] <livingbeef> Additionally I can do id3v2 -C <file> to convert v1 to v2 [00:43:40 CEST] <another> `St0ner: what ipod is it? [00:43:58 CEST] <`St0ner> iPod Touch 3G 64GB [00:44:35 CEST] <`St0ner> there was a website a few years ago that listed all the x264 details for each generation, but i don't think it's online anymore [00:45:05 CEST] <`St0ner> the website was basically a formatted ascii text file, it's probably on wayback machine if anyone remembers the url.. [02:30:11 CEST] <`St0ner> figured out a combination that works for my iPod 3G. -acodec aac -b:a 320K -c:v libx264 -crf 23 -profile:v baseline -level 3.0 -s 720x540 -preset medium -f ipod -movflags +faststart [02:30:52 CEST] <`St0ner> took about 8 hours across 2 nights of testing many different combinations [02:33:28 CEST] <`St0ner> separate question, when using -acodec ALAC, does -b:a actually do anything? [02:36:50 CEST] <`St0ner> and which AAC encoder yields highest quality? the ffmpeg documentation has conflicting information [02:38:40 CEST] <klaxa> i don't think you can hear much of a difference between the native aac encoder and libfdk [02:38:53 CEST] <`St0ner> is there a windows build with libfdk? [02:38:57 CEST] <kepstin> `St0ner: with most lossless codecs, setting the bitrate is ignored [02:39:56 CEST] <kepstin> `St0ner: ffmpeg built with fdk-aac is a non-redistributable combination of licenses, so you won't find prebuilt binaries. [02:40:34 CEST] <`St0ner> does anyone happen to have a windows x64 binary of ffmpeg with libfdk-aac? [02:42:25 CEST] <kepstin> in other words - if someone does, they can't legally share it with you. [03:20:18 CEST] <`St0ner> thanks everyone for your help, take care! [06:01:18 CEST] <MoziM> i'm using the following command to join n video clips together ffmpeg -f concat -safe 0 -i /tmp/videolist.txt -c copy output.mp4 and i'm getting the following message for each video "pts has no value" how can i log the output to a file so i can examine it later? [06:12:44 CEST] <KillerWasp> hello... [06:38:41 CEST] <MoziM> KillerWasp: hi [14:03:09 CEST] <Soni> how do I search the manual? [14:07:08 CEST] <DHE> when in doubt it's hard to go wrong searching https://ffmpeg.org/ffmpeg-all.html [14:07:28 CEST] <DHE> unless you're doing API programming in which case doxygen [14:13:30 CEST] <Soni> hm, no still a pain to use [14:13:39 CEST] <Soni> did find what I wanted on stackoverflow tho [15:55:35 CEST] <genpfault> welp, hit the button to send another trac account verification email this morning and it took less than a minute to hit gmail :) [19:15:49 CEST] <OnkelTem> Hi all [19:16:04 CEST] <OnkelTem> Guys, do you know, does Google Chrome support MPEG-2 audio streams? [19:16:31 CEST] <OnkelTem> We cannot get it working using HLS.js. Our stream works in FireFox but not in Chrome [19:16:46 CEST] <OnkelTem> And is it true, that if Chrome support MP3, it should also support MP2? [19:16:59 CEST] <ChocolateArmpits> mutually exclusive [19:17:09 CEST] <OnkelTem> oops [19:17:39 CEST] <ChocolateArmpits> nope, no mp2 https://www.chromium.org/audio-video [19:18:05 CEST] <OnkelTem> ChocolateArmpits: I see. Yeah I was on that page [19:18:14 CEST] <OnkelTem> Just thought that MP3 means MP2 also (hides) [19:18:58 CEST] Action: ChocolateArmpits searches for OnkelTem to berate him [19:19:11 CEST] <OnkelTem> =) [23:51:08 CEST] <safinaskar> what options huffyuv encoder supports? and where i supposed to find this? "man ffmpeg-all" and "man ffmpeg-codecs" don't say anything [23:58:24 CEST] <kepstin> safinaskar: most codecs that don't have a section in ffmpeg-codecs doc either only take common options or don't really have options most people need to tweak [23:58:49 CEST] <kepstin> note that you can see a list of all avoptions an encoder supports (which is *not* the same as all options) by running "ffmpeg -h encoder=huffyuv" [23:59:12 CEST] <safinaskar> kepstin: thanks [00:00:00 CEST] --- Wed Sep 25 2019

1 0

ffmpeg-devel.log.20190923
by burek 24 Sep '19

24 Sep '19

[02:46:37 CEST] <philipl> Lynne jkqxz: Progress. I'm using hwupload_cuda as a base as it's simpler, and I can get it to correctly initialise a context for the dst side and then call transfer_data_from. Got a pipeline to run producing empty output frames. Now just need to run the copy. [02:58:18 CEST] <philipl> and I have frames copied. w00t. [05:05:15 CEST] <cone-071> ffmpeg 03vectronic 07master:598962cd3a68: avformat/dashdec: fix pointer being freed was not allocated [05:05:15 CEST] <cone-071> ffmpeg 03vectronic 07master:8c90bb8ebb6e: avformat/dashdec: fix segfault when parsing segmentlist [05:14:39 CEST] <philipl> Lynne: jkqxz: BtbN: https://github.com/philipl/FFmpeg/commit/9be8e74ee9d4897614aa0d541bef05c6c0… [10:20:48 CEST] <JEEB> hmm [10:21:26 CEST] <JEEB> when a stream gets removed from PMT in MPEG-TS, should the end-of-stream come from a) demuxer b) the application should catch the "program has updated" metadata and make the EOS happen [11:07:19 CEST] <durandal_1707> Lynne: what should i read to get friendly with vulkan programming? [11:31:17 CEST] <BtbN> philipl, btw., sdk 8.0 is dead, no point in pushing patches there [11:31:44 CEST] <JEEB> the problem is, there are users that can't upgrade drivers and it seems like newer SDK requires newer drivers? [11:32:09 CEST] <JEEB> we hit that on mpv it seems, where a user had built nvdec with newer headers, and then they failed to load on his older drivers? [11:32:14 CEST] <BtbN> The oldest still supported legacy driver by nvidia supports SDK 8.1. [11:32:27 CEST] <BtbN> So there is zero point in still taking care of the 8.0 branch. [11:32:31 CEST] <JEEB> right [11:32:38 CEST] <JEEB> I forgot about 8.0/8.1 [11:33:25 CEST] <BtbN> And I'm about to bump the driver req to 435 with SDK 9.1. [12:06:28 CEST] <Lynne> durandal_1707: I don't know; its complex [12:08:36 CEST] <Lynne> as far as the lavfi framework goes: shaders (some text) has descritors (which tell what "arguments" the shader has, the type, etc), both of which get fed into the pipeline layout creation, which allocates the "arguments" of the shader, which finally gets compiled into a pipeline [12:09:18 CEST] <Lynne> a pipeline represents a single compute shader (and optionally vertex+fragment+compute shader which is done differently and there's no code for yet) [12:10:37 CEST] <Lynne> to run the pipeline, you need a command buffer which you record commands into, one of which should be to bind the pipeline layout, and then you submit that to a queue which manages behind the scenes queueing of the GPU and such [12:11:22 CEST] <Lynne> I skipped a lot but just look at something like the scale filter [12:30:20 CEST] <cone-346> ffmpeg 03Paul B Mahol 07master:eaf3c4b8bf29: avfilter/vf_v360: fix regression introduced in b342678bc47b4e5 [12:32:20 CEST] <durandal_1707> Lynne: i wish to work on that once it get pushed to ffmpeg master branch [14:24:50 CEST] <durandal_1707> i found out that my intel card is not supported by vulkan [14:26:49 CEST] <kierank> Get better one [14:28:30 CEST] <durandal_1707> kierank: which repo holds source code of vulkan? [14:29:04 CEST] <kierank> I have no idea, I don't even know what vulkan is [14:29:08 CEST] <kierank> Ask Lynne [15:02:39 CEST] <cone-346> ffmpeg 03James Almer 07master:d889ae33962e: avcodec/bsf: check that AVBSFInternal was allocated before dereferencing it [15:06:27 CEST] <Lynne> philipl: that's a much nicer solution that hwmap [15:12:59 CEST] <cone-346> ffmpeg 03James Almer 07release/3.2:7df1cb3dce1e: avcodec/bsf: check that AVBSFInternal was allocated before dereferencing it [15:13:02 CEST] <cone-346> ffmpeg 03James Almer 07release/3.4:45bf2f4d72f1: avcodec/bsf: check that AVBSFInternal was allocated before dereferencing it [15:13:04 CEST] <cone-346> ffmpeg 03James Almer 07release/4.0:693f3926ddd9: avcodec/bsf: check that AVBSFInternal was allocated before dereferencing it [15:13:07 CEST] <cone-346> ffmpeg 03James Almer 07release/4.1:11cce24e04e1: avcodec/bsf: check that AVBSFInternal was allocated before dereferencing it [15:13:09 CEST] <cone-346> ffmpeg 03James Almer 07release/4.2:25273ef23aef: avcodec/bsf: check that AVBSFInternal was allocated before dereferencing it [15:32:04 CEST] <BtbN> Nvidia is deprecating parts of cuvid in SDK 9.1, pointing to use FFmpeg instead. [15:32:19 CEST] <BtbN> SDK 9.1 is now imported into ffnvcodec [15:50:42 CEST] <philipl> BtbN: thanks. what did they deprecate? the weird parts we never used? presumably not the parser yet. [15:51:00 CEST] <BtbN> Yeah, the weird video source API [15:51:15 CEST] <BtbN> basically libavformat kinda features [16:01:59 CEST] <cone-346> ffmpeg 03Timo Rothenpieler 07master:648f5c93820e: avcodec/nvenc: add driver version info for SDK 9.1 [16:56:26 CEST] <Lynne> philipl: could the cuda->vulkan mapping also be done using the other transfer function? [17:11:51 CEST] <philipl> Lynne: yes. I'll experiment with that this week [17:14:37 CEST] <Lynne> tnx, would be nice to not have to do a pool init in the mapping function [17:17:43 CEST] <durandal_1707> the only way i can try vulkan is to enable 820M nvidia card, but i can not enable it at all [17:18:22 CEST] <Lynne> sacrifice a usb flash and install arch on it with the proprietary drivers [17:19:40 CEST] <durandal_1707> i already have drivers [17:19:49 CEST] <durandal_1707> but on ubuntu [17:50:57 CEST] <BtbN> ah shit, I typoed the windows version... [17:58:01 CEST] <jamrial> BtbN: in ffmpeg or nv-codec-headers? [17:58:37 CEST] <BtbN> ffmpeg [17:58:41 CEST] <BtbN> codec-headers is correct [19:22:53 CEST] <durandal_1707> Lynne: do you have command example how to use vulkan filters? [19:47:18 CEST] <durandal_1707> ffmpeg -i ~/Videos/red-leaf-tips.avi -init_hw_device vulkan=v:'Intel(R) HD Graphics 5500 (Broadwell GT2)' -filter_hw_device v -vf format=gbrp,hwupload,avgblur_vulkan=5:15:5,hwdownload -f null - [19:47:37 CEST] <durandal_1707> i get: Invalid output format monow for hwframe download. [19:50:49 CEST] <durandal_1707> i get zero pixels in output [19:54:33 CEST] <durandal_1707> the avgblur_vulkan filter does not work for me, scale_vulkan and chromaber_vulkan work as expected [19:55:09 CEST] <kierank> Very sad [19:57:43 CEST] <durandal_1707> it may also hang here [19:58:07 CEST] <durandal_1707> kierank: everything OK with you? [19:58:31 CEST] <kierank> Yes [20:04:27 CEST] <philipl> durandal_1707: you need a format=<something> after the hwdownload [20:05:10 CEST] <durandal_1707> yes i added it on both sided [20:05:46 CEST] <Lynne> durandal_1707: does the avgblur filter crash your gpu or what do you mean it doesn't work? [20:07:45 CEST] <Lynne> it should crash your gpu if you 1.) hwmap from vaapi/drm and 2.) the image size is not modulo 16 iirc [20:07:54 CEST] <durandal_1707> Lynne: it either gives green output or hangs gpu, so mesa reset it [20:08:27 CEST] <durandal_1707> and yes i use drm i think [20:08:58 CEST] <Lynne> yes, its a known bug, its in their issue tracker, they haven't done anything about it for a year, so I'm letting them take the shame [20:09:21 CEST] <durandal_1707> and hwupload/hwdownload overhead is sooo HUGE that it is not worth using single instance of filter (scale_vulkan in this case) [20:10:04 CEST] <Lynne> I know, right? that's why there was a lot of work to do on the interops [20:10:51 CEST] <Lynne> but nlmeans_vulkan should be faster than the software nlmeans so it would be worth it for such a filter, similar to nlmeans_opencl [20:13:33 CEST] <Lynne> also the 360 software filter should also be much slower than a hwupload/filter/hwdownload [20:14:36 CEST] <durandal_1707> Lynne: well, 360 use LUTs, i doubt vulkan one could do the same, (allocating bunch of textures) [20:15:02 CEST] <durandal_1707> actual slowdown comes from interpolation in CPU [20:16:23 CEST] <durandal_1707> if one does not do LUTs approach, but use sin/cos/tan.. functions in direct approach, how still one can gain? [20:16:49 CEST] <Lynne> bilinear interpolation is free on GPUs and anything above is hugely faster than the CPU [20:17:53 CEST] <durandal_1707> Lynne: how one do bilinear on GPUs? in 360 you give pixels with weights to use. [20:18:11 CEST] <Lynne> trig operations are also very fast, and you have a few thousand threads working on a block of the image [20:18:58 CEST] <Lynne> durandal_1707: texture(image, position) does bilinear sampling, where position is a float [20:20:39 CEST] <Lynne> so texture(image, position + vec2(0.5, 0.0)) will give you a bilinearly interpolated value of 2 horizontally neighbouring pixels [22:52:48 CEST] <cone-769> ffmpeg 03Moritz Barsnick 07master:e9e9f79a1908: avformat/hashenc: fix incorrect use of av_mallocz_array() [22:52:48 CEST] <cone-769> ffmpeg 03Limin Wang 07master:3104100a1270: avcodec/exr: cosmetics [00:00:00 CEST] --- Tue Sep 24 2019

1 0

ffmpeg.log.20190923
by burek 24 Sep '19

24 Sep '19

[02:12:58 CEST] <aaronfranke> For the Windows version of ffmpeg, instead of providing a ZIP archive, it would be nicer if there were installer executables that automatically install ffmpeg to Program Files and automatically add ffmpeg to the PATH variable. Has this been considered before? [02:14:10 CEST] <auser0> its weird that using -map_metadata -1 actually removes most of the tags, but mkvmerge says there are still some that hasn't been removed [02:35:43 CEST] <another> ffmpeg doesn't provide binaries [02:38:01 CEST] <aaronfranke> Then who runs https://ffmpeg.org/? [02:38:12 CEST] <aaronfranke> Then who runs https://ffmpeg.org/ ? [02:38:45 CEST] <Agafnd> windows binaries are at https://ffmpeg.zeranoe.com/builds/ [02:38:48 CEST] <Agafnd> so not ffmpeg.org [02:39:57 CEST] <aaronfranke> https://ffmpeg.zeranoe.com/builds/ is directly linked from https://ffmpeg.org/ [02:40:22 CEST] <aaronfranke> and https://ffmpeg.zeranoe.com/builds/ links back to this IRC and ffmpeg.org [02:46:31 CEST] <nicolas17> yes, be grateful for that convenience link :P [02:47:18 CEST] <Agafnd> I assume the people who run ffmpeg.org just link to reliable looking builds [02:47:25 CEST] <Agafnd> without actually being responsible for them [02:48:13 CEST] <aaronfranke> It's very frustrating when the site that distributes binaries links to an IRC which tells you it's not their problem... I just want to discuss this with whoever's problem it is [02:48:55 CEST] <Agafnd> https://github.com/Zeranoe has a link to that person's email, I suppose [02:49:40 CEST] <Agafnd> oh there appears to be forums on that website [02:49:41 CEST] <Agafnd> https://ffmpeg.zeranoe.com/forum/viewforum.php?f=42 ? [03:06:07 CEST] <another> aaronfranke: feel free to build an installer if you want [08:08:00 CEST] <JEEB> lain98: something something descriptor based om the pix_fmt i think [08:09:03 CEST] <JEEB> av_pix_fmt_desc_get apparently [08:09:21 CEST] <JEEB> then that has a bit depth per component [08:46:28 CEST] <lain98> thanks JEEB [11:36:42 CEST] <lain98> is it a good idea to have vp9 in mp4 container. what could go wrong ? [11:37:19 CEST] <JEEB> some players probably only added support in webm since that's what youtube did? but if everything you care about supports it in mp4, please do :) [12:04:51 CEST] <snooky> moin [13:46:19 CEST] <dongs> JEEB: beap. how do I add pts/dts to raw HEVC? [13:46:27 CEST] <dongs> do I mux it into PES or what? [13:47:38 CEST] <JEEB> in theory there's timestamp SEI messages but if you actually need timestamps you mux it into something with timestamps [13:48:09 CEST] <dongs> is this "SEI" shit something that ex. mkvmerge or etc can make use of? [13:48:12 CEST] <JEEB> no [13:48:21 CEST] <JEEB> almost nothing supports that stuff :P [13:48:28 CEST] <dongs> oh [13:48:32 CEST] <dongs> then waht should it be output as? [13:48:39 CEST] <dongs> PES seems old (ancient mpeg2 shit) [13:48:54 CEST] <JEEB> anything the following stuff can read I guess (Ž4@) [13:49:04 CEST] <JEEB> matroska, mp4, mpeg-ts... [13:49:09 CEST] <dongs> hmm.. [13:50:48 CEST] <dongs> that seems like a lot of work [13:51:32 CEST] <JEEB> if you actually want timestamps that is expected, though. some apps do read raw annex b if it has the maximum time base value set (which many read as "frame rate") [13:51:48 CEST] <JEEB> but that's just so you can then stick that into some actual container :P [13:52:07 CEST] <dongs> right now the raw hevc doens't work with anything other than mkvmerge [13:52:15 CEST] <dongs> ffmpeg does the "fuck off, no timestamp" thing [13:52:57 CEST] <JEEB> I think there was a patch on the ML regarding forcing a frame rate, wonder if that has it work any better [13:53:16 CEST] <dongs> well we have real timestamps from tlv [13:53:19 CEST] <dongs> so i'd rather use them [13:53:57 CEST] <JEEB> mmtp mp4-like stuff I guess? since it's what is carried over the TLV [13:54:12 CEST] <dongs> its just absolutely raw frames [13:54:18 CEST] <dongs> raw hevc, raw latm aac [13:54:26 CEST] <dongs> timestamps are via descriptors in sideband [13:54:38 CEST] <JEEB> vOv [13:54:39 CEST] <dongs> outer unit is MPU packet [13:54:45 CEST] <JEEB> yea [13:54:55 CEST] <JEEB> man, that stuff is such job keeping complexity :P [13:54:59 CEST] <JEEB> some people must *love* it [13:55:51 CEST] <dongs> [mpegts @ 00000296e53c6040] Timestamps are unset in a packet for stream 0. This is deprecated and will stop working in the future. Fix your code to set the timestamps properly [13:55:54 CEST] <dongs> [mpegts @ 00000296e53c6040] first pts value must be set [13:55:57 CEST] <dongs> av_interleaved_write_frame(): Invalid data found when processing input [13:56:07 CEST] <dongs> thats when I do -i raw.mpu -vcodec copy out.ts (or mkv. or etc) [13:56:24 CEST] <JEEB> yes, the raw reader doesn't set pts/dts. wonder if that frame rate forcing thing went in [13:56:29 CEST] <dongs> [hevc @ 00000296e47f9c80] Stream #0: not enough frames to estimate rate; consider increasing probesize [13:56:40 CEST] <dongs> where is it searching here? [13:56:40 CEST] <JEEB> that's bit rate probably [13:56:45 CEST] <dongs> ah bitrate [13:57:48 CEST] <dongs> -probesize 50M worked [13:57:54 CEST] <dongs> well, that aerror is gone anyway [13:57:57 CEST] <dongs> still no timestamps [13:58:02 CEST] <dongs> so where is it expecting to read those? the annexb thing? [13:59:14 CEST] <JEEB> yea [13:59:24 CEST] <JEEB> the raw hevc demuxer doesn't set them [13:59:42 CEST] <dongs> well i dont demux that. raw hevc is my input/source [13:59:48 CEST] <dongs> as it comes out of tlv/mpu. [14:00:10 CEST] <JEEB> when you are calling ffmpeg.c like that the thing that reads the I/O input is the HEVC demuxer :P [14:00:17 CEST] <JEEB> which doesn't set dts/pts [14:00:28 CEST] <JEEB> anyways, there was a patch which doesn't seem to have gotten merged that added capability of overriding that to a frame rate [14:00:29 CEST] <dongs> ah. [14:00:56 CEST] <dongs> technically if framerate is constant, we just need a single timestamp at beginning? [14:01:08 CEST] <dongs> tream #0:0: Video: hevc (Main 10), yuv420p10le(tv, bt2020nc/bt2020/bt2020-10), 7680x4320 [SAR 1:1 DAR 16:9], 59 fps, 59.94 tbr, 1200k tbn, 59.94 tbc [14:01:13 CEST] <dongs> it knows the rest of the stream params [14:01:27 CEST] <JEEB> 'grats on decrypting the 8K stuff :P [14:01:44 CEST] <JEEB> yea the decoder context has that I think [14:01:56 CEST] <JEEB> and that's where that overall info thing is showing it from [14:02:35 CEST] <JEEB> ok, I'm at $dayjob and can't find that one patch quickly from patchwork :P [14:02:48 CEST] <JEEB> so sorry, otherwise I would have linked you the patch that lets you set -r before input and define frame rate [14:03:16 CEST] <dongs> hmm but it already knows rate from hevc, why would I need to set it again? [14:03:32 CEST] <JEEB> because that's the codec decoder context [14:03:35 CEST] <JEEB> not the actual packets [14:03:51 CEST] <JEEB> the HEVC raw "demuxer" is a barebones thing that just calls the parser, basically [14:04:06 CEST] <JEEB> see libavformat/hevcdec.c [14:04:28 CEST] <JEEB> has a custom probe function and otherwise it's defined as a minimal raw format "demuxer" [14:05:16 CEST] <JEEB> if that, or the ffmpeg.c application, would set the dts/pts then it'd work [14:05:24 CEST] <dongs> https://drive.google.com/open?id=1zeg6vsdga0e990R11dqshYqMy190FSip this is what 8K bandwidth is getting wasted on [14:10:02 CEST] <JEEB> there we go https://patchwork.ffmpeg.org/patch/12552/ [14:10:04 CEST] <JEEB> finally found it :P [14:10:14 CEST] <dongs> why wasnt this merged ;d [14:10:29 CEST] <dongs> also lol a fucking simplep atch [14:10:33 CEST] <dongs> fucking commies [14:10:41 CEST] <JEEB> I'll probably poke at if anyone has any objections and will merge if nobody objects within X days :P [14:10:57 CEST] <JEEB> I'm also to blame since I'm trying to handle $dayjob etc [14:11:38 CEST] <JEEB> on $dayjob I think implement writing XML in MP4 because of :reasons: and :psyduck: [14:11:48 CEST] <JEEB> (who the hell thought that XML subtitles was a great idea) [14:11:59 CEST] <dongs> heh, i havent looked waht subs in tlv are [14:12:07 CEST] <dongs> they're something very similar to xml in mp4, prolly [14:12:10 CEST] <dongs> i know its utf8 at least [14:12:12 CEST] <JEEB> ARIB did have its own fork of TTML [14:12:15 CEST] <dongs> isntead of retardo2022 [14:12:16 CEST] <JEEB> arib-ttml or something [14:13:35 CEST] <JEEB> but yea I should finally get a patch done for efffing TTML in MP4 at some point :P mostly working around vendors' crap at the moment [14:13:52 CEST] <JEEB> (writing that is, I don't want to parse XML) [14:14:11 CEST] <Radiator> Hi all, I have an error rising from ffmpeg while encoding a video : buffer underflow st=0 bufi=XXX size=YYY [14:14:22 CEST] <Radiator> This error only appear when using the mpeg codec [14:14:28 CEST] <dongs> it means your encoding params are wrong [14:14:44 CEST] <JEEB> Radiator: you mean MPEG-PS muxing [14:14:44 CEST] <Radiator> I'm using the library in C++ btw [14:14:46 CEST] <JEEB> not codec [14:14:58 CEST] <JEEB> that message comes from libavformat/mpegenc.c [14:15:03 CEST] <JEEB> which is MPEG-PS [14:15:05 CEST] <JEEB> :) [14:15:19 CEST] <Radiator> JEEB: Surely, I'm not very familiar with the terms nor ffmpeg [14:15:34 CEST] <JEEB> video/audio/subtitle layer vs container layer [14:15:39 CEST] <JEEB> codecs are usually on the former [14:16:17 CEST] <JEEB> Radiator: basically you have set some parameters (if any? maybe you haven't set any?) for the MPEG-PS writer, and it's complaining that your encoded video does not follow those parameters [14:16:23 CEST] <JEEB> are you sure you want to be writing MPEG-PS? [14:17:40 CEST] <Radiator> Well, I'm trying to do a zero latency (or as close as possible) encoding so I don't really know if mpeg-ps would be suitable [14:17:43 CEST] <dongs> isnt this about VBV and shit [14:18:09 CEST] <JEEB> yes [14:18:15 CEST] <JEEB> buffer size and actual rate of content [14:18:18 CEST] <dongs> right [14:18:18 CEST] <JEEB> he's muxing stuff too fast [14:18:30 CEST] <Radiator> Oh ok [14:18:41 CEST] <JEEB> Radiator: not time wise. your bit rate of content is too fast :P [14:18:48 CEST] <JEEB> compared to the parameters of the muxer [14:18:50 CEST] <JEEB> if you have set any [14:18:54 CEST] <JEEB> it could as well be some defaults [14:19:06 CEST] <JEEB> but as I said, you should be pretty sure if you want MPEG-PS [14:19:09 CEST] <Radiator> I just used "preset" "veryfast" [14:19:10 CEST] <JEEB> that is not really a much used container [14:19:38 CEST] <JEEB> DVDs used to use it, and HD-DVD as well I think? [14:20:14 CEST] <Radiator> Ok, I'll try it [14:20:40 CEST] <JEEB> uhh, I have not recommended anything to you yet :P [14:20:59 CEST] <JEEB> I'm just telling you that you've not configured the container correctly and you should double check if you realyl really want MPEG-PS [14:22:37 CEST] <Radiator> Well tbh I followed the muxing example which used mpeg [14:23:15 CEST] <Radiator> Since I'm streaming through the network I don't have a set container [14:30:57 CEST] <JEEB> basically it all depends on what the receiving end takes in then :P [14:32:57 CEST] <Radiator> It can adapt so I really don't know what to do [14:56:26 CEST] <Radiator> I tried using libx264 and the frame is totaly messed up. It's all gray and pixelised. I'm converting the frame in YUV420P, would it be the problem ? [14:58:14 CEST] <Radiator> It also says the following warning : non-strictly-monotonic PTS. I read online that it is about the order of my frame that I have to order before decoding but I really have no idea [14:58:39 CEST] <dongs> yuv420 blaze it [14:59:42 CEST] <Radiator> Which format should I use then ? [15:00:51 CEST] <Radiator> Or is there a man explaining which container is suitable with each pix format ? [15:02:28 CEST] <DHE> yuv420 goes into x264, and then the h264 output goes into a container. pixel format and container are not related unless you're doing raw/uncompressed storage [15:05:21 CEST] <dongs> JEEB holy shit japan is retarded. MPT version updates on any change except if that change is MPU timestamp descriptor. then version doesnt incr. [15:05:36 CEST] <dongs> why the hell would they do this [15:05:48 CEST] <dongs> that means re-parsing entire MPT even if version is same [15:10:38 CEST] <Radiator> DHE: I'm decoding a stream in rtsp to remux and encode it in udp. I'm trying to tune it to find an efficient format, container and encoder to do the job. [15:10:57 CEST] <Radiator> RIght now the decoding is done very easily [15:11:06 CEST] <JEEB> dongs: ouch [15:30:09 CEST] <taliho> JEEB: do you have a link to the frame rate forcing patch? [15:30:22 CEST] <taliho> JEEB: I've been thinking of working on this problem for some time. There are a few tickets because raw demuxer do not output timestamps [15:30:48 CEST] <JEEB> taliho: see the patchwork link I posted [15:30:52 CEST] <JEEB> it has the patch there [15:31:05 CEST] <JEEB> https://patchwork.ffmpeg.org/patch/12552/ [15:32:04 CEST] <taliho> thanks [16:49:43 CEST] <SpeakerToMeat> Hello all. [16:50:13 CEST] <SpeakerToMeat> Question, does an (automatically started) xyz2rgb conversion include gamma conversion by default? [17:46:26 CEST] <kepstin> SpeakerToMeat: should give you something reasonableish with similarities to the sRGB gamma curve. [17:46:57 CEST] <kepstin> SpeakerToMeat: (you may wish to look into the zscale filter for more control if it supports the conversion you're trying) [18:06:22 CEST] <jpb> greetings - has anyone had problems with avformat_open_input not correctly opening an rtsp stream? [18:06:46 CEST] <jpb> i've narrowed down the problem to libavformat not correctly negotiating the rtsp opening authentication [18:16:18 CEST] <MrSassyPants> So ubuntu 19.04s version of ffmpeg doesn't have nvenc in it (anymore), a bug about this was reported in may, probably an oversight [18:17:12 CEST] <pink_mist> ffmpeg doesn't really control what distributors put in their packages ... here it's always recommended that you build it yourself from git [18:17:13 CEST] <MrSassyPants> so does someone know of a ppa that lets me get an nvenc-enabled version without trashing dpkg by having local non-package-managed versions of things [18:17:29 CEST] <MrSassyPants> I can build it from git, I did [18:17:44 CEST] <MrSassyPants> but such installations interfere with the package manager [18:17:46 CEST] <pink_mist> just package it up yourself if you want it packaged [18:17:54 CEST] <MrSassyPants> I don't know how to package things up :3 [18:18:07 CEST] <MrSassyPants> is it easy? [18:18:11 CEST] <pink_mist> I'm sure debian has resources available to learn [18:18:18 CEST] <pink_mist> maybe even ubuntu does [18:18:27 CEST] <MrSassyPants> Ah I was thinking you have like 1 command that you know off hand [18:18:31 CEST] <pink_mist> I don't use either of those distros though, so I have no clue [18:18:49 CEST] <MrSassyPants> yeah I could probably build a package if I wanted to invest a couple days [18:19:00 CEST] <pink_mist> I doubt it would take days [18:19:06 CEST] <MrSassyPants> in learning how to create ubuntu packages and integrate them with the regular stuff [18:19:21 CEST] <MrSassyPants> oh, just making a package would probably be a matter of 1 hour [18:19:33 CEST] <MrSassyPants> but then there would be bugs and it would interfere with the normal upgrade path in unexpected ways [18:19:39 CEST] <MrSassyPants> in total the process would take days, no doubt [18:20:05 CEST] <MrSassyPants> What I'm looking for is a ppa [18:20:07 CEST] <pink_mist> I suggest you try it and see [18:20:42 CEST] <pink_mist> but I'm off to a board meeting, so I'll need to leave you to your own devices [18:20:43 CEST] <pink_mist> good luck [18:20:53 CEST] <MrSassyPants> have a nice one [18:21:02 CEST] <MrSassyPants> or, sleep well, depending on the board meeting type [18:21:19 CEST] <pink_mist> not enough people in it for me to get away with that :P [18:21:22 CEST] <ncouloute> any way to get the new video frame locations after a concatanate? FFMPEG tends to shift the frames around a bit. I'm guessing I'm going to have to write something with the ffmpeg api otherwise? [18:21:30 CEST] <kepstin> also note that installing a self-built ffmpeg to /usr/local won't conflict with the package manager stuff; packages will still link to the correct older system libav*.so and the `ffmpeg` cli tool has generally beenpretty compatible. [18:21:58 CEST] <MrSassyPants> yah, but obs doesn't seem to be using the locally installed ffmpeg [18:22:22 CEST] <MrSassyPants> i.e. obs doesn't recognize the nvenc availability which it should [18:23:24 CEST] <kepstin> oh, if you explicity *want* other packages to link to your custom ffmpeg then that's more annoying, yeah :/ [18:25:35 CEST] <MrSassyPants> I don't think obs links to it [18:25:50 CEST] <MrSassyPants> tho, not sure, it might [18:27:28 CEST] <kepstin> it does. [18:27:32 CEST] <pink_mist> it's impossible to make other binary packages link to your custom-built ffmpeg. you'll need to rebuild all of them yourself and make your own packages for all of them. [18:28:22 CEST] <kepstin> it's not impossible, if you make sure that shared libraries are enabled when your build your new ffmpeg, and that the library search path is configured appropriately [18:28:23 CEST] <pink_mist> no wait, I'm wrong [18:28:37 CEST] <pink_mist> but I really need to leave now, so can't retract my statement properly [18:28:38 CEST] <pink_mist> d'oh [18:29:41 CEST] <kepstin> most of the "install ffmpeg on ubuntu" guides only build a statically linked ffmpeg cli tool, when you need the dynamic libraries. [18:29:54 CEST] <kepstin> (it's just different options to the configure script) [18:30:21 CEST] <MrSassyPants> well it appears to work [18:31:07 CEST] <MrSassyPants> yah the obs version I installed *was* from a ppa [18:31:25 CEST] <kepstin> hmm, yeah, i dunno how the ppa one is built [18:31:38 CEST] <kepstin> the one in the ubuntu repos uses system ffmpeg libraries [18:31:47 CEST] <MrSassyPants> and just uninstalling the ffmpeg dpkg and make installing my self-built ones did it [21:31:44 CEST] <safinaskar> i think i found a bug. steps to reproduce. on fresh debian sid type "ffmpeg -f x11grab -i "$DISPLAY" /o.webm". then play video with gst-play-1.0 . you will see that video became pink [21:32:02 CEST] <safinaskar> so should i report it or it is already known? [21:32:12 CEST] <safinaskar> i. e. is v9 known to be buggy? [21:32:47 CEST] <c_14> have you tried playing the video with something that isn't gst-play? [21:32:49 CEST] <safinaskar> not reproducible with v8 [21:33:00 CEST] <safinaskar> yes. i tried mpv and it works [21:33:07 CEST] <durandal_1707> vp9 works fine here [21:33:45 CEST] <safinaskar> durandal_1707: okey, so i should report it, right? [21:34:54 CEST] <c_14> safinaskar: might be a playback bug in gstreamer or it's libs [21:37:42 CEST] <c_14> if it only happens with vp9 it'd (most likely) either be that or a bug with libvpx (the vp9 encoder) [21:38:37 CEST] <c_14> you could at very least test with a static build from latest git to make sure it isn't just debian shipping an old version of ffmpeg [21:40:39 CEST] <kepstin> sid has 4.1.4, that's pretty reasonable. and their vpx is 1.8.1 [21:41:04 CEST] <c_14> ah, he did mention he was on sid [21:41:36 CEST] <c_14> safinaskar: can you do an ffprobe of the output webm? [21:41:40 CEST] <kepstin> well, assuming it's sid that's been updated recently, and not sid from 2 years ago ;) [21:41:48 CEST] <c_14> It's probably yuv444p [21:41:59 CEST] <c_14> I had that issue a while back [21:42:12 CEST] <c_14> That is that certain decoders (at the time firefox) had issues decoding 444p vp9 [21:43:17 CEST] <kepstin> weird to see an issue with gstreamer tho. depending on how it's set up, it should be either using libvpx itself to decode, or ffmpeg :/ [21:45:44 CEST] <safinaskar> c_14: ffprobe: https://zerobin.net/?1a1b6c006548cf10#JmkQ1b/QH4OPIRTrEWNpeRE3i4DbxW2rStMLC… [21:46:14 CEST] <safinaskar> kepstin: this is sid i created today [21:51:33 CEST] <c_14> is gbrp valid webm? [21:52:33 CEST] <kepstin> huh, yeah, if gbr is decoded as yuv, a strange color video would be the likely result [21:55:22 CEST] <realnx> hi this irc channel is indicated on this page https://ffmpeg.zeranoe.com/builds/ I am actually having an issue downloading ffmpeg in ShareX giving an 403 forbidden error, the owner of Sharex closed my issue https://github.com/ShareX/ShareX/issues/4386 and referred to this repository. Where can i contact the maintainer of that repo to report this problem? [21:56:19 CEST] <kepstin> realnx: the site https://ffmpeg.zeranoe.com/builds/ has a nice big "Contact" link at the top which you can use. [21:57:37 CEST] <genpfault> how long do trac account verification emails usually take to get sent? regged with a gmail address around noon CST today (~3h ago) and haven't seen anything in my inbox or spam yet [21:57:39 CEST] <realnx> Thanks! (and sorry about not paying attention) [22:00:43 CEST] <c_14> safinaskar: I had a quick gander through the gstreamer vp9 decoder plugin and it doesn't seem to support gbrp [22:02:05 CEST] <c_14> my assumption is that maybe gst-play neglects to check the return code of a function somewhere and it's trying to use a default value [22:02:32 CEST] <c_14> maybe try adding -pix_fmt yuv420p or -pix_fmt yuv444p to your command before the output file (and after the input) and see if that works? [22:07:14 CEST] <safinaskar> c_14: both variants work [22:08:29 CEST] <c_14> yeah, I'm 90% sure it's the bug with gst-play that I mentioned [22:45:15 CEST] <jpb> ha! found my answer. i was having trouble with ffplay rtsp://username:password@host/foo [22:46:45 CEST] <jpb> turns out the password can't contain the '+' character. the '+' character is replaced with a space in libavformat ff_urldecode() [22:47:12 CEST] <jpb> no idea why [22:49:26 CEST] <kepstin> hmm, strange. I'd expect that only in the parth portion of the url [22:49:40 CEST] <kepstin> you should be able to percent-escape the + sign in the password tho [22:53:09 CEST] <jpb> haven't tried that. i can still have a strong password without the '+'. [22:58:54 CEST] <BtbN> ffmpeg actually follows the http standard quite closely regarding that. [22:59:31 CEST] <BtbN> Not HTTP, but URL standard. [22:59:48 CEST] <BtbN> Also, keep in mind that some characters are outright forbidden in passwords and usernames in URLs. For example, it can't contain / [23:55:42 CEST] <`St0ner> i've spent hours trying to figure out why one x264 video loads to my iPod Touch 3G but another x264 video doesn't, even when the x264 parameters, profile, frame rate, width, height, practically everything else is exactly the same. the file that works has Codec ID = mp42 (mp42/mp41/isom/avc1) and "Overall bit rate mode" = Variable. the file that doesn't work has Codec ID = mp42 (isom/iso2/avc1/mp41) [23:55:42 CEST] <`St0ner> and mediainfo doesn't show any "Overall bit rate mode" [23:56:47 CEST] <c_14> can you upload the ffprobe output on both files to a pastebin service? [23:57:02 CEST] <c_14> it might also be the audio codec [23:57:28 CEST] <`St0ner> i removed the audio from the file that doesn't work. yes i'll upload to pastebin, hang on [00:00:00 CEST] --- Tue Sep 24 2019

1 0

ffmpeg-devel.log.20190922
by burek 23 Sep '19

23 Sep '19

[00:26:50 CEST] <philipl> Lynne: hi [14:04:20 CEST] <durandal_1707> STOP MOVING SWSCALE INTO LAVC [14:04:30 CEST] <durandal_1707> JUST STOP!!! [14:13:30 CEST] <cone-300> ffmpeg 03Paul B Mahol 07master:79d14a3cc8e9: avfilter/vf_v360: remove intermediate variables [14:13:31 CEST] <cone-300> ffmpeg 03Paul B Mahol 07master:51a2f02117df: avfilter/vf_v360: add fixed pixel padding options [15:28:18 CEST] <Lynne> durandal_1707: I was wondering if I'm the only one seeing that patch for what it was, as usual [15:31:09 CEST] <durandal_1707> Lynne: what's blocking vulkan pushing to master? [15:32:15 CEST] <JEEB> I think he was just tired of bikeshedding etc [15:32:40 CEST] <durandal_1707> i did not noticed any [16:18:36 CEST] <Lynne> who's tired of bikeshedding? [18:14:13 CEST] <kierank> durandal_1707: I AGREE [18:14:16 CEST] <kierank> BUT BIG CORPORATES DO THIS [18:18:30 CEST] <kierank> durandal_1707: I support you [18:18:43 CEST] <BBB> huh? [18:26:51 CEST] <kierank> BBB: there's a guy trying to merge colourspace conversion into an encoder [18:26:57 CEST] <kierank> that old chestnut [18:33:00 CEST] <cone-504> ffmpeg 03Andreas Rheinhardt 07master:25a501b5287b: avutil/opt: Fix type specifier [18:33:00 CEST] <cone-504> ffmpeg 03Michael Niedermayer 07master:0e8b7709a92a: avcodec/sunrast: Fix return type for "unsupported (compression) type" [18:33:00 CEST] <cone-504> ffmpeg 03Michael Niedermayer 07master:0728d644973c: avcodec/sunrast: Fix indention [19:15:36 CEST] <BBB> kierank: lol [19:15:44 CEST] <BBB> kierank: one of these stupid game codecs again? [19:15:47 CEST] <BBB> or a real codec this time? [19:15:48 CEST] <kierank> BBB: no, v210 [19:16:02 CEST] <BBB> is v210 even a codec? [19:16:13 CEST] <BBB> it's just a particular way of writing 422-10bit, right? [19:17:10 CEST] <kierank> yes [19:17:29 CEST] <kierank> debatable whether it's a codec or a pixel format [19:18:30 CEST] <BBB> i see patch now [19:18:34 CEST] <BBB> that's fun [19:18:40 CEST] <BBB> big name, expert, important [19:18:44 CEST] <BBB> all the buzzwords [19:18:46 CEST] <BBB> <3 [19:32:16 CEST] <cone-504> ffmpeg 03Paul B Mahol 07master:05a2ce932654: avformat/y4m: do not try to seek if pts is less than 0 [19:57:45 CEST] <cone-504> ffmpeg 03Paul B Mahol 07master:b4d2bea64722: avfilter/v360: reduce size of some struct members [20:49:51 CEST] <philipl> Lynne: I've been taking a look at map_to_cuda [20:49:59 CEST] <philipl> jkqxz: Could use your thoughts on this one. [20:50:42 CEST] <philipl> We want to map a vulkan hw frame to cuda. Due to a vulkan image not mapping to the same thing as what PIX_FMT_CUDA represents, we can't directly map and need to do a gpu side copy. [20:50:52 CEST] <philipl> So far so good. Lynne wrote this for going from cuda to vulkan. [20:52:11 CEST] <philipl> So, we need to allocate the dst frame from a cuda device context. The approach I've started with is to support deriving a cuda context from a vulkan context and then doing a hwmap=derive_device=cuda in the filter chain. [20:53:00 CEST] <philipl> So when I get to the map callback, I want to do av_hwframe_get_buffer, but because the cuda context is derived, it actually triggers an infinite recursion as it tries to go back to the source context to map the frame. [20:53:24 CEST] <philipl> So I want a non-derived cuda context where av_hwframe_get_buffer actually allocates. [20:54:06 CEST] <philipl> but of course I also want to make sure the cuda context is on the same physical device as the vulkan context or none of this works. [20:54:11 CEST] <philipl> Is there a way to get what I want? [22:17:22 CEST] <Lynne> philipl: sounds good, the way I did it in the vulkan hwcontext was to just manually reinitialize the internal frames pool in the map function in case it isn't [22:17:55 CEST] <Lynne> avhwdevice* are always initialized properly and that's all that's needed to init the frames pool, but I'm not sure how it is for cuda [22:20:47 CEST] <Lynne> btw why does cuda have the concept of mipmaps when its purely a compute API? [22:27:32 CEST] <philipl> Lynne: I guess because they expect to interact with textures. That basic concept has been in the language since very early on. [22:27:46 CEST] <philipl> Given that, I don't know why the interop is so awkward. [22:37:33 CEST] <cone-504> ffmpeg 03Paul B Mahol 07master:a8925d264a95: avfilter/vf_v360: fix mercator_to_xyz() [22:37:34 CEST] <cone-504> ffmpeg 03Paul B Mahol 07master:043038ea56ab: avfilter/vf_v360: fix xyz_to_mercator() [22:37:35 CEST] <cone-504> ffmpeg 03Paul B Mahol 07master:4ba45a95df1b: avfilter/vf_v360: adjust h for mercator input/output [22:38:13 CEST] <philipl> Lynne: Yeah, you can init the vulkan pool using private functions. The equivalent cuda context calls are also private. [22:38:36 CEST] <philipl> And we can't easily implement map in the cuda context because it depends on the vulkan images being initialised correctly with all the export flags, etc. [22:38:53 CEST] <philipl> Might have to brute force it and add a cuda helper to do pool init from the outside. [22:40:02 CEST] <Lynne> are you talking about initing the cuda pool with vulkan frames? [22:40:33 CEST] <Lynne> if the vulkan->cuda interop is in the cuda hwcontext there's nothing private it should need outside of the cuda hwcontext [22:41:53 CEST] <philipl> Lynne: Yeah, if I was to implement in the interop in the cuda hwcontext, I'd be able to init the pool just fine, but we couldn't guarantee that the vulkan frames were exportable. [22:42:13 CEST] <philipl> So I think we really still want the interop in the vulkan hwcontext, but then I need a way to init the cuda context's pool. [22:42:42 CEST] <philipl> Basically, I need a helper that I call to 'un-derive' the context so it uses its own allocator. [22:52:41 CEST] <Lynne> philipl: if they're not exportable then you should just error out [22:53:39 CEST] <Lynne> users can always create a avhwframescontext for vulkan themselves, with usage flags or a queue that does not permit exporting [22:54:33 CEST] <Lynne> but if you need to export internal cuda stuff for use in vulkan, you can put them in hwcontext_cuda_internal.h as we already use that for cuda->vulkan device deriving [22:55:03 CEST] <philipl> Lynne: yeah, so that's basically my plan. [22:58:34 CEST] <jkqxz> It sounds like what you want isn't really mapping as currently defined by the API. [22:58:38 CEST] <jkqxz> If the destination is just any CUDA frame with the right properties, can you derive the device followed by making a new CUDA frames context and then use transfer? [22:59:19 CEST] <jkqxz> (That case could be added to the hwmap filter if wanted.) [22:59:48 CEST] <philipl> jkqxz: Yeah, it's not really mapping. It's really a transfer, but I'm not sure how you set that up to work as-is. 'hwmap' is the only way to indicate a shift from one context to another right? [22:59:58 CEST] <philipl> We sort of need a hwtransfer filter right? [23:01:09 CEST] <jkqxz> That would work. [23:01:11 CEST] <philipl> jkqxz: so right now, I've implemented derive_device in hwcontext_cuda and then have a 'hwmap=derive_device=cuda' filter which triggers the map operation on the hwcontext_vulkan side. [23:01:47 CEST] <philipl> At the time the map is called, there's already a dervied hwframesctx which won't work. I'm not sure where I'd interpose to fix that. [23:01:49 CEST] <jkqxz> Adding it to the hwmap filter might be nice for filter graphs, but it sounds like the API-level part does need to be different. [23:01:59 CEST] <philipl> Perhaps that is not worth it and I should just write the hwtransfer filter. [23:03:09 CEST] <philipl> The brute force thing to do is have some 'underive' function that would clear the source_frames and re-init the context. [23:03:19 CEST] <philipl> which is rather silly [23:05:02 CEST] <jkqxz> Do you even want the mapping semantics from the outside? (Specifically, the behaviour where it will copy back a writable mapping when you unmap.) [23:05:17 CEST] <philipl> No [23:06:24 CEST] <philipl> The only realistic pipelines here are nvdec -> vulkan filters -> non-nvidia sink, and nvdec -> vulkan filters -> nvenc. It's all one way transfers. [23:06:53 CEST] <philipl> and due to the interop limitations, each hand off is a GPU copy [23:07:22 CEST] <philipl> but highly invasive on the vulkan hwcontext due to having a bunch of cuda-side state that must be carried with the frames. [23:10:35 CEST] <philipl> Would it do the right thing if transfer_data_from implemented a cuda path and then we added a format=cuda to the filter chain before nvenc? [23:11:06 CEST] <philipl> I guess probably not due to lack of correctly initialised dst frame [23:11:43 CEST] <jkqxz> There isn't any auto-hwtransfer anywhere. A hwupload could do it? [23:13:49 CEST] <philipl> Yeah, if the hwupload accepts the vulkan pix fmt, which it won't today. [23:13:54 CEST] <philipl> I can experiment. [23:14:34 CEST] <jkqxz> I guess you want hwtransfer which is basically identical to hwupload except accepts different input formats? Or something like that. [23:14:39 CEST] <philipl> Yeah [23:14:53 CEST] <philipl> This seems like it might lead somewhere useful. [23:15:10 CEST] <philipl> but I have to take a break now. Will come back to it later. [23:35:30 CEST] <mkver> jkqxz: Can you take a look at these two patches: http://ffmpeg.org/pipermail/ffmpeg-devel/2019-September/249814.html They are in response to ticket #8126. [00:00:01 CEST] --- Mon Sep 23 2019

1 0

ffmpeg.log.20190922
by burek 23 Sep '19

23 Sep '19

[00:04:00 CEST] <auser0> i also noticed there is something called :: ErrorDetectionType [00:04:09 CEST] <auser0> how do i avoid having this added to mediainfo? [00:50:40 CEST] <auser0> okay guess i got that [00:50:55 CEST] <auser0> is there a way to fix the placement of subtitles using ffmpeg? [00:51:11 CEST] <auser0> the ones i have ended to the top left, and i want it as the default in the bottom of the screen [01:03:14 CEST] <AiNA_TE> is you sent convert your subtitile format to .srt then it has no syles [01:03:30 CEST] <AiNA_TE> and will placed at the bottom of your screen by your media player by default [01:03:42 CEST] <AiNA_TE> styles* [01:13:37 CEST] <auser0> is there a way to do it that the default placement for it is at the bottom AiNA_TE ? [01:33:29 CEST] <furq> auser0: it's just text, there is no position information [01:33:38 CEST] <furq> you'd have to configure your player [02:07:37 CEST] <auser0> thanks furq but from my previous playbacks, i always had the subtitles by default showing at the bottom [02:07:49 CEST] <auser0> i guess thats how the CC subtitles were in that stream maybe [03:52:33 CEST] <auser0> anyone know how to avoid having this :: ErrorDetectionType in mediainfo? [07:21:11 CEST] <auser0> anyone able to tell me what does that mean [07:21:16 CEST] <auser0> Fixing malformed AAC bitstream in "xx.mp4" [07:21:16 CEST] <auser0> ffmpeg -y -loglevel 'repeat+info' -i vv.mp4 -c copy -f mp4 '-bsf:a' aac_adtstoasc xx.mp4 [07:21:39 CEST] <auser0> how does it fix it ? [10:16:07 CEST] <exs> I need help [10:16:22 CEST] <exs> I have a fast machine but when I record my screen I just get frame= 1507 fps= 21 q=24.0 size= 16896kB time=00:00:48.86 bitrate=2832.4kbits/s speed=0.69x [10:16:32 CEST] <exs> what could be the reason, the cpu also gets not more than 30% [10:18:12 CEST] <durandal_1707> slow disk? [10:18:22 CEST] <exs> I have a m2 ssd [10:18:38 CEST] <durandal_1707> which OS? [10:40:47 CEST] <exs> durandal_1707: arch linux [10:41:21 CEST] <exs> ffmpeg version n4.2.1 [10:41:45 CEST] <exs> NVIDIA Corporation GP104 [GeForce GTX 1080] [10:41:58 CEST] <exs> AMD Ryzen 7 1700X (16) @ 3.400GHz [10:42:01 CEST] <exs> 64315MiB RAM [10:42:09 CEST] <exs> Resolution: 3440x1440 [10:42:20 CEST] <exs> OS: Arch Linux x86_64 with Kernel: 5.3.0-arch1-1-ARCH [10:47:27 CEST] <exs> durandal_1707: any thoughts? [10:50:04 CEST] <durandal_1707> exs: run ffmpeg command under: "perf record", and post "perf report" results [10:50:18 CEST] <durandal_1707> only few seconds of recording [10:53:54 CEST] <exs> durandal_1707: I am not fimiliar with perf record, can you tell me what I should execute? [10:58:39 CEST] <exs> frame= 464 fps= 22 q=8.0 size= 5888kB time=00:00:14.60 bitrate=3303.7kbits/s speed=0.701x is what I have during recording [10:58:55 CEST] <exs> perf record: Captured and wrote 25.172 MB perf.data (478765 samples) [10:59:06 CEST] <exs> do you want me to upload the perf.data? never used it before [11:00:44 CEST] <exs> ah ok sry I see now [11:00:47 CEST] <exs> I send you a screenshot [11:01:15 CEST] <exs> durandal_1707: https://pasteboard.co/IyyY1if.png [11:03:15 CEST] <exs> even I move no windows I get only speed=0.926x [11:03:34 CEST] <exs> I mean I am fine when the video loose quality [11:03:46 CEST] <exs> it just need to be synced with audio for any price [11:04:08 CEST] <exs> in my current situation it unsynced after a while [11:05:02 CEST] <exs> is the script https://termbin.com/vx83 [11:18:27 CEST] <durandal_1707> exs: dunno, have you tried without audio first? [11:23:42 CEST] <exs> durandal_1707: let me try now [11:24:45 CEST] <exs> durandal_1707: its slightly better [11:24:56 CEST] <exs> I use compton can this be a reason? [11:25:52 CEST] <durandal_1707> slightly? [11:29:48 CEST] <exs> from speed=0.926x to speed=0.976x [11:30:19 CEST] <exs> its funny, when I get start to record, the first seconds are 0.5x then it start to increase slowly to 0.90x [11:30:44 CEST] <exs> now its even speed=0.99x but it took 15 seconds to get there [11:31:12 CEST] <exs> I dont understand that my cpu is at 16%, does ffmpeg use concurrency for recording? [11:31:38 CEST] <durandal_1707> it should use threads for video encoding [11:32:04 CEST] <durandal_1707> try to specify -threads 8 as input and as output option [11:35:01 CEST] <exs> durandal_1707: can you please tell me where exactly in https://termbin.com/vx83 [11:35:08 CEST] <exs> since I know the order matters [11:36:14 CEST] <exs> I mean after a while it starts to work at 0.99x but I worry that the first seconds are enough to make video and audio unsycn again [11:37:01 CEST] <exs> https://termbin.com/0gez is what I have now, is that correct? [11:38:17 CEST] <durandal_1707> add also one after -vcodec libx264 [11:38:56 CEST] <durandal_1707> perhaps you should use something else for capture, like obs [11:38:57 CEST] <exs> durandal_1707: I have 16 threads, should I add 16? [11:39:06 CEST] <exs> durandal_1707: what is obs? [11:39:19 CEST] <durandal_1707> exs: dunno, too much threads may slow more [11:39:32 CEST] <exs> ok I keep 8 then, other changes? [11:39:59 CEST] <exs> sry obs I do not know can you help me to craft a good script to record my screen with full audio, income (mic) and outcome? [11:40:06 CEST] <durandal_1707> https://obsproject.com/ [11:40:19 CEST] <exs> no I have amazing speed=0.997x [11:40:46 CEST] <durandal_1707> the ffmpeg capturing is single threaded... [11:41:04 CEST] <durandal_1707> and even capture of audio and video may be in single same thread... [11:41:20 CEST] <durandal_1707> so only video encoding use multiple threads [11:41:46 CEST] <exs> low it grow to speed=0.999x [11:41:53 CEST] <exs> low=lol [11:42:06 CEST] <durandal_1707> but how fast it grows? [11:42:16 CEST] <durandal_1707> and with audio? [11:42:21 CEST] <exs> no without [11:42:24 CEST] <exs> I test with audio now [11:42:49 CEST] <exs> btw obs is multithred? if its just a gui for ffmpeg I would stick with ffmpeg because its command line, I am not a fan of guis [11:43:08 CEST] <durandal_1707> obs is used by professionals [11:43:22 CEST] <exs> https://termbin.com/bheg this is how it looks now [11:43:23 CEST] <durandal_1707> ffmpeg is used by "amateurs" [11:43:39 CEST] <exs> durandal_1707: lol I thought its a kind of core app that most guis built on [11:43:55 CEST] <exs> durandal_1707: the benefit of command line is that I can save my files automatically [11:44:01 CEST] <exs> i just press a command and the file is saved [11:44:13 CEST] <exs> with guis you have to click around a lot. but I give it a try, thanks for recommandation [11:44:23 CEST] <exs> is the script alright like that? [11:44:39 CEST] <durandal_1707> exs: ffmpeg capturing stack is very simple and not much maintained [11:44:49 CEST] <exs> oh ok [11:45:04 CEST] <exs> durandal_1707: can you reocmmand another command line tool for screencasting? [11:45:05 CEST] <durandal_1707> exp. you might use hardware video encoders [11:45:27 CEST] <durandal_1707> exs: use something else that slow x11grab [11:46:12 CEST] <durandal_1707> there is now kmsgrab [11:46:49 CEST] <durandal_1707> and xcbgrab??? [11:47:09 CEST] <durandal_1707> well than one is same as x11grab [11:52:13 CEST] <exs> should I use kmsgrab? [11:53:43 CEST] <durandal_1707> dunno, try it, it may need special hw and setup [12:00:32 CEST] <exs> durandal_1707: wow damn it, obs studio is amazing [12:00:36 CEST] <exs> its even open source right? [12:00:45 CEST] <exs> the way it saved data is also perfect [12:00:49 CEST] <exs> with dates encoded [12:00:57 CEST] <exs> durandal_1707: can you tell me which output format I should prefere [12:01:17 CEST] <exs> mp4 was my favorit for unknown reasons, maybe because mp3 is a really common and mp4 sounds similar [12:01:29 CEST] <exs> but the app tells me that the format can become corrupted fast [12:01:35 CEST] <exs> flv I am not sure its the default [12:01:43 CEST] <exs> what would you recommend as an ffmpeg dev? [12:04:13 CEST] <JEEB> mp4 without fragments has the index at the end [12:04:22 CEST] <JEEB> so if you cut the power or the app crashes [12:04:26 CEST] <durandal_1707> mkv [12:04:30 CEST] <JEEB> you've got a bunch of data that is crap [12:15:06 CEST] <exs> JEEB: thatis what I mean, so you guys would recommend mkv? [12:15:19 CEST] <exs> why mkv over flv? or other formats? [12:15:26 CEST] <JEEB> flv is very limited [12:15:41 CEST] <exs> lol than why its the default at obs studio [12:15:42 CEST] <JEEB> you usually only use flv when pushing video through the RTMP protocol, which only supports FLV [12:15:52 CEST] <JEEB> yes, a lot of services take in live video as FLV [12:15:58 CEST] <JEEB> through RTMP [12:16:00 CEST] <exs> RTMP is a protocol for online streaming I guess? [12:16:02 CEST] <JEEB> yes [12:16:04 CEST] <JEEB> adobe based [12:16:08 CEST] <JEEB> but even adobe has given up on it [12:16:22 CEST] <JEEB> which is why we so far say no to adding random additional support for newer things in FLV [12:16:30 CEST] <exs> I prefere some kind of raw format that I can use to convert into any format I want [12:16:33 CEST] <JEEB> people should switch to fragmented MP4 or MPEG-TS or something else [12:16:41 CEST] <JEEB> (for live stream ingest) [12:16:51 CEST] <JEEB> exs: I'd say go with durandal_1707's recommendation for now [12:16:53 CEST] <JEEB> matroska/mkv [12:17:01 CEST] <JEEB> you can remux to mp4 if you need to [12:17:05 CEST] <JEEB> or re-encode [12:17:43 CEST] <exs> JEEB: ok I do and why mkv and not mov ts or m3u8? [12:18:09 CEST] <exs> just curiosity [12:19:39 CEST] <JEEB> mov without fragments has the same issues as mp4 [12:19:46 CEST] <JEEB> after all, mp4 is based upon mov originally [12:20:16 CEST] <JEEB> ts is SureFine; it's the original broadcast container. but it's made for A->B playback, not for seeking etc [12:20:40 CEST] <JEEB> m3u8 is HLS which you might want if you are feeding this to a web server to be served as a live stream [12:20:49 CEST] <JEEB> bunch of segments and playlists [12:21:19 CEST] <exs> JEEB: ok thanks for that feedback [13:25:45 CEST] <exs> durandal_1707: obs is awesome, works perfect [13:26:04 CEST] <exs> just neeed to test more if I get unsync probs but seems good now [16:18:00 CEST] <auser0> ffmpeg -y -loglevel 'repeat+info' -i vv.mp4 -c copy -f mp4 '-bsf:a' aac_adtstoasc xx.mp4 , anyone can explain to me what that does? [16:21:07 CEST] <JEEB> -y = don't ask me for confirmation [16:21:20 CEST] <JEEB> -loglevel = same as -v, sets the verbosity level for logging [16:21:39 CEST] <JEEB> then you have input and setting the encoder to "copy" for all streams [16:21:45 CEST] <JEEB> then -f sets the output format (container) [16:21:53 CEST] <JEEB> (usually guessed by the output file name) [16:22:19 CEST] <JEEB> then the bit stream filter converts AAC from the way it's in MPEG-TS to how it's stored in MP4 (which sounds weird when your input is in MP4) [16:22:28 CEST] <JEEB> also that bit stream filter addition should now be automagic [16:22:37 CEST] <JEEB> then you have the output file name [16:23:48 CEST] <auser0> ty JEEB [16:24:10 CEST] <JEEB> also please learn to look up ffmpeg-all.html [16:24:18 CEST] <JEEB> that probably can answer a lot of your questions [16:24:34 CEST] <auser0> just wondering, if i specify the container in the output, will i need -f mpf part? [16:24:47 CEST] <auser0> sure, will do that too [16:25:12 CEST] <JEEB> the -f forces it, mp4 would already get auto-picked if the file name ends with ".mp4" [16:25:36 CEST] <JEEB> usually you use -f XXX when your file name doesn't have an extension, or the extension is custom [16:25:50 CEST] <auser0> and shouldnt it be like : -c:e aac_adtstoasc -bsf:a [16:25:59 CEST] <auser0> -c:a* [16:26:28 CEST] <auser0> or its not needed to specify -c:a ? [16:28:11 CEST] <JEEB> -c:a is already set within -c [16:28:13 CEST] <JEEB> to "copy" [16:28:22 CEST] <JEEB> without a specifier the rule applies to all output streams [16:28:43 CEST] <JEEB> and the bit stream filter is called like that, so no - it's not an encoder [16:28:54 CEST] <JEEB> and as I said, that gets automagically inserted now if your input and output are like that [16:29:00 CEST] <JEEB> mp4 in and mp4 out does *not* need it [16:29:01 CEST] <JEEB> as I said [16:29:09 CEST] <JEEB> so in that sense that option is non-sensical [16:29:57 CEST] <auser0> ah i got it now, and does that part is needed to be ran by the exact order : -bsf:a' aac_adtstoasc ? [16:30:52 CEST] <JEEB> the option is bsf (limited to audio since this bit stream filter is for audio only), and the parameter is the name or names of bit stream filters [16:31:42 CEST] <JEEB> option/parameter order matters, yes. but within the command line it only matters a) if you are before/after inputs b) if you have multiple inputs or outputs, at which point you set options between those [16:32:24 CEST] <JEEB> ffmpeg OPTIONS_FOR_INPUT_0 -i INPUT_0 OPTIONS_FOR_INPUT_1 -i INPUT_1 OPTIONS_FOR_OUTPUT_0 OUTPUT_0 OPTIONS_FOR_OUTPUT_1 OUTPUT_1 [16:32:48 CEST] <JEEB> there are of course some options for ffmpeg.c itself which are global [16:32:59 CEST] <JEEB> like the verbosity option etc [16:35:46 CEST] <auser0> okay i guess i understood it now, ty [16:36:09 CEST] <auser0> last night the command i asked for : -c copy -c:s srt -bsf:v 'filter_units=pass_types=1-5' [16:37:08 CEST] <auser0> once i specified -c:s srt, doed that tells ffmpeg to start encoding subtitles to be in srt format? [16:37:14 CEST] <auser0> does* [16:37:29 CEST] <JEEB> yes, it sets the subtitle encoder to srt [16:38:10 CEST] <auser0> if i ran it without the following -bsf:v, would it still work ? [16:39:05 CEST] <auser0> or i need to specify the bsf:v part everytime i want to encode to srt [16:39:37 CEST] <JEEB> the bit stream filter only touches video streams [16:39:51 CEST] <JEEB> and it just filters the video stream to remove certain units from it [16:41:50 CEST] <auser0> ok got it [16:42:45 CEST] <auser0> is there a way to control the way those subtitles are shown in my player? [16:42:53 CEST] <auser0> using ffmpeg i mean [16:43:18 CEST] <auser0> the one i tested last night had the subtitled shown on the top left [16:43:43 CEST] <auser0> Im used to it in the middle at the bottom [16:43:45 CEST] <JEEB> output the srt into a .srt file and check the lines. [16:44:03 CEST] <another> afaik it depends on the settings in your video player [16:44:55 CEST] <JEEB> SRT can have random HTML tags because people implemented that, so it could just be converting the positioning from the input like that [16:45:02 CEST] <JEEB> just like it would put ASS tags for ASS [16:45:09 CEST] <JEEB> but simplest is to dump the SRT and see how it looks [16:46:23 CEST] <auser0> ah, that would do too, i havent checked it in an editor to see if there are weird tags, but will do [16:47:15 CEST] <auser0> another, ya i thought so too but mine was set to always show them in the middle but somehow that one showed in wrong place [16:47:37 CEST] <auser0> i never edited the position of them or anything [16:49:30 CEST] <another> well, it depends on the subtitle codec [16:50:10 CEST] <another> image based subtitles e.g. can specify the location [16:50:42 CEST] <auser0> oh, so probably that conversion caused that then [16:50:46 CEST] <another> no idea about cc608/708 [16:50:56 CEST] <JEEB> closed captions have a *lot* of stuff [16:51:09 CEST] <JEEB> there's a test sample on videolan's sample server [16:51:12 CEST] <another> i didn't even know cc embeded in the video stream was a thing [16:51:17 CEST] <another> *until recently [16:51:31 CEST] <auser0> the stream itself was playing them also on the top left iirc [16:51:58 CEST] <auser0> so youre probably right [16:52:17 CEST] <JEEB> https://streams.videolan.org/samples/sub/CC/ [16:52:21 CEST] <JEEB> see the CEA sample :P [16:52:31 CEST] <auser0> ya, i guess i will need to demux them and check them soon [16:55:07 CEST] <auser0> sure , downloading :) [16:58:58 CEST] <DHE> yeah I want to make a demuxer for CCs because filters can mangle them if the framerate changes [18:18:29 CEST] <auser0> i seen it JEEB :) [18:18:41 CEST] <auser0> that would be nice DHE [18:19:05 CEST] <auser0> anyone knows if this is something that should gets me worried in here : [18:19:06 CEST] <auser0> https://zerobin.net/?f3291ac9fa42c307#EdyHEYmeS3cd4/SLxrXRVLSu2t64W53ngvbx2… [18:19:32 CEST] <auser0> seems like the input shows 8 mins only while after fixing the audio stream, it shows 44mins [18:29:09 CEST] <DHE> auser0: my plan doesn't really require decoding. I just need to be able to dismantle them and rebuild them to the same net effect in the output. I assume there's some padding since it appears each video frame has a fixed size CC payload which I can abuse to deal with increased or decreased framerates. [18:33:47 CEST] <auser0> sounds good, so i guess if i understood correctly [18:34:10 CEST] <auser0> you want to demux those CC subs and use some other tool as subtitle editor or whatever to nicely convert them to .srt [18:34:30 CEST] <auser0> subtitle editor is gui as in windows so it wont do much of help i guess [18:35:01 CEST] <auser0> i don't know about the padding part, but i feel like you can do it :) [18:35:07 CEST] <DHE> no. I want to extract the CCs from video frames, submit them to my filter queue cleaned, then as they come out I put them back into the videos. possibly combining data from multiple frames into a single frame or otherwise slightly getting them out of order but 1 or 2 frames should be okay [18:35:33 CEST] <DHE> well not out of order, just delayed [18:36:55 CEST] <auser0> i see, well if there is something i can do to help, let me know since such a thing would helps a lot many people [18:37:25 CEST] <auser0> even cehoyos was looking for something that does that smoothly [18:38:30 CEST] <DHE> https://github.com/FFmpeg/FFmpeg/blob/master/libavcodec/ccaption_dec.c Well there is code to decode captions. I think I'll have to do a partial re-implementation, but implement more of as a demuxer than a decoder [18:40:12 CEST] <cehoyos> auser0: It is in general very difficult to calculate the duration of an mpeg stream, so the fact that the duration is wrong is by itself not even an indication for a bug. [18:40:25 CEST] <cehoyos> I don't know if in your specific case the wrong duration is a bug or not [18:40:36 CEST] <cehoyos> Why did you highlight me? I don't understand the sentence [18:41:13 CEST] <cehoyos> DHE: Are you planning to program something for yourself or something that you should be added to FFmpeg? (I know that these are not exclusive) [18:41:35 CEST] <DHE> cehoyos: I have an existing applications using libav* that I planned to add it to [18:42:08 CEST] <cehoyos> I just ask because moving a decoder into libavformat could be problematic - but for your own uses there is of course no issue. [18:42:09 CEST] <auser0> oh, i never knew that the duration of an MPEG stream isn't that well detectable, thanks mate [18:42:43 CEST] <cehoyos> It can have a timestamp wrap that can be at a random point in time [18:42:50 CEST] <cehoyos> (I am not saying it has one) [18:42:59 CEST] <auser0> I thought you could be also interested in what DHE is talking about, to convert CC to SRT precisely :) [18:43:05 CEST] <DHE> yeah mpegts has that [18:43:30 CEST] <cehoyos> DHE: What exactly does not work with current FFmpeg? (I know what does not work, I am just curious if that really hits you) [18:43:33 CEST] <DHE> auser0: that ffmpeg file I linked earlier appears to do exactly that, though not SRT specifically. the catch is that the data is stored as frame sidedata and hence must be extracted differently [18:44:10 CEST] <DHE> cehoyos: if the video framerate changes - perhaps due to inverse telecine in your filter pipeline - then some of the sidedata for closed captions is lost out the other end of the filterchain [18:44:30 CEST] <auser0> i see [18:44:48 CEST] <cehoyos> Didn't we fix that? [18:44:58 CEST] <DHE> not that I'm aware of, but I could be wrong [18:45:14 CEST] <cehoyos> No: What we fixed was that inserted frames do not duplicate Closed Captions [18:45:30 CEST] <cehoyos> If frames get dropped, there is no way we can keep there SEI (or whatever it is called) [18:45:43 CEST] <cehoyos> But I would still like to see your command line including ... [18:46:06 CEST] <DHE> right. so my app will have to pull the side data, filter, then put it back. but I'll also have to understand the format enough to reassemble it appropriately [18:46:16 CEST] <cehoyos> I would expect the split filter should allow you to work around the issue [18:46:34 CEST] <DHE> I'm not using the ffmpeg cli [18:46:43 CEST] <cehoyos> Why don't you try with ffmpeg (the command line interface) first and then try to port to your application? [18:47:15 CEST] <DHE> becuase if I blindly copy sidedata at 30fps into video running at 24fps I'll have a pretty serious sync problem [18:48:14 CEST] <cehoyos> auser0: One more thing: Why don't you keep the mpegts? Because it has reception errors (which is normal for mpegts), the mp4 may be invalid (which is not ok for an mp4 file). Afair, all Media Players accept the mpegts stream [18:48:38 CEST] <cehoyos> DHE: Why don't you try with ffmpeg (the command line interface) first and then try to avoid blindly copying sidedata? [18:49:39 CEST] <DHE> sorry what? [18:50:18 CEST] <cehoyos> Does your input contain embedded Closed Captions (not separate streams) and you want the same embedded Closed Captions in the output but with a different framerate? [18:50:24 CEST] <cehoyos> This may indeed not work yet. [18:50:30 CEST] <DHE> yes that is the goal [18:50:52 CEST] <DHE> right now they get a little bit mangled on-screen [18:50:59 CEST] <cehoyos> This may indeed not work yet with the cli, should probably work using the libraries only. [18:52:01 CEST] <DHE> it kinda works, but looks like they're being typed by someone drunk [19:17:56 CEST] <auser0> this is how i got the srt files : https://zerobin.net/?0af5a850b521786e#U4SrO22UvXyT8L+BT3qqB9hX6MOVrOxTKg5dP… [19:18:26 CEST] <auser0> and they show on the top left, maybe those tags needs some cleanups : https://i.imgur.com/pwXDyTi.png [19:20:32 CEST] <furq> oh i forgot srt actually does have alignment tags [19:20:36 CEST] <furq> i've never seen anyone use them before [19:24:56 CEST] <auser0> ya removing those an7 seems to place them right [19:25:21 CEST] <auser0> is there a way in one command line to remove them from CC or from .srt before muxing them to the mkv ? [19:25:29 CEST] <furq> apparently -c:s text will give you srt with all tags stripped [19:26:12 CEST] <auser0> oh interesting, thaks furq, will try that [19:26:22 CEST] <auser0> whats the major difference between text and srt? [19:26:27 CEST] <furq> they're the same [19:26:33 CEST] <furq> just text doesn't write any tags [19:27:28 CEST] <auser0> nice, testing it atm [19:27:30 CEST] <furq> https://github.com/FFmpeg/FFmpeg/blob/master/libavcodec/srtenc.c#L211-L227 [19:28:45 CEST] <furq> that's the only difference in the code path [19:29:47 CEST] <auser0> i see thanks, so basically the command i ran in one line is: ffmpeg -i zz.ts -f lavfi -i "movie=zz.ts[out0+subcc]" -map 0:v -map 0:a -map 1:s -c copy -c:s text -bsf:v 'filter_units=pass_types=1-5' -metadata:s:s:0 language=eng xx.mkv [19:29:59 CEST] <furq> sure [19:30:04 CEST] <furq> or you could just -i foo.srt -c:s text bar.srt [19:30:28 CEST] <auser0> sorry, where exactly and what does it mean :) ? [19:30:33 CEST] <furq> that's the entire command [19:30:42 CEST] <furq> if you already made an srt with the tags then you can just strip them [19:31:10 CEST] <furq> otherwise your command looks right [19:31:33 CEST] <auser0> oh, so that to be run on the demuxed srt to remove the alignment tags from it, then i would remux it back, right? [19:31:40 CEST] <furq> yeah [19:32:34 CEST] <auser0> nice :) [19:33:07 CEST] <auser0> i wish there was an alternative to that decoding part, on my slow machine, it slows things up [19:49:44 CEST] <auser0> furq, how do i get rid of ErrorDetectionType : Per level 1 [19:49:44 CEST] <auser0> that happens when i output to an .mkv, i see it in mediainfo for example [20:02:44 CEST] <another> why do you want to get rid of error detection? [20:07:32 CEST] <auser0> i never seen it before, so i thought it would be cleaner to get rid of it [20:08:17 CEST] <JEEB> you might want to go ask mediainfo people what that actually means [20:08:39 CEST] <auser0> okay, will do that too [20:09:49 CEST] <auser0> i noticed something, if i use the ffmpeg method with -f lavfi and -bsf:v command line, the final mkv seems to have different [20:10:07 CEST] <auser0> i know you guys explained decoding not equal to encoding but : https://i.imgur.com/tF43XXE.png [20:10:10 CEST] <furq> auser0: -write_crc32 0 [20:10:17 CEST] <furq> there's no reason to turn that off though [20:10:21 CEST] <auser0> ty furq [20:10:55 CEST] <auser0> the one on the right is the .mkv without the subtitles, while the right one is with the srt subs from the lavfi hack [20:11:05 CEST] <auser0> why does the video stream looks different ? [20:11:56 CEST] <furq> that's probably just mediainfo probing them differently [20:12:45 CEST] <auser0> oh, that probably it too [20:24:31 CEST] <cehoyos> auser0: Why do you think the video streams are different? [20:25:15 CEST] <auser0> only because i no longer see the encoder settings with the x264 core as shown on the right mediainfo in that png ^ [20:26:04 CEST] <auser0> i think mediainfo isn't parsing them similarly, since also the bitrate is not there in the mkv from the ffmpeg command [20:28:18 CEST] <auser0> is there a way to remove the global tags and tags for tracks ? [20:28:30 CEST] <auser0> i tried -map_metadata -1 [20:28:37 CEST] <auser0> but it seems to still have them [20:30:18 CEST] <cehoyos> Didn't you remove the encoder settings with the filter_units bitstream filter? [20:30:53 CEST] <auser0> oh, i didn't know that does that [20:30:58 CEST] <auser0> i want to keep them :) [20:31:10 CEST] <cehoyos> And (significantly more important): Doesn't "ffmpeg -i file" also show the difference? [20:31:24 CEST] <auser0> nope ffmpeg -i doesn't show any difference [20:32:36 CEST] <cehoyos> Indeed [20:33:23 CEST] <auser0> is it possible to run the -bsf:v without that other part? [20:33:32 CEST] <auser0> -bsf:v [20:33:47 CEST] <auser0> i don't want the bitrates/encoder settings to be removed if that is possible [20:38:24 CEST] <cehoyos> You could experiment and report back [20:39:05 CEST] <auser0> okay, i shall do that [20:39:42 CEST] <auser0> how about removing all tags (global and track tags) is that also possible ? [20:43:04 CEST] <auser0> cehoyos, if i wanted to output only subtitles from the .ts not an .mkv [20:43:11 CEST] <auser0> will this work : ffmpeg -i xx.ts -f lavfi -i "movie=xx.ts[out0+subcc]" -map_metadata -1 -map 1:s -c copy -c:s text xx.srt [20:43:11 CEST] <auser0> ? [20:45:37 CEST] <another> should work [20:45:53 CEST] <another> won't however solve you issue [20:46:59 CEST] <auser0> why not another? [20:47:14 CEST] <auser0> is the -bsf:v part necessary to make sure the srt subs are okay ? [20:51:01 CEST] <cehoyos> Why don't you try other values for filter_units? [20:57:21 CEST] <auser0> don't mind me, but i really not sure how it works at all hehe [20:57:33 CEST] <auser0> not sure what i need to pass it to work or even to try messing with it :p [20:58:06 CEST] <another> because your issue was with SEI information being stripped away [20:58:20 CEST] <auser0> what is SEI? [20:58:26 CEST] <auser0> is this the encoder settings? [20:58:29 CEST] <another> Supplemental enhancement information (SEI) [20:58:40 CEST] <another> yes [20:58:45 CEST] <auser0> ah [20:58:45 CEST] <another> among others [21:01:45 CEST] <auser0> i was told earlier :: if you want to strip the closed captions then there's a filter_units or so bit stream filter by JEEB [21:02:27 CEST] <auser0> i did try right now to do -c:s text without the -bsf:v part and compared the subtitles, they looked identical [21:02:58 CEST] <auser0> does the -bsf:v has any rule in the extracting of the CC subtitles and encoding it to srt using that lavfi hack? [21:10:52 CEST] <auser0> also, is there a way to get rid efficiently using ffmpeg from all the global and tracks tags :: https://i.imgur.com/pwZRGTi.png ? [00:00:01 CEST] --- Mon Sep 23 2019

1 0

ffmpeg-devel.log.20190921
by burek 22 Sep '19

22 Sep '19

[08:39:21 CEST] <durandal_1707> libavutil/opt.c:1203:63: warning: format specifies type 'int' but the argument has type 'int64_t' (aka 'long') [-Wformat] [09:03:47 CEST] <JEEB> someone forgot a PRId64 ? [11:03:32 CEST] <cone-341> ffmpeg 03Paul B Mahol 07master:de1d6cb34ffe: avfilter/vf_v360: fix small artifacts between corners in EAC format [11:03:32 CEST] <cone-341> ffmpeg 03Paul B Mahol 07master:76f861bbb756: avfilter/vf_v360: use lrintf() in kernel [16:37:02 CEST] <cone-627> ffmpeg 03Paul B Mahol 07master:b8dfc108a225: avfilter/vf_v360: add hammer projection [16:37:02 CEST] <cone-627> ffmpeg 03Paul B Mahol 07master:ecaef69fbb1e: avfilter/vf_v360: simplify xyz_to_ball() calculations [16:37:02 CEST] <cone-627> ffmpeg 03Paul B Mahol 07master:34046bcd5ff6: avfilter/vf_v360: simplify ball_to_xyz() calculations [17:04:42 CEST] <durandal_1707> anybody know about 2d interpolation that uses 3x3 points? [17:24:07 CEST] <durandal_1707> can i use spline for 3x3 interpolation in 2D? [17:34:04 CEST] <cone-627> ffmpeg 03James Almer 07master:1c165e19f9be: avformat/latmenc: fix context used for logging [19:09:58 CEST] <cone-627> ffmpeg 03Paul B Mahol 07master:a46ee096d1a5: avfilter/avf_showfreqs: fix check for failed allocation [20:29:58 CEST] <JEEB> rcombs: man, mp3 identification seems like it's cursed :P [20:30:11 CEST] <rcombs> are you saying it's not [20:30:28 CEST] <JEEB> not... at all :) [21:19:45 CEST] <cone-627> ffmpeg 03Paul B Mahol 07master:b342678bc47b: avfilter/vf_v360: simplify dfisheye_to_xyz() calculations [21:37:31 CEST] <cone-627> ffmpeg 03Paul B Mahol 07master:073d22696e3e: avfilter/vf_v360: simplify xyz_to_dfisheye() calculations [21:37:32 CEST] <cone-627> ffmpeg 03Paul B Mahol 07master:a1391cb942f8: avfilter/vf_v360: fix comparison with float value [21:47:16 CEST] <cehoyos> rcombs: Which samples are currently failing? [21:51:20 CEST] <rcombs> cehoyos: https://h.rcombs.me/mp3-misdetect.zip [21:58:04 CEST] <cone-627> ffmpeg 03James Zern 07master:ba24b24aab3b: tools/patcheck: remove gcc-2.95 compat check [22:15:57 CEST] <cehoyos> The misdetection is not a regression, it always failed. [22:16:17 CEST] <cehoyos> (since ten years) [22:29:44 CEST] <rcombs> right [22:34:43 CEST] <cehoyos> The file starts with 50k non-mp3, non-mpegps data. If fed with 64k, mp3 probe returns 51 [22:35:00 CEST] <cehoyos> mpeg probe returns 52, this seems wrong to me. [22:37:36 CEST] <cehoyos> Variable vid == 9, all other variables are 0 [23:35:41 CEST] <Lynne> philipl: ping [00:00:00 CEST] --- Sun Sep 22 2019

1 0

ffmpeg.log.20190921
by burek 22 Sep '19

22 Sep '19

[00:10:23 CEST] <phobosoph> hi [00:11:50 CEST] <phobosoph> I got an issue with ffmpeg and Hikvision IPCamera stream [00:11:55 CEST] <phobosoph> any ideas how to debug this? :/ [00:12:03 CEST] <phobosoph> I got the ffmpeg invocation, I see lots of stuff logged [00:14:05 CEST] <cehoyos> Would you like to share lots of stuff with us? [00:14:15 CEST] <phobosoph> yes [00:14:16 CEST] <phobosoph> :) [00:26:24 CEST] <phobosoph> ffmpeg -stats -loglevel quiet -err_detect ignore_err -rtsp_transport tcp -i rtsp://user:pass@ip:554/Streaming/Channels/1 -map_metadata -1 -metadata application=datarhei/Restreamer -metadata server=NGINX-RTMP -f flv rtmp://127.0.0.1:1935/hls/live.stream [00:27:39 CEST] <phobosoph> frame= 26 fps=4.2 q=31.0 size= 669kB [00:27:45 CEST] <phobosoph> so I get not more than 5fps [00:27:50 CEST] <phobosoph> and bandwidth is also damn lwo [00:27:51 CEST] <phobosoph> *low [00:28:02 CEST] <phobosoph> but I can connect to the ipcam with e.g. VLC player and see a fullhd stream [00:28:12 CEST] <phobosoph> also the main stream is configured to be full hd 30fps [00:28:18 CEST] <phobosoph> something is strange with the stream and ffmpeg doesn't like it [00:28:20 CEST] <kepstin> that command's re-encoding the video, so it seems like you don't have enough cpu power to encode the video with the requested codecs (which you haven't specified, so it's using random defaults) [00:28:46 CEST] <phobosoph> kepstin: when I check the cpu load and it is not 100%, would this indicate the device is fast enough for this? [00:29:22 CEST] <BtbN> is one core at 100%? [00:29:48 CEST] <phobosoph> it fluctuates a lot :/ in htop all cores got 80% and then back to 5% [00:29:49 CEST] <kepstin> default video encoder with fpv is flv1, lol [00:29:51 CEST] <phobosoph> sometimes 100% [00:29:57 CEST] <kepstin> prowith flv* [00:30:05 CEST] <kepstin> which probably isn't even multithreaded [00:30:06 CEST] <phobosoph> hm, I am using restreamer which uses ffmpeg internally (and ffmpeg does the job basically) [00:30:14 CEST] <BtbN> the hell even is that [00:30:23 CEST] <kepstin> some old piece of junk [00:30:24 CEST] <phobosoph> but ffmpeg should be used in copy mode :O [00:30:30 CEST] <phobosoph> copy is selected, not re-encode [00:30:40 CEST] <BtbN> in the comment you pasted, it's very much not copying [00:30:41 CEST] <phobosoph> :( it should just pass the h264 stream of the ipcam [00:30:43 CEST] <phobosoph> oh [00:30:44 CEST] <kepstin> phobosoph: you didn't specify copy anywhere in your command [00:30:47 CEST] <kepstin> so it won't do it [00:30:48 CEST] <phobosoph> aha [00:30:51 CEST] <kepstin> default is re-encode [00:31:44 CEST] <kepstin> stick a "-c copy" in there after the input files, before the output file. [00:32:03 CEST] <phobosoph> oh, what output file... right [00:32:13 CEST] <phobosoph> there is no output file? why is restreamer using rtmp as output? [00:32:25 CEST] <BtbN> Because of lack of alternatives. [00:32:46 CEST] <kepstin> well, by "output file" i mean "the thing on the command line that specifies where the output is going" [00:33:02 CEST] <phobosoph> repeatToLocalNginx] Spawned: ffmpeg -stats -loglevel quiet -err_detect ignore_err -rtsp_transport tcp -i rtsp://user:pass@ip:554/Streaming/Channels/1 -map_metadata -1 -metadata application=datarhei/Restreamer -metadata server=NGINX-RTMP -f flv rtmp://127.0.0.1:1935/hls/live.stream [00:33:09 CEST] <phobosoph> repeatToLocalNginx? [00:33:10 CEST] <kepstin> and by "input file" i mean "the -i option and its argument" [00:33:14 CEST] <phobosoph> ah [00:33:48 CEST] <phobosoph> kepstin: so nginx somehow should serve that stream then, prepared by ffmpeg :/ [00:34:00 CEST] <kepstin> that repeatToLocalNginx thing has nothing to do with ffmpeg, it's probably just part of some log output from something else [00:34:02 CEST] <phobosoph> fittlingly the player doesn't play it back because it can't find a livestream file [00:34:04 CEST] <phobosoph> ah [00:34:04 CEST] <phobosoph> yes [00:35:24 CEST] <kepstin> anyways, assuming you add the -c copy option, and assuming the input is something reasonable like h264, that seems like a perfectly fine command to convert input from an ip cam and send it to an rtmp streaming server. [00:35:51 CEST] <kepstin> (nginx-rtmp is a decent choice of streaming server, which can generate e.g. HLS manifests for web players) [00:36:17 CEST] <BtbN> ffmpeg is generally better at generating HLS playlists though [00:36:50 CEST] <kepstin> yeah, if you just want HLS then use ffmpeg's hls output on disk somewhere, and use e.g. nginx as a static http file server [00:37:18 CEST] <kepstin> (or even hls output to a tmpfs or something if you're booted off an sd card or something ridiculous like that) [00:39:17 CEST] <phobosoph> so one CPU core is currently at 100% [00:39:23 CEST] <phobosoph> although copy is enabled [00:39:31 CEST] <phobosoph> so restreamer uses the wrong commands [00:39:32 CEST] <BtbN> You probably did it wrong then [00:39:32 CEST] <phobosoph> damn [00:39:54 CEST] <BtbN> What even is restreamer? [00:40:12 CEST] <BtbN> Do you mean https://restream.io/ ? [00:40:22 CEST] <phobosoph> https://datarhei.github.io/restreamer/ [00:40:24 CEST] <phobosoph> that one [00:40:34 CEST] <phobosoph> https://github.com/datarhei/restreamer [00:41:14 CEST] <BtbN> So it's just a Docker-container that uses a bunch of otherwise unrelates software to setup nginx-rtmp? [00:42:36 CEST] <phobosoph> yes [00:42:40 CEST] <phobosoph> it is quite comfortable [00:43:00 CEST] <phobosoph> but it doesn't work :/ and I rather use some manual ffmpeg to get it working [00:43:01 CEST] <BtbN> seems overcomplicated to me [00:43:08 CEST] <phobosoph> so what I want: RTSP f rom IPcam to RTMP of YouTube [00:43:21 CEST] <phobosoph> with this particular IPcam Restreamer doesn't work [00:43:28 CEST] <BtbN> you can do that in a single ffmpeg command without the need for Docker at all [00:44:00 CEST] <phobosoph> wow [00:44:20 CEST] <phobosoph> until they found out/fixed the issue I would gladly do that on the CLI :D [00:44:34 CEST] <phobosoph> so the ipcam has been configured to give a youtube friendly stream [00:44:43 CEST] <phobosoph> the audio is incompatible though, it must be AAC or MP3 [00:44:53 CEST] <BtbN> I really don't see the advantage of throwing another half a dozen layers at this [00:44:57 CEST] <phobosoph> but silence would also be OK [00:44:59 CEST] <phobosoph> right [00:45:02 CEST] <BtbN> encoding aac is cheap [00:45:18 CEST] <phobosoph> what kind of command can I use then? not the one restreamer uses I guess [00:45:28 CEST] <phobosoph> I got the youtube RTMP URL with secret key [00:45:52 CEST] <BtbN> the rtsp stream as input, copy video, encode aac audio, rtmp out to youtube [00:46:10 CEST] <phobosoph> yes! [00:46:17 CEST] <phobosoph> this seems also be very lightweight to cpu [00:46:24 CEST] <phobosoph> (it runs on a raspberry pi btw, hehe) [00:46:28 CEST] <BtbN> Seems pretty much identical to what you pasted before, except output to YT [00:46:34 CEST] <phobosoph> oh [00:46:40 CEST] <phobosoph> so no copy flag needed when rtmp output is used? [00:46:47 CEST] <phobosoph> repeatToLocalNginx] Spawned: ffmpeg -stats -loglevel quiet -err_detect ignore_err -rtsp_transport tcp -i rtsp://user:pass@ip:554/Streaming/Channels/1 -map_metadata -1 -metadata application=datarhei/Restreamer -metadata server=NGINX-RTMP -f flv rtmp://127.0.0.1:1935/hls/live.stream [00:46:50 CEST] <BtbN> Well yes, you need to encode/copy correctly [00:46:53 CEST] <phobosoph> so instead -f flv I would use rtmp URL? [00:47:01 CEST] <BtbN> rtmp uses flv. [00:47:21 CEST] <phobosoph> ffmpeg -stats -loglevel quiet -err_detect ignore_err -rtsp_transport tcp -i rtsp://user:pass@ip:554/Streaming/Channels/1 [00:47:53 CEST] <phobosoph> ffmpeg -stats -loglevel quiet -err_detect ignore_err -rtsp_transport tcp -i rtsp://user:pass@ip:554/Streaming/Channels/1 -f flv rtmp://a.rtmp.youtube.com/live2/secret-key-thing [00:47:57 CEST] <phobosoph> like that? [00:48:11 CEST] <BtbN> You're still re-encoding to weird flv defaults with that [00:48:17 CEST] <BtbN> you need to copy/encode aac [00:48:22 CEST] <phobosoph> BtbN: how can I do this in ffmpeg? [00:48:30 CEST] <phobosoph> copy video, re-encode audio to AAC mono? [00:48:41 CEST] <BtbN> https://trac.ffmpeg.org/wiki/Encode/AAC [00:49:17 CEST] <phobosoph> oh, ffmpeg is so useful, it does more than just mpeg stuff, right? its name is because of legacy stuff? [00:50:02 CEST] <phobosoph> -c:a libfdk_aac -b:a 64k [00:50:07 CEST] <phobosoph> so this I add to the command [00:50:19 CEST] <phobosoph> now I need to let it copy the video - not re-encode it to youtube rtmp url [00:50:25 CEST] <BtbN> if your instance support libfdk, yes [00:51:13 CEST] <phobosoph> ah right, I check that [00:52:10 CEST] <phobosoph> BtbN: any elegant way for letting ffmpeg run and auto-restart when the connection breaks? [00:54:46 CEST] <phobosoph> wow, ffmpeg does the streaming thing now [00:54:56 CEST] <phobosoph> Past duration 0.999992 too large [00:54:57 CEST] <phobosoph> hmmm [00:55:04 CEST] <phobosoph> error while decoding MB 78 34, bytestream -7 [00:55:08 CEST] <phobosoph> then ffmpeg aborts [00:59:11 CEST] <phobosoph> BtbN: help :O [01:01:11 CEST] <another> please post the complete command line wih output to some pastebin site [01:10:50 CEST] <phobosoph> another, BtbN: https://pastebin.com/j12xeeKF [01:11:10 CEST] <phobosoph> it looks much better now, but ffmpeg warns me that it ignores the audio conversion thing [01:12:37 CEST] <phobosoph> I get very high fps now :D [01:16:16 CEST] <phobosoph> but there is still an audio issue [01:16:21 CEST] <phobosoph> youtube doesn't recognize the stream [01:16:32 CEST] <phobosoph> Codec AVOption b (set bitrate (in bits/s)) specified for output file #0 (rtmp...) [01:16:41 CEST] <phobosoph> has not been used for any stream. [01:24:25 CEST] <another> <another> please post the complete command line wih output to some pastebin site [01:25:08 CEST] <phobosoph> another: but I posted a link ! [01:25:14 CEST] <phobosoph> that's the complete command line I use [01:25:30 CEST] <cehoyos> another: There is a reason for the wording I use... [01:25:40 CEST] <cehoyos> (You know it, right?) [01:26:14 CEST] <pink_mist> phobosoph: "complete command line wih output" <-- WITH OUTPUT [01:26:15 CEST] <phobosoph> another, cehoyos: you mean not only the invocation but also what is logged afterwards? [01:26:19 CEST] <phobosoph> pardon my ignorance lol [01:29:30 CEST] <DHE> yes please. details like the exact version of ffmpeg being used, format of the inputs and outputs, and any errors/warnings may be relevant [01:30:21 CEST] <phobosoph> https://pastebin.com/pFUNWELS [01:30:23 CEST] <phobosoph> here we go [01:30:47 CEST] <phobosoph> so youtube doesn't recognize the data - and the last time this was caused by incompatible audio [01:31:00 CEST] <phobosoph> either no audio (youtube requires some kind of audio channel) or incorrectly converted audio [01:31:08 CEST] <cehoyos> Your FFmpeg version is >three years old... [01:31:44 CEST] <phobosoph> cehoyos: What would be the fastest way to get it upgraded on a raspbian? [01:31:56 CEST] <cehoyos> git clone && configure && make? [01:32:16 CEST] <cehoyos> But I wonder what your issue is now: The input has no audio, so why would the output contain audio? [01:32:23 CEST] <DHE> first this sample has only video in the source, no audio so there's no audio to output.... [01:33:18 CEST] <phobosoph> ah! [01:33:30 CEST] <phobosoph> input has no audio? but the ipcam should output audio [01:33:34 CEST] <phobosoph> I have to check that [01:33:54 CEST] <phobosoph> XD, it got a Line In [01:34:00 CEST] <phobosoph> lol, no wonder there is no audio, nothing is plugged in [01:34:10 CEST] <DHE> <voice style="mythbusters">Well there's your problem... [01:34:17 CEST] <phobosoph> can ffmpeg just inject proper still audio ? [01:34:19 CEST] <phobosoph> lol [01:34:48 CEST] <cehoyos> Only if you tell it to [01:34:50 CEST] <DHE> you can source audio from a different device/source and ffmpeg can put them together [01:35:05 CEST] <DHE> though audio sync may not be guaranteed with the video [01:36:43 CEST] <phobosoph> so just silence so youtube is happy? [01:38:52 CEST] <phobosoph> yes! yes! it works !!!! [01:38:59 CEST] <phobosoph> I added -f lavfi -i anullsrc [01:39:43 CEST] <DHE> that should do [01:40:07 CEST] <DHE> one other thing caught my eye yuvj420p - note the J [01:40:13 CEST] <DHE> I don't know if that matters though [01:41:21 CEST] <phobosoph> DHE: yuvj420p is not full hd, right? [01:41:38 CEST] <DHE> "full hd" is a matter of resolution, not colourspace [01:41:55 CEST] <phobosoph> ah XD [01:41:56 CEST] <phobosoph> I need to learn [01:42:01 CEST] <phobosoph> so much [01:42:23 CEST] <DHE> it's 1080p at 30fps, which is "full hd". but the non-J version of yuv420p is the industry standard [01:42:35 CEST] <DHE> I don't know if google's servers will convert it for you or not [01:42:44 CEST] <phobosoph> it looks okish [01:42:53 CEST] <DHE> that's probably good enough then [01:42:54 CEST] <phobosoph> hm, I hope the foggy image is from the lens :/ [01:44:04 CEST] <cehoyos> You should indeed add -pix_fmt yuv420p to get acceptable colours: It is unlikely that players correctly show yuvj420p. [01:44:18 CEST] <DHE> but then you can't use -c:v copy [01:44:25 CEST] <cehoyos> True, sorry! [01:44:40 CEST] <cehoyos> Try to change the camera configuration [01:44:48 CEST] <DHE> right now we're in the experimentation phase... [01:45:34 CEST] <phobosoph> right :) [01:45:39 CEST] <DHE> if you want to try the colourspace change, replace -c:v copy with: -c:v libx264 -preset veryfast -crf 25 -profile:v main [01:46:12 CEST] <DHE> also regardless, delete -tune zerolatency because with -c:v copy it's useless and with actual video transcoding it hurts image quality a lot and there's already a buttload of latency on youtube streaming [01:46:56 CEST] <phobosoph> is crf 25 = 25fps? [01:47:16 CEST] <DHE> it's a quality setting. higher numbers use less bandwidth but obviously lower quality. [01:47:22 CEST] <DHE> 25 is a good rough starting point [01:47:22 CEST] <phobosoph> ah, camera settings: Baseline, High profiles [01:47:54 CEST] <phobosoph> Profile: Basic profile, Main Profile and High Profile for coding are selectable. [01:48:00 CEST] <phobosoph> is this the color space thing? [01:48:17 CEST] <DHE> no... [01:48:33 CEST] <phobosoph> but for what is it used? [01:48:39 CEST] <DHE> it controls the features supported in the encoded stream, which in turn dictates the features the player needs to support [01:48:54 CEST] <DHE> eg: first gen iPhones only supported Baseline profile [01:49:24 CEST] <phobosoph> DHE: so this is irrelevant for youtube because youtube handles all this, right? [01:50:15 CEST] <DHE> I don't know. I don't youtube stream. I just know yuv420p is the the only thing guaranteed supported by players [01:51:05 CEST] <phobosoph> ah, so not yuvj420p [01:51:12 CEST] <cehoyos> I wonder if Constrained Baseline support yuvj420p... [01:51:35 CEST] <phobosoph> cehoyos: ah, this would be a way to enforce it [01:51:48 CEST] <cehoyos> It could be, worth a try imo [01:52:03 CEST] <cehoyos> Good night! [01:52:04 CEST] <phobosoph> H.264+ is enabled [01:52:08 CEST] <phobosoph> thanks and good night [01:52:09 CEST] <DHE> main and high are supposed to already be 4:2:0. there are discrete 'High 4:2:2', etc profiles [01:53:51 CEST] <phobosoph> Any nice way to let ffmpeg run in the background so I can exit the raspberry SSH session? :) [01:53:59 CEST] <phobosoph> and will ffmpeg auto-retry when the connection goes down for a moment? [01:54:56 CEST] <DHE> I like to use `screen`. Other people like tmux [01:55:13 CEST] <DHE> as for outage handling, i suggest a shell script infinite while loop.. [01:55:22 CEST] <DHE> while true; do [ffmpeg commandline here] ; sleep 5 ; done [01:57:27 CEST] <phobosoph> while true; do [ [01:57:32 CEST] <phobosoph> it doesn'T like the stuff after [ :) [01:57:40 CEST] <phobosoph> it is a multiline ffmpeg thing with \ [01:57:42 CEST] <DHE> the [ ] doesn't literally get inserted [01:57:54 CEST] <DHE> while true; do ffmeg -i ... ; sleeop 5 ; done [01:57:55 CEST] <phobosoph> ahhh [01:58:04 CEST] <DHE> fix obvious typos [01:58:35 CEST] <DHE> but if you're running this on a pi... maybe avoid transcoding for now [01:59:01 CEST] <phobosoph> thx [01:59:04 CEST] <phobosoph> it works nice [01:59:09 CEST] <phobosoph> it works wonderfully [03:22:49 CEST] <classsic> somebody know how setup RGBX_8888 pixel format? [03:22:50 CEST] <classsic> *hi [03:26:17 CEST] <another> what do you mean by setup? [03:26:29 CEST] <another> do you have a source? [03:26:30 CEST] <classsic> like -pix_fmt yuv420p [03:26:49 CEST] <classsic> rtsp input camera [03:30:35 CEST] <another> that's a little vague [03:30:47 CEST] <another> there are tons of pixel formats in ffmpeg. see https://www.ffmpeg.org/doxygen/trunk/pixfmt_8h.html#a9a8e335cf3be472042bc9f… [03:31:20 CEST] <another> depending on your use case you might prefer one over the other [03:33:04 CEST] <classsic> thanks! [03:33:07 CEST] <classsic> I will try [03:36:38 CEST] <DHE> you can also use ffmpeg -pix_fmts # to get a list [03:36:51 CEST] <DHE> similarly there's also -codecs, -formats, and a few other options for info dumps [03:38:35 CEST] <kepstin> "RGBX_8888" sounds like it would correspond to either rgb0 or 0bgr in ffmpeg (depending on endian) [03:40:46 CEST] <classsic> I try to get hw acceleration on chrome [03:41:20 CEST] <classsic> GpuMemoryBuffers Status [03:41:40 CEST] <classsic> and RGBA_8888 GPU_READ, SCANOUT [03:41:59 CEST] <classsic> the other formats .---->software decode only [03:43:14 CEST] <classsic> so...I need to set up this pix_fmt, but really, IŽm not sure. [04:04:36 CEST] <another> hw accel for what codec? [04:05:43 CEST] <another> you probably only get hw accel for yuv420 [05:12:11 CEST] <grkblood13> I have a atsc broadcast stream that I want to send through ffmpeg to ensure that it's properly encoded before saving it. my command is cat /dev/dvb/adapter0/dvr0 | ffmpeg -f mpegts -i pipe:0 -c:v mpeg2video -c:a aac -f mpeg -c copy /tmp/test5.out [05:12:36 CEST] <grkblood13> this is crapping out almost instantly with [mpeg2video @ 0x557c60ce90] Invalid frame dimensions 0x0. [05:12:36 CEST] <grkblood13> Last message repeated 28 times [05:12:41 CEST] <grkblood13> any idea why? [05:20:12 CEST] <DHE> because you joined the stream at a non-keyframe so the mpeg decoder is not happy yet [05:20:25 CEST] <DHE> this is normal and can be ignored [05:20:49 CEST] <nicolas17> you're re-encoding it to mpeg2? [05:21:01 CEST] <nicolas17> or does -c copy override the previous -c:v? [05:21:23 CEST] <DHE> seems like it. ATSC around here here tends to be mpeg2 anyway [05:29:52 CEST] <furq> i assume that -c:v mpeg2video is meant to be before -i [05:30:00 CEST] <furq> but it's picking the right decoder anyway so you could also just remove it [05:30:19 CEST] <furq> and -c:a obviously [05:30:35 CEST] <DHE> well a full transcode is necessary to decode the input for an extended period before you deem it to be "intact" [05:31:04 CEST] <furq> i'm assuming that based on him asking about that option for ffplay earlier [06:12:24 CEST] <exs> hi [06:12:34 CEST] <exs> I get the error [mp4 @ 0x56143fdd8bc0] Non-monotonous DTS in output stream 0:0; previous: 25406, current: 25188; changing to 25407. This may result in incorrect timestamps in the output file. [10:29:59 CEST] <cousin_luigi> Greetings. [10:30:15 CEST] Action: cousin_luigi keeps forgetting how to remove mp4 subtitles [10:30:42 CEST] <cousin_luigi> Something to do with -map no doubt, but how? [10:32:29 CEST] <JEEB> don't map them? [10:32:45 CEST] <JEEB> -map 0:v -map 0:a [10:35:12 CEST] <cousin_luigi> JEEB: ffmpeg -i foo.mkv -acodec copy -vcodec copy -map 0:v -map 0:a bar.mp4 <- what am I doing wrong? [10:38:11 CEST] <JEEB> then it's chapters since those get inserted as a text stream [10:39:01 CEST] <JEEB> since effectively that's how mov specified chapters :) [10:39:20 CEST] <JEEB> antways don't remember how to ignore chapters [10:39:23 CEST] <cousin_luigi> JEEB: I see EIA-708 [10:39:39 CEST] <JEEB> that's in the video then most likely? [10:39:45 CEST] <JEEB> within the video stream [10:39:48 CEST] <cousin_luigi> Yes. [10:39:53 CEST] <cousin_luigi> How can I remove those? [10:40:17 CEST] <JEEB> yes, stripping that is probably less simple although there is a header bit stream filter [10:40:34 CEST] <JEEB> which might strip that [10:40:50 CEST] Action: cousin_luigi is all ears [10:42:52 CEST] <JEEB> see filter_units example I guess in ffmpeg-all.html ? [10:43:04 CEST] <JEEB> that might be what you need [10:43:40 CEST] <cousin_luigi> JEEB: "Unknown bitstream filter filter_units" [10:43:55 CEST] <cousin_luigi> I'm on ffmpeg 3.something if that's a factor. [10:44:08 CEST] <JEEB> probably yes. have fun [10:44:20 CEST] <cousin_luigi> :'( [10:44:33 CEST] <JEEB> probably worth just locally building in your home directory :p [10:44:39 CEST] <JEEB> anyways, afk [12:06:05 CEST] <Spring> has anyone done benchmarks on ffmpeg's APNG encoding vs other programs like APNG Assembler? [12:06:15 CEST] <Spring> in terms of speed that is [12:37:11 CEST] <cehoyos> APNG Assembler does not accept random input, so the benchmark will not be useful afaict [13:13:11 CEST] <Spring> cehoyos, I'm only wondering about PNG frames -> APNG benchmarks. [13:13:55 CEST] <Spring> assuming that's what you mean about random input [13:14:15 CEST] <Spring> in that it doesn't accept non-PNG inputs [13:15:12 CEST] <Spring> I'll be trying some myself later [13:22:16 CEST] <cehoyos> Then let me rephrase: APNG Assembler cannot do what FFmpeg can do and FFmpeg cannot do what APNG Assembler can do [13:22:52 CEST] <cehoyos> If APNG Assembler does something really stupid, FFmpeg could still be faster but it is unlikely [13:40:58 CEST] <Spring> so what kind of support for APNG encoding does ffmpeg have in that case? [13:43:24 CEST] <Spring> in my previous searches I couldn't find a docs page about it, just a note in a changelog [13:45:38 CEST] <cehoyos> FFmpeg supports apng encoding, APNG Assembler supports "png to apng remuxing" which is not supported by FFmpeg [13:47:52 CEST] <cehoyos> (But as said, I don't know APNG Assembler, so maybe they only support decoding png and encoding apng, but I consider that unlikely) [13:51:43 CEST] <Spring> Takes a good while to output APNGs with APNG Assembler using larger input images, whether color palette optimizations are enabled or not which was the reason I originally wondered. Wouldn't remuxing be a lot quicker? [14:25:29 CEST] <cehoyos> Then feel free to compare with FFmpeg [14:28:39 CEST] <furq> Spring: apngasm defaults to a different (slower) DEFLATE encoder [14:28:42 CEST] <furq> so that's probably why it's slower [14:29:31 CEST] <furq> you'd hope it would be a bit faster with -z0 but probably not appreciably so [14:29:33 CEST] <greysondn> Huh [14:29:37 CEST] <greysondn> "ffmpeg.pastebin.coms server IP address could not be found." [14:30:01 CEST] <greysondn> Not what I came here for, but probably worth mentioning. [14:30:11 CEST] <Spring> furq, so not much difference between the compression types? Huh. Do you know if changing the number of iterations helps? [14:30:22 CEST] <furq> well yeah fewer iterations will be faster [14:30:30 CEST] <furq> and there is a big difference between the compression types [14:30:45 CEST] <furq> in terms of speed at least [14:30:56 CEST] <furq> they're all DEFLATE so you won't gain much from using zopfli in terms of size [14:31:29 CEST] <furq> but it will heat the room up a bit more [14:32:09 CEST] <furq> -z0 is zlib deflate which is what ffmpeg uses, so that would be the fair comparison [14:32:09 CEST] <cehoyos> greysondn: Where is this url mentioned and how could we fix it? [14:32:20 CEST] <furq> it's in the chanserv welcome message iirc [14:32:35 CEST] <Spring> you mentioned -z0 isn't appreciably faster than -z1 (the default) but -z2 would be, have I understood that right? [14:32:44 CEST] <cehoyos> I must be blind because I don't see it... [14:32:59 CEST] <greysondn> Yes. The channel topic message, to all appearances. [14:33:27 CEST] <furq> Spring: -z0 is faster than -z1 is much faster than -z2 [14:33:27 CEST] <cehoyos> You have joined the channel #ffmpeg. [14:33:35 CEST] <furq> and no it's not the topic, chanserv sends you a notice when you join [14:33:36 CEST] <cehoyos> The channel topic is "Welcome to the FFmpeg USER support channel | Development channel: #ffmpeg-devel | Bug reports: http://bit.ly/cqvkhs | Wiki: https://trac.ffmpeg.org/ | FFmpeg 4.2 is released | This channel is publicly logged". [14:33:41 CEST] <furq> it'll probably be in your status window [14:33:45 CEST] <cehoyos> The topic was set by michaelni! [14:33:52 CEST] <cehoyos> Channel modes: no colors allowed, no messages from outside, topic protection [14:33:58 CEST] <cehoyos> This channel was created on 11/26/06 7:42 AM. [14:34:04 CEST] <cehoyos> Sorry, I still don't see it [14:34:14 CEST] <greysondn> https://imgur.com/XHmU6kF [14:34:17 CEST] <Spring> I see, thanks. I'll run some tests a bit later. [14:34:20 CEST] <greysondn> I'm in web client. [14:34:28 CEST] <furq> https://clbin.com/5s3tm [14:35:06 CEST] <greysondn> So... That's not working, I'mma just use normal pastebin. I'm getting what I'm dealing with together right now. That was just "I joined and woops". [14:36:08 CEST] <furq> someone with ops needs to /msg chanserv set #ffmpeg entrymsg ... [14:36:17 CEST] <furq> or something along those lines, i've never run a channel on this network [14:36:18 CEST] <cehoyos> It appears I am only operator in -devel, [14:37:16 CEST] <furq> anyway yeah any pastebin is fine [14:40:29 CEST] <greysondn> So. I've got some python code ( https://pastebin.com/EuycVQv7 ) that calls two ffmpeg subprocesses that is giving me a headache. The first of the two just opens a video file and reads the raw frames from the video into python for me. It's just fine. I then do some work on those using python, and output them to the second one where things seem to be [14:40:29 CEST] <greysondn> going wrong. That subprocess can't run in realtime, but for whatever reason tries to. I have it muxing the original audio and the new frames (as it converts them) into a new video, and I'd really prefer to get this out of trying so hard to be realtime if at all possible because of the queue blocking when it fills. [14:41:07 CEST] <greysondn> The significant part I figure maybe you could help me with is just the arguments I'm passing to ffmpeg here. [14:41:07 CEST] <cehoyos> But his isn't python-help;-( [14:41:11 CEST] <greysondn> I don't expect you to know any python. [14:41:26 CEST] <greysondn> If I need to translate those into the ffmpeg calls going on, I can. [14:41:28 CEST] <cehoyos> Does anybody know how to change the NOTICE? [14:41:39 CEST] <durandal_1707> NOT ICE [14:41:45 CEST] <cehoyos> Then translate it and post the complete, uncut console output [14:42:05 CEST] <greysondn> Sure. There's going to be a "magic" in there, basically. [14:42:05 CEST] <furq> cehoyos: /msg chanserv set #ffmpeg entrymsg [14:42:14 CEST] <cehoyos> No [14:42:38 CEST] <cehoyos> Or yes [14:42:42 CEST] <cehoyos> there is also a notice command [14:43:36 CEST] <cehoyos> Invalid command [14:44:23 CEST] -cehoyos:#ffmpeg- When [14:52:08 CEST] <cehoyos> Sorry, I was unable to find out how to change the notice [14:52:09 CEST] <greysondn> Should be this [14:52:10 CEST] <greysondn> https://pastebin.com/7ReF1EXq [14:52:24 CEST] <greysondn> Note that I've just ran my python and dumped from there, but should be correct. [14:53:00 CEST] <greysondn> That WTF between the pipes is just where the frame edit goes. Takes in raw frames, spits out raw frames. [14:53:21 CEST] <cehoyos> This is unreadable: Assuming you have a problem with the second command, save the rawvideo stream in a file and use cat [14:54:56 CEST] <greysondn> I'm puzzled at your response. [14:55:50 CEST] <greysondn> So you have the arguments given to it and the problem I'm facing. What are you missing still? [14:55:55 CEST] <cehoyos> Don't worry: Everybody is today;-) [14:56:33 CEST] <cehoyos> What I see is the console output of two ffmpeg commands interleaved with each other: Sorry that I believe parsing this is simply too difficult [14:56:44 CEST] <greysondn> Okay. [14:56:46 CEST] <greysondn> That makes more sense. [14:56:52 CEST] <greysondn> Didn't even consider it TBH [14:56:57 CEST] <greysondn> What I can do is just trash the first one. [14:57:00 CEST] <greysondn> Would that help? [14:57:14 CEST] <greysondn> It seems to be what you're asking me to do anyway. [14:58:34 CEST] <furq> greysondn: that looks like it finished ok [14:58:39 CEST] <furq> is the output actually broken [14:58:50 CEST] <greysondn> It did. It informed me the queue blocked at least once, however. [14:59:01 CEST] <greysondn> I would assume that means it dropped a frame. [14:59:13 CEST] <furq> no the status line would tell you if it dropped frames [14:59:27 CEST] <furq> the thread message queue blocking is just a warning to tell you that it's waiting for an input frame [14:59:37 CEST] <furq> or waiting to output a frame [14:59:57 CEST] <greysondn> So I would have to wait if it was realtime, but I don't have to wait if I'm just processing a video with non-realtime constraints? [15:00:14 CEST] <furq> yeah that all looks fine to me [15:00:31 CEST] <greysondn> Dead honest, didn't even realize it was interleaved. [15:00:39 CEST] <greysondn> I thought it was showing the first and then the second when I looked at it. [15:01:33 CEST] <furq> i'm surprised it deals with a full pipe buffer so neatly [15:01:36 CEST] <furq> i would have expected that to break [15:01:38 CEST] <cehoyos> Sorry that I find this extremely unlikely (from just looking at the output and your message) [15:01:49 CEST] <cehoyos> Why? [15:02:03 CEST] <cehoyos> I believe this is a common way of using FFmpeg [15:02:27 CEST] <greysondn> It supports pipes. The only thing weird I've done is do image editing via python, but there's no reason I can't. [15:02:41 CEST] <greysondn> Just have to make it hook into the pipes, which I've done. [15:02:43 CEST] <furq> well mostly because the subprocesses are being called from python [15:02:53 CEST] <furq> but yeah i guess if you do it properly it's no different from doing ffmpeg | foo | ffmpeg [15:03:32 CEST] <Spring> For the curious ran some simple timing benchmarks on 7zip (1 and 15 iterations) vs zlib compression with both optimizations enabled/disabled. 7zip with 1 iteration was literally the same speed as zlib with optimizations enabled so I decided to use that. Saved 29 seconds over what I was using prevously. [15:03:35 CEST] <greysondn> The major fly in the ointment, if you ever hear someone trying this, is that your pipe buffer in python must be at least the size of a frame. [15:03:52 CEST] <greysondn> The default pipe buffer is quite small. [15:03:57 CEST] <Spring> *7zip with 1 iteration and optimzations enabled [15:03:58 CEST] <furq> yeah that's true of pipe buffers in general [15:09:55 CEST] <greysondn> Sorry, that's a test clip about seven minutes long. I'm just watching it now to make sure it did actually come out clean. 2 minutes and so far, so good. [15:10:44 CEST] <cehoyos> Spring: Did you compare FFmpeg with something else? [15:12:04 CEST] <Spring> Haven't tried ffmpeg yet. Since reducing the processing time with APNG Assembler though I'm in a much better spot. [15:15:34 CEST] <greysondn> Far as I can tell it did. [15:16:10 CEST] <greysondn> Might hear from me on a later date, but as it is, it appears I should apologize for wasting time just to find out everything's fine. [15:23:07 CEST] <ossifrage> Is there a way to get ffplay to visualize the h.264 block sizes? [15:24:13 CEST] <greysondn> I'mma duck out of your hair. Thanks for your time and help! [18:24:10 CEST] <cpusage> Hi. I'm trying to strip/delete/purge metadata from a bunch of older .wmv files I have [18:24:12 CEST] <cpusage> using [18:24:15 CEST] <cpusage> ffmpeg -i input.wmv -map_metadata -1 -c copy output.wmv [18:24:34 CEST] <cpusage> but I get the following error: [18:24:35 CEST] <cpusage> [asf @ 000002f0a1a58700] Timestamps are unset in a packet for stream 0. This is deprecated and will stop working in the future. Fix your code to set the timestamps properly [18:24:48 CEST] <cpusage> and the output file is larger than the input file [18:25:21 CEST] <cpusage> any ideas on what I can do to properly do this? [19:05:37 CEST] <cehoyos> cpusage: Please provide the command line you tested together with the complete, uncut console output. [19:17:32 CEST] <cpusage> cehoyos would this suffice? [19:17:34 CEST] <cpusage> https://pastebin.com/V7NpbRf9 [19:18:14 CEST] <cehoyos> What problem do you have with the output file? [19:18:22 CEST] <cpusage> resulting (output) wmv/asf file is bigger than the original (input), and no playback [19:18:53 CEST] <durandal_1707> playback in what? [19:19:22 CEST] <cpusage> the resulting wmv doesn't play in any media player. It's stated as playing, but is just paused on a frame [19:19:31 CEST] <cehoyos> Please provide an input file [19:19:39 CEST] <cehoyos> That allows to reproduce the issue [22:24:05 CEST] <Agafnd> is there a way get ffprobe to output just the resolution and framerate of a file? [22:25:29 CEST] <JEEB> use a parse'able output [22:25:31 CEST] <JEEB> like -of json [22:25:44 CEST] <JEEB> and then something like -show_format -show_streams [22:25:52 CEST] <JEEB> then you can parse the stdout [22:26:41 CEST] <Agafnd> ah yeah. I think that'll work [23:01:13 CEST] <auser0> hello everyone, what does this command do ; [23:01:14 CEST] <auser0> ffmpeg -i test.ts -f lavfi -i "movie=test.ts[out0+subcc]" -map 0 -map 1:s -c:a copy -c:s srt final.mkv [23:01:51 CEST] <auser0> i have some .ts that has some CC subtitles that i want to convert to SRT and mux into that mkv, while the audio and video streams to be untouched so no encoding of both of them [23:03:21 CEST] <JEEB> it makes two inputs. one is the actual file, and the other utilizes the movie lavfi device that has a hack to create a subtitle stream for possible closed captions [23:03:33 CEST] <JEEB> then it maps all streams from input 0 (the first one, thus the .ts file itself9 [23:03:52 CEST] <JEEB> then it maps the subtitle stream from input 1 (the second one) [23:04:08 CEST] <JEEB> and finally it sets the *audio* codec to copy and *subtitle* codec to srt [23:04:16 CEST] <JEEB> for some reason it re-encodes the video stream? [23:04:37 CEST] <JEEB> if you want to strip the closed captions then there's a filter_units or so bit stream filter [23:04:48 CEST] <JEEB> no actual re-encoding needed [23:04:58 CEST] <auser0> thanks JEEB [23:05:04 CEST] <JEEB> https://www.ffmpeg.org/ffmpeg-all.html#filter_005funits [23:05:05 CEST] <auser0> i don't want no re-encoding at all [23:05:23 CEST] <JEEB> see the example for filter_units on how to utilize it [23:05:37 CEST] <JEEB> and switch the codec definitions basically to -c copy -c:s srt [23:05:51 CEST] <JEEB> (copy everything, set subtitle codec to srt) [23:07:17 CEST] <auser0> ah, i tried that before but it ignored the subtitles [23:07:28 CEST] <auser0> would that lavfi device hack thing works fine? [23:07:45 CEST] <auser0> can't seem to understand how that filter_untis work [23:09:11 CEST] <auser0> how do i re-edit that lavfi device to not re-encode the video JEEB ? [23:09:19 CEST] <JEEB> it can't [23:09:28 CEST] <JEEB> that's why you ignore everything else from the lavfi input [23:09:55 CEST] <auser0> so for that hack to work, it will need to re-encode the video first, right? [23:11:13 CEST] <cehoyos> No [23:11:21 CEST] <cehoyos> It will need to decode the video [23:11:33 CEST] <cehoyos> If you need the subtitles as srt, you currently have to decode video [23:11:49 CEST] <auser0> ah thanks [23:11:52 CEST] <cehoyos> If you don't need the subtitles (as srt), you don't have to use the movie filter at all. [23:12:07 CEST] <cehoyos> If you need the subtitles as srt, you currently have to decode video *using the video filter" [23:12:17 CEST] <cehoyos> If you need the subtitles as srt, you currently have to decode video *using the movie filter* [23:12:22 CEST] <cehoyos> (Sorry, it's late here) [23:12:29 CEST] <auser0> np, thanks [23:13:17 CEST] <auser0> so in both scenarios whether i chose to use the lavfi hack or the video filter, it will need to re-encode the video [23:13:30 CEST] <cehoyos> If you want the video and the subtitles as srt, you can choose if you want both the video including the closed captions and the srt subtitles (-c copy) or the video without the closed captions but with the additional srt subtitles (-c copy and the bitstream filter) [23:13:46 CEST] <cehoyos> No, definitely not [23:14:39 CEST] <auser0> okay, sorry i am confused a little here [23:14:56 CEST] <auser0> i would like to extract the subs as srt and that SRT only to be muxed in the final mkv [23:15:08 CEST] <auser0> i don't want to keep the CC subs at all in the final output mkv [23:15:38 CEST] <cehoyos> Then you need a bitstream filter that JEED mentioned [23:15:44 CEST] <cehoyos> JEEB... [23:15:49 CEST] <auser0> would you be able show me an example of a CLI on how to do that filer method? [23:16:02 CEST] <JEEB> I linked to the example for it ;) [23:16:11 CEST] <cehoyos> No, I didn't suggest it because I didn't know it already existed [23:17:31 CEST] <cehoyos> Given that the filter will mostly be used for Closed Captions, an explicit example may help =-) [23:17:59 CEST] <auser0> ffmpeg -i XX.ts -c:v copy -bsf:v 'filter_units=pass_types=1-5' FINAL.mkv [23:19:11 CEST] <auser0> that didn't seem to convert those subtitles to SRT, the output has the audio and video streams only [23:19:52 CEST] <auser0> https://pastebin.com/raw/xbn6fwR0 [23:19:58 CEST] <auser0> maybe those subs are hardcoded ? [23:21:31 CEST] <cehoyos> The command line was supposed to remove the Closed Captions, did it work? [23:21:45 CEST] <cehoyos> The command line did (certainly) not create a srt stream [23:21:50 CEST] <cehoyos> in the output file [23:22:27 CEST] <cehoyos> And please understand that this is not the MediaInfo channel [23:23:14 CEST] <auser0> that didn't seem to work nope [23:23:27 CEST] <cehoyos> The Closed Captions are still in the output file? [23:23:35 CEST] <auser0> when i ran that command, the final MKV had no subtitles at all [23:23:41 CEST] <auser0> https://pastebin.com/raw/LKvLHdnR [23:24:16 CEST] <auser0> I was as if i am doing -c copy, only audio and video streams are muxed in the final output mkv [23:24:17 CEST] <cehoyos> Please do not (as in: never) post excerpts of your ffmpeg command, always post the complete, uncut console output [23:25:23 CEST] <auser0> sure cehoyos [23:25:24 CEST] <auser0> https://pastebin.com/raw/zpRkjyBR [23:25:27 CEST] <auser0> sorry, new here :) [23:26:00 CEST] <cehoyos> I assume you also want "-c:a copy" to not re-encode audio [23:26:45 CEST] <JEEB> (if you do not set a codec for a type of media it will get re-encoded by *something*) [23:26:52 CEST] <JEEB> (that is generally not something you want) [23:27:08 CEST] <JEEB> (thus, always explicitly set the encoder for everything unless you really know what you're getting is what you want) [23:27:10 CEST] <cehoyos> Because it both takes time and (even if not noticeable) reduces quality [23:27:58 CEST] <auser0> got it, thanks [23:28:10 CEST] <auser0> ya, when i ran it with -c:a copy, it didn't re-encode the audio [23:28:23 CEST] <auser0> what does "bsf:v" mean ? [23:28:51 CEST] <cehoyos> Call the video bitstream filter filter_units (that removes the non-video parts of the h.264 bitstream) [23:29:22 CEST] <JEEB> "bit stream filter : video" [23:29:24 CEST] <JEEB> basically [23:29:40 CEST] <auser0> thanks guys [23:30:40 CEST] <auser0> so after i ran that command for the bsf:v, what should i do next to get those srt subtitles? [23:31:05 CEST] <another> *sigh* [23:31:17 CEST] <JEEB> the lavfi thing is needed to get an AVStream out of the closed captions due to a limitation in the ffmpeg.c command line application [23:31:40 CEST] <JEEB> unfortunately people have priorities and things they want to hack on, so ffmpeg.c is not yet dynamic with streams :P [23:31:58 CEST] <auser0> ah i got it :) [23:32:03 CEST] <another> ffmpeg -i test.ts -f lavfi -i "movie=test.ts[out0+subcc]" -map 0 -map 1:s -c copy -c:s srt -bsf:v 'filter_units=pass_types=1-5' final.mkv [23:32:53 CEST] <auser0> thanks another, but here an encoding is going to take place, which i was trying to avoid hehe [23:33:18 CEST] <auser0> i thought the filter method could magically facilitate me getting the srt without re-encoding :p [23:33:43 CEST] <JEEB> only the subtitle stream is getitng mapped out of the second input [23:33:46 CEST] <JEEB> (1th input) [23:33:52 CEST] <JEEB> (-map 1:s) [23:34:02 CEST] <JEEB> also it has -c copy AND -c:s srt [23:34:10 CEST] <another> decoding != encoding [23:34:22 CEST] <JEEB> which copies all streams, and then sets encoder to srt for subtitles [23:34:37 CEST] <JEEB> so yes, if the movie filter decodes always then yes, it will decode. but it will not encode [23:34:44 CEST] <JEEB> the video stream from that isn't even mapped :P [23:35:10 CEST] <JEEB> or audio [23:37:00 CEST] <auser0> oh, a complete newbie here with ffmpeg, i appreciate the thorough explanation [23:37:15 CEST] <auser0> testing that atm [23:50:22 CEST] <auser0> i got some error : https://pastebin.com/raw/PHYzFPxM [23:53:01 CEST] <furq> auser0: replace -map 0 with -map 0:v -map 0:a [23:55:34 CEST] <auser0> thanks furq [23:55:42 CEST] <giaco> hello [23:57:17 CEST] <giaco> I have a mpeg-dash quesiton, not strictly related to ffmpeg. I am capturing a stream and converting it into mpeg-dash chunks, I then need to transfer them to a server that does the actual stream [23:58:50 CEST] <giaco> so is: dash_producer > <???> > web_server > client (dash_player) [23:58:54 CEST] <auser0> how do i add English as a language to every stream of those ? [23:59:19 CEST] <giaco> what would you use considering that the link between the producer and the web_server is not granted? rsync? [00:00:00 CEST] --- Sun Sep 22 2019

1 0

ffmpeg-devel.log.20190920
by burek 21 Sep '19

21 Sep '19

[01:36:57 CEST] <cone-821> ffmpeg 03Carl Eugen Hoyos 07master:f9f9d5943f79: lavc/cook: Remove an incorrect comment. [02:57:54 CEST] <cone-821> ffmpeg 03James Almer 07master:179e15dc9765: avformat/movenc: don't hardcode the colr box size [04:54:56 CEST] <nkkrishna> Hi Guys, i am seeing the following log lines in /var/log/messages, can you give some info about this log lines [04:55:06 CEST] <nkkrishna> [root@encoder13 nkrishnakumar]# cat /var/log/messages | grep error | grep ffmpeg [04:55:07 CEST] <nkkrishna> ffmpeg[38552] general protection ip:1701092 sp:7ffdb736e1e8 error:0 in ffmpeg[400000+1deb000] [04:55:08 CEST] <nkkrishna> ffmpeg[400000+1deb000] [05:54:03 CEST] <rcombs> so, to reopen the AVCodecParameters discussion [05:54:27 CEST] <rcombs> there's still some stuff in AVCodecContext but not codecpar that I need access to when probing [05:54:37 CEST] <rcombs> today the latest thing is coded_width/coded_height [05:55:05 CEST] <rcombs> (because for some reason some people code 2160p as 2176) [07:39:33 CEST] <rcombs> wait VDD is in tokyo in november [07:41:17 CEST] <rcombs> guess I'll be there then [08:54:46 CEST] <cone-259> ffmpeg 03Guo, Yejun 07master:ecd026a48d63: avcodec/libvpxenc: add ROI-based encoding support for VP8/VP9 [08:59:10 CEST] <cone-259> ffmpeg 03elliottk 07master:ea673a0edb4b: Delete unused branch in libaomenc [15:46:00 CEST] <Lynne> I just managed to generate an accurip v1/v2 checksum collision on accident [15:46:14 CEST] <JEEB> vOv [15:46:29 CEST] <Lynne> few bytes changed, 32 bit ieee is different, accurip checksums remain the same [16:02:24 CEST] <Lynne> but I don't get any credit for it since the change was in the last 2940 samples, and the checksums don't cover that region [17:07:38 CEST] <kepstin> iirc not checking some samples at the start/end is intentional to account for cd drives ripping from different offsets (with offset compensation, the missing samples usually get filled with digital silence) [17:14:51 CEST] <Lynne> yes, I know, but eac's crc32 doesn't ignore the very last non-empty samples [18:37:35 CEST] <infantilo> hi, need help with stderr output syntax [silencedetect @ 0642f440]... could anyone give advice plz. (sorry for terrible english :-( ) [19:37:26 CEST] <BBB> durandal_1707: closed-source vs. non-free is similar but not the same... but anyway, if 722.1 is closed-source, I'd likely ask for a committee vote [19:37:49 CEST] <BBB> if it's non-free, I'll still dislike it but I'll likely just curse and move on... can only take on so much [20:17:42 CEST] <cone-361> ffmpeg 03Guo, Yejun 07master:b2683c66b215: libavfilter/dnn: add layer maximum for native mode. [20:17:42 CEST] <cone-361> ffmpeg 03Guo, Yejun 07master:9ae42c130ca9: FATE/dnn: add unit test for layer maximum [20:17:42 CEST] <cone-361> ffmpeg 03Guo, Yejun 07master:75ca94f3cff8: libavfilter/dnn/dnn_backend_native: find the input operand according to input name [20:17:42 CEST] <cone-361> ffmpeg 03Guo, Yejun 07master:8f13a557cad2: libavfilter/dnn: support multiple outputs for native mode [20:46:47 CEST] <kierank> durandal_1707: https://tools.ietf.org/id/draft-df-stecker-expertenforum-payload-tetra-00.h… [20:46:48 CEST] <kierank> new audio codec [20:57:05 CEST] <durandal_1707> kierank: samples? [20:57:05 CEST] <JEEB> https://github.com/0p1pp1/FFmpeg/commit/67306b7beaa5dd59ce44af786efb2df9589… [20:57:15 CEST] <JEEB> anyone poked at vaapi can see this as useful? [20:58:00 CEST] <durandal_1707> BBB: already voted against inclusion of non-open source in codebase [21:01:01 CEST] <thardin> hmm tetra [21:02:08 CEST] <kierank> durandal_1707: imo there's a difference between public source code and blob [21:02:15 CEST] <cehoyos> Have you heard of this codec? [21:02:30 CEST] <kierank> cehoyos: https://en.wikipedia.org/wiki/Terrestrial_Trunked_Radio [21:05:03 CEST] <cehoyos> I was under the impression that this is a radio service, the ietf draft points to a codec: Is it really both? [21:05:18 CEST] <kierank> if you read the article it explains they have their own compression [21:05:34 CEST] <cehoyos> "own" can mean a lot;-) [21:06:40 CEST] <cehoyos> And it doesn't say "own compression", it says "ACELP" [21:19:15 CEST] <durandal_1707> https://sifcodec.com/ [21:29:59 CEST] <cone-233> ffmpeg 03Paul B Mahol 07master:048c712aa1f5: avfilter/vf_v360: improve description about s in comments [21:29:59 CEST] <cone-233> ffmpeg 03Paul B Mahol 07master:bc89434a0bec: avfilter/vf_v360: fix various artifacts with cubemaps [21:29:59 CEST] <cone-233> ffmpeg 03Paul B Mahol 07master:9def7efdbf6e: avfilter/vf_v360: adjust kernel after latest change [21:42:48 CEST] <cone-233> ffmpeg 03Paul B Mahol 07master:5c93a0a1e409: avfilter/vf_v360: add support for mirroring to ball input format [21:42:48 CEST] <cone-233> ffmpeg 03Paul B Mahol 07master:8b58d9cb474c: avfilter/vf_v360: remove uneeded array element [23:29:51 CEST] <cone-233> ffmpeg 03Andreas Rheinhardt 07master:86bef10e7aee: avcodec/tdsc: Fix undefined shifts [23:29:52 CEST] <cone-233> ffmpeg 03Andreas Rheinhardt 07master:54e1efda6d67: avcodec/g723_1dec: Fix invalid shift [23:29:53 CEST] <cone-233> ffmpeg 03Andreas Rheinhardt 07master:2f289ec914cc: avcodec/ac3enc: Fix invalid shift [23:29:54 CEST] <cone-233> ffmpeg 03Soft Works 07master:667df60b1474: avutil/opt: Fix printing integer option defaults [23:29:55 CEST] <cone-233> ffmpeg 03Soft Works 07master:9e0a071edec9: avutil/opt: Print out numeric values of option constants [23:29:56 CEST] <cone-233> ffmpeg 03Moritz Barsnick 07master:6390f52ac799: avformat/aiffdec: parse replaygain metadata [23:29:57 CEST] <cone-233> ffmpeg 03Moritz Barsnick 07master:270f94e132b6: avformat/rtpenc_mpegts: copy metadata to mpegts sub-muxer [23:29:58 CEST] <cone-233> ffmpeg 03Michael Niedermayer 07master:83f2555e5ff5: avcodec/utils: Check channels fully earlier [23:29:59 CEST] <cone-233> ffmpeg 03Michael Niedermayer 07master:87a7fc692d8b: tools/target_dec_fuzzer: fuzz channels and sample_rate too [23:30:00 CEST] <cone-233> ffmpeg 03Andreas Rheinhardt 07master:e6018fda14d7: avutil/encryption_info: Don't pass NULL to memcpy [23:30:01 CEST] <cone-233> ffmpeg 03Moritz Barsnick 07master:d214f216110f: avformat/hashenc: rearrange options definition [23:30:02 CEST] <cone-233> ffmpeg 03Moritz Barsnick 07master:666b4278811f: avformat/hashenc: use an array of hashes [23:30:03 CEST] <cone-233> ffmpeg 03Moritz Barsnick 07master:2f87c9f646f3: avformat/hashenc: add streamhash muxer [00:00:00 CEST] --- Sat Sep 21 2019

1 0

ffmpeg.log.20190920
by burek 21 Sep '19

21 Sep '19

[11:33:10 CEST] <termos> I'm having this problem https://ffmpeg.org/pipermail/ffmpeg-user/2016-April/031841.html with two input devices of the same name. Is there no way to separate them when capturing from the command line [11:38:45 CEST] <SimAV> termos, I'm using decklink cards too, but I didn't have two identical in the same PC yet. [11:40:06 CEST] <termos> hmm yes I didn't have a problem with just one card but now I have two DeckLink Duo cards and they just show up as DeckLink Duo (1) etc for both of the cards [11:40:32 CEST] <SimAV> I think I can remember to address one of them not by the "name", but I couldn't find evidence for that quickly... [11:41:08 CEST] <SimAV> termos, "ffmpeg -format_code pal -video_input composite -f decklink -i 1 ..." [11:41:42 CEST] <SimAV> that's from a bash_history from 2018. So there should be a way of addressing them using a number. [11:42:17 CEST] <SimAV> but currently I'm addressing them by name "ffmpeg -y -format_code Hi50 -video_input sdi -f decklink -i 'DeckLink Mini Recorder 4K'" [11:42:36 CEST] <SimAV> I don't remember whether this change was necessary... [11:42:51 CEST] <termos> interesting! will try that out [11:44:20 CEST] <termos> seems like it's not longer supported to get them by the index [11:44:42 CEST] <SimAV> ok, then it is clear why I changed that... :D [11:47:31 CEST] <SimAV> if (display_name && !strcmp(name, display_name) || unique_name && !strcmp(name, unique_name)) { [11:47:37 CEST] <SimAV> it seems as if there is a unique_name [11:49:35 CEST] <termos> yes i was looking at something similar when it lists the devices https://github.com/FFmpeg/FFmpeg/blob/master/libavdevice/decklink_common.cp… [11:52:37 CEST] <SimAV> but this suggests that the unique name should be printed, as soon as it is reported by the decklink SDK? [11:59:58 CEST] <SimAV> termos, https://www.blackmagicdesign.com/developer/support/faq/desktop-video-develo… [12:00:27 CEST] <SimAV> and from the Decklink SDK Manual: BMDDeckLinkDeviceHandle ; String ; String representing an unique identifier for the device. The format of the string is RevisionID:PersistentID:TopologicalID. [12:01:41 CEST] <SimAV> I haven't found yet how to get that... [12:04:28 CEST] <SimAV> termos, did you test whether "DeckLink Duo" and "DeckLink Duo (1)" work? [12:04:55 CEST] <SimAV> or "DeckLink Duo (1)" and "DeckLink Duo (2)" ? [12:10:30 CEST] <SimAV> termos, https://forum.blackmagicdesign.com/viewtopic.php?f=12&t=36862#p216620 lists "output of the DeviceList sample code" [12:10:59 CEST] <SimAV> do you have that? It lists Persistent ID and Topological ID, so only Revision ID would be missing [12:11:45 CEST] <termos> Oh Ill check! Looks promising [12:17:29 CEST] <snooky> moin [12:38:49 CEST] <termos> seems like these unique ids are not exposed in ffmpeg, I use DeckLink Duo (n) for port n on that card, but they are all showing up twice since I have two cards [12:39:04 CEST] <termos> so when I specify 'DeckLink Duo (2)' I don't know which card i'm accessing port 2 on [12:39:11 CEST] <SimAV> ok [12:42:19 CEST] <SimAV> termos, from the code I've seen so far I got the impression, that if the card has an unique_name, they SHOULD be printed. [12:42:25 CEST] <SimAV> *it [12:42:33 CEST] <SimAV> (looking at ff_decklink_list_devices) [12:43:13 CEST] <termos> hmm yes, maybe I need to build a new version to get this, because I see i'm running 3.0 :) [12:43:32 CEST] <SimAV> ah, no, its printing device_description [12:44:08 CEST] <SimAV> termos, so yes, it's not printed. [12:44:11 CEST] <SimAV> change av_log(avctx, AV_LOG_INFO, "\t'%s'\n", device_list->devices[i]->device_description); [12:44:23 CEST] <SimAV> to av_log(avctx, AV_LOG_INFO, "\t'%s'\n", device_list->devices[i]->device_name); [12:44:32 CEST] <SimAV> (or better: just add that line) [12:45:04 CEST] <SimAV> line 409 in decklink_common.cpp, in function "void ff_decklink_list_devices_legacy" [12:45:12 CEST] <SimAV> this should print the unique_name [12:46:02 CEST] <SimAV> anyway, i have to leave now. [12:46:41 CEST] <termos> thanks for the help, will look into doing something like this! it's not preferable to keep a custom ffmpeg branch with these changes but maybe it's the only way [12:47:18 CEST] <SimAV> termos, I think you just need that once [12:47:42 CEST] <SimAV> until you know your unique names... [12:47:57 CEST] <SimAV> but I must admit, I didn't read the code carefully [12:48:36 CEST] <SimAV> anyway, see you... [14:25:03 CEST] <Radiator> Hi everyone, I'm still working on the problem I had yesterday. I've done some big improvements yet I still have issues... This time I have an assert error : "Assertion pkt->size == ((-0x7fffffff - 1) / 3 *2 + (int)sizeof(AVFrame))" failed at libavformat/mux.c:931 [14:26:15 CEST] <Radiator> I looked upon that and it seems I have an "uncoded frame packet" even though I do pass it through my CodecContext. Anyone has an idea what I do wrong ? [14:28:17 CEST] <Radiator> The assertion is the following : av_assert0(pkt->size == UNCODED_FRAME_PACKET_SIZE); [15:50:34 CEST] <saml> -vsync cfr has any impact if I don't specify -r ? [15:50:41 CEST] <JEEB> yes [15:51:16 CEST] <JEEB> ffmpeg.c has some really basic "vsync" code in there [15:51:38 CEST] <JEEB> so it will look up the possible frame rate of the input and kind of attempt to output pictures at that rate [15:52:26 CEST] <saml> If input video has too high framerate, say above 60fps, I want to set output framerate to be 60. But, otherwise, I don't want to touch input video's framerate [15:52:38 CEST] <saml> is this sane? not touching input video's framerate if it's below 60 [15:52:49 CEST] <JEEB> depends on your requirements. [15:53:18 CEST] <JEEB> like if those are your requirements you can make that happen and cap the frame rate at 60Hz [15:53:20 CEST] <saml> i won't be doing HLS. output video will be played in modern browser in <video> tag [15:54:56 CEST] <saml> so, -vsync cfr alone vs. -vsync cfr with -r (different from input video's framerate) vs. -r alone [15:55:27 CEST] <JEEB> might as well use the fps video filter if you are going to start switching frame rates :) [15:55:53 CEST] <JEEB> (not sure if -vsync cfr will poke that filter or not) [15:56:39 CEST] <Radiator> Fixed my issue, I'm just retarded and sent the packet that failed after the avcodec_receive_packet [16:07:36 CEST] <cpplearner> Guys, I want to use avio_seek() into a keyframe offset, but judging from the parameters it doesn't affect `AVFormatContext->internal`. The actual reading is done by av_read_frame, and it pull buffers from `internal`. In that case, how can I properly seek to that offset? [16:18:33 CEST] <mantaalex> Hey i have a question i'm makeing video of all .jpg files in directory's [16:18:49 CEST] <mantaalex> it used to work on another server the same command i don't understand how its not working [16:18:57 CEST] <mantaalex> erroring i mean [16:19:30 CEST] <mantaalex> this is the command: [16:19:32 CEST] <mantaalex> ffmpeg -r 0.5 -pattern_type glob -i '/home/manta/cams/gang/*.jpg' -c:v libx264 /home/manta/cams/gang/gang_`date +%Y%m%d-%H%M`.mp4 [16:19:58 CEST] <mantaalex> ffmpeg version 2.8.15 install with yum on centos 7.7 [16:20:23 CEST] <mantaalex> [image2 @ 0x15ddbe0] Could not find codec parameters for stream 0 (Video: mjpeg, none(bt470bg/unknown/unknown)): unspecified size [16:20:26 CEST] <mantaalex> Consider increasing the value for the 'analyzeduration' and 'probesize' options [16:21:40 CEST] <mantaalex> nvm works now :) [16:23:03 CEST] <JEEB> I would definitely recommend building your own FFmpeg from a newer version [16:23:11 CEST] <JEEB> so much has changed from 2.8.x [16:54:53 CEST] <Radiator> Did it occured to others that a code using ffmpeg libs worked on unix based OS but not on windows ? [16:55:37 CEST] <JEEB> without some mistakes on my side, no [16:58:46 CEST] <Radiator> Ok because I have a code that fully works on ubuntu, when running it on windiws (with the proper ffmpeg libs indeed) i have a segmentation fault [16:59:31 CEST] <JEEB> time to either use vs debugger or gdb :p (first one if you built with msvc) [17:00:06 CEST] <JEEB> i can link a gdb binary for windows when I get home if you don't know where to find one [17:00:24 CEST] <Radiator> Yup, that's what I did and it seems to me that everything is fine [17:01:03 CEST] <Radiator> JEEB: Thanks, I have the Visual studio debugger [17:01:48 CEST] <JEEB> alright, but do remember that that one only works if you built the binaries of FFmpeg with msvc [17:04:37 CEST] <grkblood13> is there a way to define the codec to be used for playback with ffplay? [17:05:09 CEST] <Radiator> JEEB: Oh I didn't know ! Interesting ! [17:05:28 CEST] <DHE> ffplay -h full # indicates a "-codec" option [17:06:52 CEST] <grkblood13> i tried that [17:07:02 CEST] <grkblood13> ffplay -i ~/test.raw -codec:v mpeg2video [17:07:11 CEST] <grkblood13> [mpegts @ 0x7f9e0c000b80] Could not find codec parameters for stream 0 (Unknown: none): unknown codec [17:07:20 CEST] <JEEB> set the containet [17:07:27 CEST] <JEEB> -f [17:07:42 CEST] <JEEB> not sure what was the format for raw mpeg-2 video [17:08:01 CEST] <JEEB> or is it mpegts but without proper data in pmt? [17:08:11 CEST] <grkblood13> its mpegts [17:08:19 CEST] <grkblood13> no pmt apparently [17:08:27 CEST] <grkblood13> or anything it seems [17:08:45 CEST] <grkblood13> jsut raw data [17:09:06 CEST] <JEEB> then it's not mpegts then ? [17:09:35 CEST] <grkblood13> its a recording from a live atsc signal using a tuner card [17:10:46 CEST] <DHE> well that should be mpegts then [17:12:04 CEST] <grkblood13> yea, it is, i was running it agaisnt some gst commands earlier and i remember seeing pmd and some other stuff reporting as zero or noe [17:12:06 CEST] <grkblood13> none* [17:13:19 CEST] <DHE> my favourite trick is to try opening it in wireshark [17:15:03 CEST] <grkblood13> file information: https://pastebin.com/raw/jMQQavdy [18:53:40 CEST] <classsic> Hi, wich alternative to ffserver exist to create an mjpeg server? [19:17:32 CEST] <BtbN> define "mjpeg server", cause that's not a thing. [19:18:14 CEST] <JEEB> probably HTTP thing that serves chunked JPEGs [19:18:21 CEST] <JEEB> no idea though [00:00:00 CEST] --- Sat Sep 21 2019

1 0