Ffmpeg-devel-irc

Download

ffmpeg-devel-irc@ffmpeg.org

November 2016

1 participants
60 discussions

ffmpeg-devel.log.20161109
by burek 10 Nov '16

10 Nov '16

[00:17:10 CET] <kierank> Gramner: thanks, the redis crc thing works, going to implement the avx2 tomorrow [00:17:18 CET] <kierank> funnily enough ends up being 10-bit bitpacked [04:55:16 CET] <kierank> time for bed [04:55:20 CET] <kierank> all hail the trump overlord [16:43:59 CET] <J_Darnley> I'l like to pick your brains. (Particularly you Gramner, kierank said you'd help). Does the mask controlling loads in vpgather(dd) affect the speed? Will loading data I don't really need make it slower? [16:45:00 CET] <Gramner> yes [16:47:15 CET] <J_Darnley> thanks [17:50:23 CET] <cone-299> ffmpeg 03Steven Liu 07master:ab6ffc2a0800: MAINTAINERS: Add myself to flvenc [20:52:10 CET] <J_Darnley> Gramner or anyone else, do you have any tips for doing horizontal ORs and XORs of dwords or qwords? (paste of code coming shortly) [20:53:51 CET] <Gramner> move the upper half of the register to the lower half of a temporary register. perform desired operation. repeat until done. [20:55:13 CET] <Gramner> x86 SIMD simply isn't designed to do horizontal operations fast [20:55:31 CET] <J_Darnley> Yeah, I'm seeing that. [20:56:24 CET] <Gramner> it's also getting worse as vector sizes increases. Intel are aware of this. [20:56:31 CET] <J_Darnley> Would you expect it to be faster to load once and shift/shuffle than to load several times? [20:56:53 CET] <J_Darnley> Yeah, the split lanes can be a pain if you want to cross them. [20:57:04 CET] <Gramner> afaik the original Xeon Phis had horizontal sum instructions but they were removed when it turned into AVX-512 for some reason [20:57:46 CET] <Gramner> loads are generally pretty fast [20:58:03 CET] <Gramner> modern intel cpus can do 2 loads per cycle but only one shuffle per cycle [20:58:05 CET] <J_Darnley> So I better write both and bench it. [20:58:19 CET] <Gramner> also loads can oftne be OOE:d better [20:59:37 CET] <J_Darnley> Thanks again. [21:05:52 CET] <Gramner> there was actually a horizontal OR reduce instruction in KNC. https://software.intel.com/en-us/node/523756 [21:06:26 CET] <Gramner> I wouldn't be surprised to see those returned in soem AVX-512 extension at some point [21:17:27 CET] <cone-299> ffmpeg 03Andreas Cadhalpun 07master:467eece1bea5: icodec: fix leaking pkt on error [21:17:29 CET] <cone-299> ffmpeg 03Andreas Cadhalpun 07master:d54c95a1435a: icodec: add ico_read_close to fix leaking ico->images [21:17:31 CET] <cone-299> ffmpeg 03Andreas Cadhalpun 07master:226d35c84591: escape124: reject codebook size 0 [22:32:58 CET] <J_Darnley> Gramner, would you mind casting your eye over this ugly thing http://pastebin.com/2fxEQCQS [22:33:21 CET] <J_Darnley> It works but I'm looking for ways to speed it up. [22:33:53 CET] <J_Darnley> I plan to try replacing some moves with shuffles [22:34:12 CET] <J_Darnley> And perhaps interleave some distant instructions. [22:34:40 CET] <J_Darnley> By the way, how large is a typical OOE buffer? [22:34:43 CET] <Gramner> pmozx is a shuffle [22:34:50 CET] <Gramner> pmovzx* [22:37:26 CET] <J_Darnley> unpack is a shuffle too, right? [22:37:35 CET] <Gramner> yes [22:38:18 CET] <Gramner> http://agner.org/optimize/instruction_tables.pdf look for p5, that's the "shuffle unit" [22:38:33 CET] <J_Darnley> ah, thank you [22:44:04 CET] <jamrial_> J_Darnley: what is this for? [22:44:10 CET] <Gramner> might be more efficient to combine the two first gathers and shuffle the data afterwards instead of doing the unaligned pmovzx in the beginning [22:44:45 CET] <J_Darnley> An attempt at speeding up a problematic CRC for kierank [22:45:00 CET] <J_Darnley> we already managed to get the C very fast [22:45:39 CET] <J_Darnley> but this looks like a dead end from a performance viewpoint. [22:46:43 CET] <Gramner> you might be bottlenecked by memory loads which means that reducing computations with SIMD wont help [22:50:44 CET] <rcombs> why not use the CRC instruction? [22:50:54 CET] <kierank> because that's a specific crc [22:51:13 CET] <rcombs> ah, need a different poly? [22:51:15 CET] <kierank> and it's not a 10-bit crc [22:51:16 CET] <kierank> yes [22:51:38 CET] <rcombs> well then [22:51:42 CET] <rcombs> carry on [23:15:03 CET] <kierank> J_Darnley: i think you can do a load and then do a variable shift with pmulld [23:15:14 CET] <kierank> which will shift only the necessary buffers [23:15:39 CET] <J_Darnley> ah, I always forget about trying a multiply [23:16:34 CET] <Gramner> avx2 has variable shifts already though, but only for dwords [23:16:48 CET] <Gramner> and words I guess [23:16:54 CET] <Gramner> qwords* [23:17:56 CET] <J_Darnley> If I had to shift more than 2 samples I'm sure it could be helpful. [23:18:15 CET] <J_Darnley> (either a multiply or variable shift) [23:18:43 CET] <kierank> not sure if mova, punpk is better than pmovzxwd [23:19:14 CET] <J_Darnley> Was a bitwise dqword shift added at some point? [23:19:31 CET] <Gramner> J_Darnley: nope [23:20:43 CET] <Gramner> kierank: punpck and pmovzx are both shuffles. they are equally fast but the latter is shorter and doesn't require a zero register [23:21:02 CET] <kierank> interesting [23:21:42 CET] <kierank> J_Darnley: is there a reason you use xm3 and xm4 [23:22:05 CET] <Gramner> I'm actually a bit surprised that they didn't make the zero/sign extension part of the load unit like it is for scalar ones [23:22:15 CET] <Gramner> that would make it essentially free [23:22:44 CET] <J_Darnley> kierank: mostly because I don't need the upper 16bytes everywhere [23:23:12 CET] <kierank> xm3 and xm4 are the same no? [23:23:38 CET] <kierank> or i misunderstand pehaps [23:23:42 CET] <J_Darnley> Huh? [23:23:53 CET] <Gramner> gather overwrites the mask [23:23:59 CET] <kierank> ah [23:24:01 CET] <J_Darnley> Oh, the gather obliterates the mask [23:24:38 CET] <Gramner> because it can encounter faults during the instruction and have to resume. without overwriting the mask it would then need to re-fetch everything. [23:24:47 CET] Action: kierank tries to optimise c [23:25:00 CET] <J_Darnley> Ah, so there is a reason to it. [23:51:50 CET] <kierank> J_Darnley: I wonder if the dw nature of things makes the c faster [00:00:00 CET] --- Thu Nov 10 2016

1 0

ffmpeg.log.20161109
by burek 10 Nov '16

10 Nov '16

[00:08:05 CET] <Min> okey I got it working, thanks for helping guys [00:17:54 CET] <Phi> hm [00:18:01 CET] <Phi> it doesn't seem to use libmfx properly [00:18:23 CET] <Phi> it's expecting headers in "mfx/xxx", but the newest libmfx files are just in "include/xxx" [00:22:16 CET] <llogan> furq: not sure why he didn't say what he did to "get it working". probably remuxing as you suggested. [00:28:31 CET] <furq> i vaguely remember there being some longstanding bug report about this [00:35:37 CET] <furq> huh [00:35:52 CET] <furq> this only happens with libx264 [00:36:23 CET] <furq> in any container, with any audio codec [00:38:31 CET] <furq> http://vpaste.net/aBg8H [07:24:07 CET] <kepstin> furq: I assume it has to do with x264's encoder delay, you can stuff hundreds of frames into it before getting stuff out depending on settings, threads, etc. [07:25:13 CET] <kepstin> so if you're calculating shortest based on the x264 encoder's output, and stop giving it input when you hit the desired length, you still have a good chunk of frames to flush out of the encoder, i guess? [08:53:30 CET] <Chloe> does ffmpeg have anything to attempt to fix PNGs? [08:53:40 CET] <Chloe> from the CRC32 [08:54:26 CET] <Chloe> or a way I can extract the raw data of a chunk and the CRC [08:55:57 CET] <kerio> how do you fix a png from a crc32 [09:07:11 CET] <Chloe> change bits until the crc32 is correct [09:08:45 CET] <kerio> lmao [09:09:01 CET] <kerio> how big is a chunk? [09:09:52 CET] <kerio> like [09:09:56 CET] <kerio> how much data is checksummed? [09:18:23 CET] <Chloe> large [09:19:01 CET] <Chloe> but it's ok, I'll try it anyway >.< I just need to be able to extract the chunk data mainly [11:56:21 CET] <deadbeefy> Hi, I'm having trouble when recording the screen, only a few frames are captured and then it starts dropping all frames. The number of frames captured seem to be constant when not changing parameters but vary depending on the capture resolution. I'm on ubuntu/xenial and using ffmpeg 3.2.1 from ppa jonathonf/ffmpeg-3. [11:56:40 CET] <deadbeefy> here's the command I'm using: ffmpeg -video_size 64x64 -framerate 25 -f x11grab -i $DISPLAY -y output.mp4 [12:42:59 CET] <n4zarh> hello there, I might have a question about ffmpeg in c/c++ and memory-based crashes [12:46:03 CET] <n4zarh> I use C functions of ffmpeg in a native library for my android app. I use threads for every single encoder/decoder (for example one thread per audio and video decode). Each thread inits decoder once, proceeds with decoding, and after being interrupted it uninits decoder. [12:47:11 CET] <n4zarh> While it worked perfectly on sample application, on main app (where I want to use that library) uniniting decoders tends to crash whole app [12:49:08 CET] <n4zarh> it happens on closing/freeing codec, freeing frames and sometimes on single non-ffmpeg function which clears some memory as well. I doubt it's ffmpeg fault (because of last instance), but I want to make sure: is it possible that it's ffmpeg fault at those crashes? [12:50:48 CET] <bencoh> don't you have a demux (avformat) as well? [12:51:30 CET] <n4zarh> nope [12:51:32 CET] <bencoh> where does it live, and how do you init/close it? [12:51:32 CET] <bencoh> ah [12:52:17 CET] <n4zarh> only on closing [12:52:25 CET] <bencoh> ? [12:52:52 CET] <n4zarh> avcodec_close(c) or av_free(c) (where c == codec context) [12:53:21 CET] <n4zarh> the thing is, it happens quite randomly, that's why I doubt it is ffmpeg fault, but I need to make sure [12:54:16 CET] <bencoh> well you could try running a unit-test of your native library through valgrind [12:54:41 CET] <bencoh> (with just a few samples / small frames as an input) [12:59:20 CET] <nonex86> using ffmpeg in my multithreaded application without any problem [12:59:43 CET] <nonex86> btw its 24/7 software [12:59:43 CET] <bencoh> me too (at least did so in the past) so ... [13:01:58 CET] <ritsuka> bencoh: check av_lockmgr_register() [13:02:03 CET] <relaxed> deadbeefy: try using -framerate 60 for the input and lower the framerate for the output with -r [13:02:18 CET] <nonex86> n4zarh: and, btw, your scaler code you post before contains a bug, thats why you did not get your frame properly [13:02:43 CET] <ritsuka> opps I meant n4zarh [13:04:15 CET] <nonex86> n4zarh: agree with ritsuka, you can provide to ffmpeg lock manager or just use critical sections/mutex/futex yourself to guard some ffmpeg calls [13:04:31 CET] <nonex86> n4zarh: usually thats stated clearly in documentation/header files [13:04:48 CET] <n4zarh> I see [13:43:19 CET] <deadbeefy> relaxed: 60 fps didn't help, but 80 fps seems to do the trick [13:43:33 CET] <deadbeefy> relaxed: any idea as to why this is necessary? [13:51:09 CET] <deadbeefy> hmm, actually it doesn't work :( [13:53:23 CET] <ciawal> I still have a problem with ffmpeg not seeing the first frame in a video I'm working with [13:53:23 CET] <ciawal> when I dump all the frames its not there, and when I perform a filter into another video file its not included [13:53:23 CET] <ciawal> can anyone suggest anything? [13:54:57 CET] <jkqxz> "-vsync 0" [13:56:17 CET] <jkqxz> (If the frame is actually in the file and gets decoded, then the only way you should normally be able to lose it after that is by timestamp issues. That should turn off ffmpeg doing anything funny there.) [14:07:38 CET] <deadbeefy> to clarify my problem, this captures 5 frames, then it just drops every frame: ffmpeg -f x11grab -i $DISPLAY output.nut [14:08:07 CET] <deadbeefy> while the same ffplay command works fine: ffplay -f x11grab -i $DISPLAY [14:13:15 CET] <jkqxz> deadbeefy: Is your setup the same as <http://ffmpeg.org/pipermail/ffmpeg-user/2016-November/034260.html>? [14:32:03 CET] <deadbeefy> jkqxz: yes! [14:36:00 CET] <relaxed> deadbeefy: you can try my static build, https://www.johnvansickle.com/ffmpeg/ [14:36:01 CET] <jkqxz> See <http://ffmpeg.org/pipermail/ffmpeg-user/2016-November/034286.html> and reply for some investigation. [14:36:04 CET] <jkqxz> Does it work if you build your own ffmpeg without the 9001 configure options? [14:37:46 CET] <deadbeefy> relaxed: sure, one moment! [14:38:59 CET] <deadbeefy> works like a charm [14:42:45 CET] <ciawal> jkqxz: I'm running this: ffmpeg -vsync 0 -i input.mp4 -f image2 thumb%02d.jpg [14:42:45 CET] <ciawal> but the frame is still missing :( [14:44:37 CET] <ciawal> I believe the first frame is identical to the last one, I wondered if there might be some special feature being used which references it [14:44:37 CET] <ciawal> it doesnt show in vlc either [14:45:14 CET] <relaxed> vlc uses ffmpeg's libs [14:47:18 CET] <jkqxz> You could remux the video stream to something else without moving the metadata and see if the frame is there or not. [14:47:27 CET] <deadbeefy> relaxed: thanks for the help, much obliged! [14:49:09 CET] <relaxed> deadbeefy: You're welcome. You should file a bug report on debian's ffmpeg (or add to an existing bug report) [14:53:00 CET] <relaxed> ciawal: have you tried remuxing the input into mkv first? [14:54:20 CET] <ciawal> I'd tried avi previously, just tried mkv but same result [14:54:39 CET] <ciawal> the frame number is wrong if I use ffprobe also [14:54:44 CET] <ciawal> 68 when 69 is expected [15:21:44 CET] <brontosaurusrex> an easy cli way to get duration of video file in hh:mm:ss:ff (ff=frames)? [15:31:42 CET] <brontosaurusrex> or should i make some math like frames=round(miliseconds/fps)? [15:33:52 CET] <nonex86> fps can be variable... [15:37:57 CET] <brontosaurusrex> hm, yeah, whats the math in that case? [15:38:35 CET] <brontosaurusrex> i guess none, doesnt make sense [16:06:26 CET] <termos> Is there an easy way to get access to a parent this pointer in the function I set as av_log_set_callback? I have problems using std::bind because of the va_list argument [16:06:59 CET] <termos> std::bind(&Class::NonStaticMember, this) was my attempt [16:57:01 CET] <brontosaurusrex> another q: if i assume 25 fps, what is that called? 25fps base? [17:39:21 CET] <brontosaurusrex> got some bash working https://raw.githubusercontent.com/brontosaurusrex/singularity/master/bin/du… [17:39:39 CET] <brontosaurusrex> not very fast, but does the trick. [18:34:49 CET] <Croepha> Hi, trying to get to use ffmpeg api to play video and I seem to get the first frame of the video, but each frame after seems to be the same thing as the first video frame... any newb mistake I might be making? [18:52:10 CET] <Croepha> this is my code btw: http://dpaste.com/25050T2 [19:50:45 CET] <kaipee> hey gus, I'm getting an error : "Could not write header for output file #0 (incorrect codec parameters ?): Invalid data found when processing input" [19:51:08 CET] <kaipee> prior to that is : "[matroska @ 0x80a84a000] Invalid chapter start (6154481000000) or end (6131965000000)." [19:51:32 CET] <kaipee> does anybody have an idea as to what the problem is? [19:51:45 CET] <kaipee> (more importantly, how to resolve it?) [19:53:32 CET] <furq> pastebin the full command and output [19:53:46 CET] <furq> based on that it looks like you have a chapter which ends before it starts [19:55:11 CET] <kaipee> ah OK [19:55:19 CET] <kaipee> any way to edit (or ignore) that? [19:57:52 CET] <Phi> on a unrelated note, anyone know how to get Intel QSV working on a Windoze environment? [19:58:59 CET] <Phi> the example uses libav, but is that even for Windows? [20:03:44 CET] <kaipee> furq: how do I output to a log file? (BSD commandline) [20:05:35 CET] <kaipee> furq: pastebin > http://pastebin.com/kCqymjQm [20:06:20 CET] <furq> Chapter #0:19: start 6154.481000, end 6131.965000 [20:06:24 CET] <furq> yeah that's no good [20:06:31 CET] <furq> you can disable chapters with -map_chapters -1 [20:07:00 CET] <furq> if you want to keep them you'll have to fix them in the source with mkvtoolnix or something [20:07:55 CET] <DHE> start is after the end... [20:08:16 CET] <furq> the chapter before that has the same end time, [20:08:25 CET] <furq> so i'm guessing those times are just swapped for whatever reason [20:32:05 CET] <Phi> any idea for my scenario folks? [20:36:16 CET] <Phi> Intel QSV on Windows [20:36:37 CET] <Phi> the QSV example uses libva, which looks like a Linux-style only [20:38:26 CET] <furq> don't you need libmfx on windows [20:43:18 CET] <kaipee> ok cool, thanks furq [20:43:45 CET] <kaipee> would disabling chapters prevent skipping? or what side-effect would it have ? [20:48:19 CET] <furq> it would prevent seeking to chapter markers [20:48:27 CET] <furq> it won't prevent seeking in general [20:48:39 CET] <Phi> I have libmfx linked up [20:48:59 CET] <Phi> in the configure I've got --enable-libmfx, and the enable-encoder/decoder/etc for h264_qsv [20:49:09 CET] <Phi> that all turned out find, I got it all linked [20:49:26 CET] <Phi> ... as in, configured, maked, linked to my VS project [20:49:39 CET] <shincodex> tell me why -Wl,-rpath,./ does not load the stupid shared library from ,. [20:49:40 CET] <shincodex> ./ [20:49:51 CET] <shincodex> and ldd says library is being looked for in ../ [20:50:24 CET] <furq> kaipee: you probably just want to edit the chapters in mkvtoolnix [20:50:59 CET] <kaipee> ok ta [20:51:01 CET] <furq> either with the gui or with mkvpropedit [20:51:09 CET] <BtbN> you don't ever want to have . in your rpath [20:51:13 CET] <BtbN> that's as bad as security can get. [20:51:15 CET] <furq> probably the former because the latter looks like it requires hand-editing xml [20:51:20 CET] <BtbN> most loaders block it now [20:51:29 CET] <kaipee> great, thanks [20:56:11 CET] <gentwo> ffprobe ffprobe -v error -select_streams v:0 -show_entries stream=avg_frame_rate -of default=nw=1:nk=1 -show_entries name=filename SampleVideo_720x480_1mb.mp4 [20:56:35 CET] <gentwo> ^ how do i get just the filename with show_intries ? [20:59:37 CET] <furq> -show_entries format=filename [21:00:33 CET] <shincodex> yeah [21:00:36 CET] <shincodex> well [21:00:47 CET] <gentwo> ffprobe -v error -select_streams v:0 -show_entries stream=avg_frame_rate,format=filename, -of default=nw=1:nk=1 SampleVideo_720x480_1mb.mp4 [21:00:50 CET] <shincodex> ./sofolder/ [21:01:08 CET] <gentwo> furq: still no filename outputs [21:01:29 CET] <furq> replace the first , with : [21:01:33 CET] <furq> and get rid of the second one [21:01:52 CET] <furq> it seems like you already have the filename though [21:02:40 CET] <gentwo> furq: thanks, that does it [21:07:16 CET] <gentwo> now my results are getting on two lines, is there a way to get this in one single line? may be printformat? [21:13:16 CET] <shincodex> pretty certain ld is just broken [21:13:29 CET] <shincodex> and . is broken [21:23:12 CET] <ffnoob> hello everyone, im looking for a media streaming server and wondered if ffmpeg can be used for this purpose. this is to serve rtsp streams from an encoder to end users. thanks [21:25:59 CET] <Mavrik> nope [21:26:10 CET] <Mavrik> ffmpeg is the encoder :) [21:26:29 CET] <DHE> ffmpeg can feed something like an RTSP server which will do the distribution [21:26:48 CET] <DHE> or is it RTMP... I get those confused [21:27:08 CET] <Mavrik> both are protocols :) [21:27:40 CET] <ffnoob> I understand ffmpeg can be used as an encoder..but wanted to know if it can be used as a server [21:28:19 CET] <Mavrik> And the answer for that is "nope" :) [21:28:21 CET] <ffnoob> so to distribute it I will need a media server like wowza? [21:29:06 CET] <Croepha> anyone know why when I call avcodec_decode_video2 with new packets I dont get new frames? [21:30:37 CET] <Phi> you'd need a key frame in there first [21:57:39 CET] <DHE> can bitstream filters be chained? [22:00:02 CET] <BtbN> isn't there even a meta bsf to do just that? [23:12:19 CET] <DHE> BtbN: turns out the -help does in fact say "A comma-separated list of bitstream filters", so that's what I get for not checking first... [23:19:11 CET] <Croepha> where can I look for options to make ffmpeg/ffplay look ahead more when streaming in content... specifically I am trying to play an hls stream, but there is a hangup when it tries to load in chunks [23:44:27 CET] <bens218> hi [00:00:00 CET] --- Thu Nov 10 2016

1 0

ffmpeg-devel.log.20161108
by burek 09 Nov '16

09 Nov '16

[00:02:43 CET] <llogan> do we have a coverity access policy? I tell random users "no" (in a nice way) but i'm not sure about what others do. I assumed we let only developers have access but the list shows many users i've never heard of. [00:07:32 CET] <michaelni> some people with access are gsoc/outreachy students, some people i dont know and dont know why they have access though either [00:10:02 CET] <DHE> Zeranoe: holy crap, 2 TB a day average? (roughly) [00:11:23 CET] <Zeranoe> DHE: something was hitting a specific build and racking up traffic. [00:13:03 CET] <rcombs> bintray has a free host for FOSS packages, but it's limited to 1TB/mo [00:13:40 CET] <michaelni> Zeranoe, ask Raz- or try something like SF for hosting large binaries [00:13:50 CET] <Zeranoe> If you remove the traffic for that file it was roughly 15TB [00:13:53 CET] <Zeranoe> 25TB * [00:14:09 CET] <Zeranoe> I'm going to avoid SF [00:15:06 CET] <rcombs> SF's under new management and they're less evil now [00:15:22 CET] <rcombs> Zeranoe: wait, 12TB with that file and 25TB without? [00:16:12 CET] <Zeranoe> rcombs: Sorry, that's 12TB for Nov (including that file), and 25TB for Oct excluding it [00:16:26 CET] <rcombs> ah [00:17:25 CET] <rcombs> might want to poke bintray and see if they'd be willing to do a higher limit; I think homebrew has an arrangement like that [00:19:12 CET] <Zeranoe> I poked my current host telling them I removed that file, but if they don't play ball I'll need to move. I'm not terribly concerned about losing old builds, unless someone else is. [00:19:34 CET] <Zeranoe> The forum is backed up [00:21:57 CET] <rcombs> CloudFlare isn't technically supposed to be a binary download host, but they're generally pretty cool about people using them that way [00:23:03 CET] <llogan> michaelni: maybe ill purge the list of unknowns. [00:24:30 CET] <michaelni> llogan, send me the list you want to purge privatly so i can double check [00:24:52 CET] <Zeranoe> Fosshub has reached out to me in the past. They could probably hook me up too [00:25:04 CET] <llogan> michaelni: sure. may not get to it until later in week. [00:25:13 CET] <michaelni> no hurry [00:27:35 CET] <llogan> Zeranoe: i was always curious what your numbers would be but didn't think it would be that high [00:28:20 CET] <Zeranoe> llogan: I doubt that's all organic. I bet people program to download from it [00:30:46 CET] <llogan> i wonder how it compares with builds from relaxed [00:31:16 CET] <Zeranoe> relaxed builds? [00:31:32 CET] <llogan> yes, what the traffic is [00:31:50 CET] <Zeranoe> I'm not familiar with his builds? [00:32:15 CET] <llogan> http://johnvansickle.com/ffmpeg/ [00:32:49 CET] <Zeranoe> Ah [01:06:53 CET] <Compn> Zeranoe : i was curious how much people were downloading your builds and how you were hosting it monetarily wise [01:06:54 CET] <Compn> ehe [01:07:11 CET] <Compn> 10tb+ month, thats crazy... [01:07:32 CET] <Compn> i remember when we were a smaller project :P [01:13:48 CET] <Zeranoe> Like I said though, a lot of that is clearly automated downloads [01:14:09 CET] <Compn> 3rd party apps directly downloading your builds ? [01:14:11 CET] <Compn> you think? [01:14:15 CET] <Compn> probably yes [01:14:49 CET] <Compn> you could probably dissuade downloads like that by enforcing referrers [01:14:54 CET] <Compn> at least temporarily [01:14:54 CET] <Zeranoe> Well I doubt 130k visitors a month are accounting for all that :) [01:15:42 CET] <nevcairiel> i wouldnt be surprised if some third party sites deep link directly, or even worse, put links into their apps for client-local downloads or stuff like that [01:15:59 CET] <Zeranoe> Maybe, but there is always ways around that stuff unless you enforce IP session blocking or other complex countermeasures. Even a simple curl command can fake a server to think it's a browser requesting. [01:16:19 CET] <Compn> right [01:16:53 CET] <Compn> are there any web based free mirrors still left? like nyud.net ? :P [01:16:56 CET] <nevcairiel> well sure, but maybe they are not aware its a bad thing :p [01:17:28 CET] <Compn> .nyud.net:8090 [01:18:02 CET] <Zeranoe> Using something like reCAPTCHA could be a solution, but I don't really want to block people downloading them directly from a console app. [01:20:01 CET] <Zeranoe> All of this depends on my hosting providers response. I can do nothing but wait right now. [01:20:05 CET] <Compn> i hate recaptcha [01:21:48 CET] <Zeranoe> Everyone does [01:24:28 CET] <Zeranoe> OVH has some interesting plans.... [01:26:25 CET] <Zeranoe> Unlimited bandwidth but only 40GB of storage.. I'd have to remove older builds [01:32:26 CET] <haasn> Huh. I've added a breakpoint on matroska_read_header and webm_dash_manifest_read_header but neither of them get hit when playing a webm file in mpv [01:33:50 CET] <haasn> oh nvm, I think mpv uses its own demuxer here [01:44:22 CET] <rcombs> haasn: if you're trying to test the demuxer, you can force it with an option (I forget the name, though) [01:52:22 CET] <haasn> rcombs: more importantly, where the heck is the matroska EBML tree documented? [01:52:34 CET] <haasn> I can see lots of hard-coded magic numbers in the libavformat matroska demuxer [01:52:42 CET] <haasn> but I can't find the specification that defines any of these [01:52:54 CET] <haasn> for example that the video color track has ID 0x55B0 [01:52:57 CET] <haasn> seems to just come out of nowhere [01:53:55 CET] <rcombs> that's from https://matroska.org/technical/specs/index.html [01:54:06 CET] <haasn> `55B0` no hits :( [01:54:15 CET] <haasn> oh [01:54:15 CET] <rcombs> [55][B0] [01:54:18 CET] <haasn> they're extremely autistic [01:54:33 CET] <haasn> thanks [01:54:34 CET] <TD-Linux> or more recently the matroskaorg github / ietf draft [01:56:14 CET] <cone-023> ffmpeg 03Rostislav Pehlivanov 07master:0cf685380467: aacenc: quit when the audio queue reaches 0 rather than keeping track of empty frames [01:56:56 CET] <cone-023> ffmpeg 03Steven Liu 07master:acd87dfc05b5: cmdutils: add show_demuxers and show_muxers [02:59:58 CET] <llogan> Zeranoe: https://lists.ffmpeg.org/pipermail/ffmpeg-user/2016-November/034253.html [03:00:21 CET] <llogan> (you can click on the sender's email address link to reply if you want to) [03:18:10 CET] <CFS-MP3> Is there a non-hackish way for a filter to find out the name of the output file? [03:18:25 CET] <CFS-MP3> Suppose I want to add the output filename to showinfo, specifically [03:19:12 CET] <Zeranoe> llogan: I think I might just change the 403 page to read: "zeranoe.com is down due to HostGator being terrible. Please contact me with new hosting offers." [03:19:45 CET] <Zeranoe> Until they resolve it there's nothing I can do... [03:36:59 CET] <llogan> Anything with the name "gator" in it is usually terrible [03:38:06 CET] <CFS-MP3> Zeranoe why not just use a leaseweb or OVH server? Those really have unlimited traffic and are cheap [03:38:19 CET] <Zeranoe> llogan: Haven't had an issue since the beginning until now. [03:38:53 CET] <Zeranoe> CFS-MP3: I talked about that here before. I think I might move to an OVH VPS when this settles down, or doesn't resolve. [03:41:28 CET] <llogan> maybe you can talk hostgator into sponsoring the site. [03:44:44 CET] <Zeranoe> I'm already taking steps to get away from them. This issue is enough for me to run the other direction. [03:53:11 CET] <CFS-MP3> if it's just static file storage you can also use amazon S3, which solves lots of problems too [03:54:25 CET] <Zeranoe> CFS-MP3: Isn't that pretty pricy? I need bandwidth in the magnitude of ~30TB [03:59:08 CET] <llogan> its actually not that bad. [04:01:59 CET] <Zeranoe> Wow, apparently you get faster support via Twitter... https://twitter.com/Zeranoe/status/795820067359432704 [04:07:39 CET] <c_14> 30TB traffic on amazon will cost you 2-3000 US (according to the price lists and the amazon easy usage calculator) [04:10:10 CET] <rcombs> yeah, companies tend to react quickly to public shaming [04:10:35 CET] <Zeranoe> c_14: Oh is that all? Pass [04:12:39 CET] <Zeranoe> rcombs: The reply was quick, but we shall see how fast the resolution is. [06:10:40 CET] <Zeranoe> Looks like we're back, for now. [11:42:49 CET] <kierank> Gramner: ping [11:44:27 CET] <Compn> Zeranoe : its annoying when one form of support works quickly (twitter) while email goes 48 hours lol [11:44:51 CET] <BtbN> public Twitter is also the best way to get support from my ISP [11:45:13 CET] <BtbN> Tweet at them that stuff is slow, and as they already know who I am, stuff gets fixed [11:46:12 CET] <nevcairiel> presumably different support people, the interactive nature of twitter messaging might make it more real-time [11:46:51 CET] <BtbN> than a phone call? [11:46:52 CET] <funman> Gramner: http://pastebin.com/H2bKxj8G how can we make that fast ? [11:47:18 CET] <BtbN> If you call them and tell them stuff is broken they get defensive and blame shit on you [11:47:34 CET] <nevcairiel> like every support person, they think you are an idiot [11:47:42 CET] <funman> presumably since you can tweet they assume you're not mad because no internet at all [11:47:43 CET] <nevcairiel> which is probably true for 99.9% of thjeir cases anyway [11:47:50 CET] <BtbN> I also have an ongoing issue with IPv6 not working 75% of the time [11:48:18 CET] <BtbN> They even sent a technician, who connected his iPhone to the Wifi, did a speedtest, and told me if I call again they will bill me for his time. [11:49:00 CET] <BtbN> And by not working I mean their router does hand out addresses, but no data goes through. So it messed stuff up quite badly [11:49:38 CET] <nevcairiel> when I enable v6 my phone randomly loses connectivity over wifi, i'm still not sure if the router/ap is to blame or the phone, because there is some shit going in on the phone where it drops broadcast announces in the wifi firmware to "save power" [11:50:19 CET] <BtbN> Still upset with that technician. He didn't even bother trying to reproduce the issue. Just blamed it on our equipment. [11:55:00 CET] <Compn> an iphone isnt the best tool to test problems with :P [11:55:01 CET] <Compn> ehe [11:57:07 CET] <BtbN> dude didn't even bring a laptop, and had no idea what I was talking about when I said I suspect a routing misconfiguration on their end. [12:38:20 CET] <cone-199> ffmpeg 03Reynaldo H. Verdejo Pinochet 07master:311107a65d01: ffserver: check for codec match using AVStream.codecpar [12:38:20 CET] <cone-199> ffmpeg 03Reynaldo H. Verdejo Pinochet 07master:689f648a9596: ffserver: use .codecpar instead of .codec in print_stream_params() [12:38:21 CET] <cone-199> ffmpeg 03Reynaldo H. Verdejo Pinochet 07master:1323349befd3: ffserver: get time_base from AVStream in print_stream_params() [12:38:22 CET] <cone-199> ffmpeg 03Reynaldo H. Verdejo Pinochet 07master:afcbadf0eda3: ffserver: use AVStream.codecpar in find_stream_in_feed() [12:38:23 CET] <cone-199> ffmpeg 03Reynaldo H. Verdejo Pinochet 07master:822e3e2ddb8a: ffserver: user AVStream.codecpar in compute_status() [12:38:25 CET] <cone-199> ffmpeg 03Reynaldo H. Verdejo Pinochet 07master:6f0a1710d77d: ffserver: use AVStream.codecpar in open_input_stream() [12:40:08 CET] <wm4> wait what does this mean [12:45:35 CET] <BtbN> where to these even come from? [14:07:59 CET] <cone-199> ffmpeg 03Vittorio Giovara 07master:a765ba647d3d: avformat/mov: Read multiple stsd from DV [15:18:56 CET] <Gramner> kierank: what about the "divide and conquer" (e.g. libavutil/crc or whatever redis does) method? with some modifications. or does that cause issues with non-power of 2? y+c can be SIMD:ed with avx2 gathers, but I'm not sure if that even gains anyting if L1 latency is the bottleneck due to dependencies between iterations [15:19:10 CET] <kierank> Gramner: doesn't work I believe because it's a 10-bit crc [15:19:17 CET] <cone-199> ffmpeg 03Rostislav Pehlivanov 07master:0660a09dd1d1: opus: move all tables to a separate file [15:19:18 CET] <cone-199> ffmpeg 03Rostislav Pehlivanov 07master:317be31eaf4f: opus: move the entropy decoding functions to opus_rc.c [15:20:38 CET] <kierank> Gramner: hmm not sure actually [15:20:39 CET] <kierank> might work [15:21:31 CET] <Gramner> I'd try to get that working first [15:21:39 CET] <Gramner> probably the most potential gain [15:22:04 CET] <Gramner> that method should be SIMD:able as well in the same fashion with gathers [15:22:22 CET] <kierank> i thought avx2 gather was not a simd operation [15:22:25 CET] <kierank> until future processors [15:22:35 CET] <Gramner> gather is avx2, scatter is avx-512 [15:22:56 CET] <Gramner> it's not faster than doing individual loads, but you can do the address calculation in SIMD [15:23:02 CET] <kierank> oh [15:23:03 CET] <Gramner> which is the entire point of gathers, really [15:24:11 CET] <Gramner> but as I said, if the load latency is the bottleneck then reducing the number of arith ops in half wont really make any difference [15:29:17 CET] <kierank> the load of the pixel data or from the LUT? [15:30:14 CET] <Gramner> the non-divide-and-conquer method could probably be SIMD:ed with something like this (completely untested, even for syntax errors) http://pastebin.com/x5t1DMta and the same concept would apply to a potentially better algorithm as well. [15:30:52 CET] <Gramner> the lut. the index is dependant on the previous interation so it can't be OOE:ed [15:36:25 CET] <Gramner> the "dec" should obviously be "inc" btw [17:45:22 CET] <cone-199> ffmpeg 03James Almer 07master:70c6a1bcf021: avformat/matroskadec: fix DiscardPadding element parsing [17:49:14 CET] <atomnuker> woot, more vp9 avx2 [17:49:25 CET] <atomnuker> BBB: what are the performance gains like? [18:00:36 CET] <wm4> BBB: does vp9 support colorspace and mastering bitstream flags? [18:00:50 CET] <wm4> asking because youtube is apparently using dumb webm container tags [18:01:06 CET] <wm4> not even sure whether they're standardized in matroska and I sure wouldn't expect them in webm [18:01:12 CET] <BBB> wm4: yes, but it merges trc+space+matrix together in one item instead of specifying all 3 separately [18:01:30 CET] <BBB> so I guess they decided to also put it in the container in case people want them separated? [18:01:37 CET] <BBB> atomnuker: not much b/c adst is intra-only [18:01:59 CET] <nevcairiel> vp9 only has the "normal" flags for the 3 variables, not mastering data [18:02:04 CET] <nevcairiel> hence why it goes in the container [18:02:07 CET] <wm4> lol [18:02:09 CET] <wm4> thanks [18:02:55 CET] <nevcairiel> someone asked me to support that, I should poke that some day [18:03:42 CET] <wm4> or you could just use libavformat (same goes to me) [18:04:28 CET] <nevcairiel> i kinda do! my custom demuxer is inside there =p [18:04:41 CET] <nevcairiel> just need to copy-paste the metadata handling over there [18:04:42 CET] <nevcairiel> :D [18:05:08 CET] <nevcairiel> but my real problem is communicating that up the chain from the demuxer to the renderer, usually there is no direct metadata line there for me [18:05:42 CET] <BBB> atomnuker: checkasm numbers: [18:05:42 CET] <BBB> vp9_inv_adst_adst_16x16_add_8_avx: 726.2 [18:05:43 CET] <BBB> vp9_inv_adst_adst_16x16_add_8_avx2: 393.8 [18:05:48 CET] <nevcairiel> neat [18:05:50 CET] <BBB> compared to idct: [18:05:50 CET] <BBB> vp9_inv_dct_dct_16x16_add_8_avx: 466.8 [18:05:51 CET] <BBB> vp9_inv_dct_dct_16x16_add_8_avx2: 271.2 [18:05:57 CET] <BBB> adst is slower, which is expected [18:06:07 CET] <BBB> overall impact is low b/c intra-only, like I just said [18:09:22 CET] <wm4> nevcairiel: you can't transport packet side data? [18:10:18 CET] <cone-199> ffmpeg 03Tom Butterworth 07master:0a2458758874: avcodec/hap: pass texture-compression destination as argument, not in context [18:28:31 CET] <jamrial> wm4: they are going to be standarized in webm. see https://www.webmproject.org/docs/container/ [18:53:18 CET] <wm4> jamrial: do you know how they are going to specify container and decoder metadata conflicts? [18:56:25 CET] <jamrial> wm4: no, sorry [19:13:39 CET] <nevcairiel> wm4: isnt that stream global data, i would probably need to hook something up for that [19:13:47 CET] <nevcairiel> also the side data probably doesnt cover the matrix, or does it? [19:14:40 CET] <wm4> you can inject global stream side data into packets [19:14:53 CET] <wm4> maybe there's even API for that [19:16:41 CET] <nevcairiel> probably [19:16:48 CET] <nevcairiel> i downloaded on the youtube streams for testing [19:16:52 CET] <nevcairiel> +of [19:17:03 CET] <nevcairiel> probably my first 10-bit stream of vp9 ever =p [19:18:57 CET] <BBB> I encoded one at some point [19:35:08 CET] <jamrial> youtube is using 10bit vp9 now? [19:35:15 CET] <nevcairiel> for HDR content, yes [19:35:42 CET] <nevcairiel> https://www.youtube.com/playlist?list=PLyqf6gJt7KuGArjMwHmgprtDeY8WDa8YX format 337 with youtube-dl [19:37:09 CET] <nevcairiel> (although their sdr conversions are a tad bit over-saturated) [19:38:03 CET] <wm4> mpv already supports it lol [19:38:30 CET] <nevcairiel> chromecast ultra presumably does as well if you have a hdr display [19:47:11 CET] <gnafu> nevcairiel: Neat! [19:47:37 CET] <gnafu> I don't even have hardware capable of playing back UHD right now, much less 10-bit HDR, but it's fun to look at such a file :-). [19:48:04 CET] <gnafu> (Actually, my phone might be able to do 8-bit UHD VP9 in hardware, but I'm not sure if it's not quite new enough or not.) [19:48:21 CET] <gnafu> (I know it does hardware VP9 decoding, but it might only be up to 1080p.) [19:50:19 CET] Action: gnafu downloads 315, the high-bitrate 8-bit version, to test. [19:52:36 CET] <gnafu> Wait, looks like mine can do UHD in some format, but not VP9. [19:52:46 CET] <gnafu> https://en.wikipedia.org/wiki/List_of_Qualcomm_Snapdragon_devices#Snapdrago… [19:53:53 CET] <BBB> so I suppose you need new type of videocards etc. to display 10bit as-is? [19:54:01 CET] <BBB> like, for example, if I have a nice sweet macbook pro [19:54:10 CET] <BBB> (and Im displaying stuff using opengl, in floats) [19:54:15 CET] <BBB> does it support 10bit? or not? [19:54:40 CET] <nevcairiel> for nvidia you need a maxwell or pascal card, and i dont know about opengl, opengl 10-bit has in the past required "pro" cards, ie quadro or firepro [19:54:59 CET] <BBB> oh you need a video display mode I guess, right? [19:54:59 CET] <nevcairiel> for direct3d you use fp16 surfaces with 1.0 at 80 nits [19:55:22 CET] <nevcairiel> and the driver converts that to the proper hdr hdmi format [20:02:20 CET] <wm4> yeah what you actually need is signaling the HDR metadata [20:02:57 CET] <nevcairiel> nvidia defined some in their NvAPI package [20:03:20 CET] <nevcairiel> https://developer.nvidia.com/displaying-hdr-nuts-and-bolts [20:04:03 CET] <nevcairiel> (actual metadata setting not shown, but its part of NV_HDR_COLOR_DATA) [20:21:27 CET] <cone-199> ffmpeg 03Andreas Cadhalpun 07master:ff100c9dd97d: matroskadec: fix NULL pointer dereference in webm_dash_manifest_read_header [20:31:38 CET] <gnafu> Confirmed: My LG G3 cannot play the 8-bit UHD version of that video, neither in the stock video player nor VLC. [20:32:02 CET] <gnafu> I believe I have successfully played 1080p VP9 ony my phone, but that was presumably software-decoded (which is pretty neat). [20:32:12 CET] <gnafu> s/ony/on/ [22:28:57 CET] <cone-199> ffmpeg 03Andreas Cadhalpun 07master:1bbb18fe82fc: mpegts: prevent division by zero [22:42:29 CET] <nevcairiel> the last commit in the avconv merge batch is annoying .. the ffmpeg filtering code is way different, and has an architectural conflict - in short, the commit is trying to delay initializing filters until input was decoded (which makes sense, only then do we have the real info) - but our code tries to probe the filters to find out which files need to be read further, so clearly delaying their creation stands in direct competition to [22:42:29 CET] <nevcairiel> that [22:43:31 CET] <JEEB> ugh [22:43:36 CET] <JEEB> yeah, that's problematic [22:47:48 CET] <nevcairiel> will probably need proper re-implementation instead of just merging. maybe we should just skip that one for now and work on that in parallel instead of blocking on it, it shouldn't block any further merges except a few "fixups" to that code [22:49:51 CET] <nevcairiel> anyway all the commits in the batch up to that one are here: https://github.com/Nevcairiel/FFmpeg/commits/merge-avconv .. fate passes with one tiny change (timebase precision changes) [22:50:06 CET] <nevcairiel> (top 3) [23:02:54 CET] <jkqxz> Does that mean that ffmpeg would still have the lavfi reinit on hardware transcode? [23:03:19 CET] <nevcairiel> i guess? [23:03:21 CET] <jkqxz> Making cuvid or qsv work in the intermediate state would probably be a pain. [23:03:29 CET] <nevcairiel> feel free to fix it quicker =p [23:03:42 CET] <jkqxz> Maybe just ignore that... [23:04:12 CET] <BtbN> cuvid hwtranscode is already broken on master [23:07:46 CET] <nevcairiel> hm maybe i can hack around the limitation [23:13:20 CET] <nevcairiel> or maybe not [23:16:38 CET] <nevcairiel> well it works for some tests, many other still broken [23:16:41 CET] <nevcairiel> but i guess its progress [23:18:25 CET] <nevcairiel> put the WIP also on the branch, will try more tomorrow [23:20:56 CET] <wm4> "which files need to be read further" wat? [23:21:29 CET] <nevcairiel> its for multiple-input things, it tries to figure out from the filter graph which files need to be read next, by checking which inputs are starving [23:21:53 CET] <nevcairiel> it really makes sense, just conflicts a bit with this idea of lazy graph init [23:23:22 CET] <nevcairiel> but i worked around that by just assuming that a input which wasnt initialized yet is by definition starving [23:25:19 CET] <dkc> in AVPixelFormat, there are some definitions for planar YUV with various bpp, but those variants doesn't exist for packed YUV. In my case I'd like to support UYVY with 20bpp, how should I handle that? [23:25:36 CET] <wm4> nevcairiel: that sounds like a proper solution [23:29:31 CET] <nevcairiel> dkc: support in what? usually pixel formats are created as there is a need for them, and apparently noone needed this one [23:30:31 CET] <dkc> the use case is to receive raw video through RTP (RFC 4175: https://tools.ietf.org/html/rfc4175) [23:32:23 CET] <dkc> there are different combinations of sampling and depth [23:33:37 CET] <wm4> treat it as raw codec and repack to sane formats manually [23:34:28 CET] <nevcairiel> those formats dont really represent typical pixel formats, that the 422 case happens to match UYVY layout is probably more of an accident then design - none of the others match any common formats whatsoever [23:40:30 CET] <dkc> what do you mean? The point of this RFC is more or less to send over IP sensors' outputs, so I think it's by design [23:41:34 CET] <nevcairiel> the 420 format just looks very bizzare, never seen it in a YYYYCbCr layout before [23:41:46 CET] <nevcairiel> and the 444 version also doesnt look like a component order i have seen before [23:42:20 CET] <TD-Linux> you are going to need lots of copies anyway because there are going to be many packets per frame [23:42:33 CET] <JEEB> is that in blocks of 2x2? > YYYYCbCr [23:42:52 CET] <JEEB> weird stuff [23:43:05 CET] <nevcairiel> JEEB: they want all parts of one pixel always together so that missing packets only corrupt that particular part of the image [23:43:16 CET] <JEEB> right, so yeah :D [23:43:18 CET] <nevcairiel> so yeah, 4 Y samples and their chroma [23:43:21 CET] <nevcairiel> in 2x2 [23:44:10 CET] <nevcairiel> but TD-Linux is probably right, while you are re-assembling it into a continous image, you might as well just convert it into a format we have [23:44:24 CET] <JEEB> yup [23:47:53 CET] <dkc> hum okay, I'll have to discuss that with my team, we might have overlooked a quite big issue [23:49:06 CET] <dkc> Just for info, this RFC is used in TR-03: http://www.videoservicesforum.org/download/technical_recommendations/VSF_TR… [23:49:29 CET] <dkc> thanks for your help! [23:54:53 CET] <kierank> JEEB: it's bitpacked uyvy [23:56:06 CET] <kierank> looks quite complex but it's quite easy to simd [00:00:00 CET] --- Wed Nov 9 2016

1 0

ffmpeg.log.20161108
by burek 09 Nov '16

09 Nov '16

[00:15:22 CET] <yuhl_rec> hi, I'm trying to record 2 video streams and 1 audio to a mkv. It does works well videos and audio are no syncronized [00:19:08 CET] <yuhl_rec> http://pastebin.com/zL3XsYLN [00:20:44 CET] <yuhl_rec> I do use a logitech C920 and a datapath video grabber from http://www.datapath.co.uk/video-capture-cards/vision-range/visionrgb-e1s [00:21:11 CET] <yuhl_rec> Is there some stuff I can do to tell ffmpeg to keep it syncronized ? [01:18:15 CET] <CFS-MP3> is it possible to generate a reencoded file (let's suppose a ts, but probably irrelevant) and at the same time a report of frame types for that file? [01:18:40 CET] <CFS-MP3> Obviously ffprobe can do it in a 2nd pass, but the idea here is to generate the video and the report at the same time, live [01:21:51 CET] <furq> CFS-MP3: https://ffmpeg.org/ffmpeg-filters.html#showinfo [01:24:12 CET] <furq> no idea if it's possible for output frames [01:24:29 CET] <furq> i doubt it though [01:33:56 CET] <CFS-MP3> you're mistaken but as usual you sent me to the right place [01:34:17 CET] <CFS-MP3> where in the world are you? You've earned at least a lunch [01:35:59 CET] <CFS-MP3> -filter_complex "fps=12,scale=416:-2:flags=lanczos,showinfo" -map v -codec:v libx264 -f segment -segment_time 6 -segment_list_type m3u8 -segment_list wwseg.m3u8 wwseg%05d.ts [01:36:10 CET] <CFS-MP3> that seems to be in the right direction [01:36:25 CET] <CFS-MP3> it reports size being 416 so clearly it refers to the output stream [01:38:09 CET] <CFS-MP3> is there a way to somehow write each of the lines to a file that is named following the segments? Ideally if I have wwseg00004.ts the report (or console dump, whatever yhou want to call it) should be wwseg00004.txt [01:41:01 CET] <deweydb> i have an ffmpeg process running inside a celery instance, it will run for a bit at about 100% cpu but eventually it will just hang, and it always hangs on message like "write(2, "Past duration 0.988655 too large"..., 33 " (when i inspect the thread with strace) [01:41:11 CET] <deweydb> what does that mean? past duration too large? [01:41:56 CET] <DHE> thats' not what matters. file descriptor 2 is stderr, usually your console. something jamming up the controlling terminal? [01:42:31 CET] <deweydb> interesting. [01:43:07 CET] <deweydb> i don't really know what you mean by :"jamming up the controlling terminal" [01:43:53 CET] <DHE> well stderr is traditionally the console, unless you explicitly redirect it. [01:44:02 CET] <DHE> so what are you running it under? linux text console? xterm? screen? ssh? [01:44:32 CET] <deweydb> its being run as a subprocess from python, within a celery task. [01:44:44 CET] <deweydb> the whole thing is then run with supervisor. [01:45:00 CET] <DHE> is there an interface in said process or supervisor to collect the output of ffmpeg? [01:45:23 CET] <deweydb> i think output is, but not stderr. [01:45:26 CET] <deweydb> checking.. [01:49:21 CET] <deweydb> interesting. yeah, so i have stderr passing to subprocess.PIPE i'll have to check with the #python guys why this would be an issue. thanks for giving me a lead. [01:49:48 CET] <deweydb> i removed that part, and ran it. [01:49:56 CET] <deweydb> i see how. there's literally 100000s of those messages: [01:52:08 CET] <deweydb> DHE: http://termbin.com/4aix [01:52:22 CET] <deweydb> is that normal, that ffmpeg would report this error so many times during a render? [01:53:08 CET] <deweydb> so i guess that because this error happened so many times the stderr=subprocess.PIPE got overloaded. funny. [01:53:18 CET] <DHE> yeah that's probably the cause [01:53:31 CET] <DHE> as for the past duration, I'm not sure. check the output to see if it's okay [02:29:51 CET] <deweydb> DHE: thank you for your help, you saved me hours of headaches. [02:37:43 CET] <DHE> it's what I do [02:40:52 CET] <_2016Trump> a 10 page beginner's guide to ffmpeg would be nice [02:40:57 CET] <_2016Trump> I'll make it for $100 [02:41:28 CET] <_2016Trump> I think government grants should fund software development [02:41:49 CET] <CFS-MP3> _2016Trump And what would you cover? You probably need half a page for each of the 10000 different things FFmpeg can do [02:42:52 CET] <_2016Trump> Just start out by showing how to encode in to the most common formats [02:43:12 CET] <_2016Trump> How to set audio / video bit rates, frame rates, sample rates [02:44:05 CET] <_2016Trump> how to do video capture and screen captures as inputs [02:44:11 CET] <_2016Trump> setting resolution [02:44:19 CET] <_2016Trump> and a table of the optimal bit rates for the various formats [02:44:36 CET] <CFS-MP3> _2016Trump all that is well document [02:44:47 CET] <CFS-MP3> to be honest what we need is a collection of advanced recipes [02:44:56 CET] <_2016Trump> For a DVD, 4mbps will look great with MPEG2, but at 2mbps MPEG4 will be better than mpeg2 [02:45:34 CET] <CFS-MP3> show me a way to do what I need to do right now without coding and I'll give you those $100 :-) [03:11:39 CET] <_2016Trump> CFS-MP3, I wish I could! [03:34:49 CET] <fling> Is there an audio format with which it is possible to convert stereo to mono dropping one of the channels with '-c copy'? [03:42:43 CET] <_2016Trump> fling, I sort of thought ffmpeg can't change audio channels [03:42:51 CET] <_2016Trump> there's an audio transforming tool called sox though [03:44:29 CET] <fling> _2016Trump: are you saying it can't change audio channels in general? [03:45:00 CET] <llogan> sure it can, if you're willing to re-encode: http://trac.ffmpeg.org/wiki/AudioChannelManipulation [03:45:16 CET] <fling> The question was about '-c copy' [03:45:17 CET] <_2016Trump> I'm not sure [03:45:44 CET] <_2016Trump> Well MP3 is joint stereo [03:45:55 CET] <fling> _2016Trump: not always. [03:45:55 CET] <_2016Trump> so they aren't really separate channels [03:46:23 CET] <fling> _2016Trump: but you still _can_ copy the left channel. It is encoded the regular way. [03:46:34 CET] <fling> In theory ^ [03:46:51 CET] <_2016Trump> It's not a mixture of the left and right channels? [03:48:24 CET] <fling> _2016Trump: It is probably two streams. First is the left channel, second is the stream of frames calculated with $(left - right) [03:49:33 CET] <fling> The idea is to encode less data in the second stream if there is any mono in the stream. [03:49:58 CET] <fling> But this is kinof inefficient for true stereo streams where is no mono. [03:50:30 CET] <fling> Idk if it is possible to have both joind and non-joint chunks in a single mp3 stream. [03:50:35 CET] Action: fling is not using mp3 much anyway :P [03:51:02 CET] <fling> There are mostly flac, opus and different variants of pcm [03:51:19 CET] <fling> aac sometimes& [04:01:16 CET] <_2016Trump> You can have it not joint stereo as I think you said [04:04:35 CET] <fling> Sure. [04:26:23 CET] <xiaowl> hi, ffmepg keeps dropping frames when using x11grap on Debian Jessie, using version 3.14. command: ffmpeg -video_size 1024x768 -framerate 25 -f x11grab -i :0.0+100,200 output.mp4 [04:26:32 CET] <xiaowl> any ideas? [04:26:48 CET] <fling> xiaowl: slow storage? [04:27:14 CET] <fling> xiaowl: slow encoding? [04:28:12 CET] <fling> xiaowl: try tuning libx264 preset [04:28:15 CET] <xiaowl> no sure, this debug message keeps printing: [04:28:17 CET] <xiaowl> [rawvideo @ 0x2117a80] PACKET SIZE: 3145728, STRIDE: 4096 [04:28:19 CET] <xiaowl> *** dropping frame 3 from stream 0 at ts 0 [04:28:21 CET] <xiaowl> cur_dts is invalid (this is harmless if it occurs once at the start per stream) [04:28:47 CET] <xiaowl> Intel CPu i7 + SSD, shouldn't be encoding or storage issue? [04:29:19 CET] <fling> xiaowl: just try it, don't guess. [04:29:41 CET] <xiaowl> the same command works well on another laptop with i5 + normal disk [04:30:01 CET] <fling> xiaowl: -c libx264 -crf 28 -preset ultrafast [04:32:21 CET] <xiaowl> same output, no luck:( ffmpeg -s 1366x768 -f x11grab -framerate 10 -i :0.0 -c libx264 -crf 28 -preset ultrafast output.mp4 [04:32:45 CET] <xiaowl> how can i get more output logs? trying debug flag seems not help [04:32:57 CET] <xiaowl> or maybe i missed some import logs? [04:32:57 CET] <fling> pastebin the command and the full output [04:34:51 CET] <xiaowl> http://pastebin.com/DNGkzqtp [08:35:49 CET] <ElAngelo> there is no demuxer for mov in ffmpeg? [08:37:38 CET] <bencoh> ElAngelo: mp4? [08:37:52 CET] <ElAngelo> yeah... no [08:38:04 CET] <ElAngelo> i have a couple of mov's that don't want to play with ffplay [08:38:27 CET] <ElAngelo> if i play them with mplayer they don't play as well... just untill i tell it explicitely to use the mov demuxer [08:38:31 CET] <JEEB> there is a demuxer just fine for mov/mp4/derivatives [08:38:40 CET] <JEEB> in FFmpeg [08:38:59 CET] <ElAngelo> then they play just fine [08:38:59 CET] <JEEB> thus either you found a bug, missing feature or the file is bonkers [08:39:18 CET] <ElAngelo> is there a way to figure out what it is? [08:39:29 CET] <JEEB> feel free to post a sample on the tra [08:39:33 CET] <JEEB> *trac [08:39:40 CET] <JEEB> in an issue [08:39:49 CET] <ElAngelo> would love too... only these are not files i own :( [08:40:14 CET] <JEEB> also the demuxer lately got "support" for edit lists from google [08:40:14 CET] <ElAngelo> anything i can figure out myself? [08:41:09 CET] <JEEB> boxdumper from l-smash, ffprobe etc [08:41:16 CET] <bencoh> does mplayer use avformat to demux mov/mp4 or does it have its own demux? [08:41:40 CET] <ElAngelo> i asked at the #mplayer channel and they tell me mplayer comes with it's own mov demuxer [08:41:44 CET] <ElAngelo> not sure if that is correct [08:41:54 CET] <JEEB> also try downgrading FFmpeg to before el google's edit list stuff [08:42:26 CET] <JEEB> because that broke a lot of stuff because they stuffed presentation level shit into a demuxer [08:43:05 CET] <JEEB> and now bugs keep getting reported and nobody wants to revert that stuff for whatever reason [08:43:10 CET] <ElAngelo> i have the issue both with an old ffmpeg version (2014) and a new one (3.2....) [08:43:27 CET] <JEEB> instead more and more hacks keep getting added :v [08:43:42 CET] <ElAngelo> when was that stuff merged? [08:43:44 CET] <ElAngelo> rougly? [08:44:24 CET] <JEEB> very lately. not sure if 3.2 has it [08:45:23 CET] <JEEB> http://git.videolan.org/?p=ffmpeg.git;a=commit;h=ca6cae73db207f17a0d5507609… [08:46:01 CET] <ElAngelo> ok, then that can't be the problem [08:46:10 CET] <ElAngelo> most of the time i'm using a version from may [08:47:10 CET] <JEEB> also of course that thing does "fix" some files as long as you don't even think of -c copy and are using ffmpeg cli ... (api users had their stuff broken without compilation breaking) [08:47:41 CET] <JEEB> so I guess it's worth trying the latest head :p [08:48:43 CET] <ElAngelo> ok let's see if i can get a version from git [08:49:16 CET] <JEEB> and in the end unless you want to find out whether it was a bad file or missing support for x yourself you'll have to come up with a sample that you can share [08:50:03 CET] <ElAngelo> yes i figure so. i'm going to check with my client if there is any movie they wouldn't mind sharing [08:50:32 CET] <JEEB> well it doesn't mean movie, just a file that's created the same way. the smaller the better actually [08:50:53 CET] <JEEB> as long as it gets created the same way [08:50:59 CET] <JEEB> and is affected [08:51:30 CET] <ElAngelo> nod [08:51:40 CET] <ElAngelo> the one i have now is only 1 minute long [08:51:43 CET] <ElAngelo> but it's 1.8G [08:51:50 CET] <ElAngelo> rawvideo [08:51:57 CET] <ElAngelo> av1x codec [08:52:02 CET] <ElAngelo> pcm audio [08:52:13 CET] <ElAngelo> and actually a very low resolution [09:40:38 CET] <sopparus> hello, is there a way to "restream" dash like hls? I noticed .mpd as input didnt work very well [09:41:32 CET] <sopparus> probably what I want is https://trac.ffmpeg.org/ticket/5269 then [09:44:59 CET] <JEEB> I think people are waiting on which XML parser FFmpeg will end up with so that DASH demuxer can be written [09:48:32 CET] <sopparus> ok so its work in progress then, thanks [10:14:51 CET] <sopparus> think we're talking within six months? [13:28:20 CET] <JonG> I am trying to make use of the FFmpeg MediaCodec decoder for Android that is in the current FFmpeg git master branch. I have built FFmpeg for Android using a variation of the JNI example from https://github.com/appunite/AndroidFFmpeg. I have written some code to take an command string and run the ffmpeg.c main function. I have used the --enable-jni [13:28:21 CET] <JonG> and --enable-mediacodec options in my configuration script. I am now wondering how I know whether MediaCodec is being used or not? [13:28:26 CET] <JonG> Here is paste of my command and output (-loglevel debug): http://pastebin.com/JaTC5hSc [13:28:33 CET] <JonG> I was expecting to see some mention of MediaCodec, such as: "h264_mediacodec" (https://github.com/FFmpeg/FFmpeg/blob/d5082a2ce776a2e43e25998aa07541c9ab7af…) or "MediaCodec started successfully, ret = %d\n" (https://github.com/FFmpeg/FFmpeg/blob/d5082a2ce776a2e43e25998aa07541c9ab7af… [13:28:33 CET] <JonG> 2), but I don't see either. [14:01:45 CET] <leotreasure> Hello, can someone please help me turn pictures into a movie? [14:02:08 CET] <leotreasure> i have them in this format: crazydancehair_00001.png [14:02:16 CET] <leotreasure> all the way to crazydancehair_00120.png [14:02:45 CET] <leotreasure> I tried this, but it didnt work: ffmpeg -f image2 -i crazydancehair%d_00120.png /tmp/crazyhair.mpg [14:03:16 CET] <JonG> Have you seen this: http://trac.ffmpeg.org/wiki/Create%20a%20video%20slideshow%20from%20images [14:06:45 CET] <leotreasure> JonG: I hadnt thanks this has a png example - perfect! [14:07:10 CET] <squ> how select audio channel with filter_complex [14:07:31 CET] <squ> .mkv file contains multipe audio streams and subtitles [14:12:06 CET] <JonG> @squ You could use the -map command: https://trac.ffmpeg.org/wiki/Map [14:12:31 CET] <squ> JonG: can you explain? [14:12:45 CET] <ciawal> Im trying to add an overlay image to only the last frame of a video without altering the video at all, but it seems whatever I do is resulting in the video sometimes being altered [14:12:46 CET] <ciawal> could anyone take a look at my script and see if theres a way to improve it? https://gist.github.com/ciaran/f37d71332d764d46aba6e1655099a4af [14:13:35 CET] <squ> -filter_complex [0:a] ... [14:14:03 CET] <squ> JonG: how to specify a stream here: -filter_complex [0:a] ... [0:v] ... [14:17:20 CET] <JonG> squ: If your file has e.g. 3 streams: [video, audio, audio], I think you can reference them as [0:0], [0:1] and [0:2] respectively. [0:a] chooses the best quality audio stream. [14:17:36 CET] <klaxa> to select the first audio stream: [0:a:0] [14:17:55 CET] <klaxa> that works with -map at least not sure if -filter_complex does too (but i'd think so) [14:18:18 CET] <squ> klaxa: how to do it with -map and keep filter_complex too? [14:18:55 CET] <klaxa> i don't understand [14:20:29 CET] <squ> JonG: so I need just replace [0:a] and [0:v] with [0:x] and [0:y] inside filter_complex, where x and y are streams [14:20:48 CET] <furq> yes [14:21:32 CET] <squ> is video always at 0? [14:21:39 CET] <squ> may I keep [0:v]? [14:21:42 CET] <furq> 13:17:36 ( klaxa) to select the first audio stream: [0:a:0] [14:25:22 CET] <furq> ciawal: what does "the video sometimes being altered" mean [14:27:35 CET] <ciawal> furq: e.g. sometimes the output video was clipped or out of sync [14:27:44 CET] <ciawal> current example I have is seemingly skipping the first frame [14:30:30 CET] <ciawal> it's a shame there's no way to get the frame count in the enable= expression, this approach seems so clumsy and brittle [15:07:01 CET] <pomaranc> does hls allow to somehow reset media sequence to 0? [16:16:35 CET] <nacho2k> hi, little question, I am trying to make a video of sequence pictures, but they have been made with different orientation, and the resolution is different, I have try to convert all pictures to the same resolution, but when I create the video it has pictures with different orientation, I have try to rotate them with convert, but it still the issue. Any advice? regards! [16:22:00 CET] <JonG> nacho2k: You can use the transpose filter to rotate content, e.g: http://stackoverflow.com/a/9570992/1847734 . I thought more recent versions of FFmpeg did this automatically based on meta data though ... [16:22:27 CET] <nacho2k> Jong: Thanks ! I will try it! [16:34:35 CET] <nacho2k> JonG: Actually, the issue is with the pictures, that I have mixed, in different Orientations, I have checked them with exiftool, and I have 2 kind of orientations, any idea? [16:43:36 CET] <JonG> I am certain there is a better way that this if you have a large number of images, or need to automate the function, but you can rotate individual images with e.g: 'ffmpeg -i input.png -vf "transpose=1" output.png'. [16:44:54 CET] <furq> you can try ffmpeg -i in.png -metadata:s:v:0 rotate=0 out.png [16:45:21 CET] <furq> if you have jpegs then you'd need to convert them to png if you use ffmpeg [16:45:26 CET] <furq> i imagine there are tools that can strip that metadata inplace [17:14:26 CET] <ciawal> I'm dumping all(?) the frames from a video like this: ffmpeg -i foo.mp4 thumb%02d.jpg [17:14:26 CET] <ciawal> but when I look at the first image, its not the image I see when I load up the video in quicktime and look at the first frame [17:14:26 CET] <ciawal> can anyone think of a reason why that could be? is it not exporting every frame? [17:22:48 CET] <relaxed> ciawal: use png [17:23:40 CET] <relaxed> ciawal: oh, try using thumb%04d.jpg [17:24:25 CET] <ciawal> no change [17:35:31 CET] <bencoh> then which frame do you see? [17:37:33 CET] <ciawal> it seems to start from the 2nd frame [17:38:24 CET] <ciawal> hm but VLC also skips this first frame [17:38:38 CET] <ciawal> what would quicktime be doing differently/special? [17:40:19 CET] <ritsuka> maybe quicktime handles the edit list properly, I don't know if it was eventually implemented in ffmpeg or not [17:43:17 CET] <bencoh> you mean that https://ffmpeg.org/pipermail/ffmpeg-devel/2016-August/197509.html ? [17:53:24 CET] <TheHumanPerson> Hello guys! [17:53:49 CET] <nacho2k> JonG,furq: thanks for your advice, I have fixed it, removing the exif rotation value to each .jpg [17:53:55 CET] <TheHumanPerson> I'm a bit of a newb at ffmpeg, used some GUIs before, but there's nothing better than the real deal. [17:54:40 CET] <klaxa> i do agree, i used guis before diving into the command line too [17:56:20 CET] <TheHumanPerson> Anyways, I want to ask, this is going to sound stupid, but bear with me, I'm looking to encode a video to h265 in a pure, true losless quality for certain reasons, and at the same time, very weirdly, I still want to achieve the lowest possible filesize while keeping the quality truly 100% losless. I know there are much better options out there, bu [17:56:20 CET] <TheHumanPerson> t I want to do this and I know what I'm doing. Any ideas? [18:00:10 CET] <TheHumanPerson> I think I could do:ffmpeg -i INPUT.mkv -c:v libx265 -preset placebo -x265-params lossless=1 OUTPUT.mkv [18:00:46 CET] <TheHumanPerson> however im not sure if this will generate the lowest file size at such quality settings. particularly, I'm not sure if this option uses 2 pass encoding or not [18:02:35 CET] <klaxa> you think correctly [18:02:50 CET] <klaxa> lossless encoding does not need 2 pass though [18:03:21 CET] <dl2s4> isnt the parameter for lossless -qp 0? but i guess most people dont now what they are doing when they encode to x265 [18:03:24 CET] <klaxa> i *think* in some rare cases -preset veryslow can yield better results than placebo but i'm not sure [18:03:32 CET] <klaxa> dl2s4: not for x265 [18:03:36 CET] <dl2s4> kk [18:03:39 CET] <dl2s4> ok [18:03:40 CET] <furq> -qp 0 is for x264 [18:03:47 CET] <furq> i'm not sure if -qp is even mapped for x265 [18:05:00 CET] <furq> you'll probably want to encode a short sample with veryslow and placebo [18:05:26 CET] <furq> i expect placebo is significantly slower so you probably want to make sure it's worth the marginal filesize reduction [18:05:27 CET] <klaxa> i would also advise to compare to lossless x264 encodes [18:05:42 CET] <furq> he seems pretty set on using hevc [18:06:17 CET] <klaxa> right [18:06:52 CET] <furq> if you just want lowest filesize then obviously do check other codecs [18:13:05 CET] <TheHumanPerson> furq: im not familiar with other codecs. I don't wish to use those weirdly named ones [18:13:33 CET] <TheHumanPerson> x265 is quite popular, has passed the test of time, and has better compatibility on some systems [18:13:45 CET] <furq> are you thinking of x264 [18:13:54 CET] <TheHumanPerson> no, x264 is old school now :P [18:14:03 CET] <furq> x265 doesn't meet the last two criteria [18:14:13 CET] <TheHumanPerson> which ones? [18:14:26 CET] <furq> the ones which aren't the first one [18:14:33 CET] <TheHumanPerson> -_- [18:14:41 CET] <furq> and it's not that popular really [18:15:13 CET] <furq> it's still unclear whether there'll ever be major uptake of hevc because of the insane licensing situation [18:15:14 CET] <TheHumanPerson> Well, h265 is quite popular [18:15:38 CET] <TheHumanPerson> But as a specific encoder, then I don't know how much popular x265 is. [18:15:45 CET] <furq> i'm talking about h265 [18:16:01 CET] <furq> x265 is the most popular h265 encoder by miles [18:16:09 CET] <TheHumanPerson> I thought so as well. [18:16:40 CET] <furq> and yeah try decoding any >1080p hevc content with libavcodec [18:16:42 CET] <TheHumanPerson> So, just to be sure, do I go ahead and select the options I showed you before? [18:16:58 CET] <furq> probably [18:17:01 CET] <TheHumanPerson> furq: also libavacodec doesn't count, ffmpeg devs take half of it every release [18:17:12 CET] <furq> well it's used by a lot of media players [18:17:22 CET] <TheHumanPerson> I wouldn't be surprised if libavacoded == ffmpeg [18:17:42 CET] <TheHumanPerson> they take so much code from each other anyways [18:17:47 CET] <furq> uh [18:17:55 CET] <furq> libavcodec is part of ffmpeg [18:18:00 CET] <TheHumanPerson> wtf [18:18:15 CET] <furq> are you thinking of libav [18:18:20 CET] <TheHumanPerson> yes [18:18:24 CET] Last message repeated 1 time(s). [18:18:24 CET] <TheHumanPerson> libav [18:18:24 CET] <furq> the badly-named ffmpeg fork which burned out [18:18:38 CET] <TheHumanPerson> um, first of all the name isnt too bad, and secondly its fine [18:18:48 CET] <furq> the name is bad because it's needlessly confusing [18:18:53 CET] <furq> as we just found out [18:19:09 CET] <TheHumanPerson> in fact, before it "burned out", ffmpeg devs took lots of its code. they wouldn't have done that if it sucked [18:19:32 CET] <furq> they took pretty much all of its new code [18:19:39 CET] <TheHumanPerson> and its not burned out or anything, version 12.0 was released 21 days ago [18:20:12 CET] <furq> all the major distros have switched back to ffmpeg though [18:20:45 CET] <TheHumanPerson> You might be thinking of the time libav changed the command names to "avconv", but that's a loong time ago [18:21:08 CET] <furq> huh [18:21:09 CET] <TheHumanPerson> However, I agree with you, Debain Strech has switched to ffmpeg back then [18:21:16 CET] <TheHumanPerson> Stretch* [18:21:28 CET] <bencoh> (finally!) [18:22:08 CET] <bencoh> furq: libav "burned out"? [18:22:26 CET] <furq> most people seem to have stopped using it [18:22:32 CET] <TheHumanPerson> bencoh: nah, it's up and running [18:23:07 CET] <TheHumanPerson> furq: if most people counts as "Debain Stretch switched to ffmpeg, and Gentoo is one of the distros that lets you choose on your own", that's hardly most people [18:23:28 CET] <furq> are there any major distros which use libav [18:23:38 CET] <TheHumanPerson> furq [18:23:41 CET] <TheHumanPerson> yeah [18:23:52 CET] <TheHumanPerson> Gentoo lets you pick between libav and ffmpeg: https://packages.gentoo.org/packages/virtual/ffmpeg [18:24:09 CET] <furq> i said major [18:24:48 CET] <TheHumanPerson> Since when is Gentoo "minor"?? [18:25:09 CET] <TheHumanPerson> By that logic, Ubuntu isn't major, it's based off Debian [18:25:15 CET] <TheHumanPerson> So Debian is "more [18:25:21 CET] <TheHumanPerson> "more" major than Ubuntu [18:25:31 CET] <TheHumanPerson> which doesn't make sense [18:25:32 CET] <TheHumanPerson> anyways [18:25:56 CET] <TheHumanPerson> I'm going to encode my video with ffmpeg -i INPUT.mkv -c:v libx265 -preset placebo -x265-params lossless=1 OUTPUT.mkv [18:26:15 CET] <TheHumanPerson> this is going to take a looooooong time, so I want to make sure my settings are right [18:27:08 CET] <furq> like i said, encode a sample first [18:27:46 CET] <TheHumanPerson> well, i dont have a sample [18:27:50 CET] <furq> -t 60 [18:27:55 CET] <TheHumanPerson> ?? [18:28:10 CET] <furq> add -t 60 to that command and it'll encode the first minute [18:28:25 CET] <TheHumanPerson> I found a random 100KB video [18:28:27 CET] <TheHumanPerson> ill try that [18:28:38 CET] <furq> you probably want it to be a representative sample [18:28:44 CET] <TheHumanPerson> okay [18:28:47 CET] <TheHumanPerson> ill do that now [18:28:48 CET] <furq> and yeah i would compare it with other presets and codecs [18:29:04 CET] <furq> i have no idea how good x265's lossless mode is [18:29:33 CET] <TheHumanPerson> "no such file or directory" [18:29:36 CET] <TheHumanPerson> weird error [18:30:01 CET] <furq> -t 60 goes after the input filename [18:30:57 CET] <bencoh> silly question, but .... why would you encode a video using x265/placebo? [18:31:34 CET] <TheHumanPerson> bencoh: Because I'm crazy and I want to do it. [18:31:48 CET] <bencoh> I mean ... I get the idea of lossless compression and storage size, but .... it'll just take ages. is it worth it? [18:32:27 CET] <bencoh> TheHumanPerson: most people around here aren't exactly sane either, but ... :D [18:32:36 CET] <TheHumanPerson> bencoh: :D [18:32:46 CET] <TheHumanPerson> How do I know when the command is finished processing? [18:32:55 CET] <bencoh> when it returns (?) [18:33:30 CET] <TheHumanPerson> Okay :P [18:33:43 CET] <TheHumanPerson> im a newb at this stuff :P [18:35:10 CET] <TheHumanPerson> I'm a large proponent of graphical interfaces, but sometimes graphics cannot provide what the classic command-line can do [18:35:49 CET] <TheHumanPerson> And it took me a while to understand this, no amount of trying and installing and uninstalling software will ever beat just using raw ffmpeg [18:39:34 CET] <furq> guis are nice if you need to do one very specific thing [18:39:58 CET] <TheHumanPerson> bencoh: I read the numbers wrong, thought I was 100 fps, but I'm 0.4 fps :D [18:40:18 CET] <furq> if it's even possible to write a gui tool that covers everything ffmpeg does then i'm pretty sure it'll be unusably complex [18:41:15 CET] <furq> and yeah you might want to use -t 30 for your sample [18:41:30 CET] <TheHumanPerson> I did [18:41:41 CET] <TheHumanPerson> GPU temp was 85C so I closed ffmpeg :P [18:41:59 CET] <furq> gpu? [18:42:03 CET] <TheHumanPerson> yeah [18:42:10 CET] <TheHumanPerson> doesnt ffmpeg use gpu video block? [18:42:16 CET] <furq> not unless you tell it to [18:42:23 CET] <TheHumanPerson> I'm running a very old GPU though [18:42:25 CET] <furq> and certainly not for encoding with x265 [18:42:39 CET] <furq> it could maybe use it for decoding, depending on the codec [18:42:54 CET] <TheHumanPerson> so you mean in this case it _wasn't_ heating up my GPU? [18:42:59 CET] <furq> i doubt it [18:43:11 CET] <TheHumanPerson> All I know is the fan noise was super crazy [18:43:32 CET] <TheHumanPerson> I shouldn [18:43:35 CET] <TheHumanPerson> shouldn [18:43:35 CET] <furq> maybe x265 automatically uses some opencl stuff [18:43:41 CET] <TheHumanPerson> I don't know [18:43:50 CET] <furq> i know x264 can use it for lookahead but you need to explicitly enable it [18:43:57 CET] <TheHumanPerson> I shouldn't try doing losless encoding on such an old system, but im a crazy person [18:44:06 CET] <furq> how old is this system [18:44:12 CET] <furq> by which i mean what cpu is it [18:44:14 CET] <TheHumanPerson> Sandy Bridge [18:44:27 CET] <furq> that's not that old [18:44:29 CET] <TheHumanPerson> mobile CPU [18:44:35 CET] <TheHumanPerson> i5-2540M [18:44:38 CET] <furq> yeah that's a much bigger problem [18:45:00 CET] <TheHumanPerson> it was a fast, high-end model back at the time it was made [18:45:17 CET] <TheHumanPerson> at least that's what notebookcheck says [18:45:19 CET] <furq> x265 placebo on a 35W dualcore sounds like a thoroughly miserable experience [18:45:41 CET] <furq> i guess at least your source isn't 4k [18:46:08 CET] <bencoh> :D [18:46:13 CET] <TheHumanPerson> it's 720p. I would have downloaded the 4k version now that you reminded me, but the file size was just stupid [18:46:15 CET] <TheHumanPerson> :P [18:46:35 CET] <bencoh> are you actually decoding from h265 as well? [18:46:36 CET] <furq> well yeah you'd be lucky to get 0.1fps with a 4k source [18:46:43 CET] <furq> especially if the source is hevc [18:46:51 CET] <TheHumanPerson> bencoh: No idea what the source file is. [18:46:53 CET] <furq> i get about 5fps decoding 4k hevc with ffmpeg on a desktop [18:47:01 CET] <TheHumanPerson> furq: what specs? [18:47:06 CET] <bencoh> furq: 5fps? nice :p [18:47:11 CET] <bencoh> could be worst [18:47:16 CET] <bencoh> -t+e [18:47:16 CET] <furq> i7 920 at 3.4 or so [18:47:24 CET] <TheHumanPerson> bencoh: mine was 0.4 fps [18:47:29 CET] <TheHumanPerson> could be worse, eh? [18:47:36 CET] <furq> i'm talking about decoding [18:47:57 CET] <TheHumanPerson> ah well then time to upgrade, folks! [18:48:01 CET] <TheHumanPerson> i7 920 is old school [18:48:08 CET] <furq> it's not really the cpu's fault [18:48:13 CET] <furq> ffmpeg's hevc decoder is really slow [18:48:28 CET] <TheHumanPerson> If you're really on a budget, get a good quality used 3rd gen CPU [18:48:30 CET] <bencoh> s/slow/still not heavily optimized/ :) [18:48:36 CET] <TheHumanPerson> Or better yet, a used Haswell i7 [18:48:49 CET] <TheHumanPerson> that thing will CRUSH your i7 920\ [18:48:51 CET] <furq> bencoh: isn't it that they don't want intrinsics in the core [18:49:23 CET] <TheHumanPerson> Sorry, I mean destroy, not crush. [18:49:41 CET] <furq> the problem with sandy bridge and newer is that you need to pay the intel overclocking premium [18:49:49 CET] <TheHumanPerson> lol [18:49:56 CET] <furq> which is something i'd rather not do [18:49:57 CET] <TheHumanPerson> then keep your years old cpu [18:50:08 CET] <TheHumanPerson> and browse facebook [18:50:15 CET] <furq> i'm still looking for a xeon x5670 [18:50:44 CET] <TheHumanPerson> furq: Wait until the Kaby Lake Xeon 3 is released, and then read the reviews. [18:51:23 CET] <TheHumanPerson> x5670 is EOL [18:51:33 CET] <TheHumanPerson> And costed $1440.00 - $1443.00 at release [18:51:43 CET] <furq> yeah it costs £60 now and fits in my motherboard [18:51:47 CET] <TheHumanPerson> Now it's much cheaper, maybe, but wait for Kaby Lake [18:51:47 CET] <furq> and it's 6-core and clocks to 4ghz [18:51:52 CET] <TheHumanPerson> Now it's much cheaper, maybe, but wait for Kaby Lake [18:52:53 CET] <TheHumanPerson> bencoh: so uh, if I don't do losless, what other settings would your recommend that offer a good balance? [18:53:12 CET] <kerio> TheHumanPerson: can't you just use ffv1 [18:53:21 CET] <furq> ^ [18:53:55 CET] <TheHumanPerson> Keridos: not sure if ffv1 is even better than h265 in the first place [18:54:06 CET] <TheHumanPerson> h265 will have better support than ffv1 ever will, as a plus [18:54:09 CET] <kerio> "better" how [18:54:23 CET] <kerio> ffv1 is lossless [18:54:27 CET] <kerio> does h265 even have a lossless mode? [18:54:31 CET] <klaxa> on my laptop i get a hole 15 fps decoding :P [18:54:39 CET] <klaxa> kerio: yes [18:54:40 CET] <kerio> does it support high bpc? [18:54:56 CET] <kerio> like grayscale 16bpp [18:55:44 CET] <TheHumanPerson> kerio: anyways how I do encode ffv1.3, then? [18:55:54 CET] <kerio> -c:v ffv1 -level 3 [18:57:29 CET] <TheHumanPerson> kerio: Thanks. However, I read this article ( http://blogs.library.duke.edu/bitstreams/2015/05/08/the-pros-and-cons-of-ff… ) and it claims, just claims, that ffv1 may not be truly lossless [18:58:03 CET] <TheHumanPerson> and, seems like ffv1.3 , latest version, no updates since THREE years ago [18:58:16 CET] <TheHumanPerson> But I'm willing to deal with that for the timebeing [18:58:48 CET] <furq> In testing, it is difficult to verify that a file converted (compressed) to FFV1 and then converted back (decompressed) is an identical file to its original state. [18:58:51 CET] <furq> what [18:58:57 CET] <kerio> like [18:59:00 CET] <kerio> -f hash [18:59:03 CET] <kerio> :\ [18:59:03 CET] <furq> this article is bad [18:59:33 CET] <furq> the comments are pretty good though [19:00:00 CET] <furq> The concept of lossless compression is mysterious, and seemingly a paradox. How can it make a file smaller, without eliminating or irreversibly altering any data? [19:00:05 CET] <furq> who wrote this [19:00:17 CET] <bencoh> :D [19:00:27 CET] <TheHumanPerson> furq: Yeah, but the point was he claims ffv1 might not be truly lossless [19:00:35 CET] <TheHumanPerson> is this, truly true? (heh) [19:00:55 CET] <furq> well sadly he also claims there is no way of testing [19:00:59 CET] <furq> so i guess we will never know for sure [19:01:17 CET] <TheHumanPerson> you've not answered me [19:01:23 CET] <TheHumanPerson> and yes, you can easily test it [19:01:38 CET] <TheHumanPerson> take the pixel difference between two files [19:02:34 CET] <TheHumanPerson> anyways ill try it [19:02:36 CET] <furq> why are you giving credence to his other claims if you know that claim is bullshit [19:02:39 CET] <TheHumanPerson> ffmpeg -i <input_video> \ [19:02:40 CET] <TheHumanPerson> -acodec copy \ [19:02:40 CET] <TheHumanPerson> -vcodec ffv1 \ [19:02:40 CET] <TheHumanPerson> -level 1 \ [19:02:40 CET] <TheHumanPerson> -coder 1 \ [19:02:40 CET] <TheHumanPerson> -context 1 \ [19:02:40 CET] <TheHumanPerson> -g 1 \ [19:02:41 CET] <TheHumanPerson> <output_video> [19:04:25 CET] <TheHumanPerson> doesn't work [19:04:26 CET] <TheHumanPerson> weird [19:04:51 CET] <TheHumanPerson> looks like ffv1 has 2 pass though, should I use that? [19:08:00 CET] <TheHumanPerson> sorry for spamming too many lines :P [19:08:44 CET] <TheHumanPerson> furq: im getting significantly higher fps with ffv1 than x265, which is good, but how can this be? [19:09:05 CET] <TheHumanPerson> what does ffv1 do differently to achieve this? [19:11:20 CET] <furq> i imagine it doesn't compress as well [19:12:48 CET] <TheHumanPerson> perhaps, but at least it seems a bit faster [19:12:55 CET] <furq> it should be several orders of magnitude faster [19:13:07 CET] <bencoh> the usual tradeoff, yeah :p [19:13:48 CET] <furq> also you were using x265 placebo [19:14:02 CET] <bencoh> size, quality, speed, pick two [19:14:05 CET] <TheHumanPerson> on an unrelated note, im trying to check the total frames of the video, is this correct? ffprobe -v error -count_frames -select_streams v:0 -show_entries stream=nb_read_frames -of default=nokey=1:noprint_wrappers=1 input.mkv [19:14:23 CET] <furq> the slower presets will use significantly slower methods which don't compress that much better (if at all) [19:15:14 CET] <TheHumanPerson> Is x265 lossless ultrafast preset, actually ultrafast? [19:15:17 CET] <furq> e.g. x264 placebo will try to use 16 consecutive bframes, which takes much longer to analyse, even though it's very rare you'll ever see as many as 8 consecutive bframes [19:15:38 CET] <furq> i usually use -bf 5 because everything i've ever analysed drops to <1% after that [19:16:07 CET] <furq> -bf 16 on a clip which never uses more than 8 is spending a bunch of time effectively doing nothing [19:17:38 CET] <furq> and yeah at least with x264 lossless the slower presets are slower and compress better [19:17:49 CET] <furq> they won't set the same set of options because some of them don't make sense for lossless [19:18:08 CET] <furq> the same set as they would for lossy, that is [19:18:14 CET] <TheHumanPerson> So with x265, going slower doesn't bring as much benefits than it did with x264? [19:18:27 CET] <furq> i have no idea, i've used x265 once in my life to test something [19:19:02 CET] <furq> i assume the concepts are roughly the same [19:21:04 CET] <furq> fwiw if i was encoding a 720p clip lossless i'd probably use x264 with default settings [19:21:31 CET] <furq> it's not worth using anything slower than that unless you're encoding a ton of stuff and/or you're starved for space [19:23:32 CET] <TheHumanPerson> Basically, my 720p file is 1.5GB. I'm just testing what settings produce almost identical quality at a smaller size [19:23:42 CET] <JonG> furq: Someway back in the conversation you mention that FFmpeg won't use the GPU unless you tell it to... What is the option I should be looking at for that? I just wondered if it might be related to my question from earlier today about trying to use MediaCodec and the GPU on Android. [19:23:52 CET] <furq> nothing lossless will come out smaller [19:24:04 CET] <TheHumanPerson> Initially I'm trying true lossless, but next time when I encode I'll hopefully be much more reasonable :D :P [19:24:08 CET] <bencoh> TheHumanPerson: then don't use lossless encoding [19:24:13 CET] <bencoh> :) [19:24:20 CET] <furq> even x265 lossless with placebo will be orders of magnitude larger [19:24:32 CET] <furq> assuming that even gives the best compression [19:24:50 CET] <TheHumanPerson> bencoh: Like I said, im just trying it out. Sadly, wish I could figure out if the source video was already compressed or not, and uncompress it first [19:25:04 CET] <furq> JonG: -hwaccel [decoder-name] [19:25:06 CET] <bencoh> uncompress? [19:25:20 CET] <TheHumanPerson> bencoh: I mean extract the raw video [19:25:24 CET] <furq> i have no idea how that would work on android though [19:25:48 CET] <furq> TheHumanPerson: you can extract the raw video with -i src.mkv -c:v rawvideo dest.mkv [19:25:52 CET] <furq> you probably don't want to do that though [19:25:55 CET] <TheHumanPerson> furq: why? [19:25:57 CET] <TheHumanPerson> :D [19:26:02 CET] <dl2s4> huggeee :) [19:26:10 CET] <TheHumanPerson> I can already tell [19:26:13 CET] <furq> because it's the exact same video you already have [19:26:20 CET] <furq> it won't be "uncompressed" [19:26:25 CET] <furq> it'll still have all the compression artifacts [19:26:31 CET] <furq> just without any of the compression [19:26:53 CET] <furq> much like decoding an mp3 to wav [19:27:16 CET] <TheHumanPerson> furq: Actually, if the source video im testing is already compressed in some shape or form, then im wasting my time. hence the suggestion to extract the raw video first. otherwise if I can find from the metadata what compression it uses, i'd love that [19:27:26 CET] <furq> use mediainfo or ffprobe [19:28:01 CET] <furq> ffprobe -show_streams -select_streams 0:v -show_entries stream=codec_name foo.mkv [19:28:21 CET] <TheHumanPerson> Thanks! [19:29:02 CET] <bencoh> if you dont have the source then you're doomed. there is no such thing as "extracting the raw" :) [19:29:42 CET] <TheHumanPerson> furq: looks like the source video is h264. audio is aac 48khz [19:30:42 CET] <TheHumanPerson> Which means I partly wasted my time, however I've never really tried ffmpeg (the raw CLI) before, so I guess I'm doing it for science!! [19:30:50 CET] <TheHumanPerson> :D [19:30:52 CET] <JonG> Well maybe it is a naive assumption, but this: https://github.com/FFmpeg/FFmpeg/blob/21b68cdbae6576bb58c19dd44c0669293f7d0… suggests mediacodec gets registered as a possible hwaccel option. And MediaCodec seems to be the thing for HW acceleration on Android... Only one way to find out! Thanks. (That and I emailed the de [19:30:52 CET] <JonG> veloper who wrote the mediacodec code for FFmpeg) [19:31:02 CET] <JonG> furq** [19:31:10 CET] <TheHumanPerson> JonG: ask on #ffmpeg-devel, not here [19:32:07 CET] <JonG> OK, I assumed that was for FFmpeg developers only ... but I'll give it a try. [19:32:18 CET] <bencoh> actually it is, somehow [19:32:19 CET] <TheHumanPerson> Well, if you have a dev question, yes [19:32:36 CET] <TheHumanPerson> dev questions go to #ffmpeg-devel if I understand correctly. [19:32:41 CET] <bencoh> ffmpeg-devel is focused on ffmpeg internal dev [19:33:53 CET] <TheHumanPerson> ohhh okay then [19:34:02 CET] <JonG> I am not an ffmpeg internal developer (yet) :) [19:34:38 CET] <TheHumanPerson> :) [19:41:59 CET] <TheHumanPerson> bencoh: So, if I want to encode something in general in the future, what is a good balance of quality and speed for my poor hardware? [19:42:57 CET] <furq> x264 with the slowest preset you can tolerate [19:43:06 CET] <furq> except placebo which is a waste of time [19:43:12 CET] <furq> as the name implies [19:43:21 CET] <TheHumanPerson> furq: why not x265 [19:43:43 CET] <furq> if x264 veryslow is faster than you're willing to put up with then try x265 [19:43:47 CET] <furq> otherwise there's no point [19:44:30 CET] <furq> last time i checked, at the same encoding speed they performed roughly the same [19:44:46 CET] <furq> x265 only pulls ahead when you use the really slow presets [19:45:26 CET] <furq> and h264 is generally much nicer to work with than h265 [19:49:11 CET] <TheHumanPerson> furq: Guess what? source is 1.5GB, encoded is 27GB :P [19:49:31 CET] <TheHumanPerson> ive deleted it [19:49:43 CET] <TheHumanPerson> silly me for encoding an already h264 compressed video [19:50:23 CET] <TheHumanPerson> anyways, the source video is h264, around 58 minutes. im looking to reduce that 1.5GB to something quite less as a test, with similar quality [19:51:38 CET] <TheHumanPerson> not sure what presets to use [19:53:51 CET] <furq> -crf controls filesize [19:53:57 CET] <furq> higher value = smaller file [19:55:13 CET] <TheHumanPerson> furq: okay, but how do I let ffmpeg understand the video is already encoded in h264? I don't want to encode over that, I just want to modify the encoding [19:57:34 CET] <furq> you can't do that [19:58:12 CET] <TheHumanPerson> then what do I do? [19:58:24 CET] <TheHumanPerson> I've decided I want a CRF of 18 by the way, seems like a sane value [19:58:27 CET] <furq> either deal with the current filesize or deal with the generation loss [19:59:30 CET] <TheHumanPerson> I can deal with the current filesize, I'm just testing the options that's all [20:00:15 CET] <TheHumanPerson> You told me I shouldn't ever try to extract the raw video, which I supposed I would then encode again [20:01:18 CET] <furq> well that is what you're doing but ffmpeg does it internally [20:01:51 CET] <kerio> TheHumanPerson: ffmpeg -i sourcevideo.whatever -c:a copy -c:v h264 -crf 18 destvideo.mkv [20:06:01 CET] <TheHumanPerson> Thanks! [21:24:15 CET] <Phi> Moop [21:24:36 CET] <Phi> any idea how to detect if a hardware accelerated thing is usable on the platform? [21:24:59 CET] <BtbN> if it works, it's probably supported. [21:37:57 CET] <Phi> *faceplants* [21:38:22 CET] <Phi> I would hope so, but I'd rather detect if it doesn't work so I can set up fallbacks [21:38:30 CET] <kepstin> in other words, you check whether it's usable by trying to use it. If it doesn't work, do something else. [21:39:43 CET] <pgorley> there's nothing like that with the hwaccel api [21:40:19 CET] <pgorley> the way i do it in our own application is through a call to av_hwdevice_ctx_create and if it returns 0, it should work [21:40:55 CET] <pgorley> if it still doesn't work, i've set up my own fallback method that reinits the AVCodecContext [21:49:54 CET] <Phi> right-o [21:49:56 CET] <jkqxz> If you are only decoding with proper hwaccels (dxva, vaapi, vdpau) attached to normal ffmpeg decoders, then it is already set up to use the hardware when it can and otherwise transparently fall back to using software. [21:50:20 CET] <Phi> well, I'm going to be using Intel QSV, and if not, fall back on libx264 [21:50:44 CET] <jkqxz> The standalone decoders (cuvid, qsv, ...) do require you to do that yourself manually, since they don't work alongside the ffmpeg software decoders. [22:03:17 CET] <Phi> the thing is, Intel QSV will fall back by itself if it's not supported [22:15:47 CET] <Phi> which is bad, because then libx264 won't be used, sob sob [22:24:39 CET] <Phi> okay, I did configure with enable-hwaccel=h264_qsv, but nothing [22:25:16 CET] <jkqxz> What platform are you building on? [22:25:23 CET] <Phi> MinGW, the funnest one [22:25:31 CET] <jkqxz> (The option you want is --enable-libmfx.) [22:25:34 CET] <Phi> Windoze [22:25:41 CET] <Phi> I'll chuck that in there too [23:02:17 CET] <dkc> is there a pixel format for packed YUV422 10bpp? I only see 16 or 12bpp in AVPixelFormat [23:05:22 CET] <kerio> does hevc support gray16? [23:07:56 CET] <kepstin> no, but nothing stopping you from putting 16-bit grey into the luma of color video with the chroma channels left neutral. [23:08:18 CET] <kepstin> oh, wait, i'm wrong [23:08:27 CET] <kepstin> there is a monochrome 16 profile, huh [23:08:28 CET] <kepstin> neat [23:08:57 CET] <kepstin> whether you can find an encoder that supports it is the other question :) [23:12:06 CET] <kerio> x265 doesn't :< [23:12:27 CET] <kerio> hm what's a color profile with 16bit luma? [23:12:45 CET] <kepstin> looks like x265 tops out at 12bit support right now [23:13:34 CET] <kepstin> or i'm wrong about that too, need to stop looking at outdated docs [23:13:48 CET] <furq> kepstin: yuv4**p16le [23:13:53 CET] <furq> s/kepstin/kerio/ [23:14:04 CET] <Min> hey guys, i could need some help with a little issue ;) [23:14:57 CET] <kerio> can i assume that a constant chroma will result in negligible space increase? [23:15:33 CET] <furq> probably [23:15:33 CET] <kepstin> interesting, according to the docs on https://x265.readthedocs.io/en/default/cli.html#input-output-file-options you might actually be able to do 16 bit monochrome? [23:15:50 CET] <kepstin> they list 4:0:0 monochrome as a supported chroma subsampling [23:16:46 CET] <kepstin> oh, the output depth does still max out at 12 [23:16:52 CET] <kerio> rip [23:16:55 CET] <kepstin> you can give it 16bit input :/ [23:16:56 CET] <Min> so i got the following command: ffmpeg -r 1 -loop 1 -i picture.jpg -i audio.m4a -c:v libx264 -tune stillimage -c:a aac -b:a 256k -r 1 -shortest "video.mp4" Its working, but my output file has an extra one minute longer without audio in the end [23:17:21 CET] <Min> *is [23:17:26 CET] <furq> Min: use a higher -r [23:17:47 CET] <furq> or a newer ffmpeg since iirc this was fixed [23:18:39 CET] <Min> higher -r muliplies the render time ;I downloaded the lastest windows version today [23:19:10 CET] <Min> but thanks, ill try other values for -r [23:24:35 CET] <furq> oh nvm i remember how i worked around this [23:25:03 CET] <furq> for some reason if you encode the audio in the same command it gets the length wrong [23:25:13 CET] <furq> if your source is m4a then the audio is probably already aac, so use -c:a copy [23:27:41 CET] <Min> I get "Could not find tag for codec alac in stream #1, codec not currently supported in container Could not write header for output file #0 (incorrect codec parameters ?): Invalid argument" using copy instead of acc [23:28:19 CET] <Min> already tried that earlier [23:39:47 CET] <kibibyte> hi [23:40:08 CET] <kibibyte> whats -bf option for ffmpeg ? [23:41:40 CET] <llogan> Min: a dumb workaround is to add -g 1 [23:41:54 CET] <kibibyte> is it video bit frames ? [23:43:02 CET] <llogan> "set maximum number of B-frames between non-B-frames" [23:43:25 CET] <kibibyte> but its only for specific codec ? [23:43:41 CET] <kepstin> it affects multiple codecs, in different ways [23:44:31 CET] <kibibyte> hm but is same as -bsf:v ? [23:44:47 CET] <kibibyte> bit stream filter [23:44:52 CET] <kepstin> no, they're completely unrelated [23:45:35 CET] <kibibyte> ok [23:46:27 CET] <llogan> most likely you can ignore -bf. if you're encoding with x264 the preset will automatically handle it anyway [23:46:55 CET] <Min> llogan: extra minute goes down to ~15 seconds [23:47:12 CET] <llogan> i guess i was wrong. [23:47:41 CET] <furq> encode it to a separate file first and then mux it in [23:47:55 CET] <furq> alternatively, if this is for youtube then just mux the existing file into mkv because youtube don't care [23:48:30 CET] <llogan> oh, i misread the duration when I did my single lazy ass test [00:00:00 CET] --- Wed Nov 9 2016

1 0

ffmpeg-devel.log.20161107
by burek 08 Nov '16

08 Nov '16

[00:54:03 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:2eb05eaa682e: adxdec: validate sample_rate [00:54:04 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:872fcfcc0f01: bfi: validate sample_rate [00:54:05 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:0009457dc097: ffmdec: validate sample_rate [00:54:06 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:fc022e77eed6: mov: validate sample_rate [00:54:07 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:a398f054fdb9: mov: validate time_scale [00:54:08 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:9d36602abc4e: mov: validate sidx timescale [00:54:09 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:074775462283: mpeg4audio: validate sample_rate [00:54:10 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:2c818c3821e4: mvdec: validate sample_rate [00:54:11 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:f748e3b5a219: nuv: validate sample_rate [00:54:12 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:eb205eda3fec: oggparsespeex: validate sample_rate [00:54:13 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:54b8fbbc5dc9: voc_packet: validate sample_rate [00:54:14 CET] <cone-499> ffmpeg 03Andreas Cadhalpun 07master:34aeb5dbc4fe: xmv: validate sample_rate [01:17:15 CET] <Chloe[m]> Jeez, that couldn't be squashed? [07:56:16 CET] <FishPencil> The Wiki for seeking <https://trac.ffmpeg.org/wiki/Seeking> says "HH:MM:SS:Microseconds", but I'm pretty sure that should be milliseconds [08:23:37 CET] <rcombs> FishPencil: you can use either [08:24:08 CET] <rcombs> it's really "fractional seconds" [08:24:25 CET] <rcombs> the "unit" is determined by the number of digits, effectively [08:29:19 CET] <FishPencil> rcombs: I see [08:29:59 CET] <FishPencil> So what does 'Chapter #0:0: start 0.000000, end 0.000247' mean then? Is 1.0 considered a second? [08:30:45 CET] <rcombs> that'll be in fractional seconds, yes [08:31:52 CET] <FishPencil> And if so, something isn't right with parsing ;FFMETADATA1 without a TIMEBASE, since I have START=0\nEND=247080 [08:34:02 CET] Action: rcombs shrugs [08:34:49 CET] <FishPencil> Not a big deal since I can add TIMEBASE=1/1000 to get ms, but the docs do say "If the timebase is missing then start/end times are assumed to be in milliseconds." [13:16:33 CET] <WonChulJo> hello developers [13:16:58 CET] <WonChulJo> I have a question about build ffmpeg in android with libopus [13:23:08 CET] <WonChulJo> is there anybody help me? [13:27:52 CET] <Compn> WonChulJo : i think everyone is asleep right now. try asking in #ffmpeg too [13:27:56 CET] <Compn> or one of the ffmpeg forums [13:30:52 CET] <WonChulJo> ok Thankyou [17:46:55 CET] <kierank> peloverde: are you involved in vp9-pq? [17:58:56 CET] <cone-527> ffmpeg 03Anssi Hannula 07master:9a51cd35b87d: avformat/hls: Factor copying stream info to a separate function [17:58:56 CET] <cone-527> ffmpeg 03Anssi Hannula 07master:3d2f636497f7: avformat/hls: Fix probing mpegts audio streams that use probing [17:58:56 CET] <cone-527> ffmpeg 03Anssi Hannula 07master:e2193b53eab9: avformat/hls: Add missing error check for avcodec_parameters_copy() [18:08:04 CET] <peloverde> kierank: no [18:08:24 CET] <kierank> ok I am curious how they are going to integrate proprietary dolby stuff into vp9 [18:08:34 CET] <kierank> and why [18:11:38 CET] <cone-527> ffmpeg 03Anssi Hannula 07release/3.2:32ceeb579e81: avformat/hls: Factor copying stream info to a separate function [18:11:39 CET] <cone-527> ffmpeg 03Anssi Hannula 07release/3.2:fc20e300580c: avformat/hls: Fix probing mpegts audio streams that use probing [18:11:40 CET] <cone-527> ffmpeg 03Anssi Hannula 07release/3.2:7568b0f553b0: avformat/hls: Add missing error check for avcodec_parameters_copy() [18:21:56 CET] <TD-Linux> kierank, according to https://source.android.com/devices/tech/display/hdr.html [18:22:02 CET] <TD-Linux> vp9-pq lacks all of the Dolby metadata [18:22:30 CET] <TD-Linux> it is effectively HDR10 with VP9 [19:28:43 CET] <cone-527> ffmpeg 03Michael Niedermayer 07master:4c8d69e534f1: ffserver: drop a bunch of apparently unneeded internal headers [19:28:44 CET] <cone-527> ffmpeg 03Michael Niedermayer 07master:f67d22f0fd6b: ffserver: Throw ffm.h out its not used except for a constant that is part of the format [19:28:45 CET] <cone-527> ffmpeg 03Michael Niedermayer 07master:10019c13e003: ffserver: use av_freep() for a case that is not clearly at the end of a function [19:28:46 CET] <cone-527> ffmpeg 03Michael Niedermayer 07master:ecb40dee3284: ffserver: eliminate hardcoded literal and duplicated array size [19:28:47 CET] <cone-527> ffmpeg 03Michael Niedermayer 07master:73ce1d864c6a: ffserver: steal SPACE_CHARS, remove one reason for internal.h that way [19:51:05 CET] <cone-527> ffmpeg 03Andreas Cadhalpun 07master:5249706e9d2e: mpegaudio_parser: don't return AVERROR_PATCHWELCOME [19:51:06 CET] <cone-527> ffmpeg 03Andreas Cadhalpun 07master:a305e0e5c0de: hls: move hls_close above hls_read_header [19:51:07 CET] <cone-527> ffmpeg 03Andreas Cadhalpun 07master:2c90316b46fc: hls: fix leaking avio_opts on hls_read_header error [19:57:58 CET] <microchip_> michaelni: wasn't ffserver being dropped? if so, why still update it? [19:59:08 CET] <atomnuker> it is being moved to another repo by reynaldo [19:59:17 CET] <atomnuker> (speaking of which, what's up reynaldo?) [20:00:43 CET] <microchip_> ic [20:01:36 CET] <michaelni> microchip_, it gets more testing here than in some repo noone knows of [20:42:43 CET] <cone-527> ffmpeg 03Michael Niedermayer 07master:ef6a786401f8: ffserver: Fix one AVFormatContext misuse [23:06:02 CET] <Zeranoe> So, little bit of an issue... My hosting provider decided that I was using too much bandwidth on their "unlimited bandwidth" plan... [23:07:24 CET] <DHE> I suspect you want #ffmpeg as this channel is for patches, bugs, etc [23:07:47 CET] <Zeranoe> Well this does concern the state of the Windows builds... [23:07:57 CET] <DHE> oh.. oh! [23:08:01 CET] <DHE> sorry didn't click [23:10:24 CET] <nevcairiel> Unlimited are often more of a "fair use" Deal, if it's too high they cut you off [23:11:15 CET] <Zeranoe> Everything has been fine for years. This is really frustrating [23:13:43 CET] <Zeranoe> Of course when you talk to them they think you know nothing "try enable caching".... Oh sure, I'll just cache archives, that will reduce bandwidth [23:14:28 CET] <DHE> what, browser side content cache? [23:15:51 CET] <RiCON> ovh's kimsufi is at 4TB, you could use the cheapest as cdn [23:17:47 CET] <Zeranoe> RiCON: That's a month? [23:18:46 CET] <Zeranoe> Because I'm already at 12TB for November... [23:21:30 CET] <ubitux> http://arstechnica.com/information-technology/2016/11/adobe-voco-photoshop-… [23:21:33 CET] <ubitux> fun stuff [23:22:31 CET] <ubitux> too bad the audience is full of braindead ppl [23:32:54 CET] <Zeranoe> Where does FFmpeg get their hosting/bandwidth from? [23:33:40 CET] <nevcairiel> Sponsoring usually [23:42:11 CET] <CFS-MP3> /join #ccetractor [23:42:21 CET] <CFS-MP3> Sorry [00:00:00 CET] --- Tue Nov 8 2016

1 0

ffmpeg.log.20161107
by burek 08 Nov '16

08 Nov '16

[00:12:34 CET] <Phi> make still insists there's nothing to do [00:12:54 CET] <andai> when doing multi-pass encoding the output file doesn't seem to matter. Where is the information used in subsequent passes going? [00:13:26 CET] <c_14> Phi: what if you `make ffmpeg' [00:13:43 CET] <c_14> andai: ffmpeg2pass.log or something along those lines [00:13:52 CET] <c_14> Look at the docs for passlogfile [00:15:20 CET] <andai> c_14: oh neat, it just goes in the folder. Thanks! [00:18:47 CET] <Phi> make: *** No rule to make target 'ffmpeg'. Stop. [00:21:32 CET] <Phi> ._. [00:22:03 CET] <c_14> When's the last time your computer's had an exorcism? [00:23:09 CET] <SchrodingersScat> you need an old priest and a young priest [00:23:39 CET] <Phi> another set? [00:23:46 CET] <Phi> I can't find the last two priests [00:24:08 CET] <SchrodingersScat> they often flatten themselves and go into the wall, like cats [00:25:11 CET] <Phi> http://pastie.org/private/hlozbhpxhzefcjxe2d9ww [00:26:13 CET] <c_14> tried updating mingw? [00:29:31 CET] <Phi> update [00:29:37 CET] <Phi> do it, earth program [00:31:24 CET] <andai> Oh it looks like multi pass encoding for x265 doesn't work yet [00:31:29 CET] <andai> the log file is always empty [00:31:40 CET] <andai> works when i change x265 to x264 [00:31:47 CET] <Phi> just did it [00:31:52 CET] <Phi> no updates to do :3 [00:37:20 CET] <Phi> T.T [00:48:35 CET] <Phi> pacman updating is failing [00:48:48 CET] <Phi> says corrupt database [00:49:39 CET] <Phi> where's my sherlock holmes hat [01:17:25 CET] <Phi> ...what's the point in an "init" function that depends on being init'd already [01:17:32 CET] <Phi> pacman get your swizzlesticks together [01:33:30 CET] <DHE> Phi: ?? [02:09:34 CET] <Phi> I did pacman-something --init [02:09:43 CET] <Phi> it complained it needed something to init [02:09:48 CET] <Phi> facepalm.com [02:13:12 CET] <Phi> I ended up doing a full uninstall/install of msys2 [02:14:00 CET] <Phi> then after that, rename link.exe from msys folder to msys-link.exe so MSVC compiling doesn't use wrong linker [02:14:15 CET] <Phi> then run it under VS 2013 Cmd Pmpt [02:14:25 CET] <Phi> seems to be happier now [02:14:41 CET] <furq> apparently they fixed msys2 pacman fucking up your install every time you do anything with it [02:14:47 CET] <Phi> (the configure is taking forever like normal, instead of failing quickly) [02:14:59 CET] <furq> i'm sure you'll get some more fun from it though [02:15:03 CET] <Phi> defo [02:15:09 CET] <Phi> ffmpeg is just full of surprises [02:15:18 CET] <furq> well this isn't really ffmpeg's fault [02:15:25 CET] <Phi> lies [02:15:43 CET] <furq> it's your fault for wanting to use msvc like some dirty perv [02:15:46 CET] <Phi> if I didn't have to compile ffmpeg through msys2, none of this could happen [02:16:11 CET] <Phi> well, I'm producing a DLL for a project that someone else made [02:16:14 CET] <Phi> so it's their fault [02:16:36 CET] <furq> i think ultimately it's msvc's fault [02:16:45 CET] <furq> not for the first time [02:17:28 CET] <Phi> well, no, it'd be msys2 [02:17:43 CET] <Phi> msvc was fine with compiling, the configure failed [02:18:50 CET] <Phi> and failed in a way that seemed like there was no errors [02:21:12 CET] <Phi> just the make seemed to be happy with producing nothing :p [02:21:37 CET] <Phi> time to try my longer configure [02:24:37 CET] <Phi> if all else fails, at least there's creamed corn [03:56:47 CET] <kbarry> I seem unable to view icecast title metadata (visible in VLC) while using ffplay. I can't seem to find a switch to display it. Is icecast metadata fully supported in ffplay? [04:56:38 CET] <DeadSix27> @kbarry did you try the -report switch? [04:57:43 CET] <DeadSix27> I never worked with icy icecast, but a quick test of a random stream i found with -report on ffplay shows me icy-name/icy-genre/etc. [05:00:22 CET] <Phi> moop [05:00:46 CET] <Phi> well I had to reconfig and remake libx264 or ffmpeg couldn't see it, but it's working now [05:01:01 CET] <Phi> catcha later folkses [05:55:23 CET] <FishPencil> Is it possible to include chapter information in an mkv? [06:57:18 CET] <FishPencil> I really wish there was more documentation for -map_chapters [08:47:15 CET] <kerio> what's the point of nut when mkv exists? [08:47:17 CET] <kerio> efficiency? [08:48:41 CET] <dl2s4> FishPencil, it is, but dont ask me how to do it with ffmpeg [08:56:41 CET] <s0126h> kerio nut? [09:35:54 CET] <dl2s4> FishPencil, i just investigated a bit because i needed to know.. as usual mostly it helps to read the doc http://www.ffmpeg.org/ffmpeg-formats.html#Metadata section 5, works perfect, tested [11:44:41 CET] <Betablocker> hey ho - anybody got a hint for me how to dewarp a 360° fisheye video ? [11:45:05 CET] <Betablocker> the source video ist 1:1 eg. 1920x1920 [11:46:00 CET] <Betablocker> i found this is the solution & https://trac.ffmpeg.org/wiki/RemapFilter [13:24:10 CET] <kdehl> Greetings. [13:30:30 CET] <kdehl> So, when I send H.264 data over RTP I should not include the NAL unit start prefix? [13:30:54 CET] <nonex86> kdehl, no :) [13:31:33 CET] <nonex86> kdehl: if you talk about annex-b startcode (0x000001/0x00000001) then no, you dont need to include it in packet [13:32:04 CET] <kdehl> nonex86: Hello again! :) [13:32:11 CET] <kdehl> nonex86: Okay. That's what I suspected. [13:32:14 CET] <nonex86> kdehl, hey :) [13:32:24 CET] <kdehl> We had a bad consultant here, it seems. [13:32:45 CET] <nonex86> kdehl: ? :) [13:32:48 CET] <kdehl> nonex86: But what about the emulation prevention bytes? Do I need to bother about that? [13:33:09 CET] <nonex86> kdehl: they are already in stream, isnt it? [13:33:16 CET] <kdehl> nonex86: A consultant wrote a routine that sends H.264 data over RTP including the start prefix. [13:33:36 CET] <jkqxz> You need to do quite a bit of work to send H.264 over RTP. See <https://tools.ietf.org/html/rfc6184>. [13:34:01 CET] <nonex86> kdehl: i saw many ip cameras - bosch, axis, sanyo, sony - noone is include startcode in the packet [13:34:14 CET] <kdehl> nonex86: I guess. It confuses me, how do you ever know whether it _is_ an emulation prevention byte, or just valid H.264 data from a codec? [13:34:26 CET] <kdehl> nonex86: That's exactly what I wanted to hear! [13:34:39 CET] <kdehl> jkqxz: Thanks. That's the one I'm trying to interpret. :) [13:36:02 CET] <kdehl> But it seems that the codec still needs the prefix when you send data that is to be decoded. So I have to remove it when I send data over RTP and then re-insert it in front of all NAL units before I send data for decoding. [13:36:06 CET] <kdehl> It seems. [13:36:07 CET] <jkqxz> If you have stream with single-NAL-unit packets only then it's easy. You just take the packets and add a start code to each one and you have an annexb stream. [13:36:36 CET] <nonex86> hdehl: correct, you should append start codes to the stream [13:36:47 CET] <nonex86> *kdehl [13:36:51 CET] <nonex86> sorry :) [13:36:56 CET] <kdehl> jkqxz: Indeed. That is the packetization type I'm using now. But faultily so. [13:36:56 CET] <jkqxz> If you have any of the other modes then you need to do quite a bit of work to reconstruct the stream, removing the extra headers, pasting together fragments and adding start codes as appropriate. [13:38:12 CET] <kdehl> jkqxz: Right. I plan to do non-interleaved mode now. But I just realized that those prefix headers are still there on the RTP packets, which confuses Wireshark, so I figured the single-NAL unit packetization mode was faulty as well. [13:38:35 CET] <kdehl> Have to fix that first. [14:54:29 CET] <FishPencil> How could I add a chapter to a FFMETADATA1 file that goes from x to the end of the stream? Do I actually need the time of the stream to give the chapter an END [15:05:24 CET] <FishPencil> Do the start and end points even matter, or could I just make them the same? [17:42:16 CET] <nyuszika7h> how do I mark a stream as NOT default? [17:42:30 CET] <nyuszika7h> I marked one explicitly as default but it seems it resulted in two default audio tracks [17:43:00 CET] <furq> -disposition:a:0 0 [17:43:09 CET] <nyuszika7h> ok, thanks [17:44:10 CET] <jason__> What is usually the bottleneck in extracting thumbnails from video with ffmpeg? CPU? [17:45:29 CET] <bencoh> most probably yeah [17:49:36 CET] <jason__> could any sort of GPU acceleration help with this? I'm looking at the options on the docs and cuvid seems like a possibility [17:49:50 CET] <furq> probably [17:50:23 CET] <furq> if you're using the thumbnail filter then it has to buffer a bunch of raw frames in memory, so i guess the bottleneck could be swapping [17:50:38 CET] <furq> but otherwise it'll be decode speed [17:50:55 CET] <jason__> I'm not. Just extracting blindly at a certain interval. [17:50:57 CET] <bencoh> furq: not necessarily, depending on how this filter works [17:51:08 CET] <furq> it mentions in the docs that it does that [17:51:15 CET] <bencoh> :/ [17:51:20 CET] <bencoh> it doesn't have to though :( [17:51:28 CET] <furq> yeah i don't get it either [17:52:12 CET] <furq> hopefully it does it so that it can make a better choice about which is the best frame [17:52:28 CET] <furq> but it wouldn't surprise me to find it was because of some fundamental libavfilter design constraint [17:52:40 CET] <bencoh> maybe [17:53:24 CET] <jason__> bleh building ffmpeg on windows looks like a PITA [17:53:30 CET] <furq> libavfilter can't seek backwards can it [17:53:35 CET] <furq> that's presumably the reason [17:54:16 CET] <furq> i know the reverse filter requires you to buffer the entire clip as rawvideo [17:57:24 CET] <jkqxz> Is speed the important thing here? Scaling before the thumbnail filter is possible, and in that case you can do it in hardware. [17:57:52 CET] <furq> jason__: you can try -skip_frame nokey [17:57:59 CET] <jkqxz> For example, something like "ffmpeg -hwaccel vaapi -hwaccel_output_format vaapi -i input.file -vf scale_vaapi=160:90:rgba,hwdownload,thumbnail=100 output.file". [17:58:42 CET] <furq> can you do that with dxva2 [17:59:02 CET] <jason__> I'm not really using the thumbnail filter. [17:59:07 CET] <jkqxz> Oh, Windows. [17:59:29 CET] <jkqxz> You might be able to do it with qsv there? dxva2 is only for decode, it isn't really used in lavfi. [18:01:06 CET] <jkqxz> jason__: Perhaps it might help if you explain better what you want as the output? [18:02:59 CET] <jason__> Nothing complicated really. I have a bunch of videos. I want to extract a bunch of frames from the videos. I don't care if it extracts frames that are similar, I just want to do it quickly. [18:03:28 CET] <furq> use -skip_frame nokey then [18:04:15 CET] <jkqxz> Yeah, if your input has key frames at the right sort of interval that's totally the best way. [18:04:37 CET] <jason__> Yeah I'm not sure about that. How frequent are key frames usually? [18:05:01 CET] <furq> depends on a lot of things but probably about 10 seconds [18:05:16 CET] <furq> if you need exact times then that's not going to work [18:05:16 CET] <jason__> yeah that's probably fine [18:05:25 CET] <furq> it should be much faster [18:08:30 CET] <jason__> will the -vf fps option cooperate with that? [18:09:06 CET] <jason__> like potentially skip keyframes? [18:09:15 CET] <jason__> without trying to get exact timing? [18:09:32 CET] <furq> i have no idea how you'd use -vf fps while extracting thumbnails [18:10:28 CET] <jason__> the examples have like -vf fps=1/60 to extract one frame per minute [18:11:08 CET] <furq> oh [18:11:16 CET] <furq> i think that'll work [18:11:33 CET] <jason__> I just wondered if it would get bogged down if the fps and keyframe rate weren't compatible [18:12:04 CET] <jason__> like if it would prioritize the fps over ignoring non keyframes [18:12:20 CET] <furq> -skip_frame is applied before the filterchain afaik [18:12:56 CET] <furq> so the filters will just process a stream of decoded keyframes [18:13:38 CET] <jason__> So if I tell it extract one frame per 60 seconds and it gets a keyframe every 7 seconds, what does it do? [18:13:50 CET] <furq> no idea [18:13:59 CET] <furq> there's one way to find out [18:14:00 CET] <jason__> I guess I'll find out [18:14:02 CET] <jason__> yeah :P [18:14:36 CET] <furq> i assume it'll just use the frame with the closest pts [18:14:53 CET] <jason__> yeah it just seems like it could be tricky [19:45:51 CET] <ElAngelo> with mplayer it's possible to specify custom codecs on runtime (playing with codecs.conf) is that possible for ffmpeg too? [19:54:44 CET] <jason__> If I'm extracting frames from a video is it possible to get the timestamps at which the frames were extracted so that it may be mapped to the file name somehow? [19:56:15 CET] <furq> i think you could if you were extracting to images [19:58:36 CET] <jason__> I am, but the format specifier options for the filename are pretty limited though. [19:59:04 CET] <jason__> it seems like you can only get an (optionally 0 padded) sequence from 1 to N or a timestamp [20:09:46 CET] <furq> yeah i don't see anything [20:09:58 CET] <furq> i'm probably thinking of drawtext [20:26:57 CET] <greatt> hi [20:27:31 CET] <greatt> im trying to add custom subtitles to video [20:28:10 CET] <greatt> and i got this error: Cannot find a matching stream for unlabeled input pad 0 [20:30:33 CET] <greatt> im using this command: http://pastebin.com/HiT4kdVy [20:32:02 CET] <furq> replace -filter_complex with -vf [20:32:08 CET] <furq> and replace that ; with a , [20:33:32 CET] <greatt> omg it worked [20:33:38 CET] <greatt> but i cant see subtitles [20:35:41 CET] <greatt> i copied font config from somewhere else [20:35:43 CET] <mdijkerman> hello there, I'm trying to download the windows build but the webserver gives me a 403. Is there a backup webserver for it or can I download it using bittorrent somewhere? [20:36:11 CET] <greatt> maybe this is the problem [20:43:55 CET] <greatt> ok i find the issue [20:44:39 CET] <greatt> it takes subtites start time as 0 instead of 5:58 [20:45:53 CET] <ElAngelo> i have a video file which i can't decode with ffmpeg but i can with mplayer if i specify -demuxer=mov [20:46:01 CET] <ElAngelo> is there any analog option in ffmpeg for this? [20:46:10 CET] <ElAngelo> i've tried -f mov [20:46:14 CET] <ElAngelo> but that didn't help a bit [20:48:15 CET] <notmateo> Hey is there a reason that the Windows builds part of the download page is returning a 403? [20:48:54 CET] <mdijkerman> thanks notmateo for confirming it is not just me. I have the same problem [20:49:14 CET] <greatt> zeranoe.com was down 30mins ago [20:49:21 CET] <notmateo> oh really? [20:49:42 CET] <greatt> http://zeranoe.com/ [20:50:18 CET] <notmateo> well damn, guess i'm stuck waiting [20:51:22 CET] <greatt> found my issue.. https://trac.ffmpeg.org/ticket/2067 [20:53:52 CET] <foobarfreak> Hi everyone. Is there a way I can generate the HLS m3u8 playlist using ffmpeg without transcoding the video and then later create the segments on demand? [20:55:06 CET] <foobarfreak> I've tried generating a playlist programatically with fixed lengths (10 seconds) and then using ffmpeg -ss and -t but I think that is wrong because the segments it generates are not always exact [20:55:22 CET] <foobarfreak> (I'm trying to write a simple HLS streamer for fun) [21:01:27 CET] <greatt> Anyone knows how to style subtitles? [21:04:49 CET] <klaxa> greatt: get aegisub [21:05:23 CET] <klaxa> the most sophisticated subtitle editor i know of [21:08:47 CET] <greatt> klaxa [21:08:54 CET] <greatt> yep , thanks [21:14:02 CET] <dsc_> can ffmpeg detect a codec from a bytestream (4mb) [21:14:08 CET] <dsc_> so that's without a file header etc [21:14:15 CET] <dsc_> and without a moov atom (?) [21:16:57 CET] <jkqxz> dsc_: Not really; many compressed bitstreams have few distinguishing features to tell you what they might be. The probe features of ffmpeg can try to guess what something is, but aren't necessarily correct in all cases. [21:18:50 CET] <dsc_> jkqxz: Thanks for the info [21:19:38 CET] <dsc_> can I ask you a long question [21:19:41 CET] <jkqxz> That isn't to say that it doesn't work in many cases, but don't rely on anything there. [21:19:52 CET] <dsc_> right, i wont then [21:20:36 CET] <dsc_> my other question was; say you have a .mp4 h264 hosted remotely which you can download over HTTP and you want to generate some screenshots from it [21:20:49 CET] <dsc_> without donwloading the whole thing [21:21:07 CET] <dsc_> http (and other protocols) allow you to download parts (accept-range i think) [21:21:25 CET] <dsc_> my question becomes; How much does ffmpeg need to have in order to make a screenshot [21:22:13 CET] <dsc_> if I make the assumption that the file is indeed h264 can I just get a random part of the movie file and look for keyframes then make a screenshot from it? [21:22:15 CET] <jkqxz> Yeah, you need accept-ranges support on the server for the seeking to work. [21:22:48 CET] <dsc_> jkqxz: lets say seeking works [21:23:40 CET] <jkqxz> If it were a raw h264 file (annex B bytestream) then that would probably work. If it's an mp4 then vital data is in the container, so you really need to read that - just picking random data out of the middle will not succeed. [21:24:14 CET] <dsc_> jkqxz: valuable information once again. So the objective would become obtaining that vital information first. Which is located .. in the file header? [21:25:05 CET] <dsc_> and by vital information do you mean "where all the keyframes are" and "what demuxer to use" [21:25:10 CET] <dsc_> idk if demux is the right word [21:25:28 CET] <jkqxz> I'll admit I don't actually know how good ffmpeg support for seeking over HTTP is, it might already work on an mp4. I suggest you just try it - "ffmpeg -ss 01:00 -i http://server/file". [21:27:21 CET] <jkqxz> No, I mean the sequence parameters of H.264 - those are separated into the container extradata in mp4. That also includes the packet boundaries, since what is actually stored is not an annex B bytestream. [21:27:47 CET] <dsc_> right [21:28:43 CET] <dsc_> I'm all for using that CLI thing you just suggested but ofc. rather use the ffmpeg library inconjunction with some python bindings and do my own thing because at one point making screenshots would not be the only thing, perhaps also retreive resolutions, bitrates. etc. [21:29:04 CET] <dsc_> -1 english [21:30:02 CET] <greatt> dsc_ im trying to do same thing [21:30:04 CET] <jkqxz> If it works on the CLI, then you can work out how to do it with the libraries. [21:30:04 CET] <dsc_> Your CLI thing just assumes the movie has a length of 01:00, which isnt always the case ^^ [21:30:17 CET] <dsc_> jkqxz: true true .. [21:30:34 CET] <jkqxz> Yeah, it was just arbitrary - I was wondering whether seeking to one minute in would work, or whether it would fetch all of the data in between. [21:30:58 CET] <dsc_> right, yeah. [21:31:02 CET] <greatt> best way is downloading file completly [21:31:10 CET] <dsc_> that's really not an option [21:31:15 CET] <dsc_> in my scenario [21:31:26 CET] <greatt> :( [21:31:32 CET] <dsc_> unfortunately ^^ [21:31:41 CET] <greatt> i got my files in server ^^ [21:31:57 CET] <dsc_> greatt: what about other servers? [21:32:05 CET] <greatt> but how can i add a png with overlay [21:32:30 CET] <greatt> dsc_: its only for our site so no other servers [21:32:52 CET] <dsc_> making screenshots from local files would be easy I guess. [21:33:17 CET] <greatt> yep [21:33:30 CET] <dsc_> m1dnight_: good to see you here m8 :) [21:34:03 CET] <greatt> im trying video to gif now [21:55:12 CET] <m1dnight_> oi [21:55:27 CET] <m1dnight_> Nice to see you again! :D [21:55:43 CET] <dsc_> m1dnight_: you keep appearing on random networks, in random channels :) [21:56:03 CET] <m1dnight_> Well, that's what you think! :p [21:56:09 CET] <m1dnight_> Where have you spotted me before? [21:56:16 CET] <dsc_> quakenet [21:56:22 CET] <dsc_> and ofc #python [21:56:23 CET] <dsc_> :) [21:56:33 CET] <m1dnight_> Ah yes, python. The weirdest language of them all. [21:56:52 CET] <m1dnight_> I gotta run [21:56:59 CET] <dsc_> yeah, you better run if you keep going like that [21:57:00 CET] <dsc_> later :) [21:57:02 CET] <m1dnight_> QT with the misses! See you around dsc_ ;D [22:09:53 CET] <greatt> hey [22:10:06 CET] <greatt> -i watermark-r1-720.png -vf "scale=720:404,overlay=660:350, [22:10:23 CET] <greatt> im trying to add overlay like this but its wrong i guess [22:10:53 CET] <furq> you need filter_complex for overlay [22:11:24 CET] <greatt> i must add that after or before this -vf thing? [22:13:24 CET] <greatt> -vf/-af/-filter and -filter_complex cannot be used together for the same stream. [22:13:25 CET] <greatt> ops [22:14:30 CET] <greatt> here is the full command http://pastebin.com/nQtGUu8n [22:19:29 CET] <greatt> anyone have idea? [22:22:04 CET] <c_14> replace ' -vf ' with ',' [22:27:55 CET] <greatt> c_14: its working but still waiting with no output [22:30:01 CET] <greatt> probably encoding 100% of video instead of 12 secs [22:30:48 CET] <greatt> and yes [22:33:54 CET] <greatt> thanks everyone! [22:34:08 CET] <greatt> have a nice day/night [22:34:13 CET] <greatt> love you [23:26:59 CET] <Polochon_street> Hi! I'm trying to decode songs using this https://github.com/Polochon-street/bliss/blob/master/src/decode.c, but the got_frame here https://github.com/Polochon-street/bliss/blob/master/src/decode.c#L259 gives me a « resource temporarily unavailable » for a specific .m4a file, do you have any idea of where that could come from ? [23:35:36 CET] <furq> Polochon_street: https://ffmpeg.org/doxygen/3.1/group__lavc__encdec.html [23:35:45 CET] <furq> The AVERROR(EAGAIN) return value means that new input data is required to return new output. In this case, continue with sending input. For each input frame/packet, the codec will typically return 1 output frame/packet, but it can also be 0 or more than 1. [23:37:31 CET] <Guest2286> Having issues escaping apostrophes when using the drawtext filter. Been bashing my head against the keyboard and that didnt seem to work so thought I'd come here lol. Anyone mind taking a look? http://pastebin.com/V0PGgwnS [23:42:31 CET] <c_14> Guest2286: text='MESSAGE DALERTE ou ALERTE'" [23:42:43 CET] <c_14> If at first you don't succeed, use something in unicode that is almost but not exactly what you're looking for. [23:43:08 CET] <c_14> (U+2019 RIGHT SINGLE QUOTATION MARK) [23:44:55 CET] <furq> it's all about U+02BC MODIFIER LETTER APOSTROPHE [23:45:20 CET] <furq> s/it's/itŒs/ [23:46:06 CET] <Guest2286> so how do i insert that to replace it in text? [23:46:46 CET] <furq> copy and paste [23:47:03 CET] <c_14> Either copy what furq or I wrote here, grab the character from fileformat or find out how your terminal supports unicode input [23:48:15 CET] <Guest2286> text='MESSAGE D (U+2019 RIGHT SINGLE QUOTATION MARK) ALERTE ou ALERTE'" ? [23:48:53 CET] <Polochon_street> furq: I'm decoding, not encoding [23:49:37 CET] <c_14> Guest2286: before that, when I wrote text= [23:49:46 CET] <furq> Polochon_street: my reading of that is that you need to call avcodec_send_packet again [23:50:54 CET] <Guest2286> derp. ill give it a go [23:51:21 CET] <Guest2286> thatnks c_14 and furq :D [23:52:35 CET] <Polochon_street> furq: well I'm on an a loop, so I'm calling avcodec_send_packet as long as av_read_frame() returns true [23:52:42 CET] <Polochon_street> so I call it over and over again [23:56:31 CET] <Polochon_street> okay, nevermind, the error comes from before [23:56:55 CET] <Polochon_street> I have "Invalid data found when processing input" for the avcodec_send_packet just before [00:00:00 CET] --- Tue Nov 8 2016

1 0

ffmpeg-devel.log.20161106
by burek 07 Nov '16

07 Nov '16

[02:13:35 CET] <cone-574> ffmpeg 03Michael Niedermayer 07master:c3e0755663a8: MAINTAINERS: Replace QSV maintainer [02:45:25 CET] <cone-574> ffmpeg 03James Almer 07master:e8a39f584a97: avformat/framehash: also print channel layout as a string [02:45:26 CET] <cone-574> ffmpeg 03James Almer 07master:75a13115cd6b: avformat/mux: remove unnecessary autobsf hack [13:20:57 CET] <cone-509> ffmpeg 03Stefano Sabatini 07master:9ccdf04bd0aa: lavf/mpegtsenc: move putstr8 definition up [13:20:57 CET] <cone-509> ffmpeg 03Stefano Sabatini 07master:755472124fb6: lavf/mpegtsenc: write metadata descriptor for timed ID3 packets [13:23:48 CET] <RiCON> BtbN: could you do a rebase of the cuda patches? [14:30:20 CET] <BtbN> RiCON, done. [14:30:47 CET] <BtbN> cuvid full hw transcoding seems to be broken though. Was the breaking change already merged, nevcairiel? [14:33:08 CET] <nevcairiel> no [14:33:23 CET] <BtbN> hm [14:33:30 CET] <BtbN> something else already broke it then [14:33:43 CET] <BtbN> it's complaining about runtime changes to filter graph parameters not being supported [14:36:25 CET] <nevcairiel> its possible that some other commit broke it, this entire thing was meant as a set, but due to the qsv thing in between it didnt go forward for a while, but i should be able to finish the last commit later today hopefully [14:37:04 CET] <BtbN> the non-hwaccel path still works fine, and isn't noticably slower, so it's not too horrible [15:26:39 CET] <cone-509> ffmpeg 03Derek Buitenhuis 07master:b54fd42d7d75: doc/libx26[45]: Add documentation for forced-idr [15:54:14 CET] <cone-509> ffmpeg 03DeHackEd 07master:6d50dff816ec: ffmpeg: parameters for filter thread counts [17:21:23 CET] <cone-509> ffmpeg 03Michael Niedermayer 07master:0e66dcd733d0: avfilter/af_volumedetect: Remove dependency on channel layout [21:31:21 CET] <cone-499> ffmpeg 03Michael Niedermayer 07master:c6bc823eea6a: avformat/tcp: Put struct sockaddr_in6 under #if [22:46:44 CET] <cone-499> ffmpeg 03fuqiuping 07master:2d72ea68f0ed: avformat/flvenc: support mp3 audio with 48khz [23:17:39 CET] <cone-499> ffmpeg 03Alex Agranovsky 07master:88db6c2890a6: avfilter/vf_drawtext: fixed default/flt formatting ignoring offset parameter [00:00:00 CET] --- Mon Nov 7 2016

1 0

ffmpeg.log.20161106
by burek 07 Nov '16

07 Nov '16

[00:09:30 CET] <Phi> there may be an mjpeg stream [00:09:36 CET] <Phi> http://pastie.org/private/iu93shhihljsj6f7inhg is still producing nothing to make [00:16:07 CET] <furq> i guess pastebin the output of configure [00:41:35 CET] <Phi> http://pastie.org/private/s4rvfax14wq1swjwrbidgw [00:41:51 CET] <Phi> (why does configure take like 5 minutes to produce anything >.> ) [00:42:18 CET] <__jack__> 5min ? creepy PC ? [00:43:12 CET] <__jack__> takes 7sec on mine [00:43:36 CET] <Phi> nah [00:43:53 CET] <klaxa> 15 seconds on mine [00:43:56 CET] <Phi> for me it takes that long, it doesn't write a line to output until several minutes have passed [00:44:20 CET] <BtbN> it doesn't output anything until it's done. [00:46:02 CET] <Phi> I'm running it under MinGW [00:46:14 CET] <Phi> so maybe that's it? [00:47:07 CET] <Phi> ironically it's a very fast PC [00:47:21 CET] <Phi> i7, 32GB, SSD [01:00:21 CET] <bplaxco> Does any one know why `execl("/usr/bin/ffmpeg", "-f", "alsa", "-i", "hw:1", "-y", out);` fails with "hw:1: Protocol not found" when just running the command in the command line works fine? [01:07:28 CET] <c_14> you have to pass /usr/bin/ffmpeg twice [01:07:30 CET] <c_14> afair [01:07:33 CET] <DHE> because you specify the executable to run and its argv[0] as distinct parameters [01:07:47 CET] <DHE> yeah, it's being run effectively as: ffmpeg alsa -i hw:1 -y $out [01:07:56 CET] <DHE> see the man page for execl [01:08:19 CET] <c_14> >The first argument, by convention, should point to the filename associated with the file being executed [01:08:42 CET] <c_14> so "/usr/bin/ffmpeg", "ffmpeg" [01:09:24 CET] <furq> also the last argument needs to be 0 [01:09:29 CET] <DHE> oh, quite right [01:09:38 CET] <DHE> NULL or (char*)0 [01:18:02 CET] <bplaxco> oh derp, thanks :) [02:15:08 CET] <Phi> is there a way for the most recent git to be broken? [02:15:19 CET] <Phi> or is it unit-tested to death before commits? [02:16:30 CET] <DHE> there's a test system called "fate" [02:16:43 CET] <c_14> But it doesn't catch everything [02:16:47 CET] <c_14> Regressions do occasionally pop up [02:17:22 CET] <Phi> mm [02:17:40 CET] <Phi> it looks like the ./configure I had without all the disables produces something worthwhile for make [02:17:57 CET] <Phi> so now I've got to work out what specifically is required for a valid make [02:18:19 CET] <Phi> ...no [02:18:26 CET] <Phi> it looks like the most recent ffmpeg doesn't make at all [02:18:53 CET] <Phi> even with the cmdl that was working a couple weeks back [02:19:33 CET] <Phi> with make, I get "Nothing to be done for 'all'," and with make install, I get "make: *** No rule to make target 'install-libs-yes', needed by 'install-libs'. " [02:20:06 CET] <Phi> ima do another git pull and try again :p [02:21:30 CET] <Phi> ...well the only change was "MAINTAINERS", so I guess that's not going to fix it [02:21:31 CET] <ldiamond> I'm trying to "mass convert" my AVI to Matroska. First attemp gives this: [matroska @ 0x55d2b33e8400] Can't write packet with unknown timestamp [02:21:57 CET] <ldiamond> I used the following: ffmpeg -i filename.mkv -c:v copy -c:a copy output.avi [02:22:00 CET] <ldiamond> well [02:22:07 CET] <ldiamond> invert the .mkv and .avi [02:23:19 CET] <Phi> input, output :p [02:23:20 CET] <ldiamond> I'm using 3.1.4 [02:24:19 CET] <Phi> try using setpts filter? [02:24:43 CET] <DHE> Phi: disable-everything literally does that. no ffmpeg binary, no codecs supported, no file formats supported. some core things like the "file" protocol are actually really important. [02:24:49 CET] <ldiamond> I haven't tried. I have no clue at all about ffmpeg. [02:24:56 CET] <ldiamond> I'll read on it I guess [02:25:22 CET] <ldiamond> ffmpeg -i my.avi output.mkv seems to work. I assume I'll lose quality though [02:25:32 CET] <DHE> addyour copy commands [02:25:39 CET] <DHE> between in and out [02:26:39 CET] <Phi> yea, but I'm not using disable-everything [02:26:53 CET] <ldiamond> DHE: was that directed at me? [02:26:55 CET] <Phi> is it just a replacement for all the disable-somethingplurals? [02:27:07 CET] <DHE> it's "--disable-x" for all x [02:27:26 CET] <DHE> but some of that stuff is really important, and you probably shouldn't disable protocols, decoders and programs unless you're really sure what you need [02:28:15 CET] <Phi> I know what I need, explicitly, I dunno what ffmpeg requires [02:28:32 CET] <Phi> it does turn on some things, but it doesn't complain "x requires y" [02:29:34 CET] <ldiamond> Filtering and streamcopy cannot be used together. [02:31:56 CET] <ldiamond> Phi: -fflags +genpts is what I needed [02:34:04 CET] <Phi> ah [02:34:14 CET] <Phi> I did mention genpts [02:34:37 CET] <Phi> tbh I've only used ffmpeg in C++ style, not commandlines [02:34:59 CET] <Phi> back to my dead make problem, I'm reverting the code to branch release/3.2 [02:39:31 CET] <ldiamond> Would you guys rather converting to .mp4 or .mkv? [02:39:32 CET] <ZimbabweLegit> hi, I'm doing an rtmp stream, and can't figure out why I'm getting such shit quality back with this; [02:39:33 CET] <ZimbabweLegit> -vcodec libx264 -maxrate 1200k -s 1280x720 -strict -2 -acodec aac -ab 190k -ar 48000 -ac 2 -f flv [02:41:21 CET] <Phi> your call ldiamond, although I do note mkv is sharper pictures [02:45:42 CET] <Phi> release 3.2 has the same issue :p [02:47:33 CET] <Phi> any idea what git command I should use to roll back to an older branch? [02:55:26 CET] <klaxa> Phi: using mkv or mp4 makes no difference to the video stream. it does not influence picture quality [02:56:01 CET] <klaxa> ldiamond: depends on your target device, mkv is more flexible, mp4 more widely supported [03:02:44 CET] <DHE> Phi: avoid --disable-everything unless you want to audit all the output of ./configure and decide each and every thing you want [03:03:21 CET] <DHE> personally I disable muxers, encoders and filters. then enable the ones I know I will need [03:15:40 CET] <Phi> hmm, I'll keep that in mind [03:15:55 CET] <Phi> atm I'm doing ./configure --toolchain=msvc --enable-libx264 --disable-swscale-alpha --enable-gpl [03:16:17 CET] <Phi> and that's producing the same "make: Nothing to be done for 'all'." [03:16:25 CET] <Phi> no idea why it's doing this [03:16:33 CET] <Phi> but it's 2am so I'm going to vamoosh [03:51:21 CET] <ItWasntMe2013> anyone know why when encoding to AC3 audio that I get a longer duration of audio file? about 2 second [03:51:49 CET] <ItWasntMe2013> and then when I mux that with a video, my video file gets a start: 1.400 roughly added as a delay to each file [04:14:10 CET] <ldiamond> klaxa: yea I guess "transcoding" (I don't know if transcontainerizing is a word :) isnt expensive either so w/e [04:14:24 CET] <ldiamond> I might go for mp4 just for the sake of support on many devices [04:15:01 CET] <ldiamond> Might be running the server on a udoo x86 so might as well save the "transcontainerizing" part [04:15:37 CET] <DHE> we usually call it remuxing [04:53:08 CET] <infinito> i finnally ready whit my ffmpeg cdn origin/border ready [04:53:20 CET] <infinito> 100 cores and a quadro m4000 [04:54:23 CET] <infinito> for my, quadro m4000 supports 20 ffmpegs processes, but i have to use 18 to aviod artifacts [04:55:20 CET] <infinito> was 6 or 7 monts of work :) [04:56:47 CET] <infinito> obviously, cnd is running 24/7 live whit a lot of tv channels like a charm [04:56:56 CET] <infinito> ffmpeg rules [04:58:28 CET] <SchrodingersScat> odoyle rules! [05:01:28 CET] <infinito> LOL [05:03:16 CET] <infinito> just see adam sandler video on vlc on windows :E [05:03:42 CET] <infinito> 800% cpu after a crash [05:04:40 CET] <ItWasntMe2013> anyone know why when encoding to AC3 audio that I get a longer duration of audio file? about 2 second [05:04:43 CET] <ItWasntMe2013> and then when I mux that with a video, my video file gets a start: 1.400 roughly added as a delay to each file [05:09:55 CET] <infinito> ItWasntMe2013 pastebin your command [05:17:42 CET] <ItWasntMe2013> infinito: http://pastebin.com/PC90UjrV [05:19:02 CET] <ItWasntMe2013> and this one infinito http://pastebin.com/n36B0GBu [05:19:27 CET] <ItWasntMe2013> no matter what I do..... I get a delay added to all my files [05:19:39 CET] <ItWasntMe2013> http://pastebin.com/uQNWLq1L [05:19:56 CET] <ItWasntMe2013> and http://pastebin.com/QF3mvnB9 [05:34:00 CET] <infinito> yes i think is because filter complex [05:34:52 CET] <infinito> im not an expert on that, but i think when you merge the audio channels are delayed [05:45:54 CET] <ItWasntMe2013> infinito: yeah... dont think so, but thanks for trying. [05:46:02 CET] <ItWasntMe2013> I think its a bug [05:46:17 CET] <ItWasntMe2013> going to have to dust of the programming skills to figure it out [11:29:32 CET] <kerio> should stuff like creation_time go with the stream or with the container? [13:42:54 CET] <maxmuster> i have some problems compiling ffmpeg on centos, anyone here who could help? [13:44:14 CET] <c_14> If you describe your problem. [13:44:24 CET] <maxmuster> beside that some libs seem to not compile, ffmpeg does not find yasm (on centos 32 bit) and does not find libopus (on 64 bit), even tho i use the commands from the webpage pointing to the lib install dirs. im currently re-compiling to get the logs back, then i can tell more [13:44:39 CET] <maxmuster> (libopus did compile and yasm too) [13:49:37 CET] <maxmuster> following this guide: https://trac.ffmpeg.org/wiki/CompilationGuide/Centos [14:00:19 CET] <maxmuster> from ffmpeg config.log i see that configure seems to not check for yasm in the right directory [14:00:31 CET] <maxmuster> how would i tell configure where to look for it? [14:00:42 CET] <c_14> What's your current configure command? [14:00:50 CET] <maxmuster> wait a sec.. [14:01:51 CET] <maxmuster> PKG_CONFIG_PATH=\"" + install_dir + "ffmpeg_build/lib/pkgconfig\" && ./configure --prefix=\"" + install_dir + "ffmpeg_build\" --extra-cflags=\"-I" + install_dir + "ffmpeg_build/include\" --extra-ldflags=\"-L" + install_dir + "ffmpeg_build/lib -ldl\" --bindir=\"" + install_dir + "bin\" --pkg-config-flags=\"--static\" --enable-gpl --enable-nonfree --enable-libfdk-aac --enable-libfreetype --enable-libmp3lame --enable-libopus - [14:01:51 CET] <maxmuster> -enable-libvorbis --enable-libvpx --enable-libx264 [14:02:08 CET] <maxmuster> from my self written script [14:02:57 CET] <c_14> yasm has to be in your PATH [14:02:59 CET] <maxmuster> its the one from the ffmpeg page [14:03:20 CET] <maxmuster> its in install_dir + "bin" [14:03:39 CET] <maxmuster> and same command works on centos 64 for yasm detection [14:04:24 CET] <c_14> try export PATH=install_dir + "bin" [14:04:28 CET] <c_14> and then rerun the other command [14:05:01 CET] <maxmuster> like in the ubuntu section on the ffmpeg page? [14:05:09 CET] <maxmuster> PATH="$HOME/bin:$PATH" [14:05:22 CET] <maxmuster> i'll try [14:07:41 CET] <furq> whatever language that is looks like a bad replacement for sh [14:08:17 CET] <furq> i take it you need fdk-aac [14:08:25 CET] <furq> otherwise you'd be better off using the static binaries [14:08:39 CET] <kerio> if fdk-aac is the best, why isn't it included with ffmpeg? [14:08:46 CET] <furq> because it's gpl incompatible [14:09:32 CET] <furq> https://github.com/mstorsjo/fdk-aac/blob/master/NOTICE#L57-L58 [14:09:35 CET] <furq> specifically that clause there [14:10:13 CET] <furq> it's also not the best unless you're only counting open-source encoders [14:10:24 CET] <kerio> what about encoders in ffmpeg [14:10:32 CET] <furq> what about them [14:11:21 CET] <kerio> is it the best of the encoders supported by ffmpeg? [14:11:25 CET] <maxmuster> furq: yes it is a bad sh replacement cause im new to python, but python is crossplatform and sh is not native for all [14:13:17 CET] <furq> kerio: probably, there's not been a proper listening test done [14:13:28 CET] <furq> it supports more stuff than the builtin encoder and i don't know if any other aac encoders are still supported in 3.x [14:13:35 CET] <furq> maxmuster: you can run sh on windows [14:14:01 CET] <maxmuster> i know [14:14:06 CET] <kerio> does autoconf even work in windows [14:14:07 CET] <maxmuster> but i wanted to test out python [14:14:11 CET] <furq> fair enough [14:14:51 CET] <furq> well you should probably use one of python's many string interpolation methods rather than doing all that concatenation [14:15:25 CET] <maxmuster> yeah. i wanted to make that script nicer when it compiles ffmpeg correctly [14:15:41 CET] <maxmuster> for centos64 ffmpegs configure does not find opus, [14:16:09 CET] <maxmuster> "package opus was not found in the pkg-config search path" [14:16:23 CET] <maxmuster> do i need to add something to the opus configure? [14:17:38 CET] <c_14> Is opus.pc in the PKG_CONFIG_PATH? [14:17:48 CET] <furq> if it is then check config.log [14:18:10 CET] <furq> a lot of errors will give that message [14:19:00 CET] <furq> kerio: it does under msys [14:19:33 CET] <maxmuster> i'll check [14:19:33 CET] <furq> you need msys or cygwin to build ffmpeg on windows anyway [14:19:46 CET] <furq> or preferably a linux box with mingw [14:19:46 CET] <maxmuster> yeh im not testing on windows atm [14:19:59 CET] <maxmuster> just for centos, ubuntu already working [14:20:36 CET] <furq> i should really push my changes to this tower of babel mingw cross-compile thing [14:20:54 CET] <furq> stupid cmake [14:21:24 CET] <kerio> why not MSVC :^) [14:21:29 CET] <maxmuster> opus.pc is in PKG_CONFIG_PATH [14:21:42 CET] <furq> check config.log then [14:21:47 CET] <maxmuster> the error is from there [14:22:02 CET] <furq> there'll be another error a bit above that [14:22:18 CET] <furq> pastebin the last 100 lines or so if you can't find it [14:22:20 CET] <maxmuster> okay let me see.. [14:22:58 CET] <furq> i'm guessing the header/library paths aren't set up right [14:24:00 CET] <maxmuster> i create the paste cause im not seeng any error [14:24:02 CET] <maxmuster> wait a sec [14:25:43 CET] <maxmuster> http://pastebin.com/fHeWB1gy [14:27:19 CET] <cousin_luigi> Greetings. [14:27:34 CET] <cousin_luigi> Trying to add a srt to a mkv container but I can't understand the error "Subtitle codec 94213 is not supported. [14:27:37 CET] <maxmuster> the lib paths are correct [14:29:51 CET] <maxmuster> adding the PATH to the centos 32 script worked, now i got the same opus not found error there [14:31:47 CET] <furq> oh cool pastebin isn't working [14:32:30 CET] <maxmuster> ? [14:32:40 CET] <maxmuster> lool [14:33:22 CET] <maxmuster> reload [14:34:15 CET] <maxmuster> furq [14:34:20 CET] <furq> check near the top of config.log [14:34:28 CET] <furq> it should have printed the PKG_CONFIG_PATH that configure is using [14:34:36 CET] <furq> i'm guessing that's not the same as the one you're setting [14:35:09 CET] <kerio> cousin_luigi: that's so weird tho [14:35:19 CET] <kerio> i thought mkv could house mostly everything [14:35:19 CET] <furq> oh [14:35:28 CET] <furq> maxmuster: remove the && from your script [14:35:54 CET] <maxmuster> which && [14:36:01 CET] <furq> the one after you set PKG_CONFIG_PATH [14:36:14 CET] <furq> it should be PKG_CONFIG_PATH=/foo/bar ./configure ... [14:36:27 CET] <maxmuster> >.< ty for pointing that out [14:36:29 CET] <maxmuster> :D [14:36:31 CET] <maxmuster> ima try it [14:36:51 CET] <kerio> cousin_luigi: is the .srt actually a srt? [14:36:52 CET] <kerio> ffprobe it [14:45:26 CET] <cousin_luigi> kerio: "Stream #0:0: Subtitle: subrip" [14:47:18 CET] <furq> pastebin the ffmpeg command and full output [14:51:14 CET] <cousin_luigi> furq: http://pastie.org/pastes/10956916/text?key=5qjowdu2fwr8tysoagmbg [14:52:16 CET] <kerio> ...why mov_text? [14:53:02 CET] <cousin_luigi> kerio: that's the example I found on the ffmpeg site. I don't know how else to have a language tag for the subtitle. [14:53:28 CET] <furq> well yeah mkv doesn't support mov_text [14:53:30 CET] <kerio> same thing but don't put a -c:s:0 mov_text [14:53:34 CET] <kerio> i guess [14:53:52 CET] <furq> yeah it should work without that [14:54:50 CET] <cousin_luigi> yes, it worked [14:54:51 CET] <cousin_luigi> thanks [14:55:16 CET] <cousin_luigi> bbl [14:55:35 CET] Action: kerio is baffled sometimes [14:56:27 CET] <furq> that was a textbook example of why you should ask for the full pastebin [14:56:40 CET] <furq> people are weird [14:57:01 CET] <kerio> also why doesn't ffmpeg default to -codec copy whenever possible? [14:58:08 CET] <BtbN> because it more frequently doesn't work than it does work, and there is no way for it to know in advance. [14:58:28 CET] <kerio> but it's so cheap tho ;o [14:58:47 CET] <c_14> And if you added it now, the behavior would change. [14:58:56 CET] <c_14> And therefore break users' scripts. [14:59:11 CET] <furq> if it was too easy then there'd be nothing for us to do [14:59:27 CET] <BtbN> not specifying a codec is pretty much guaranteed to work [14:59:35 CET] <BtbN> -c copy easily fails [15:05:58 CET] <kerio> did anyone here test avfoundation on macos? [15:06:10 CET] <kerio> the video i'm getting from my webcam has some very noticeable delay [15:21:08 CET] <dlb76> hey guys can pls tell me since then winxp not supported in ffmpeg.exe so i can find last build where it was still supported ? [15:21:40 CET] <dlb76> 0222 was still ok [15:21:44 CET] <dlb76> 1104 is not [15:21:47 CET] <dlb76> 3.1.5 is not [15:27:06 CET] <dlb76> 3.0.1 even not ok heh [15:29:09 CET] <dlb76> :( [15:35:01 CET] <klaxa> you will probably have to try for yourself and see when support stopped [15:35:17 CET] <klaxa> or migrate to a non-abandoned operating system? [15:35:24 CET] <klaxa> i would suggest the latter [15:35:40 CET] <furq> ffmpeg still supports winxp [15:35:48 CET] <furq> the zeranoe builds haven't since they started bundling libmfx [15:35:56 CET] <klaxa> ah [15:36:00 CET] <furq> if you want a more recent version, build it yourself [15:36:27 CET] <dlb76> yeah thanks [15:36:33 CET] <dlb76> reading it up now on forums [15:40:57 CET] <kerio> xp reached EOL over two years ago [15:41:30 CET] <SchrodingersScat> luckily fly by night companies like pcmatic say they can continue support of xp indefinitely [15:41:53 CET] <DHE> if it compiles and the os-specific APIs keep working, why complain? [15:42:23 CET] <kerio> i'm not complaining about ffmpeg still working on xp [15:42:27 CET] <kerio> i'm complaining about people using xp [18:52:25 CET] <kerio> x265 is so slooooooooow [18:53:15 CET] <bencoh> kerio: got some spare time to write SIMD code? ;p [18:56:12 CET] <JEEB> I think x265 needs more than just SIMD :P [18:56:46 CET] <JEEB> it needs logical optimizations regarding what steps it can skip without affecting the quality in the encoding process :P [18:57:10 CET] <kerio> hevc was sliiiiiiiiiightly better [18:57:13 CET] <kerio> on this particular video [18:58:02 CET] <bencoh> JEEB: prolly both :) [18:58:11 CET] <JEEB> for decoding you can just stick in SIMD [18:58:17 CET] <JEEB> and make it go fastah [18:58:34 CET] <JEEB> for encoding SIMD helps, but it seems like MCW is already writing random SIMD :) [19:32:04 CET] <maxmuster> furq you there? [19:32:32 CET] <maxmuster> configure of x264 does not find the yasm installed in $HOMEDIR/bin [19:32:41 CET] <maxmuster> how to tell him [19:36:40 CET] <relaxed> maxmuster: yasm must be in your $PATH [19:36:53 CET] <maxmuster> yeah i try it atm :P [19:36:56 CET] <maxmuster> forgot bout it [19:38:03 CET] <relaxed> there might be a configure option to pint to it [19:38:09 CET] <relaxed> point* [19:38:50 CET] <kerio> also to be fair i was confronting default settings for x264 and x265 at -crf 18 [19:39:03 CET] <kerio> maybe they're "wrong" [19:53:57 CET] <furq> kerio: they don't map exactly [19:54:07 CET] <furq> apparently 28 in x265 is roughly equivalent to 23 in x264 [19:54:23 CET] <furq> so i guess you want 22-23 [19:59:33 CET] <kerio> but i want 18 :< [19:59:37 CET] <kerio> oh i see what you mean [23:06:55 CET] <Phi> sempais [23:06:58 CET] <Phi> my make is ded [23:07:05 CET] <Phi> halp [23:07:30 CET] <Phi> ./configure --toolchain=msvc --enable-libx264 --disable-swscale-alpha --enable-gpl [23:07:51 CET] <Phi> produces "make: Nothing to be done for 'all'." [23:08:03 CET] <Phi> I've tried rolling back the code, but it's the same issue [23:20:04 CET] <JEEB> Phi: git clean -dfx [23:20:09 CET] <JEEB> clears your whole darn thing [23:20:27 CET] <JEEB> also out-of-tree builds are recommended so you can just nuke the build dir [23:41:47 CET] <Phi> tactical nuke incominggg [23:52:36 CET] <Phi> didn't fix JEEB [00:00:00 CET] --- Mon Nov 7 2016

1 0

ffmpeg-devel.log.20161105
by burek 06 Nov '16

06 Nov '16

[00:44:46 CET] <cone-653> ffmpeg 03Carl Eugen Hoyos 07master:222f59afd1cf: lavf/movenc: Do not print an error when muxing gray8 rawvideo. [02:14:59 CET] <cone-653> ffmpeg 03James Almer 07master:6005c7e65696: Revert "avformat/mux: split side data earlier in av_write_frame and av_interleaved_write_frame" [02:15:00 CET] <cone-653> ffmpeg 03Michael Niedermayer 07master:9e5881251931: avformat/mux: split side data before internal auto BSF [02:15:01 CET] <cone-653> ffmpeg 03James Almer 07master:7e603fb322d1: fate: add bsf tests for ticket 5927 [03:40:37 CET] <cone-653> ffmpeg 03James Almer 07master:51e329918dc1: avcodec/rawdec: check for side data before checking its size [04:07:17 CET] <cone-653> ffmpeg 03James Almer 07master:55061bbc558e: ffmpeg: don't overwrite av_bsf_receive_packet return value before checking it [08:08:34 CET] <cone-150> ffmpeg 03Matt Oliver 07master:85553b42f924: compat/w32dlfcn.h: Add safe win32 dlopen/dlclose/dlsym functions. [08:08:34 CET] <cone-150> ffmpeg 03Matt Oliver 07master:85db1f97eb50: avutil/hwcontext_dxva.c: Use new safe dlopen code. [08:08:34 CET] <cone-150> ffmpeg 03Matt Oliver 07master:d6f85ec2700e: avformat/avisynth.c: Use new safe dlopen code. [08:08:34 CET] <cone-150> ffmpeg 03Matt Oliver 07master:6ead033bca9a: avcodec/nvenc.c: Use new safe dlopen code. [10:26:15 CET] <cone-150> ffmpeg 03Michael Niedermayer 07master:5dc237aaaab6: fate: add zombie test [15:08:41 CET] <jamrial> michaelni: the ref file for the mov-zombie test you added has a line with the full path to the samples directory from your computer [15:11:49 CET] <jamrial> which is weird, as other tests have a relative path instead [15:16:49 CET] <nevcairiel> the fate stations also print a local path in there, something about the sample must be weird [15:17:19 CET] <nevcairiel> or nothing else uses -show_format [15:21:13 CET] <jamrial> no, looks like other test use -show_format, but using a relative path for a generated file in the build folder rather than the samples directory [15:21:28 CET] <nevcairiel> ah ok [15:21:37 CET] <nevcairiel> generated using relative makes sense [15:29:06 CET] <cone-574> ffmpeg 03Michael Niedermayer 07master:26e9efab1ecc: fate: Remove show_formats for fate-mov-zombie test [16:02:26 CET] <kierank> so a friendly outreachy student emailed me, what do I suggest they work on? [16:16:38 CET] <jamrial> kierank: there's no suggested projects page this time? [16:16:56 CET] <kierank> she missed the deadline and just wants general stuff to look at [18:10:31 CET] <cone-574> ffmpeg 03Timur Aydin 07master:774610322719: avformat/rtpdec: Add support for 24 bit RTSP audio playback [18:27:45 CET] <michaelni> jamrial, outreachy project: page but its way after all deadlines, http://trac.ffmpeg.org/wiki/SponsoringPrograms/Outreachy/2016-12 [18:48:30 CET] <cone-574> ffmpeg 03liu jc 07master:c0f6eff6a7ed: avformat/tcp: workaround for IOS9 getaddrinfo in IPv6 only network use hardcode IPv4 address can not resolve port number. [19:02:59 CET] <cone-574> ffmpeg 03Andreas Cadhalpun 07master:bb6a7b6f75ac: lzf: update pointer p after realloc [19:03:00 CET] <cone-574> ffmpeg 03Andreas Cadhalpun 07master:0efb6106118c: mxfdec: fix NULL pointer dereference [21:55:00 CET] <cone-574> ffmpeg 03James Almer 07master:9ea69f480892: avutil/softfloat_ieee754: make all functions inline [00:00:00 CET] --- Sun Nov 6 2016

1 0

ffmpeg.log.20161105
by burek 06 Nov '16

06 Nov '16

[03:13:37 CET] <ossifrage> I'm looking for a minimal dependancy h.264 systems library for taking elementary frames (with PTS as metadata) and generating a something that ffplay will play properly. [03:14:38 CET] <ossifrage> Just giving ffplay the '-framerate' option works if the source is constant frame rate, but I'm starting to work with variable frame rate video [04:10:51 CET] <DHE> any reason you can't put it in some kind of container? mkv or maybe mpegts could work [04:15:52 CET] <t4nk820> Getting Past duration too large errors when converting from mp4 to xvid using -vcodec mpeg4 -vtag xvid [04:22:55 CET] <ossifrage> DHE, yeah I'm looking for a lazy way to add the container (because it is only for debug) [04:24:19 CET] <ossifrage> The hardware spits out a PTS and a list of NALs (size + offset) and a block of data, I've just been shoving the block of data at ffplay [04:26:55 CET] <DHE> ossifrage: mpegts is used in broadcast, has a fixed timebase of 1/90,000, and has a somewhat high muxing overhead. but the (de)muxer is pretty straight-forward and is a widely supported format. [04:27:06 CET] <DHE> mkv is much more complex, but is also more feature complete [04:27:46 CET] <ossifrage> Yeah, I've been through the joy that is transport streams in a previous life... [04:29:50 CET] <ossifrage> I am heavily resource limited because some hardware guy thought that 13MB of flash was plenty. [04:31:20 CET] <ossifrage> I guess gpac or Bento are an option depending on how many deps they have in the minimal config. [06:29:16 CET] <Delicates> Trying to build ffmpeg 3.2 on Gentoo with CUDA and CUVID support... I have nVIDIA CUDA Toolkit installed, but ./configure says it can't find CUDA. Any ideas? [07:19:19 CET] <Alex> Delicates: WHat's the exact error from configure, OOI? [07:27:37 CET] <Delicates> Alex: fatal error: cuda.h: No such file or directory [07:28:14 CET] <Delicates> This is with nVIDIA CUDA Toolkit v8.0.44 installed btw [07:30:53 CET] <Alex> I think I ended up using -I /usr/local/cuda/include/ (or my equivalent dir) as it wasn't being pulled in properly. [07:42:21 CET] <Delicates> Alex: considering I'm trying to use Gentoo ebuild system, I'd rather fix the environment whatever might be wrong with it [07:54:54 CET] <kerio> ossifrage: 13mb? [07:55:09 CET] <kerio> shit, i think a daily log file from my desktop takes more than that [07:56:02 CET] <ossifrage> Yeah, I use a compressed filesystem (otherwise nothing fits) [07:58:17 CET] <ossifrage> But a 16MB SPI flash part is really cheap. But you could buy alot of flash parts for the time I spent working around the lack of space... [10:57:19 CET] <jcath> hello, friends [11:00:16 CET] <jcath> the filter minterpolate, may i use it to do ntsc to pal framerate conversion? [11:03:40 CET] <jcath> it seems that the filter is very slow, i only get about 1 fps with my i5 cpu, is it ok? [12:29:46 CET] <fling> I have a video which is longer than audio and I want to sync them. There is a lot of dupe frames here and there and some of them need to be dropped to shorted the video by about 30 seconds in total [12:29:57 CET] <fling> Which tool to use to accurately calculate frame numbers? [12:36:08 CET] <fling> Do I need avidemux or openshot? [12:36:52 CET] <fling> I can try to determine frame numbers with avidemux or to just drop the pieces with openshot& [12:37:02 CET] <fling> What would be the proper way of syncing? :P [13:58:41 CET] <fling> How is audio/video syncing happening when capturing from multiple live inputs at once (eg webcam with sound)? Is ffmpeg just dropping frames and inserting dupes sometimes to match the video timestamps to audio? [13:59:36 CET] <fling> Then can't I easily capture a multiangle video? [14:03:53 CET] <BtbN> there are no timestamps on webcam or audio captures [14:04:04 CET] <BtbN> ffmpeg just makes them up [14:05:32 CET] <fling> BtbN: right, so are they in sync because of the frame dropping and/or dupe injecting? [14:05:40 CET] <BtbN> what? [14:06:10 CET] <BtbN> they are in sync because they happen at the same time? [14:07:16 CET] <fling> BtbN: no, they are should be running at different speed because webcam clock is not in sync with soundcard clock! [14:07:32 CET] <BtbN> there is no "webcam clock" or "soundcard clock" [14:07:33 CET] <fling> For example a capture from multiple sound cards will be out of sync [14:07:41 CET] <fling> BtbN: ohh there is! [14:08:17 CET] <JEEB> maybe there is, but whether or not that means anything in the FFmpeg context is a whole separate discussion [14:08:21 CET] <fling> BtbN: I have a clock connector on my soundcard for syncing for other hardware for this purpose [14:08:31 CET] <BtbN> that's nice for your soundcard. [14:08:37 CET] <fling> JEEB: you are right [14:08:38 CET] <BtbN> ffmpeg still uses unix time based timesamps. [14:08:45 CET] <BtbN> +t [14:08:53 CET] <fling> BtbN: ok, you are not getting what I'm asking about. [14:09:05 CET] <JEEB> well, it uses whatever the "demuxer" or whatever utilizes [14:09:20 CET] <BtbN> https://github.com/FFmpeg/FFmpeg/blob/master/libavdevice/alsa_dec.c#L40 [14:09:37 CET] <JEEB> basically if you use input modules X,Y,Z [14:09:44 CET] <JEEB> you have to check WTF they do for their timestamps if anything [14:09:45 CET] <furq> "an unix" [14:09:52 CET] <fling> BtbN: the question is how it is getting the video to be synced with this ^ alsa captured audio? [14:10:18 CET] <BtbN> how would it not be like that? [14:10:33 CET] <fling> The webcam can't know the alsa times so it outputting the stream with it's own speed and ffmpeg is syncing this mess somehow together. [14:10:43 CET] <BtbN> why would it need to know that? [14:10:58 CET] <fling> It would not need [14:11:01 CET] <BtbN> If a new frame comes from the webcam, ffmpeg puts a timestamp based on the current unix time in microseconds on it. [14:11:07 CET] <BtbN> It does the same for audio frames. [14:11:07 CET] <fling> It is ffmpeg's job to sync things together [14:11:16 CET] <fling> ahh [14:11:29 CET] <BtbN> So as long as your webcam or soundcard doesn't do stupid shit like delaying frame output, everything just works. [14:13:23 CET] <fling> ok, so for example the alsa is running at system clock and the camera is a bit behind returning like 14.99 fps while we are capturing at 15 fps. What will ffmpeg do to sync this? Will it insert a dupe to the videosteam sometimes to get 15fps for the target? [14:14:51 CET] <BtbN> https://github.com/FFmpeg/FFmpeg/blob/master/libavdevice/v4l.c#L292 [14:18:15 CET] <fling> Does this mean we will always have 15 frames per second in the output container? [14:19:17 CET] <BtbN> If the webcam advertises a framerate and then delivers frames faster than that, it's plain broken and results might be undefined [14:19:25 CET] <BtbN> If it's slower, ffmpeg will drop a frame at some point [14:20:00 CET] <BtbN> ffmpeg assumes the webcam to be correct about its framerate. [14:20:23 CET] <BtbN> No idea if there is a simple vfr mode [14:20:26 CET] <BtbN> but doesn't look like it [14:21:07 CET] <fling> Then is it possible to captre multiangle video easily if there is not much to get in sync? [14:21:18 CET] <fling> I can have multiple ffmpegs running then. [14:21:41 CET] <fling> for example for two webcams setup I will need two ffmpegs capturing the webcams and outputting to pipes [14:21:53 CET] <BtbN> make sure to use wallclock for timestamps if you want to sync multiple independend files. [14:22:01 CET] <fling> then another ffmpeg will be reading from pipes and muxing all the things together [14:22:03 CET] <BtbN> and make sure that wallclock is accurate [14:22:26 CET] <fling> wallclock? [14:30:04 CET] <iive> system clock. [14:30:25 CET] <fling> I don't get this. [14:30:43 CET] <iive> audio cards have their own sampling clocks and they might drift (e.g. different temperature) [14:30:46 CET] <fling> >> multiple independend files. [14:30:49 CET] <fling> ^ no [14:31:01 CET] <iive> same applies to video and it's timing. [14:31:15 CET] <fling> right, this is why I'm asking how is ffmpeg syncing them. [14:31:20 CET] <iive> since you capture from different sources, there might be small differences. [14:31:31 CET] <fling> There should be! [14:32:23 CET] <fling> iive: so is it just dropping frames and inserting dupes when needed to sync to the audio right? [14:32:41 CET] <fling> Is not there such thing as a timestamp of each frame in the video stream for syncing to audio instead? [14:33:00 CET] <iive> donno, ask BtbN ;) what options make you use system clock . [14:33:20 CET] <iive> i thought that v4l2 already provides monotonic timestamps. [14:33:41 CET] <fling> Are timestamps going into the target container? [15:21:08 CET] <fling> ffv1 is bad about dupes! [15:21:28 CET] <fling> I mean it is compressing dupes bad [15:41:50 CET] <kerio> ? [15:52:00 CET] <xeuari> Hello all and happy Saturday [15:52:14 CET] <xeuari> I have a question about compilation [15:54:08 CET] <xeuari> I would like to use the vidstabdetect and vidstabtransform filters to stabalize a video. The documentation says 'To enable compilation of this filter you need to configure FFmpeg with --enable-libvidstab'. Does this mean I will need to compile a fresh copy and get rid of my apt installed version? Or is there a runtime configuration option somewhere? [15:54:21 CET] <furq> you need to rebuild [15:54:30 CET] <xeuari> nooo !!! [15:55:40 CET] <kerio> rip [15:55:48 CET] <xeuari> okay, I'm pretty shaky with building software .. could you point me in the direction of a makefile and such ..? [15:56:05 CET] <kerio> https://ffmpeg.org/download.html [15:56:13 CET] <kerio> check if you have the filter, tho [15:56:20 CET] <furq> apt-get install build-essential; apt-get build-dep ffmpeg [15:56:32 CET] <furq> then just copy the configure line from your installed ffmpeg and add --enable-libvidstab [15:57:49 CET] <xeuari> oo ok .. where would I find the config line? [15:57:55 CET] <furq> ffmpeg -version [15:58:07 CET] <xeuari> ah .. quality [15:58:12 CET] <furq> it doesn't look like vidstab is in apt so you'll need to install that separately [15:58:24 CET] <xeuari> yeah .. thought so [15:58:57 CET] <xeuari> does that need a separate download of transform or vid.stab then? [15:59:45 CET] <furq> i assume you need vid.stab, yeah [15:59:50 CET] <furq> i've never used it [16:00:10 CET] <kerio> apparently deb-multimedia.org's ffmpeg has vidstab enabled by default [16:00:27 CET] <furq> i wouldn't install anything from there [16:00:37 CET] <xeuari> cool, I'm still getting to grips with what a library is [16:00:42 CET] <kerio> furq: bad packages? [16:01:15 CET] <furq> i forget the reason now but i remember them causing some problems in the past [16:02:27 CET] <xeuari> I've read there's a package called transform which includes the vid.stab (plugin|library)? [16:02:50 CET] <xeuari> reckon I can apt that and reference in at build time? [16:03:07 CET] <furq> there's no debian package with that name [16:03:59 CET] <xeuari> hmm seems you're right [16:04:39 CET] <furq> https://packages.debian.org/search?searchon=contents&keywords=libvidstab.h&… [16:05:06 CET] <xeuari> (I meant transcode) [16:06:10 CET] <xeuari> I'm looking at Georg Martius' GitHub: https://github.com/georgmartius/vid.stab [16:06:51 CET] <furq> i think that just means you can't do a single pass with ffmpeg [16:06:57 CET] <furq> the transcode package on debian doesn't have vidstab [16:07:30 CET] <xeuari> well that is but a minor inconvenience [16:08:20 CET] <xeuari> This is pretty much a 1 off, I'm not building anything for other users [16:11:06 CET] <xeuari> Am I right in assuming that, if I make this vidstab plugin, and put it somewhere in my $_PATH, ffmpeg will find it at build time without needing an absolute path? [16:12:17 CET] <furq> it doesn't go in your $PATH [16:12:34 CET] <furq> just run make install [16:13:22 CET] <xeuari> (( noob confusion )) [16:14:13 CET] <xeuari> but then how will ffmpeg locate in with a simple flag `--enable-libvidstab`? [16:15:03 CET] <furq> it'll install to /usr/local/lib [16:15:12 CET] <furq> although annoyingly on debian that's not in gcc's default search path [16:15:15 CET] <kerio> xeuari: https://www.johnvansickle.com/ffmpeg/ has some static libs [16:15:17 CET] <kerio> er [16:15:18 CET] <kerio> static bins [16:15:30 CET] <furq> oh hey that has vidstab [16:15:37 CET] <furq> well that's a time saver [16:15:55 CET] <kerio> indeed! [16:16:23 CET] <furq> in that case disregard everything i said and just install those binaries [16:16:37 CET] <kerio> put them in ~/bin/ [16:16:44 CET] <kerio> so you don't touch anything of the system [16:16:55 CET] <furq> it's debian, you can safely mess with /usr/local/bin [16:17:03 CET] <furq> and/or ubuntu [16:17:03 CET] <kerio> still [16:17:18 CET] <furq> you'll also want to uninstall ffmpeg from apt [16:17:33 CET] <furq> i don't think debian puts ~/bin on your PATH by default [16:17:44 CET] <furq> not that it's hard to change, but you might as well use /usr/local [16:17:57 CET] <kerio> i think it does, if it exists [16:18:15 CET] <kerio> sure, just keep track of what you put in there [16:18:18 CET] <kerio> in general [16:18:25 CET] <kerio> this time it's easy, it's just a handful of binaries [16:26:17 CET] <xeuari> haha .. sweet. So i'm on Mint is Mint debian? [16:26:43 CET] <furq> regular mint is based on ubuntu [16:26:47 CET] <dl2s4> mint is ubuntu based afaik nd ubuntu is debian based [16:26:47 CET] <xeuari> I'm happy with /usr/local [16:26:49 CET] <dl2s4> afaik [16:26:50 CET] <furq> there's a version which is based on debian [16:26:53 CET] <dl2s4> ya [16:27:05 CET] <xeuari> right ..there's some history I'm missing here methinks [16:27:08 CET] <furq> but yeah they use fundamentally the same packaging and filesystem layout and whatnot [16:27:15 CET] <xeuari> cool ok [16:27:22 CET] <furq> debian packages will never touch /usr/local so you can do with it as you see fit [16:27:50 CET] <furq> most stuff you compile yourself will go to /usr/local by default [16:28:21 CET] <furq> this is one of the things that baffles me about the ffmpeg wiki ubuntu compilation guide which does some insane gymnastics to keep everything in ~ [16:28:27 CET] <furq> just put it all in /usr/local [16:28:53 CET] <JEEB> well I would totally recommend a custom prefix and not everyone has access to root (/usr/local is indeed root-only by default in most systems) [16:29:06 CET] <JEEB> for example I have ~/encoding_prefix myself, or ~/ownapps [16:29:09 CET] <furq> i guess [16:29:22 CET] <JEEB> those are also things I can relatively simply just clean up [16:29:23 CET] <JEEB> if needed [16:29:49 CET] <JEEB> but that is also OK for /usr/local as long as you know what exactly you've put there [16:30:02 CET] <xeuari> sounds sensible .. my /usr/local only contains a few shell scripts and my stuff anyway .. [16:47:08 CET] <xeuari> success!! .. but (grr) I can't call it from command line without specifying /usr/local/bin/. Is there any harm in just whacking it straight into /usr/bin? [16:48:24 CET] <c_14> xeuari: Just adjust your PATH? [16:49:11 CET] <c_14> In ~/.profile or ~/.bash_profile or ~/.zprofile or whatever [16:49:17 CET] <xeuari> hmm .. but can run my shell scripts just find from /usr/local/ .. so it must already be in there right? [16:49:33 CET] <c_14> echo $PATH [16:49:46 CET] <xeuari> null [16:49:56 CET] <c_14> What shell is this? [16:50:01 CET] <xeuari> oh i did echo $_PATH [16:50:10 CET] <xeuari> *facepalm* [16:51:09 CET] <kerio> xeuari: also try rehash [16:51:18 CET] <kerio> or restarting your shell [16:51:34 CET] <xeuari> nah they are all there .. [16:51:38 CET] <kerio> there's no way /usr/local/bin is not in the path [16:51:43 CET] <xeuari> is $PATH non recursive? [16:51:47 CET] <c_14> yes [16:52:35 CET] <furq> what happens if you just run ffmpeg [16:52:42 CET] <furq> if it's running debian's ffmpeg then you'll want to uninstall that [16:54:11 CET] <xeuari> no that's been purged .. it's fine, this is just a matter of tidiness now [16:55:49 CET] <xeuari> thank you for your ninja linking people .. !! [16:58:27 CET] <kerio> ^^ [17:58:39 CET] <BtbN> Anyone happens to know if ffmpeg is able to connect to a spice server? [18:17:56 CET] <xeuari> hmm ffmpeg is complaining about trailing options on the command line .. [18:18:38 CET] <xeuari> command: ffmpeg -i lightStrDemo.mp4 -vf vidstabdetect [18:20:16 CET] <BtbN> looks more like you are missing the output. [18:22:10 CET] <xeuari> yeah that too (The documentation says there is a default result file), but still trialing options for.. [18:22:16 CET] <xeuari> command: ffmpeg -i lightStrDemo.mp4 -vf vidstabdetect=result="lightStrDemoTransf.trf" [18:23:14 CET] <furq> add -f null - [18:24:15 CET] <BtbN> the vf does not apply to anything. So it's dangling and does nothing anyway. [18:24:22 CET] <BtbN> You need an output [18:24:25 CET] <xeuari> oh I see. ffmpeg itself needs an outputfile [18:24:31 CET] <furq> BtbN: that filter writes a pass file [18:24:45 CET] <xeuari> yeah it's a two pass thng [18:24:56 CET] <BtbN> furq, so? Without an output it still does nothing. [18:25:03 CET] <furq> oh [18:25:10 CET] <furq> i thought you were saying a null output wouldn't work [18:25:33 CET] <BtbN> ffmpeg options allways apply to the input or output that comes after them [18:25:44 CET] <BtbN> so without such an option after it, it does nothing and is dangling [18:25:44 CET] <xeuari> ffmpeg is currently running successfully after specifying an output [18:26:25 CET] <xeuari> thanks for the vf tip.. I'm just being cautious and coying what i read in docs [18:26:52 CET] <furq> https://github.com/georgmartius/vid.stab#usage-instructions [18:27:26 CET] <furq> that looks more useful than the ffmpeg filter docs [18:28:17 CET] <DHE> but the examples still specify an output file at the end of the command-lines [18:28:29 CET] <DHE> even if it is 'null to stdout' [18:29:42 CET] <furq> they sure do [18:29:53 CET] <xeuari> aw yea, that looks good .. I'm there [19:15:18 CET] <FishPencil> What are some decent commandlines for ripping Bluray's? I'm thinking x264 in an mp4 container, or maybe mkv [19:19:33 CET] <larsi> mkv is the format that is more widely used FishPencil [19:19:40 CET] <FishPencil> It'd like to retain as much quality as possible, so prettymuch visually lossless [19:20:06 CET] <JEEB> FishPencil: you mean re-encoding a blu-ray? with libx264 it's hard to go wrong. first of all, do you have any plastic boxes or mobiles that you want this to play on? that usually sets you some limits more than else [19:21:19 CET] <JEEB> otherwise it's the vlassic "encode 2500 frames or so with defaults" (preset medium and crf 23), then go down on crf if it looks bad, up otherwise [19:21:32 CET] <FishPencil> JEEB: I'll probably be sticking to a laptop so VLC. Ripping a high quality bluray to a mobile phone size doesn't make sense to me. [19:22:06 CET] <larsi> mp4 will most likely work better on low end devices and on old devices, but mkv is far better in terms of quality [19:22:13 CET] <JEEB> and when you have found the highest crf value that still looks good enough you poke the presets to see which is the slowest stillnfast enough for you [19:22:16 CET] <JEEB> eh [19:22:26 CET] <JEEB> matroska has nothing to do with quality [19:22:37 CET] <JEEB> it's a container for gawd's sake [19:22:40 CET] <FishPencil> larsi: "quality"? The container doesn't effect the quality [19:23:17 CET] <JEEB> what you stick into the container is what matters and there are of course more things you can stick into matroska than mp4 [19:23:37 CET] <larsi> yes, but an mkv container supports far more formats than mp4 [19:23:45 CET] <JEEB> whuch might or might not matter in your use case [19:23:59 CET] <FishPencil> JEEB: If I don't care about encoding times, any reason to not just do -p:v veryslow [19:24:14 CET] <JEEB> nope [19:24:32 CET] <FishPencil> Do the tune settings matter anymore? Like -tune film [19:24:36 CET] <JEEB> just stick to guesstimating your crf then [19:24:51 CET] <JEEB> it can help, but not necessary [19:25:15 CET] <FishPencil> How about audio? Blurays come as 5.1 DTS [19:25:42 CET] <JEEB> unless you want to re-encode, just do -c:a copy [19:26:03 CET] <JEEB> audio shouldn't be too big in general.compared to video [19:26:33 CET] <JEEB> and I'm not sure if lavf can write dts into mp4 yet. it's specified but not often used [19:26:46 CET] <microchip_> it can't [19:26:54 CET] Action: microchip_ dislikes DTS with a passion [19:27:01 CET] <FishPencil> I'd like to compress everything as much as I can but maintaining visually/audibly lossless [19:27:27 CET] <microchip_> FishPencil: use E-AC-3 @ 768 kbits. Virtually transparent [19:27:33 CET] <JEEB> well dts-hd ma isn't too bad as lossless audio :p [19:27:55 CET] <furq> FishPencil: -tune does make a big difference [19:28:10 CET] <furq> film and grain will obviously retain a lot more fine detail at the cost of higher bitrate [19:28:14 CET] <FishPencil> E-AC-3 is better in compression than AAC LE? [19:28:24 CET] <FishPencil> Or is it HE? [19:28:24 CET] <furq> no [19:28:30 CET] <JEEB> furq: but if you're eyeballing the crf anyways [19:28:35 CET] <microchip_> FishPencil: it isn't, but it provides better quality than HE-AAC [19:29:07 CET] <microchip_> FishPencil: you should not use HE-AAC unless you have a good reason. If you need AAC, stick to LC-AAC [19:29:18 CET] <FishPencil> microchip_: But if it's worse compression wouldn't you still want to do HE-AAC @ 768? [19:29:27 CET] <furq> yeah there's no point encoding to ac3 in 2016 [19:29:36 CET] <microchip_> FishPencil: no, as that's overkill for HE-AAC [19:29:44 CET] <furq> keep it if you already have it, otherwise use aac or opus or something less ancient [19:29:51 CET] <JEEB> furq: you will get somewhat different coding of course (less deblock etc), but in the end when you're setting your crf you will get the quality you want [19:29:57 CET] <FishPencil> I do like Opus [19:30:19 CET] <microchip_> JEEB: problem is, many receivers don't do Opus or Vorbis :( [19:31:02 CET] <furq> i was going to say "something more modern" but aac turns 20 next year [19:31:22 CET] <FishPencil> So what about: subtitles, playlists, multiple audio tracks [19:31:33 CET] <furq> mkv can handle all of that [19:31:35 CET] <furq> mp4 can handle some of that [19:31:56 CET] <furq> it definitely does multiple audio tracks and plaintext subtitles [19:32:08 CET] <FishPencil> I'll do mkv, but how should I be selecting the streams or playlists? Can FFmpeg even read a bluray playlist? [19:32:19 CET] <furq> i don't even know what a bluray playlist is [19:32:28 CET] <microchip_> FishPencil: I don't support playlists in ffx264 (i have no need for them) but you can try https://sourceforge.net/projects/ffx264/ :P [19:33:05 CET] <furq> is this like a pgc [19:33:30 CET] <FishPencil> Why wouldn't you just use FFmpeg directly microchip_ [19:34:49 CET] <microchip_> FishPencil: ffx264 uses ffmpeg..... directly using ffmpeg for what I need will make my life harder by specifying all the options manually for it, so I wrote ffx264 [19:35:10 CET] <furq> https://www.ffmpeg.org/ffmpeg-protocols.html#bluray [19:35:18 CET] <furq> looks like it's supported anyway [19:35:40 CET] <furq> not sure if that still needs libbluray [19:35:48 CET] <FishPencil> It does [19:35:51 CET] <microchip_> furq: probably [19:36:28 CET] <furq> it sure would be nice if there was an equivalent for reading pgcs with libdvdread or whatever [19:37:02 CET] <FishPencil> I know this is #ffmpeg, but are there better tools to convert/encode a bluray? [19:38:50 CET] <furq> i wouldn't use anything else for converting an m2ts with x264 [19:39:00 CET] <furq> as far as the actual ripping of the bluray i have no idea [19:40:19 CET] <microchip_> furq: MakeMKV (if you don't mind losing playlists, etc) [19:40:45 CET] <furq> makemkv just remuxes [19:41:00 CET] <furq> oh nvm you meant the second sentence [19:41:10 CET] <furq> i guess that works [19:41:44 CET] <microchip_> furq: I just need the movie, so MakeMKV works for me :) [19:42:27 CET] <furq> it's a shame there's no linux version [19:42:55 CET] <microchip_> furq: sure there is http://www.makemkv.com/forum2/viewtopic.php?f=3&t=224 [19:43:12 CET] <furq> oh neat [19:43:12 CET] <FishPencil> For subtitles I'm guessing I just rip them to an .srt and choose to not map the subtitle stream during the encode? [19:43:24 CET] <furq> let's see if this builds on freebsd [19:43:39 CET] <microchip_> furq: builds on Linux. Never tried on BSD [19:43:55 CET] <furq> i take it you can just build the cli version [19:44:13 CET] <microchip_> FishPencil: disable subs with -sn, then add your own during encoding [19:44:49 CET] <furq> i'm pretty sure mkv can store pgs subtitles [19:44:56 CET] <microchip_> yes [19:44:58 CET] <furq> there's no harm in ripping them though [19:44:59 CET] <FishPencil> microchip_: That'd be part of the same command right? I'd like to do it all in one step to avoid any sync issues [19:45:23 CET] <furq> FishPencil: sure [19:45:42 CET] <microchip_> FishPencil: yes.... -sn -map 0:2 (example) -c:s copy [19:54:28 CET] <FishPencil> Does Opus provide the best compression to bitrate ratio of audio encoders? [19:59:02 CET] <microchip_> FishPencil: it's one of the best, currently. Problem is, there's still (very) poor support for it, especially on hardware devices like receivers [20:00:29 CET] <FishPencil> I think most software players can handle it though [20:00:37 CET] <microchip_> yes [20:01:01 CET] <microchip_> at least those based on libavcodec, or ones that link to libopus [20:02:23 CET] <microchip_> FishPencil: 5 years ago I never imagined I'd get a surround receiver. Now I have one, and am glad I did all my encodes in AC3/EAC3 since it supports them, like the majority of other receivers [20:03:11 CET] <microchip_> FishPencil: while AC3/EAC3 doesn't provide the best compression, I'm fine in trading that for compatibility with virtually everything [20:03:41 CET] <microchip_> FishPencil: besides, you'd need serious equipment and golden ears to hear the difference [20:04:34 CET] <FishPencil> microchip_: I can tell the file size difference :) [20:04:53 CET] <microchip_> FishPencil: sure,... but i quote "storage is cheap" :P [20:04:53 CET] <furq> it's not going to add a significant amount to a decent 1080p x264 encode [20:05:04 CET] <furq> also makemkv does not build under freebsd [20:06:27 CET] <FishPencil> I might as well not even encode the audio if I'm using AC3 [20:06:48 CET] <microchip_> furq: got Linux around? Try it there. I package MakeMKV myself for openSUSE. Very few things need to be patched for compilation, if at all [20:07:07 CET] <furq> it's for my nas, so no [20:07:15 CET] <microchip_> ah [20:10:51 CET] <s0126h> what is the difference between 8 bit color vs 6-bits + FRC [20:11:23 CET] <microchip_> FishPencil: well, if you get DTS HD MA/TrueHD, you can encode it to EAC3 (or AC3) and at high enough bitrates and you won't tell the diff, while at them same time save space :) [20:12:11 CET] <microchip_> FishPencil: of course, doing AAC at lower bitrates than AC3/EAC3 will give you more space saving [20:12:45 CET] <FishPencil> I'm leaning to Opus, x264, mkv, srt. I don't need hardware player support [20:13:35 CET] <microchip_> FishPencil: go for it. Just remember, you can't predict the future. Just like I said to myself 5 years ago that I'll never get a receiver :) [20:14:26 CET] <FishPencil> I'd use x265, but I think the consensus is it isn't always better than x264 [20:15:11 CET] <microchip_> FishPencil: from personal experience (and this is really personal, my opinion is that x265 needs a lot of work to reach x264 details [20:15:30 CET] <FishPencil> microchip_: With that same logic receivers might support Opus in the future :) [20:16:07 CET] <microchip_> FishPencil: *might*. Most still don't support Vorbis and Vorbis is getting old ;) [20:16:58 CET] <c_14> It's also rather academic because your receiver will support raw pcm and decoding audio to pcm isn't really that much effort [20:17:16 CET] <microchip_> unless Opus becomes widely popular, I doubt receivers will add support for it. Be who knows? :) [20:18:20 CET] <microchip_> c_14: i can do it (am a power user), but most folks won't be able to unless they get into it, which most won't [20:19:33 CET] <c_14> People who aren't power users also don't tend to hook up their receivers to their computers. They'll just use their smart TVs and smart BluRay players etc [20:20:01 CET] <FishPencil> Not really #ffmpeg related, but why even use a receiver? Couldn't you just use a cheapo laptop with HDMI and software players? [20:20:04 CET] <microchip_> c_14: right :) [20:20:17 CET] <FishPencil> I despise dedicated optical players [20:20:57 CET] <FishPencil> Laptop with a USB DAC that supports whatever you need? [20:21:13 CET] <microchip_> FishPencil: my receiver is multi-functional, like most current ones are, so I use it to listen to radio, stream music from PC with DLNA, listen to internet radio, stream music from mobile devices with bluethooth, etc :) [20:21:47 CET] <c_14> You can do everything with a pc as well, the receiver is just easier to set up (at least it's supposed to be) because you don't have to do anything yourself but can just plug it in and it should work" [20:21:56 CET] <furq> don't use x265 unless you're prepared to encode at 2fps to get a noticeable compression improvement over x264 [20:22:30 CET] <furq> and also know that you can decode it [20:22:49 CET] <c_14> in at least realtime [20:23:12 CET] Action: c_14 still remembers the few times he's tried playing 4k HEVC [20:23:25 CET] <microchip_> :D [20:23:26 CET] <furq> libavcodec's hevc decoder is pretty slow, so if you're using anything based on that (e.g. mpv) then don't even bother [20:23:31 CET] <FishPencil> Yeah, I think x264, Opus, srt, mkv are the winning combo. [20:23:57 CET] <microchip_> Opus offers best compression atm, so yes [20:24:06 CET] <furq> i tend to use aac with x264 but it's probably less of a concern for a bluray rip [20:24:50 CET] <FishPencil> Should Opus be used in a -q:a or -b:a way? [20:25:01 CET] <s0126h> why opus [20:25:25 CET] <furq> i don't think -q is mapped for opus [20:25:35 CET] <microchip_> s0126h: he wants best possible compression with best possible quality [20:25:41 CET] <furq> iirc -b is quality-based with opus anyway [20:25:50 CET] <furq> it's not abr [20:26:29 CET] <FishPencil> And 128k Opus is probably pretty audibly lossless? [20:26:44 CET] <microchip_> 192 [20:27:00 CET] <microchip_> depending on your ears ;) [20:27:11 CET] <FishPencil> microchip_: For what channel config [20:27:12 CET] <furq> like how lame -V0 is nominally 256k, libopus -b 128k is nominally 128k [20:27:17 CET] <furq> also yeah 64k per channel [20:27:29 CET] <furq> if it's 5.1 you'll want at least 384k [20:27:40 CET] <FishPencil> Opus support 5.1 I take it [20:27:42 CET] <microchip_> FishPencil: stereo, for 5.1 try 256 [20:29:28 CET] <FishPencil> Huh, mp4 "officially" support Opus [20:29:48 CET] <c_14> I don't think the draft is final yet, is it? [20:30:18 CET] <microchip_> no idea.i don't really follow Opus much yet [20:30:21 CET] <FishPencil> Perhaps not [20:30:58 CET] <c_14> https://www.opus-codec.org/docs/opus_in_isobmff.html [20:31:01 CET] <c_14> says incomplete [20:33:52 CET] <microchip_> https://commons.wikimedia.org/wiki/Opus_(audio_codec) [20:33:53 CET] <s0126h> how does smartphones handle opus [20:34:06 CET] <furq> android does, iOS probably doesn't [20:34:20 CET] <furq> all non-apple browsers support opus now [20:35:56 CET] <FishPencil> I wonder why that is... Apple must be too busy designing phones without 3.5mm audio jacks [20:36:24 CET] <microchip_> FishPencil: NIH (Not Invented Here). Only add it if enough pressure [20:36:31 CET] <furq> because opus goes with vp9 which is google [20:37:04 CET] <furq> also because they're doubtless paying a lot of money for aac and would prefer it if everyone else had to pay for it too [20:37:21 CET] <FishPencil> What a mess [20:37:49 CET] <BtbN> There are decent hardware decoders for AAC [20:37:50 CET] <microchip_> one of the reasons why I avoid anything Apple [20:37:53 CET] <BtbN> but not for opus [20:38:09 CET] <furq> that's true but that doesn't explain its absence from desktop safari [20:38:09 CET] <BtbN> So playing Opus+VP8 or even 9 is fully on to the CPU [20:39:30 CET] <microchip_> wasn't Apple involved in initial AAC development, or am I thinking of the mp4 container? [20:39:37 CET] <furq> mp4 is based on mov [20:39:38 CET] <furq> so yeah [20:40:09 CET] <furq> apple weren't involved in initial development but they do have a best-in-class implementation [20:40:23 CET] <microchip_> better than fdk? [20:40:28 CET] <furq> yeah [20:40:33 CET] <microchip_> crap :p [20:40:39 CET] <MrMonkey31> really quick q - for libx265, I'm not seeing a ffmpeg2pass-0.btree : why aren't I seeing a ffmpeg2pass-0.btree? [20:40:42 CET] <furq> fdk isn't even the best fraunhofer aac encoder [20:41:30 CET] <microchip_> furq: fraunhofer has another one? [20:41:37 CET] <furq> fhg [20:41:41 CET] <MrMonkey31> it's on the second pass now [20:41:44 CET] <microchip_> oooooo [20:42:05 CET] <furq> i think fdk is a spinoff from fhg but nobody's really sure [20:44:19 CET] <microchip_> furq: well no linux version :( (at least not from Googeling for a minute) [20:44:47 CET] <furq> i don't think you can get a standalone version [20:45:01 CET] <microchip_> this all i found http://www.andrews-corner.org/linux/fhgaacenc.html [20:45:06 CET] <furq> winamp paid for the right to bundle it [20:49:15 CET] <microchip_> MrMonkey31: wild guess... because x265 doesn't need it? :P [20:49:32 CET] <microchip_> smae thing for .mbtree ? [20:49:46 CET] <MrMonkey31> WHAT?! Sorcery! [20:50:50 CET] <microchip_> MrMonkey31: x265 is quite different to x264. It took x264 code base a&s source, but diverged heavily since HEVC and AVC differ a lot [20:51:36 CET] <MrMonkey31> oh right, I omitted an m... oh well I only tried the former once or twice before [20:51:40 CET] <microchip_> MrMonkey31: anyways, you should be using CRF encoding, if you don't shoot for target file size :p [20:51:45 CET] <MrMonkey31> good to know it's working proper [20:51:50 CET] <BtbN> x265 was based of on x264? [20:52:06 CET] <MrMonkey31> I shoot for efficiency, always. is there any reason not to? [20:52:19 CET] <microchip_> BtbN: yes, the initial codebase... they took it from x264 [20:52:34 CET] <microchip_> or something like that :p [20:52:47 CET] <MrMonkey31> by this I mean highest possible compression while keeping desired fidelity [20:53:48 CET] <furq> 2-pass is only useful if you want to hit a specific bitrate [20:54:01 CET] <microchip_> MrMonkey31: problem with 2-pass, is you have to guess the bitrate. And bitrate requirements differ from video to vide, so you can't use the same bitrate for all videos, unless you use some insane value but then you're overshooting and wasting bits. Better let the encoder handle that for you [20:54:03 CET] <MrMonkey31> but if I'm hering you right crf is the go-to right now with x265? [20:54:16 CET] <furq> yes [20:54:20 CET] <furq> and x264 fwiw [20:54:28 CET] <iive> hevc and AVC differ far less than ASP and AVC [20:54:39 CET] <furq> 2-pass is a relic of the bad old days of xvid [20:54:51 CET] <iive> nah... [20:55:08 CET] <furq> and the bad current days of vp9 [20:55:11 CET] <iive> 2 pass is how you get consistent quality for fixed size. [20:55:51 CET] <furq> those two criteria disagree with each other [20:56:11 CET] <MrMonkey31> interesting.... I did not find those days *so* bad. well my feeling would be that it depends on the clip length also. the longer the source, the more fudging you can afford in the bit rate guess [20:56:31 CET] <microchip_> iive: hevc is an evolution of avc, but has enough differences to say it differs enough from avc [20:56:42 CET] <MrMonkey31> but I'm also inclined to take heed to what I hear here :-p [20:56:57 CET] <furq> well crf isn't an abr mode at all [20:57:11 CET] <furq> it'll use as much bitrate as it feels it needs to use [20:57:28 CET] <microchip_> MrMonkey31: rule of thumb is: if you shoot for roughly equal quality in all you encodes, use same CRF. If you need to target specific file size, use 2-pass [20:57:43 CET] <MrMonkey31> furq, then aye, that sounds like something to look into [20:57:43 CET] <furq> pretty much that with an emphasis on roughly [20:57:44 CET] <iive> furq: that's why you need 2 passes to satisfy both. [20:58:02 CET] <furq> iive: if you mean consistent quality within the file then sure [20:58:24 CET] <MrMonkey31> although... if you think about that logically.... [20:58:53 CET] <MrMonkey31> it would stand to reason you let crf do its estimation thing, then use the resulting value for a 2 pass encode [20:58:59 CET] <MrMonkey31> for max efficiency [20:59:16 CET] <microchip_> MrMonkey31: you'll get almost same result [20:59:38 CET] <microchip_> MrMonkey31: except with CRF you only need it once, instead of doing it in 2 passes [20:59:59 CET] <furq> 2-pass only makes sense for abr [21:00:30 CET] <furq> or if there's some kind of upper limit on how much bitrate you can use [21:01:06 CET] <BtbN> not so much bitrate, but filesize [21:01:19 CET] <BtbN> like, if you want a movie to fit perfectly on a DVD, that's what you use. [21:01:26 CET] <microchip_> yup [21:02:11 CET] <furq> the problem with 1-pass abr is that you can run out of bitrate towards the end of a file [21:02:42 CET] <furq> so you need to analyse the whole file so the encoder can make the best decision about where to use the bits [21:03:02 CET] <furq> if you're not trying to hit an average then it makes no difference, it'll just use what's needed [21:03:18 CET] <furq> assuming it's a halfdecent encoder ofc [21:03:24 CET] <BtbN> ABR doesn't usually mean "average over the whole file" [21:03:27 CET] <BtbN> that's what 2pass is for [21:03:38 CET] <BtbN> ABR usually just means average over a given timeframe [21:04:12 CET] <MrMonkey31> I don't wanna spout steam here but where you said those two criteria disagree with each other, I highly disagree. that's probably why we aren't understanding one another [21:04:44 CET] <furq> i took "consistent quality" to mean across a variety of sources [21:04:54 CET] <MrMonkey31> compression to me has always meant that "look, we're trying our best while being small" (if files could talk) [21:05:01 CET] <furq> one of the nice things about crf is that you can pretty much pick one value and use it for everything [21:05:16 CET] <MrMonkey31> ah ah ic [21:05:34 CET] <BtbN> if you don't care too much about the final filesize [21:05:39 CET] <furq> sure [21:06:07 CET] <furq> but i've had as many things turn out unexpectedly small as i have unexpectedly large [21:10:19 CET] <MrMonkey31> I guess for personal stuff (read: pron! j/k!) I prefer a compromise. I mean if a video has ninja stars flying across the entire screen, the entire time, then I'll allow an extra 100-150 mbs, but then again I might just elect to only take a clipping, ie a sample of the ninja stars. a shuriken's a shuriken after all [21:10:26 CET] <MrMonkey31> /wisdom [21:10:57 CET] <furq> are you encoding the godfrey ho criterion collection or something [21:11:23 CET] <MrMonkey31> quiet! I have been charged this task by ALL HUMANITEY! [21:12:09 CET] <MrMonkey31> but yeah: file size, clip length, clip prettiness, everything is subject to compromise. I'm just glad the technology's around to give me more options [21:13:03 CET] <furq> i just stopped caring about filesize [21:13:26 CET] <furq> if a 24 minute tv show comes out at 600MB or something then i'll probably denoise it but otherwise idc [21:13:55 CET] <kerio> qp 18 or bust [21:13:55 CET] <microchip_> MrMonkey31: we're not here to convince you, but some of the people in here know a lot about compression. In no way are we claiming that 2pass is not to be used. Just that common consensues is, that if you want specific quality and don't care too much about file size, CRF is the way to go. Of course, you decide :) [21:14:18 CET] <furq> crf is also obviously much faster [21:14:54 CET] <furq> the fact that it's better for what i care about is secondary to that [21:15:16 CET] <MrMonkey31> alright microchip_, thanks for the sum-up [21:27:03 CET] <larsi> is there a way to limit the download speed per second when seeking an online stream? e.g if I have 100MB of data and want to ratelimit it to x amount of bytes per second? [21:43:25 CET] <larsi> using command: ffmpeg -ss %(seekto)s -i %(url)s -to %(captureTo)s -c copy %(outfile)s -y [22:11:49 CET] <Luc__> Can somebody pleasse help me with the sysntax of ffmpeg? I would like to stream audio and video from a Raspberry Pi running Raspbian. [22:18:03 CET] <kerio> isn't the pi camera just v4l2 [22:18:15 CET] <JEEB> also original rpi was way too underpowered [22:18:32 CET] <kerio> i think there's some way to pull out h264 out of the gpu [22:18:35 CET] <kerio> maybe [22:18:37 CET] <JEEB> even if you got the video done with hw encoding it was actually too slow to encode the audio [22:18:43 CET] <JEEB> which was pop corn time for me [22:20:02 CET] <kerio> the raspberry pi is so overhyped [22:20:48 CET] <JEEB> rpi3 is kind of usable for some things [22:21:05 CET] <JEEB> original rpi was OK as a random hacking machine, but holy begesus do not try to do multimedia with it :D [22:21:32 CET] <xeuari> I <3 rpi's ..but they are not audio visual platforms [22:22:06 CET] <xeuari> unless you get a graphics card of equal physical size [22:23:04 CET] <xeuari> you could certainly perform non realtime operations on video, but streaming might be asking too muxch [22:23:29 CET] <furq> you can use the builtin h264 encoder with ffmpeg now [22:23:37 CET] <furq> it's not very good though [22:23:45 CET] <kerio> however, it's "free" [22:23:50 CET] <kerio> from a performance standpoint [22:24:03 CET] <JEEB> and then you will want to do audio and you go :sadface: [22:24:07 CET] <kerio> hm can't you stream raw data [22:24:18 CET] <furq> depends what comes out of the camera [22:24:30 CET] <furq> and what you're streaming to [22:24:44 CET] <kerio> furq: something with an actual cpu, clearly :^) [22:27:28 CET] <kerio> 40MB/s is doable on gigE right [22:27:36 CET] <xeuari> I'm user of what you want to do Luc__ but if it's webcam streaming, you can use fswebcam for ras pi (sudo apt-get install fswebcam) [22:27:38 CET] <kerio> raw yuv420p [22:27:47 CET] <furq> the rpi doesn't have 1G [22:27:51 CET] <Waddledee> Is there a way to set FFmpeg to _reduce_ an input's FPS, but not to _increase_ it? In this case, I want to encode a video to 25 FPS at _most,_ but if it's already 25 FPS or less I don't want to change its frame rate. [22:28:02 CET] <kerio> furq: yes it does [22:28:07 CET] <kerio> the pi3 at least [22:28:12 CET] <kerio> ...i think [22:28:52 CET] <furq> it's still 10/100 [22:29:24 CET] <furq> although it's usb anyway so you could always just get a 1G or 802.11ac dongle [22:29:38 CET] <kerio> *300mbps [22:29:47 CET] <kerio> with like a best case scenario of 480 [22:29:50 CET] <furq> i wouldn't trust the throughput on either of those though [22:29:52 CET] <kerio> usb 2 is shite [22:30:08 CET] <furq> yeah it's still usb2 sadly [22:42:29 CET] <Luc__> @xeuari: i want to stream audio and video from an USB capture device to a RTMP server. [22:44:36 CET] <kerio> does it get detected as v4l2? [23:32:42 CET] <Phi> yo folks [23:33:17 CET] <Phi> I'm trying to cut down the ./configure command so the output .libs are small [23:33:47 CET] <Phi> for some reason "make" is saying there's nothing to produce for "all" [23:34:07 CET] <furq> did you run make clean [23:34:24 CET] <Phi> no diff [23:34:31 CET] <Phi> I'm probably disabling everything somehow [23:34:59 CET] <Phi> http://pastie.org/private/h5j0cysrqado1pqhdqffq [23:35:19 CET] <Phi> it's RTSP or HTTP H264 -> H264 MP4 file [23:45:06 CET] <furq> you don't need any decoders or encoders to do that [23:46:04 CET] <Phi> I'm not flat copying, I'm decoding then re-encoding in a new profile [23:46:27 CET] <Phi> also I'm converting the YUV to display formats [23:46:29 CET] <furq> you'll need the h264 decoder then [23:47:31 CET] <furq> you probably also want --enable-protocol=file [23:48:04 CET] <Phi> do I need the h264 decoder if I'm using libx264? [23:48:12 CET] <furq> x264 is an encoder [23:48:33 CET] <Phi> so I use --enable-encoder=x264? [23:49:28 CET] <furq> --enable-decoder=h264 [23:50:01 CET] <c_14> you'll need the parser too [23:50:01 CET] <Phi> moop [23:50:26 CET] <furq> you probably also want the rtp/rtsp protocols [23:52:24 CET] <Phi> they're in there [23:52:38 CET] <furq> you've got the demuxers, not the protocols [23:52:49 CET] <furq> i assume you need both [23:53:16 CET] <furq> fwiw --enable-demuxer=rtsp,rtp,sdp works [23:53:29 CET] <Phi> http://pastie.org/private/k4r0fw7d7voqghprnborw [23:53:34 CET] <Phi> that might help cut it down XD [23:54:38 CET] <furq> i take it you need that mjpeg/mpeg4 stuff [00:00:00 CET] --- Sun Nov 6 2016

1 0