[Ffmpeg-devel-irc] ffmpeg.log.20130528

[burek]

[04:08] <fling> How to properly rip a dvd?
[04:08] <fling> I do not want to concat vobs.
[04:08] <fling> There are multiple tracks on it.
[04:24] <Hans_Henrik> ffmpeg -i 'http://www.4chanx.org/torrentdownloads/After%20Class%20Lesson%20%5bBlitz%5d%20%28DVD%20XviD%20MP3%29/%5bBlitz%5d%20After%20Class%20Lesson%20-%2002%20%28640x480%20DVD%20XviD%20MP3%29%20%5b70EDFF1E%5d.avi' -c:v libx264 -preset ultrafast pipe:1 2>&1
[04:24] <Hans_Henrik>  Unrecognized option 'preset'.
[04:25] <Hans_Henrik> what am i doing wrong? x.
[04:25] <Hans_Henrik> (about 5 hours past bedtime, i may just be ubberderping atm, but i thought it was called "preset" ?)
[04:26] <Hans_Henrik> it talks about preset option right here! http://ffmpeg.org/trac/ffmpeg/wiki/x264EncodingGuide
[04:29] <Hans_Henrik> ah, removing the preset, i get "Unable to find a suitable output format for 'pipe:1'
[04:29] <Hans_Henrik> "      x.x
[04:29] <Hans_Henrik> i want the output format to be x264 godam
[08:57] <Keshl> I'll just leave this here because I can't figure out mailing lists.. http://pastebin.com/WSLmj1H4
[09:01] <JEEB> Keshl, setting -r wouldn't do what you'd want it to do anyways :P Also I can't see anything specifically wrong in your terminal output.
[09:02] <Keshl> Then how come I get the wrong framerate? .É.
[09:02] <JEEB> upload a sample :P
[09:02] <JEEB> of the source
[09:02] <Keshl> Kaisec oÉo
[09:02] <JEEB> you can just dd like the first 100MB or so of the input
[09:02] <Keshl> That's actually what I did.
[09:03] <Keshl> Source is a few gigs and my upload speed's 250 kb/sec. <.<
[09:03] <JEEB> because mpeg-ts is supposed to be like that
[09:03] <JEEB> (you can cut it at arbitrary points more or less)
[09:03] <Keshl> ftp://iforgotmyIponesec
[09:03] <Keshl> ftp://broken@24.115.184.198:21918 -- Hug crop_b.ts , any password.
[09:04] <JEEB> grabbing
[09:09] <JEEB> ok, grabbed it
[09:09] <JEEB> let's see how this goes
[09:10] <Keshl> Luck oÉo
[09:10] <JEEB> ok...
[09:10] <JEEB> I think I notice the problem
[09:10] <Keshl> oÉo?
[09:10] <JEEB> > dup and drop values increasing
[09:11] <Keshl> They do that even on the converted video that works.
[09:11] <Keshl> I have wondered why though, it doesn't make sense..
[09:14] <JEEB> hmm
[09:14] <JEEB> playing it back looks fine tho
[09:14] <Keshl> o.O
[09:14] <Keshl> It looks 60 FPS?
[09:14] <JEEB> and sure is higher than 24fps
[09:14] <JEEB> yes
[09:14] <Keshl> o.O'
[09:14] <Keshl> What build you on?
[09:14] <Keshl> Maybe it got fixed and I just got my build the day beofre it did or something?
[09:14] <Keshl> *before
[09:15] <JEEB> I have an old zeranoe build from april 12th here
[09:15] <Keshl> Here, just to make sure, grab OUT.mp4 and see if that plauys bad.
[09:15] <Keshl> *plays
[09:15] <Keshl> If it plays good then at least I know it's a *really* weird decoder issue.
[09:16] <JEEB> ok
[09:16] <JEEB> the drops and dups are still weird, but *shrug*
[09:16] <JEEB> it looks like it has enough frames :D
[09:16] <JEEB> as in, definitely not 24fps
[09:16] <Keshl> ...Looking again, now it looks like it's 15 FPS to me. OUT.mp4, I mean.
[09:17] <Keshl> If it plays good, would'ya mind uploading your build? <É<
[09:17] <JEEB> yup, works here too @ OUT.mp4
[09:17] <Keshl> o.O'
[09:17] <Keshl> Wait, that's at 60 too?
[09:17] <JEEB> sure looks like higher-than-24 :P
[09:17] <Keshl> ...
[09:17] <Keshl> Wtf?
[09:17] <JEEB> and the renderer is set to 60000/1001fps
[09:17] <JEEB> seems to be getting samples with that kind of rate
[09:18] <Keshl> ... Wtf?
[09:18] <JEEB> what are you trying to play it with? :D
[09:18] <Keshl> i7-3840qm with 32 gigs of RAM and an nVidia 680m.
[09:18] <JEEB> uhh, software-wise
[09:18] <Keshl> VLC, smplayer.
[09:19] <JEEB> lol
[09:19] <Keshl> Both play at like 15 FPS.
[09:19] <Keshl> (Yet, they play it just fine if I convert the video twice..)
[09:19] <Keshl> What do you use, oÉo?
[09:20] <JEEB> VLC started supporting multithreading some time ago (IIRC), but I have no idea how its renderer deals with the high rate. smplayer's default is mplayer, and that one sure doesn't use multithreading or anything good rendering-wise
[09:20] <Keshl> Just tried WMP, same result.
[09:20] <JEEB> I'm on my LAV Filters + MPC-HC setup, as usual
[09:20] <Keshl> JEEB: I was laughed at for using VLC and told to use SMplayer, because "it's the best thing out there" o.O
[09:20] <Keshl> ...Mind explaining how to get that?
[09:21] <JEEB> if you want it simple you could just install the newest CCCP RC
[09:21] <JEEB> that is exactly the setup I have
[09:21] <JEEB> http://www.cccp-project.net/beta/
[09:21] <Keshl> I do need simple. The videos in question are being sent to people who don't know what FPS stands for. <.<
[09:21] <JEEB> oh
[09:22] <Keshl> ... What happens if I touch "do_not_touch_this"?
[09:22] <JEEB> it has some random builds that weren't yet ready for general usage
[09:22] <Keshl> Oh, kai, gud.
[09:23] <JEEB> anyways, I'm pretty much on defaults here with that setup (I've only enabled yadif deinterlacing from LAV Video and made MPC-HC limit the window resize to the dimensions of the video)
[09:24] <Keshl> JEEB: Same result.
[09:25] <Keshl> Definitely 15 FPS or so.
[09:25] <JEEB> funky, what's under play->filters when you play that file in the player that came with that thing I linked?
[09:26] <Keshl> Things. (More useful information: http://pasteall.org/pic/52340 )
[09:26] <JEEB> yeah, that looks all right
[09:26] <Keshl> If you hit F5, I've thrown an NFO with my system information up there.
[09:27] <Keshl> Might be some hardware incompatibility, oÉo? Which, makes, no sense, but there.
[09:27] <JEEB> how does it look when you press ctrl+J after you open that file in the player?
[09:28] <Keshl> Says the frame rate is 58.944, oÉo.
[09:28] <JEEB> are the green and red lines close to each other?
[09:28] <Keshl> But it's very clearly not.
[09:28] <Keshl> Very.
[09:28] <Keshl> Almost blocking each other.
[09:29] <JEEB> if they're in the middle and close that's exactly how it is here... am I really that bad at seeing frame rates? Because this clearly is smooth
[09:29] <JEEB> and the renderer is set to receive samples at 60/1.001fps
[09:29] <Keshl> It could be me. I'm really, really sensitive to changes in it.
[09:29] <Keshl> Like, I can see a difference between 60 and 120, and it drives me nuts when stuff's not at 120. Even these videos bother me. <.<
[09:30] <Keshl> ...How do I disable hardware acceleration
[09:30] <Keshl> *?
[09:31] <JEEB> it should be off by default
[09:31] <Keshl> It appears to be oin.
[09:31] <JEEB> then it's a misdetection
[09:32] <JEEB> when playing click LAV Video's properties in play->filters
[09:32] <JEEB> it should show you which decoder is used in the property sheet
[09:32] <Keshl> Kay, it is none..
[09:33] <Keshl> Enabling it don't help.
[09:33] <JEEB> it says which decoder is used under "Active Decoder:" anyways
[09:33] <Keshl> ... Hey
[09:33] <Keshl> hey
[09:33] Last message repeated 1 time(s).
[09:33] <Keshl> this makes sense.
[09:33] <Keshl> With that red stuff up with the lines.
[09:33] <Keshl> I see a thing that says "buffered 0", then "free 5", and current surface counting up from 0 to 4 over and over.
[09:33] <Keshl> 60 / 5 is 12.
[09:33] <Keshl> Which is about what I'm seeing.
[09:33] <Keshl> Is this relivant, oÉo?
[09:34] <JEEB> nope
[09:34] <Keshl> D:
[09:34] <JEEB> it just notes that it works :P
[09:34] <Keshl> By the way, it's counting up at 60 FPS.
[09:35] <JEEB> also the timestamps for your video seem just fine, and 60/1.001 http://up-cat.net/p/a7bc71d6
[09:35] <JEEB> see how timescale is 60000 and every frame goes one up with 1001
[09:35] <Keshl> So it's definitely something with my system, then..
[09:36] <JEEB> of course that only tells us that the timestamps for every frame are like that
[09:36] <Keshl> ... Uh, this is weird.
[09:36] <Keshl> So I'm using "step" to see if there's even all the frames there.
[09:36] <Keshl> No. They're not.
[09:36] <Keshl> There's far too much motion between frames for this to have 60 actual unique frames per second.
[09:37] <Keshl> And what's weird is, the first click advances it, the second does too, and the third is identical to the second, then it repeats this.
[09:37] <JEEB> you could grab something ffms2-using like current Aegisub in order to check it frame by frame regarding what the frames' contents are
[09:37] <JEEB> it's possible that there are duplicates for whatever reason
[09:37] <Keshl> ... -Picks an ear.- What?
[09:38] <Keshl> Thatwentottallyovermyhead <É<
[09:38] <JEEB> I just note that Aegisub seems to be the simplest application for you that you can open files with ffms2 :P
[09:38] <JEEB> I don't want to start teaching you Avisynth and friends
[09:38] <Keshl> ...Or, or, I can just ffmpeg -i OUT.mp4 frame_%d.png
[09:38] <JEEB> http://www.aegisub.org/
[09:38] <JEEB> Keshl, takes space?
[09:39] <Keshl> JEEB: Ramdisks ftw.
[09:39] <JEEB> anyways, I recommend having something at hand that lets you step forward and backwards in a file with relatively good frame accuracy :P
[09:39] <JEEB> and ffms2 gives you that
[09:39] <JEEB> and Aegisub has nice menu entries and so forth
[09:39] <JEEB> (and a portable version)
[09:40] <Keshl> Oh, ish program OÉO
[09:40] <Keshl> Somehow thought iut was a library, my bad.
[09:40] <Keshl> >w>
[09:40] <JEEB> ffms2 is a library
[09:40] <Keshl> ... Okay you're bad at reading framerates.
[09:40] <Keshl> ffmeg has the same output.
[09:40] <Keshl> step, step, stall, definitely too much motion between frames.
[09:41] <JEEB> which can be used as such, or from avisynth or an app that uses it
[09:46] <Keshl> ...Yeah no way, I'm sure now.
[09:46] <Keshl> See, the score counts up at each frame. <.<
[09:46] <JEEB> yeah, then ffmpeg decided to encode dupes or so
[09:46] <Keshl> I'm seeing it skip 3 at a time, then freeze, then skip, etc.
[09:50] <JEEB> I'll use a completely separate to ffmpeg parser/decoder combo for the sample you gave me, and see what kind of frames are to be found there
[09:51] <Keshl> Shiny, oÉo. Lemme toss upload privlages on that folder..
[09:51] <Keshl> There'ya go, oÉo.
[09:53] <JEEB> not that I need them :P
[09:54] <Keshl> ....Scarihaxxour, oÉo?
[09:56] <Keshl> And come to think of it, no, it doesn't duplicate frames like this and stuff on the other videos. It's just the .ts's.
[09:57] <JEEB> OK... it seems like ffmpeg is doing something /very/ derpy here
[09:57] <Keshl> Si D:
[09:59] <Keshl> Should someone file a bug, oÉo? Iunno how to do this cuz big attachments D:
[10:01] <JEEB> you should :P Try with an up-to-date zeranoe's build first tho. It should be rather simple to test for "ffmpeg -i input.ts -c:v libx264 -an out.mp4" should still give the same result
[10:03] <JEEB> or heck, probably any video encoder should do the same
[10:03] <JEEB> not sure if specific to mp4 output, either
[10:04] <Keshl> I'll try stuff, oÉo. In the meantime, have at thee! http://weebls-stuff.com/songs/badgers/
[10:05] <JEEB> http://x264.fushizen.eu/random/out_lsmash.mp4
[10:05] <JEEB> this should have "some" more frames :D
[10:05] <Keshl> That plays fine.
[10:05] <JEEB> anyways, that was encoded via avisynth and a non-ffmpeg reader
[10:09] <Keshl> Recent build has the same results.
[10:09] <Keshl> ...I think I figured out why.
[10:09] <Keshl> It seems to run at 60 when there's very little motion. The countdown, for example, is fluid. However the race isn't, and I think it's dropping frames to keep[ up with some CRF that it's mis-calculating.
[10:10] <JEEB> I have no idea what it is /trying/ to do
[10:10] <JEEB> but it should not be doing anythin related to the encoding speed, that's for sure
[10:10] <JEEB> anyways, time for you to hit the trac :P
[10:11] <Keshl> Where's that at, oÉo?
[10:11] <JEEB> https://ffmpeg.org/trac/ffmpeg/
[10:13] <Keshl> Dankies ^É^
[10:13] <JEEB> also make sure it reproduces with the simple command I noted
[10:13] <JEEB> with the newest zeranoe build
[10:14] <JEEB> and stick the log of it into the issue as well
[10:17] <Keshl> It does reproduce with your command, oÉo.
[10:17] <JEEB> yeah, it should be nice n' simple :P
[10:17] <JEEB> just note that it most probably isn't limited to libx264, and that most re-encoding options probably would lead to the same result
[10:19] <Keshl> Well lesse if mpeg2 does crap, owoi.
[10:19] <Keshl> ...Okay mpeg4.
[10:20] <Keshl> ...Worse.
[10:20] <Keshl> Far far worse.
[10:20] <Keshl> XD
[10:20] <Keshl> I'm seeing under 10 fps.. XD
[10:20] <JEEB> the encoding fps has nothing to do with the actual result
[10:20] <JEEB> unless you are doing realtime streaming, which you sure shouldn't be doing with a simple command line :P
[10:21] <Keshl> No, I mean the result.
[10:21] <Keshl> I'm really, really sensitive to framerates. Believe me, this is under 10 FPS.
[10:21] <Keshl> Encoding was over 120.
[10:22] <JEEB> anyways, create an issue on the tracker, with a sample linked/uploaded, and a brief explanation of it (ffmpeg creates unneeded dupes) and an example log with f.ex. libx264 or so (that simple line I posted f.ex.)
[10:24] <Keshl> Log of the output or just the input, oÉo?
[10:24] <JEEB> command line and the output basically
[10:25] <Keshl> oÉo.
[10:25] <Keshl> Is there any chance you can host CROP_B.ts for me?
[10:26] <JEEB> http://ffmpeg.org/bugreports.html
[10:26] <JEEB> see the "submitting sample media" part
[10:26] <Keshl> What component would I put this under, oÉo?
[10:26] <Keshl> (Danks, oÉo, was /just/ about to get to reading that.)
[10:30] <Keshl> Problem: The output of -v 9 -loglevel 99  is so long that it goes off-screen, and windows is stupid. How do I actually /log/ this?
[10:31] <JEEB> no need for that really :P
[10:31] <Keshl> But it says what frames were dropped and dupped, oÉo. You sure?
[10:32] <JEEB> hmph
[10:32] <JEEB> you can then just add 2> welp.txt to the end
[10:32] <JEEB> and it will output stderr to welp.txt
[10:32] <Keshl> Did, welp.txt is empty.
[10:32] <Keshl> Oh.
[10:32] <Keshl> *OH.
[10:32] <Keshl> Squee ^É^
[10:33] <Keshl> You knows window-secrets. <wwwww<
[10:37] <JEEB> Keshl, and for stdout AND stderr into one file you can try > welp.txt 2>&1
[10:37] <JEEB> > is stdout
[10:37] <JEEB> 2> is stderr
[10:38] <Keshl> Shiny, oÉo.
[10:38] <Keshl> <É< Does one simply walk into windows 8?
[10:38] <Keshl> (Sorry, had to.)
[10:39] <Keshl> [mpegts @ 00000000025da620] Non-increasing DTS in stream 0: packet 87 with DTS 180220450, packet 88 with DTS 180217447
[10:39] <Keshl> ...Is that normal?
[10:39] <Keshl> There's a crapton of that.
[10:39] <Keshl> 3 frames apart each.
[10:40] <JEEB> means that there is a time discontinuity in the stream
[10:40] <JEEB> the video stream I would guess
[10:40] <JEEB> as you can see, packet 87 has a larger time stamp (should be decoded later than) than packet 88
[10:40] <JEEB> and that doesn't make sense
[10:41] <Keshl> Could be the issue, oÉo..
[10:41] <Keshl> Every 3 frames is when it appears to stop.
[10:41] <Keshl> (As in, has a duplicate frame)
[10:41] <JEEB> yeah, probably related
[10:41] <JEEB> anyways, I recommend making that issue there
[10:42] <JEEB> in the worst case you get yelled at a bit, but I don't think that'll happen :P
[10:42] <JEEB> it seems like a valid issue, even if the stream is kind of wonky possibly
[10:43] <Keshl> Si, oÉo. I mean, wonky or not, it's still getting far below what it should.
[10:46] <Keshl> ...Says the sample needs to be under 10 megs, according to the link you sent. Should I compress it?
[10:47] <Keshl> (using 7zip, I mean.)
[10:47] <Keshl> Scratch that, compression ain't doing much.
[10:47] <JEEB> "Movie files which have been compressed (rar,7z,gzip,...) will be deleted without being examine"
[10:47] <Keshl> Would they yell at me for putting up 63.5 megs? x.x
[10:47] <JEEB> just upload it :P
[10:48] <JEEB> "...or the ftp on Submitting a Bug Report (there is no hard filesize limit). "
[10:49] <Keshl> Oh, kai, gud -É-
[10:51] <Keshl> What do I set the "component" to, oÉo?
[10:52] <Keshl> And version? git-master?
[10:53] <JEEB> libavformat or libavcodec or the ffmpeg command line app if that's there
[10:53] <JEEB> because the error comes from mpegts , which is libavformat
[10:55] <Keshl> oÉo.
[10:55] <Keshl> Did I get the version right, at least?
[10:56] <JEEB> no idea
[10:57] <JEEB> oh tgat
[10:57] <JEEB> yes
[10:59] <Keshl> Squee ^É^
[11:05] <Keshl> Er, what's the FTP server's address? xwx
[11:08] <JEEB> Keshl, mplayerhq.hu ?
[11:09] <Keshl> Danks, oÉo.
[11:10] <JEEB> hmm
[11:10] <JEEB> mplayerhq.hu's documentation points towards ftp://upload.mplayerhq.hu/incoming/
[11:10] <JEEB> anyways, as long as you can upload it somewhere *shrug* :D
[11:14] <Keshl> Si oÉo
[11:23] Action: Keshl put'd. -É-.
[11:23] <Keshl> https://ffmpeg.org/trac/ffmpeg/ticket/2607 looks reasonable right, oÉo? I get all self-conscious when submititng bugs. <É<'
[11:23] <Keshl> *submitting
[11:33] <Keshl> JEEB: Dang, they're fast. https://ffmpeg.org/trac/ffmpeg/ticket/2607 -- What codec should I be using, oÉo?
[11:33] <Keshl> (See the comment)
[11:34] <JEEB> why didn't you put in the log?
[11:34] <Keshl> I did, read the report. oÉo
[11:34] <Keshl> "I have uploaded a sample from the .ts file to upload.mplayerhq.hu/incoming/ under the filename "Ffmpeg_duplicates_and_drops_frames_when_given_a_ts_stream.ts", along with a .txt file with the same name that contains a copy of this bug and the output of "ffmpeg -v 9 -loglevel 99 -i CROP_B.ts -c:v libx264 -an test.mp4". The log has an identical name to the .ts file, except with .txt at the end."
[11:34] <JEEB> ok
[11:35] <Keshl> Did he just not read that? oÉo?
[11:35] <JEEB> you could have just posted it on the trac instead
[11:35] <JEEB> he probably did not :P
[11:35] <Keshl> But it's like 24 kb, I'd feel like I'm spamming .É.
[11:35] <Keshl> *158 kb
[11:35] <JEEB> heh, trac attachment if you can do those?
[11:35] <JEEB> also you might be able to note that it happens with mpeg4 if you have tried with it :P
[11:36] <JEEB> anyways, I have a feeling that cehoyos didn't want to go grab the file to just read the log
[11:37] <Keshl> Ah, oÉo.
[11:39] Action: Keshl put'd. -É-.
[11:40] <JEEB> I hope cehoyos will read a bit more this time :)
[11:41] <Keshl> Same -É- It bothers me when devs do that. I've been thwacked before since I tend to write long stuff (usually a /lot/ longer than that) and the devs are all "tl;dr" D:
[12:31] <Keshl> JEEB: He replied 'gain! 'cept he wants a non-libx264 thingy. What thingies can I use that aren't external thingies, oÉo?
[12:31] <Keshl> (...I don't know if I can word that better. <.<')
[12:33] <JEEB> if just "ffmpeg -i welp.ts -c:v mpeg4 -q:v 2 -an out.mp4" derps too, just give him that, and say that it happened with every encoder you tried if the case is so :P he seems to be following some blind protocol and not seeing what you really are replying :P And tell him that the actual problem doesn't change if the output is similar, and that the log is in the damn attachment
[12:34] <JEEB> -q:v 2 because I think the mpeg-4 encoder has something like 300kbps default or so, and I have no idea if that does framedrops if it can't hit the bit rate or what
[12:34] Action: JEEB shrugs
[12:35] <JEEB> or remove -q:v 2 if it doesn't do anything weird
[12:35] <JEEB> without it
[12:35] <Keshl> Funny. I did exactly that earlier, -qscale 2 and all.
[12:36] <JEEB> also joining #ffmpeg-devel could be a good idea, cehoyos is there from time to time so you can poke him in person
[12:36] <JEEB> or just ask if someone *except* cehoyos who could look at the issue :P
[12:37] <JEEB> because it seems pretty clear that he didn't read much
[12:39] <Keshl> Oh, good. <É< I was afraid they're all like that.
[12:39] <Keshl> Anyway, I'll head there. I have this feeling that this won't end, or at best he'll just go "meh, it's real, lemme get someone from #ffmpeg-devel"
[12:39] <JEEB> I like mini and ubitux at the very least
[12:40] <Keshl> ...Just hit me, if I ask about it then isn't it like I'm an attention whore?
[12:40] <Keshl> <.<
[12:41] <ubitux> oh i e you too JEEB
[12:41] <JEEB> <3 glad to know it's mutual
[12:41] <ubitux> :))
[12:41] <Keshl> Oh, oh good. ubitux, save me from making a fool of myself D:
[12:41] <Keshl> Cuz http://www.thinkgeek.com/product/1271/ D:
[12:41] <Keshl> ...Wrong paste.
[12:41] <Keshl> Cuz https://ffmpeg.org/trac/ffmpeg/ticket/2607 D:
[12:42] <ubitux> i'm on it
[12:42] <Keshl> Squee OÉO
[12:42] Action: Keshl huggles OÉO
[12:42] <ubitux> i won't solve it though :p
[12:42] <JEEB> seems like yer usual case of derpy DTS
[12:42] <ubitux> btw, you can't share the sample i guess? (yup i haven't read too)
[12:42] <Keshl> At the very least you can poke it and see if I'm doing something horribly stupid, oÉo.
[12:42] <Keshl> (Did already)
[12:42] <JEEB> he has uploaded it to the mplayerhq ftp
[12:42] <Keshl> (Look under incoming, same name as the bug.)
[12:42] <Keshl> (Same with -v 9 --loglevel 99)
[12:43] <ubitux> ah my eyes didn't match the link in the trac
[12:43] <Keshl> XD
[12:45] <Keshl> Also, this is unrelated, ish, but I'm trying to explain something to a friend of mine. Do you prefer KDE or Gnome, ubitux, oÉo?
[12:45] <Keshl> And JEEB.
[12:45] <ubitux> if i prefer kde or gnome to explain something to a friend?
[12:45] <Keshl> Yes.
[12:45] <JEEB> the last time I tried KDE I couldn't get the damn thing to select a pretty font fallback for CJK
[12:45] <JEEB> in terminal that is
[12:45] <Keshl> It's... Complicated.
[12:46] <JEEB> and setting up ibus-anthy/mozc was a pain in the arse
[12:46] <Keshl> Involves long stories that aren't important right now. <É<
[12:46] <JEEB> so in the end I went back to Ubuntu's standard stuff because it was a single click away
[12:46] <JEEB> instead of editing config files and *hoping* that it'll work
[12:46] <JEEB> (the best part was when only /some/ apps got the IME, and others didn't)
[12:47] <Keshl> oÉo... I meant as an orginization, not the DE's alone.
[12:48] <Keshl> Like, the groups, their philsophies, the way they handle things like bug reports and feature requests from "irrational" users, etc.
[12:48] <JEEB> All FOSS communities are a mess, and it just depends on which part of the mess you happen to have jumped to
[12:49] <Keshl> And how you define "mess". XD
[12:49] <ubitux> depending on the container, you have either drop or dup & drop
[12:49] <Keshl> ubitux: oÉo?
[12:49] <JEEB> ubitux, yeah
[12:49] <JEEB> I checked with dg's decoder with avisynth and that was able to decode all frames, so the picture data /is/ there
[12:50] <JEEB> just that ffmpeg is handling the DTS derpyness in a wonky way
[12:50] <Keshl> ...That feeling when you check another channel and click back but ever so slightly miss and start wondering "what the freak is this guy on about..?"
[12:50] <ubitux> the output looks ok after a quick look though
[12:50] <Keshl> ubitux: Did you take the output apart and look at it frame-by-frame, and do the same with the source video?
[12:51] <JEEB> basically from dg if he used the same sample it should look like this frame-wise http://x264.fushizen.eu/random/out_lsmash.mp4
[12:51] <ubitux> not really
[12:51] <ubitux> but from a playback PoV it looked OK
[12:51] <JEEB> the ffmpeg's output looks surprisingly good, yes
[12:52] <JEEB> but it does dupe/drop frames :s
[12:52] <Keshl> Yes, but it looked okay to JEEB, too.
[12:52] <JEEB> compare with that thing I linked if he used the same part of the sample
[12:52] <Keshl> Until he looked at it frame-by-frame he didn't see what I was on about (Correct me if I'm wrong)
[12:52] <JEEB> well, it surely didn't look non-fluid, that's what it was for me at first :P
[12:52] <JEEB> checking properly was the next step
[12:52] <Keshl> Well, maybe if it was a faster part of the game it'd make it more obvious.
[12:53] <Keshl> Just can't figure out how to /cut/ that out properly.
[12:53] <Keshl> Long story short though, after about ten minutes it goes about 6-10 times faster. It's very, very., very noticable then. <.<
[12:54] <ubitux> you should provide a sample where it's noticeable, but well that drop isn't justified anyway
[12:54] <Keshl> ubitux: I don't know how, though.
[12:54] <JEEB> it's pretty notice'able after you compare to f.ex. my encode using another decoder/splitter
[12:54] <Keshl> I can't just start the game at that speed. It has to work up there, and I can't focus enough to start a recording mid-game.
[12:54] <JEEB> but yeah, the dupes/drops are the problem
[12:54] <JEEB> not "if it's notice'able"
[12:54] <Keshl> And, I don't know how to sanely split a .ts file starting at the middle. xwx
[12:54] <ubitux> record the whole session and cut the fast part?
[12:55] <Keshl> ubitux: How?
[12:55] <Keshl> Wion't that break the .ts file such that it won't even open?
[12:55] <Keshl> Or does it not have header information, oÉo?
[12:55] <JEEB> nope
[12:55] <ubitux> ffmpeg -i in.ts -ss 12:34 -c copy out.ts
[12:55] <ubitux> maybe
[12:55] <ubitux> but that will do some remuxing and alter the timestamps
[12:55] <JEEB> mpeg-ts is just a collection of streams with possibly some metadata
[12:55] <Keshl> Okay, lemme try that.
[12:55] <ubitux> dd should be ok otherwise
[12:55] <Keshl> Wait, yeah, nevermind bad idea.
[12:56] <Keshl> Okay, I'll kjust dd at it and see if I can get it, gimme a bit..
[12:56] <ubitux> anyway, how can i help with the ticket?
[12:56] <ubitux> i don't think i'm qualified enough to solve that problem
[12:56] <JEEB> even commenting on it so that cehoyos can see that all the information for replication already is there :P
[12:56] <ubitux> it looks like something for mini :)
[12:56] <JEEB> because it seems like he's on auto-mode
[12:56] <JEEB> not really reading
[12:56] <ubitux> ah right
[12:56] <Keshl> ubitux: Get it in the hands of someone who can help, oÉo? Mainly I just feel like I'm an attention whore if I ask for someone who isn't Cehoyos to help.
[12:57] <JEEB> I wonder if the PTS is wonky, too
[13:00] <Keshl> ...Hey I found the video I lost, kewl, thanks =D! (Unrelated. Just saying.)
[13:04] <ubitux> just commented the ticket, not sure if that will help
[13:05] <Keshl> Hey someone else replied =D!
[13:05] <Keshl> Oh!
[13:05] Action: Keshl huggles ubitux -É-
[13:11] <Keshl> KAI. Should I toss the thing in the same place as the other thing, oÉo?
[13:12] <Keshl> (On another note: Yes, there is a header. It contains the icon used to make the preview thumbnail, appearntly.)
[13:12] <JEEB> nah
[13:12] <JEEB> at least proper mpeg-ts streams don't have anything like that
[13:13] <JEEB> and I'm pretty sure the windows thumbnailing thingy doesn't save stuff in the files
[13:13] <Keshl> Interesting.
[13:13] <Keshl> With -vfr or whichever it was, all the frames /are/ there.
[13:13] <Keshl> And correct.
[13:13] <Keshl> Appearntly.
[13:14] <Keshl> Nope.
[13:14] <Keshl> They're not.
[13:14] <Keshl> ...Nope.
[13:14] <Keshl> They are.
[13:14] <Keshl> Stupid me.
[13:14] <Keshl> But that's from the .ts file.
[13:15] <JEEB> which means ffmpeg can at least decode all of the pictures
[13:15] <Keshl> Si, oÉo.
[13:16] <Keshl> When I do it on the re-encoded video, it does what I said before. The tap-tap-stall thing.
[13:16] <Keshl> JEEB: How do I reply to this and mention that? I feel weird replying to ubitux and not that other guy, even though I essentially wanna tell him "sit down and read before typing" <.<
[13:17] <JEEB> you're overthinking :P
[13:17] <Keshl> I do dat sometimes. xwx
[13:17] <JEEB> just reply to a reply in the topic
[13:18] <ubitux> i'm not asking any question Keshl :)
[13:18] <ubitux> just providing various random info
[13:18] <ubitux> which anyone debugging that problem could do to reproduce anyway
[13:19] <ubitux> i'm mainly doing this so cehoyos can flag the ticket as reproduced by developer, ticket open, and tag with random keywords
[13:19] <ubitux> he might even do some git bisect at some point if you say it's a regression
[13:20] <Keshl> What is a regression exactly, oÉo?
[13:20] <JEEB> "used to work, doesn't with newer versions"
[13:20] <Keshl> Oh. I don't know if it ever used to <É<
[13:21] <t4nk379> hi
[13:21] <Keshl> Ohai oÉo
[13:21] <Bor0> regression can be thought of as, you spawn a hydra (bug), you solve it (kill it, but not kill its spawns), so it causes regressions
[13:21] <Keshl> Because dragons make everything sexier.
[13:21] <t4nk379> does anyone know the layout of /tmp/feed1.ffm
[13:21] <t4nk379> how to map the /tmp/feed1.ffm to the struct FFMContext?
[13:22] <t4nk379> I got a problem for a long time, ffmdec.c get error when ffserver try to read data from input stream
[13:23] <t4nk379> because the element "header" in struct FFMContext has wrong data
[13:23] <t4nk379> any ideas?
[13:23] <t4nk379> help help
[13:25] <Keshl> JEEB: Video's at ftp://Broken@24.115.184.198:21918 if you wanna mess around with it, just hug CROP.ts . Any password works.
[13:26] <t4nk379> any one have ideas?
[13:26] <Keshl> ubitux: cc     <-- Totally new technical term. A-yep.
[13:26] <t4nk379> hi burek
[13:26] <t4nk379> here?
[13:26] <Keshl> t4nk379: If nobody replies to you, it means nobody knows how to help. IRC would get really full of spam if everyone kept saying "sorry, can't help".
[13:27] <t4nk379> mmmmm
[13:28] <t4nk379> Can someone explain how ffserver get video data from ffmpeg?
[13:30] <durandal_1707> t4nk379: what are you trying to do?
[13:32] <durandal_1707> and what error you get from ffmdec ?
[13:33] <t4nk379> yes
[13:33] <t4nk379> thanks durandal_1707
[13:33] <t4nk379> can I explain more detail?
[13:34] <t4nk379> I use ffserver + ffmpeg to output video data to remote VLC client
[13:34] <durandal_1707> and what exact error you get, and how your ffm file looks like?
[13:34] <t4nk379> then VLC client can not get video data any more when ffserver read wrong ffm header
[13:34] <durandal_1707> use pastebin or similar
[13:35] <t4nk379> I do not know the format about /tmp/feed1.ffm
[13:36] <t4nk379> in ffmdec.c, it says the ffm header size is wrong
[13:37] <t4nk379> in ffmdec.c,ffm_read_packet(), it says the ffm header size is wrong
[13:38] <durandal_1707> t4nk379: look in ffmenc.c to see what ffmpeg writes
[13:38] <t4nk379> but I see ffserver , in receive data , it write data about 4096byte to feed_fd
[13:39] <t4nk379> yes, I see ffmenc.c, it is write by ffmpeg,
[13:39] <durandal_1707> ffmpeg sends stuff to .ffm and ffserver reads from that stream
[13:39] <t4nk379> but I do not know why ffserver get wrong size when size = AV_RB24(ffm->header + 2);
[13:40] <durandal_1707> you sure ffserver version is same as ffmpeg one?
[13:41] <t4nk379> but I see ffserver receive by socket 127.0.0.1 and ffserver write data 4096 bytes to feed_fd, and I do not see where to link header and packet dta
[13:41] <t4nk379> where is the link about ffm header and packet data
[13:41] <durandal_1707> packet data as stored in .ffm ?
[13:42] <durandal_1707> FFM (FFserver live feed)
[13:42] <t4nk379> yes, but ffm header is in stream priv_data pointer
[13:43] <t4nk379> the content of priv_data , where to update the element? and who update the element?
[13:44] <t4nk379> I can see ffserver will update the index "feed_write_index", when receive data and prepare data
[13:44] <durandal_1707> but why you need that?
[13:44] <t4nk379> but when I see the error occurs, there is no any method to recover
[13:45] <t4nk379> because I port ffmpeg on my embedded system, when the error occurs, then the stream is stop
[13:45] <durandal_1707> error happens because there is no enough data, (guessing) so failure should be handled by user
[13:45] <durandal_1707> by caching data or something
[13:46] <t4nk379> no enourgh data?
[13:46] <t4nk379> do you mean there is no enough data from webcam?
[13:46] <durandal_1707> ffmpeg did not finish writing data to .ffm (guessing)
[13:47] <durandal_1707> you are looking for help in very special are nobody works on now
[13:48] <t4nk379> sure,is it the special case?
[13:48] <t4nk379> the error occus easily when I use wifi client to get stream
[13:49] <t4nk379> very easily to get error
[13:50] <t4nk379> I can see ffmpeg will write data with the format struct FFMContext
[13:51] <durandal_1707> but how do you create .ffm and read from .ffm ?
[13:52] <t4nk379> .ffm will be crated by ffserver
[13:53] <t4nk379> the command line I used ffmpeg -f video4linux2 -r 30 -s 1280x720 -input_format mjpeg -i /dev/video0 -vcodec copy http://localhost:8090/feed1.ffm
[13:53] <durandal_1707> really? afaik .ffm is created by ffmpeg
[13:53] <t4nk379> sure?
[13:53] <t4nk379> but ffmpeg does not know where to create the file
[13:54] <t4nk379> in ffserver.conf I set the path is /tmp/feed1.ffm
[13:54] <t4nk379> right?
[13:54] <t4nk379> the size of ffm file is 5M
[13:55] <durandal_1707> have you read documentation about ffserver?
[13:55] <t4nk379> which one ?
[13:56] <durandal_1707> http://ffmpeg.org/ffserver.html
[14:08] <t4nk379> it does not seem help for my serious problem
[14:09] <durandal_1707> when you get error?
[14:11] <t4nk379> just a few seconds when I try to use wifi client to connect ffserver
[14:11] <t4nk379> about 10 ~15 seconds
[14:11] <t4nk379> then Wifi client can not get any data from ffserver
[14:12] <t4nk379> then ffserver will get the state:wait data available
[14:12] <t4nk379> because ffserver does not get the vaild ffm header
[14:12] <durandal_1707> what version of FFmpeg you use?
[14:12] <t4nk379> v1.2
[14:14] <t4nk379> i have try the version "master"
[14:14] <t4nk379> it is the same result
[14:15] <t4nk379> and
[14:16] <durandal_1707> pastebin all commands so it can be reproduced
[14:16] <t4nk379> if I try to connect ffserver via ethernet
[14:16] <t4nk379> it is ok
[14:16] <durandal_1707> then i don't think its ffserver problem
[14:16] <t4nk379> do you mean ffserver + ffmpeg command to pastebin?
[14:17] <durandal_1707> not needed, as via ethernet it works
[14:17] <t4nk379> yes
[14:17] <t4nk379> it is very strange
[14:18] <t4nk379> if the network is more smoothly then the stream ismore stable
[14:18] <t4nk379> if the network is not very smoothly , then ffserver can not output stream smoothly
[14:19] <t4nk379> because in the air, wifi client will get more collision than ethernet
[14:20] <t4nk379> do you have any idea?
[15:13] <hackeron> hey, I installed the latest x264 from git but when I run configure in ffmpeg, I get ERROR: libx264 not found -- any ideas?
[15:14] <klaxa> did you comiple shared libraries? are they in the library path?
[15:14] <hackeron> hmm, no I did ./configure --enable-static -- was I not meant to?
[15:15] <klaxa> only if you build a static ffmpeg (i think?)
[15:15] <ubitux> --enable-static is the default, and it only affects the FFmpeg libraries
[15:15] <hackeron> ubitux: I meant I compiled x264 with ./configure --enable-static
[15:16] <klaxa> yeah that'S how i understood it
[15:21] <mpfundstein_work> sudo ldconfig
[15:21] <mpfundstein_work> aber installing
[15:21] <mpfundstein_work> after
[15:43] <hackeron> mpfundstein_work: made no difference
[15:43] <hackeron> anyone able to compile latest ffmpeg from git against last x264 from git today?
[15:45] <JEEB> pastebin your config.log
[15:46] <hackeron> JEEB: http://pastebin.com/G0EwX9UA
[15:47] <JEEB> pretty obvious
[15:47] <JEEB> the x264.h header is not in your search path
[15:47] <hackeron> hmm, but it's in /usr/local/include/x264.h
[15:47] <hackeron> is that not searched?
[15:48] <hackeron> oh wait, the file isn't there despite checkinstall saying it is, hmmm
[15:48] <JEEB> you could see what places it checks like this http://gcc.gnu.org/ml/gcc-help/2007-08/msg00125.html
[15:50] <hackeron> gmm, gcc -print-search-dirs doesn't show /usr/local/include/ or /usr/include/ -- how would I add these?
[15:50] <JEEB> uhh
[15:51] <hackeron> oh wait, the config.log changed after I re-compiled x264
[15:52] <hackeron> JEEB: what do you make of this? < http://pastebin.com/QJEpeRmm
[15:55] <JEEB> ah, the usual
[15:55] <JEEB> x264's opencl support needs --extra-ldflags="-ldl"
[16:01] <hackeron> JEEB: I tried to recompile x264 passing --extra-ldflags="-ldl" to configure, ffmpeg still not finding libx264
[16:01] <JEEB> ...
[16:01] <hackeron> what am I missing?
[16:01] <JEEB> le sigh
[16:01] <JEEB> you give that extra flag not when building x264
[16:01] <JEEB> but when linking x264 to ffmpeg
[16:01] <JEEB> :V
[16:02] <hackeron> JEEB: what do you mean? - I don't have ffmpeg installed yet
[16:02] <JEEB> --extra-ldflags="-ldl" to ffmpeg's configure
[16:02] <JEEB> that's what I mean
[16:03] <hackeron> does this look reasonable? < ./configure --extra-ldflags="-ldl" --enable-gpl --enable-libfaac       --enable-libfdk-aac --enable-libmp3lame --enable-libopencore-amrnb       --enable-libopencore-amrwb --enable-librtmp --enable-libtheora       --enable-libvorbis --enable-libx264 --enable-nonfree --enable-version3
[16:03] <JEEB> why do you have both faac and fdk-aac?
[16:03] <JEEB> do you really need the opencore amr things?
[16:03] <JEEB> do you need librtmp? ffmpeg has an internal implementation now
[16:04] <hackeron> don't think so, but I record from various IP cameras and they have weird audio encodings like ulaw and g726 - don't think I've seen amr yet -- and didn't know that, removed
[16:04] <hackeron> anyway, that configure line still says lib264 not found
[16:04] <JEEB> config.log pastebin, as usual
[16:05] <hackeron> JEEB: http://pastebin.com/uE69FST5
[16:05] <JEEB> you didn't add the extra ldflag :P
[16:06] <hackeron> I didn't?
[16:06] <JEEB> see the first line
[16:06] <hackeron> it says: # ./configure --extra-ldflags=-ldl
[16:06] <JEEB> no it does not
[16:06] <JEEB> see your damn pastebin
[16:06] <JEEB> oh wait
[16:06] <JEEB> it is there
[16:06] <JEEB> meh
[16:06] <hackeron> lol
[16:07] <hackeron> any ideas?
[16:08] <JEEB> it can't find those symbols, which generally should be exported by the dl library
[16:08] <hackeron> what could be causing that?
[16:09] <JEEB> I have no idea, generally -ldl should get it done :s
[16:10] <JEEB> what distro are you on?
[16:10] <hackeron> ubuntu
[16:10] <JEEB> which version?
[16:10] <hackeron> latest Ubuntu 13.04
[16:10] <JEEB> http://manpages.ubuntu.com/manpages/raring/man3/dlsym.3.html :s
[16:11] <JEEB> should be with -ldl
[16:12] <hackeron> :/
[16:13] <JEEB> check if you have libc6-dev installed
[16:13] <hackeron> what can I try?
[16:13] <hackeron> libc6-dev is already the newest version.
[16:13] <JEEB> although I think that should have come with build-essential(s) :S
[16:13] <JEEB> I don't have 13.04 around atm so I can't just try it out here
[16:14] <hackeron> JEEB: I compiled about a month ago on 13.04 (before it was officially released) and it compiled just fine. After git pull to the latest releases of x264 and ffmpeg, ffmpeg can no longer find libx264
[16:14] <JEEB> but that error means that the linker cannot find the functions that are noticed
[16:15] <JEEB> hackeron, it's x264 because the opencl patch now uses runtime loading instead of linking the opencl library to the binary
[16:15] <hackeron> maybe something changed in the x264 git master?
[16:15] <JEEB> yes
[16:15] <JEEB> but it should work, I've seen people having it work with -ldl added
[16:15] <hackeron> hmm
[16:16] <JEEB> hackeron, also you could just disable opencl support in x264 :P
[16:16] <JEEB> but it should work with -ldl
[16:18] <hackeron> JEEB: hmmm, it doesn't :/ - let me try disabling opencl in x264
[16:18] <JEEB> well yes, and I'm at loss at why :P
[16:18] <JEEB> and yes, opencl support getting disabled means no need for those functions
[16:18] <hackeron> I compile x264 with ./configure --enable-static -- maybe that is an issue?
[16:18] <JEEB> no
[16:21] <JEEB> basically see the output of ./configure --help of x264's and see the option to disable opencl :P That will make it not need dlopen and friends
[16:22] <hackeron> yeh, trying that now :)
[16:23] <hackeron> yeh, that did the trick - but would be nice to get it working with opencl too :)
[16:23] <hackeron> any ideas at all what else I could try?
[17:08] <saste> michaelni, any idea about the diffing mcdeint issue?
[17:09] <saste> ^^ sorry wrong channel
[17:46] <murgle> I used ffmpeg to compress a .mov taken with a camera to a .mp4 (h.264/mp3).  When I try to play the compressed file in Quicktime, I don't get any sound.  My ffmpeg command:  ffmpeg -i input.MOV -vcodec libx264 -preset slow -crf 26 -acodec libmp3lame -ab 192k -ar 44100 -ac 2 -ss 19 -t 30 output.mp4
[17:58] <relaxed> murgle: Looks fine. What about when using vlc or something else?
[18:05] <murgle> hm, works fine in vlc.  must be some quicktime issue =\
[18:07] <Mavrik> does quicktime even like mp3 audio in mp4 files?
[18:07] <JEEB> it might have random limitations
[18:08] <JEEB> I'm pretty sure if you changed a small lot of things that go into the file by switching to mov it'd possibly play it
[18:08] <JEEB> lol
[18:08] <JEEB> ("mp4" type files are basically based on the MOV format, which is the "AVI of Apple")
[18:10] <Mavrik> JEEB, yep
[18:10] <Mavrik> and Apple went and changed MOV right after MP4 standardization
[18:10] <Mavrik> so those aren't compatible anymore ;)
[18:11] <JEEB> which standardization lololol. You've got like three or four specs by now that define a similar container in MPEG-4
[18:11] <JEEB> Plus 14496-15 being the mapping of AVC in those and such
[18:12] <Mavrik> JEEB, IIRC apple started changing MOV format spec after the first MPEG-4 spec got the "finalized" tag on it :)
[18:12] <Mavrik> so there was never a point where mp4 players could read mov
[18:12] <Mavrik> later changes and differences for mappings in MPEG4 not included ;)
[18:13] <JEEB> well, they always were different even if it was just on a metadata scale
[18:13] <Mavrik> mhm
[18:15] <murgle> quicktime seems to play aac in a mp4 file, but not mp3
[18:46] <jxself_> Hello - http://dpaste.com/1202606/ I assume that there's something wrong with my input files because FFmpeg seems to not do anything but drop frames although I'm not sure why or where to begin. I'm hoping that someone might be able to point me in the right direction of being able to determine what's wrong with them.
[18:51] <jxself_> I increased loglevel to debug, in case it's helpful: http://dpaste.com/1202624/
[18:52] <relaxed> yadif should be first in the filter chain
[18:53] <jxself_> Ah? OK. I'll go move that...
[18:53] <relaxed> yadif,setpts,scale
[18:54] <relaxed> I doubt you need setpts at all
[18:55] <jxself_> Perhaps. I run into audio/video sync issues and it's the result of my experimenting trying to solve it.
[18:56] <jxself_> Ok - The latest attempt: http://dpaste.com/1202627/
[18:56] <jxself_> It seems to have stopped at *** 811090 dup!
[18:58] <relaxed> why are you changing the framerate?
[18:59] <jxself_> I've found that it produces slightly smaller files than if I don't.
[18:59] <jxself_> Fewer frames to encode, I suppose.
[19:01] <relaxed> the correct way to get 23.* fps is -r 24000/1001
[19:01] <jxself_> OK. I can change that.
[19:02] <jxself_> And the result appears to be the same: http://dpaste.com/1202632/
[19:03] <jxself_> There is no further output after *** 811091 dup!
[19:03] <relaxed> maybe, -vf yadif,fps=24000/1001,scale
[19:03] <relaxed> and lose "-r 24000/1001"
[19:04] <relaxed> because with "-r 24000/1001" I'm not sure where it's instered in the filter chain.
[19:04] <relaxed> inserted*
[19:05] <jxself_> OK. Now it doesn't seem to hang but does continue dropping every frame: http://dpaste.com/1202634/
[19:05] <relaxed> when troubleshooing ffmpeg it's best to use the most basic command and build from there.
[19:06] <relaxed> it has to drop frames to achieve your desired framerate.
[19:06] <relaxed> omit fps and see what happens.
[19:08] <jxself_> And it seems to freeze: http://dpaste.com/1202635/
[19:09] <relaxed> please drop all options except -vf and the codecs and output
[19:09] <jxself_> And indeed it does need to drop frames. The thing I noticed was in the little progress thing it stayed at frame= 1 while the number number of dropped frames just kept increasing as the time went up.
[19:10] <jxself_> Ok Just a moment.
[19:12] <jxself_> And it has frozen: http://dpaste.com/1202638/
[19:13] <relaxed> jxself_: not try with -vf "yadif, scale=1280:720"
[19:13] <relaxed> now*
[19:14] <relaxed> maybe the defaults will work better for you.
[19:14] <jxself_> It appears to be the same: http://dpaste.com/1202641/
[19:14] <jxself_> The last thing shown is the *** 1013863 dup!
[19:17] <relaxed> I'm not really sure what's going on. Maybe remuxing to somehting like matroska might help?
[19:18] <jxself> Ok. I can try that.
[19:19] <relaxed> mkvmerge input -o output.mkv
[19:21] <jxself> It seems to not make a difference: http://dpaste.com/1202652/
[19:23] <relaxed> Stick around, maybe someone else will have an idea.
[19:23] <jxself> OK
[19:29] <llogan> jxself: can you provide a short sample?
[19:29] <jxself> Sure.
[19:36] <jxself> llogan: 10MB from dd - http://files.jxself.org/sample.mpg
[19:37] <llogan> thanks
[19:58] <the_gamer> is it possible to create an mp4 out of jpgs with ffmpeg?
[19:58] <the_gamer> like a diashow?
[19:59] <ubitux> yes
[19:59] <ubitux> it's in the FAQ
[20:03] <the_gamer> can't find it :(
[20:04] <llogan> http://ffmpeg.org/faq.html#How-do-I-encode-single-pictures-into-movies_003f
[20:04] <the_gamer> ffmpeg -i PICT0005.JPG -i PICT0006.JPG -framerate 25 -r 50 show.mp4
[20:04] <the_gamer> produces an empty file
[20:04] <the_gamer> oh, thanks :)
[20:05] <the_gamer> llogan, output is still an empty moviefile :(
[20:06] <the_gamer> using ffmpeg 1.2
[20:07] <the_gamer> http://pastebin.com/Zpa1fChA
[20:09] <the_gamer> without -r and -framerate same issue
[20:09] <the_gamer> it creates a videofile but the length is 0
[20:10] <llogan> what are you trying to do? you want to make a video from two images?
[20:10] <llogan> where did you get the idea to use -framerate?
[20:11] <llogan> duration isn't 0. it's 00:00:00.02
[20:11] <the_gamer> llogan, i want to tell ffmpeg how long one of the pice should be shown in the video
[20:11] <the_gamer> a have more images
[20:11] <the_gamer> two are for testing only
[20:12] <ubitux> why didn't you try reading the 2 first line of the faq entry?
[20:12] <the_gamer> i did. does %d make that much of a difference? ffmpeg finds the pictures?
[20:12] <ubitux> yes
[20:13] <ubitux> using two -i means 2 different streams
[20:13] <zap0> don't cross the streams!!!!11
[20:13] <the_gamer> two different streams?
[20:14] <the_gamer> oh get it, wait
[20:16] <the_gamer> ubitux, is there something like -end_number, too?
[20:17] <ubitux> -frames:v 123
[20:17] <ubitux> (number of output frames)
[20:18] <the_gamer> number of output frames would be bad& 1 picture 1frame would mean each picture is shown 1/25th of a second, right?
[20:18] <the_gamer> way too short
[20:20] <the_gamer> ffmpeg -start_number 5 -frames:v 70 -f image2 -i PICT%04d.JPG -framerate 25 -r 4 show.mp4
[20:20] <the_gamer> crashes in the middle of creating the video without notice
[20:20] <ubitux> -frames:v 70 is an output option
[20:21] <ubitux> for the crash
[20:21] <the_gamer> i still don't get how to say how long each picture should be shown
[20:22] <the_gamer> i guess i try something else :( thanks for your efforts
[20:22] <llogan> by default the inputs get -r 25.
[20:23] <llogan> you can declare an input -r and an output -r if you want to vary it (but that will result in duplicated or dropped frames)
[20:24] <ubitux> your command doesn't crash here btw
[20:25] <ubitux> even if it's definitely wrong
[20:25] <llogan> if you want each input to be shown for one second but your output to have frame rate of 25 (for compatibility with retarded players or whatever): ffmpeg -r 1 -i PICT%04d.JPG -r 25 output.mp4
[20:26] <llogan> -r can also use fractions
[20:27] Action: llogan leaves for camping trip.
[20:27] <the_gamer> great, thanks :) sry that i didn't manage it on my own
[20:34] <relaxed> There must be a bug because the ffmpeg's output with '-r 1 -i input.jpg -r 25' is 40ms while avconv's a 1 second.
[20:35] <ubitux> isn't 40 ms the correct thing?
[20:36] <ubitux> you specify that your input.jpg is at rate 1 fps, and then you want it at 25 fps
[20:36] <ubitux> since you have only one frame at 25fps, you get a 40 ms output
[20:37] <ubitux> relaxed: what's wrong with that?
[20:37] <durandal11707> its bug
[20:37] <ubitux> why is that wrong?
[20:37] <ubitux> looks correct to me
[20:39] <relaxed> I want 1fps in and 25fps out. What am I missing?
[20:39] <ubitux> isn't it exactly what you get?
[20:40] <relaxed> unless I'm missing something, 40 ms is not 1 second.
[20:40] <ubitux> one frame at 25 fps has a duration of 1/25.
[20:40] <ubitux> which is 40ms
[20:41] <ubitux> what you may want is repeating input.jpg at 25 fps (so repeat the picture 25 times on a time basis of 40ms to get one second)
[20:41] <relaxed> shouldn't it take each frame a dup to 25 frames, thus giving me one second?
[20:41] <ubitux> and in that case yo have a loop option
[20:42] <ubitux> relaxed: so you're in that case with multiple frames? -i %04.jpg?
[20:43] <Hans_Henrik> ffmpeg -i 'http://www.4chanx.org/torrentdownloads/Blood%2b/%5bShinsen-Subs%5d_Blood%2b_01_%5bC70CF80C%5d.avi' -c:v libx264 -f libx264 output pipe:1 2>&1
[20:43] <Hans_Henrik> [NULL @ 0x15f80c0] Requested output format 'libx264' is not a suitable output format
[20:43] <Hans_Henrik> output: Invalid argument
[20:43] <ubitux> libx264 is not a format
[20:43] <ubitux> it's a codec
[20:44] <Hans_Henrik> right.. "-f x264"   then?
[20:44] <ubitux> no, h264 i would guess
[20:44] <ubitux> x264 is an encoder
[20:44] <ubitux> a software name
[20:45] <Hans_Henrik> dam ,now the error changed to Unknown encoder 'libx264'
[20:46] <Hans_Henrik> guess that means by compiled file is fked
[20:46] <Hans_Henrik> thanks tho
[20:46] <durandal11707> Hans_Henrik: why are you sharing pirated content here?
[20:46] <ubitux> doesn't matter, the anime sux anyway
[21:15] <Hans_Henrik> durandal11707, ah don't worry, i have permission from Production I.G to share that link irc://irc.freenode.net/#ffmpeg
[22:32] <bunniefoofoo> Hi ffmpeg developers. I am updating code to use ffmpeg 1.2.x (have been using 0.8 for a long time) and I have run into the infamous x264/ffmpeg compatibility issue. I have relied on FF to pass rate control settings through to libx264, in addition the x264opts to be passed as well. Apparently this will not work anymore and I have to use presets. That isn't a great option since I have countless hours invested in tuning profiles. Any
[22:35] <jxself> Your message appears to be cut off after "profiles. Any"
[22:35] <bunniefoofoo> Too long.. ook here is the rest
[22:35] <bunniefoofoo>  Any opinons on what I should do? Go through x264 directly instead of libav ?
[22:38] <bunniefoofoo> Is it possible to pass all settings through x264opts (rate control and the whole shebang) so I won't need to use the api equivalenet of -vpre (assuming there is one)
[22:40] <durandal11707> api equivalent is seting options
[22:42] <bunniefoofoo> you mean passing AVDictionary with option keys/values when opening the codec, and those correspond to the key/values in ffpreset?
[22:44] <durandal11707> there is x264-params
[22:44] <durandal11707> actually same can be done with ffmpeg tool
[22:44] <durandal11707> so what options you cant pass?
[22:46] <bunniefoofoo> I can't tell, x264 check for about 10 different things and then aborts "broken ffmpeg default settings detected"
[22:47] <bunniefoofoo> looks like "x264opts" has been renamed to "x264-params"
[22:49] <durandal11707> both are available
[22:50] <bunniefoofoo> yeah I just saw that, trying to figure out the differences
[22:53] <bunniefoofoo> they look equivalent? I guess this is not the problem
[00:00] --- Wed May 29 2013