[Ffmpeg-devel-irc] ffmpeg.log.20120712

[burek]

[00:03] <Orbixx> burek: Yeah :/
[00:04] <heth> llogan, is there any information/ basic develompment or any alpha builds on that
[00:05] <llogan> heth: some irc logs: http://pastebin.com/wcWjkQrF
[00:06] <llogan> you can try using a faster preset
[00:12] <heth> llogan, I see. Thanks.
[02:20] <r4y> 	When I use try to convert flv to avi using transmageddon the audio becomes out of sync. If I use ffmpeg the picture quality is boxy.
[02:41] <r4y> Sorry I was trying to come up with a good example: http://pastebin.com/hQFL1qA3
[02:43] <r4y> This is where the video came from: http://www.youtube.com/watch?v=DB3pMlHUgTo
[02:44] <r4y> I am not sure how I downloaded that video back a few years ago
[02:45] <r4y> I just know I have tried many times to use ffmpeg before to convert videos to avi and it was boxy
[02:46] <r4y> The reason I convert videos to avi is because I use avidemux to edit out parts
[03:01] <r4y> How do I set the codec?
[03:12] <burek> r4y, why do you want to convert it to avi
[03:12] <burek> just ignore me
[03:12] <burek> i just saw your next line :)
[03:13] <r4y> I did "ffmpeg -formats" I want an example of how to use the codecs shown?, As you can see I know how to choose the file format for the input and output. How for instance can I choose both the codec and the file format for the input and output?
[03:13] <burek> well, it would be best to just decode flv format (audio and video streams inside) to raw streams
[03:13] <burek> it will be big
[03:13] <burek> but you'll have the least losses possible
[03:14] <burek> btw, format and codecs are not the same
[03:14] <llogan> yes, for editing it is best to use a temporary lossless video if the editor can't properly deal with the original file
[03:14] <burek> format/container is like an envelope in which you put your letter(s) or your streams (audio/video/subtitles/data)
[03:15] <burek> so it is kinda wrong to say "convert" to avi.. rather maybe remux to avi (since avi is a format, not a codec)
[03:15] <llogan> and the reason it is "blocky" is because the default settings are crappy for lossy outputs
[03:15] <llogan> *for most lossy outputs
[03:16] <burek> try ffmpeg -i Siteswap_423.flv -c copy out.avi
[03:16] <burek> just for fun :D
[03:17] <llogan> i assume avidemux should support the original file.
[03:17] <burek> when it fails, try this :) ffmpeg -i Siteswap_423.flv -vcodec rawvideo -acodec pcm_s16le out.avi
[03:17] <llogan> his ffmpeg is too old for -c copy
[03:18] <r4y> sorry I have too many windows open lol
[03:18] <llogan> i'd recommend huffyuv over rawvideo
[03:19] <burek> why's that
[03:20] <llogan> more manageable file size i'd guess
[03:21] <burek> there was some avi codec or M$ something, that was using only RLE, without loss of quality
[03:21] <burek> is there a way to select from ffmpeg -codecs only lossless codecs?
[03:21] <burek> or those available to encode losslessly
[03:21] <llogan> no
[03:22] <llogan> but other options are ffv1, libx264, ffvhuff, libutvideo, and probably some others i'm forgetting
[03:22] <r4y> Thank you, that seemed to work
[03:22] <burek> is it pointless to ask for such a feature in that listing
[03:23] <burek> r4y, what exactly worked :D
[03:23] <r4y> It's not boxy
[03:23] <burek> I mean what command line
[03:23] <r4y> "ffmpeg -i Siteswap_423.flv -vcodec rawvideo -acodec pcm_s16le out.avi"
[03:24] <llogan> burek: i don't think it would be pointless, but i don't know if it would be implemented
[03:24] <r4y> I need to study the command you just gave me
[03:24] <llogan> r4y: avidemux doesn't like Siteswap_423.flv?
[03:25] <r4y> flv files just don't do well in avidemux. I think it's made for avi file formats
[03:26] <burek> too bad they didn't upgrade their software to use libavformat, at least to extract streams and edit them
[03:26] <r4y> I should say though since I am new to this though. I've been converting and editing videos for less then a year
[03:26] <r4y> I meant I shouldn't say, oops typo
[03:27] <r4y> Maybe I should go ask them to?
[03:27] <llogan> http://www.avidemux.org/admWiki/doku.php?id=general:input_formats
[03:27] <llogan> "flash video" is listed
[03:28] <r4y> I must be doing something wrong. I just remember having trouble
[03:28] <burek> is that container or a codec :)
[03:29] <llogan> i assume they mean the container in that context
[03:32] <r4y> I could try editing an flv video with avidemux directly but the problem is choosing what the output with be, so it could take a while to see how things go
[03:33] <r4y> At least I have a video I wanted to edit that is an flv video file
[03:33] <llogan> what output do you want it to be? it has a good number of options.
[03:34] <r4y> I normally choose xvid
[03:34] <burek> btw, you can also save it as a lossless video
[03:34] <burek> and then use ffmpeg to encode h264
[03:34] <burek> which would be the best choice imho
[03:35] <r4y> lossless?, in avidemux?
[03:35] <burek> well, I dunno.. maybe there is something.. :) when everything else fails, try reading the documentation :D
[03:35] <llogan> it should allow you to export huyyuv in mkv or avi
[03:35] <llogan> *huffyuv
[03:36] <r4y> I could try just copying maybe, but I am not sure how that will go
[03:37] <r4y> I will try both
[03:40] <r4y> O, you mean edit then re-encode?
[03:43] <r4y> I've never heard of mkv but you are right, it has the option to export as mkv or avi
[03:45] <r4y> I tried huffyuv and it works, so I guess there is no need for converting videos from flv to avi. One of the things I was shooting for was creating more space, so I guess I should compare file size right now
[03:46] <burek> when you are done editing
[03:46] <r4y> Holy smokes, it went from 19.6 MB to 956.5 MB
[03:47] <burek> then just: ffmpeg -i input -vcodec libx264 -preset slow -crf 20 -acodec ... output.mp4
[03:47] <burek> or flv
[03:47] <burek> choose acodec you like, might even be -acodec copy
[03:47] <burek> to preserve original
[03:48] <burek> you might also do that anyway, to keep the audio always in original form, no matter what you encode video to
[03:50] <r4y> What do you think I choose for video output .flv?, mp4?, avi?,
[03:50] <burek> well, mp4 is good for file storage
[03:50] <burek> flv is also streamable and also popular
[03:50] <r4y> Cool, sorry, I just had to ask
[03:50] <burek> avi is.. well.. old
[03:51] <burek> avoid it if you can
[03:52] <burek> I personally use flv more, because I sometimes like to stream media to remote machines
[03:52] <burek> so I can stream flv without any problems
[03:52] <burek> but mp4 is ok too
[03:52] <burek> just, it can't be easily streamable
[03:52] <burek> it wasn't designed with that in mind
[03:53] <r4y> "ffmpeg: unrecognized option '-preset'" the command I used:
[03:53] <r4y> "ffmpeg -i Rubik_s_Cube_Notation_All_Notation_.flv -vcodec libx264 -preset slow -crf 20 -acodec pcm_s16le output.mp4"
[03:53] <burek> ugh
[03:53] <burek> FFmpeg version SVN-r0.5.9-4:0.5.9-0ubuntu0.10.04.1
[03:53] <burek> that's really really really old
[03:54] <burek> and since you appear to be on ubuntu
[03:54] <burek> they removed ffmpeg from their repository
[03:54] <r4y> I didn't like the new gui that put out with Ubuntu 12
[03:54] <burek> so, most probably you will have to build it from source
[03:55] <burek> things are even worse.. in new ubuntu you don't even have ffmpeg
[03:55] <burek> you have avconv that is really buggy
[03:55] <burek> so, choose your own way
[03:55] <burek> I suggest compilation
[03:56] <burek> you have some tutorials there
[03:56] <burek> it's not that difficult
[03:56] <burek> and is always better to have it compiled from git, since you can always easily update it
[03:56] <burek> and stay up-to-date
[03:57] <burek> just every couple of months you do: make uninstall, make distclean, git pull, ./configure ..., make, make install
[03:57] <burek> and that's it
[03:57] <burek> you can even create a script for that :)
[04:00] <r4y> It's a good thing I've installed from make packages before, but I wondering why I don't just adjust the command I posted. I was manning it in the terminal
[04:01] <r4y> From the command you gave me I mean
[04:06] <r4y> I just learned that by hitting "/" in the terminal that I can search for a word I am looking for, so I am studying what I found under the word preset for the ffmpeg manpage
[04:10] <r4y> I have to leave for a little. I will need to read up on the commands you gave me
[04:44] <Synthead> I'm trying to make a video with an audio track and a static image.  Here's my command and output: http://pastebin.com/Lp9pnYEV  Specifically, I'm getting "Output file #0 does not contain any stream".  What am I doing wrong?
[04:46] <newl> what does it give without all that stuff between the wav and the mkv
[04:48] <Synthead> newl: what do you mean?
[04:50] <newl> just the -i and output file
[04:51] <newl> make it simple and find what breaks
[04:52] <Synthead> oops
[04:52] <Synthead> I forgot the second -i
[04:52] <Synthead> pebkac
[05:17] <smokeyrat> Whoa.
[05:17] <smokeyrat> Did NOT expect this to have an IRC channel, much less such a big one.
[05:17] <smokeyrat> Well, this is nice.
[05:18] <smokeyrat> Well, I have the latest stable ffmpeg on 64-bit Windows 7. Thanks to the .bat files I created, I am able to drag and drop any video file into them and ffmpeg spits out an .avi. However, when I drag and drop .flv files to the following, the file size is much smaller in the resulting .avi and the quality is worse.
[05:18] <smokeyrat> Can you explain why, and help me remedy it? "C:\Program Files (x86)\ffmpeg\bin\ffmpeg" -i %1 %1.avi
[05:19] <smokeyrat> It seems to me that by default it should use the same output quality as the input, no?
[05:43] <r4y> I've used open movie editor to do what I think you are trying to do
[05:45] <r4y> Here's my example: http://www.youtube.com/watch?v=Fk0_Ux5VrhM
[05:50] <smokeyrat> Did anyone see what I asked earlier and replied?
[05:52] <r4y> I am sorry I was talking to Synthead, but I guess he figured out what to do. I am not the person to ask questions here. I am sorry
[05:55] <smokeyrat> =S
[05:59] <lake> hi all, i'm encoding video from a V4L2 device. My goal is to create dvds of the content. Each video is ~120 minutes long and I am encoding them in MPEG2 and a bitrate small enough to fit on a dvd (~4.5gb).
[05:59] <lake> I have found major loss in quality when I put the image on dvd. it is very noticeable.
[06:00] <lake> How can i ensure that I have the best looking encoding?
[06:00] <lake> my bitrate is set to around 5000kbs
[06:19] <Derakon> I have a raw video file exported from an SNES emulator. It's about 15 minutes long, and by the end the audio is about 1s in advance of the video. I suspect this is because the emulator runs at 59.94FPS but the video is marked as being 60FPS. I want to fix the video rate so it matches the audio rate.
[06:19] <Derakon> I tried using the -r command to set the input/output FPS but it doesn't seem to actually do anything, as far as I can tell.
[06:20] <Derakon> For example, I would expect "ffmpeg -r 59.94 -i input.mov -r 80 output.mov" to cause the video to rapidly get well in advance of the audio, but I still get exactly the same desynchronization.
[06:43] <smokeyrat> Seems like this channel is all questions and no answers.
[06:45] <Derakon> Unfortunate.
[06:46] <zap0> smokeyrat  -sameq
[06:46] <smokeyrat> zap0: Why on Earth is that not the default?
[06:46] <zap0> cause everyone that made ffmpeg is from mars
[06:47] Action: smokeyrat tests to see if it works.
[06:47] <smokeyrat> ffmpeg is an odd name and an obscure app (since it has escaped my knowledge for so long), yet apparently extremely advanced.
[06:47] <smokeyrat> But I'll be damned if the FOSS folks ever make a video editor that is usable.
[06:48] <smokeyrat> Ah. The -sameq thing worked. Thanks.
[06:48] <smokeyrat> Much bigger files, though.
[06:48] <zap0> Derakon, set the audio correctly as per the data.  set the video rate correctly as per the data.   then its the players problem to sync it.
[06:49] <smokeyrat> zap0 cleans up the list of pending problems.
[06:49] <Derakon> zap0: got some search terms I could use to find the appropriate arguments?
[06:50] <zap0> Derakon, hwo do you know it wasn't the player out of sync
[06:51] <Derakon> Because I've tried several players and they all behave identically.
[06:55] <smokeyrat> Seriously, though... Why would anyone, ultra-geek or not, consider it reasonable to make the CONVERSION of a video file to another video file NOT use the same quality *by default*?
[06:55] <smokeyrat> I mean, that really is the assumption one has.
[06:55] <drv> -sameq is unfortunately named and doesn't do what you probably think
[06:56] <drv> the q means "quantizer", not "quality", so it doesn't make any sense across different codecs
[06:56] <smokeyrat> =S
[06:56] <smokeyrat> Well, it seemed to do what I expected it to.
[06:56] <smokeyrat> Why must everything be non-straight-forward and complex?
[06:57] <Derakon> Because it was written by people who understood exactly what they were doing, for other people who understood exactly what they were doing.
[06:57] <smokeyrat> So basically, you are saying that my .flv-to-.avi conversion worked out of luck?
[06:57] <Derakon> And with little mind for people who didn't.
[06:58] <smokeyrat> Wouldn't it make more sense for the binary to be a "magic black box" where you input any video and it spits out the same thing in the format you want?
[07:07] <Derakon> Well, I found a setting in my emulator that makes the source movie have A/V sync, so I guess I don't need ffmpeg any more.
[07:07] <Derakon> Oh well.
[07:08] <smokeyrat> You just have to replay it.
[07:09] <smokeyrat> Sounds annoying to me.
[07:09] <Derakon> Not so much, actually; the emulator was playing back a recording of controller inputs, and outputting the video and audio streams.
[07:09] <Derakon> So I just have to replay back the inputs, which I'd saved to a separate file.
[07:09] <Derakon> (Though, I'll be replaying anyway since I wasn't satisfied with the play quality)
[07:10] <smokeyrat> =S
[07:10] <smokeyrat> Which game?
[07:10] <smokeyrat> Super Mario Kart?
[07:10] <Derakon> Super Metroid. Actually, a hack thereof.
[07:11] <Derakon> As a proof of concept I ran through a boss rush hack; later I want to record a playthrough of a full-conversion hack.
[07:12] <smokeyrat> Well...
[07:12] <smokeyrat> All hope of impressing people is lost due to this asshole: http://www.youtube.com/watch?v=4xyvQQSbX5o
[07:13] <Derakon> Why, you think that's the most impressive a hack can be?
[07:14] <smokeyrat> Pretty much.
[07:14] <smokeyrat> Must've taken forever.
[07:14] <Derakon> Ehh.
[07:14] <Derakon> I've seen more effort-intensive stuff.
[07:14] <Derakon> There's no graphical or code changes in that, for example.
[07:14] <smokeyrat> Seems impossible. :S
[07:15] <smokeyrat> Sure, he's probably "just" used some level editor, but the sheer length and complexity...
[07:15] <Derakon> Oh, it's impressive, I grant.
[07:15] <smokeyrat> And timed with the music.
[07:15] <smokeyrat> Not really sure how Mario can move like that, though. Even though I know Super Mario World has an impressive physics engine.
[07:15] <smokeyrat> That is, the controller is not touched.
[07:15] <Derakon> There's blocks that set him to max forward speed.
[07:16] <smokeyrat> Are those in the actual game?
[07:16] <Derakon> Dunno; even if they aren't used in the base game they could just be dummied-out.
[07:17] <smokeyrat> All chances of finding secret stuff in these old games are gone due to digital archeologs having dug through them and analyzed every block.
[07:17] <Derakon> Well, for the popular games anyway.
[07:18] <smokeyrat> I kept hoping for a long time that Super Mario Kart would have hidden netplay code for future hardware.
[07:18] <Derakon> Heh.
[07:18] <Derakon> The Internet wasn't so much of a thing when the SNES was made.
[07:18] <smokeyrat> They had some kind of networked SNES in the USA for a while. Third party add-on.
[07:18] <smokeyrat> I forget its name...
[07:19] <smokeyrat> And Japan of course had Satellaview...
[07:19] <Derakon> That's different -- they distributed content over it, but it was a one-way link.
[07:19] <Derakon> No way to use that for netplay.
[07:19] <smokeyrat> The Satellaview? Yeah.
[07:19] <Derakon> ZSNES has a netplay option though.
[07:19] <smokeyrat> But the first one did have actual netplay.
[07:19] <smokeyrat> ZSNES hasn't been touched for ages.
[08:32] <dan_> nvm about the ac3 questions... it turned out that the recording had one commercial which was broadcast with stereo audio, whereas the whole rest was in 5.1
[08:32] <dan_> and the splice points had odd PTS gaps
[10:33] <r4y> Is huffyuv lossless?
[10:34] <r4y> I am trying to dissect what was said before. It gets confusing.
[10:36] <md_5> I'm using ffmpeg to concert a stream rtsp, 3gp stream to 128x128 frames at X fps. Issue is sometimes ffmpeg just goes crazy and creates over one hundred thousand, garbage frames
[10:36] <md_5> Thinking it may be something to do with it handling latency in the stream
[10:37] <r4y> Using avidemux to edit what codec should I choose to get close to lossless from an flv file then use ffmpeg to convert that video to mp4 for storage using the codec h264
[10:39] <r4y> I should do testing, sorry for the trouble
[11:49] <md_5> burek hmm maybe I'll try manually running it until that happens later, I was running it in a headless environment so had no actual output
[11:50] <md_5> command I used was ffmpeg -i rtsp://<url>.3gp -s 128x128 -r 18 Frame%d.png
[11:51] <burek> and what version of ffmpeg are you using
[11:51] <md_5> ffmpeg version 0.8.3-4:0.8.3-0ubuntu0.12.04.1,
[11:51] <burek> 0.11.x is current
[11:51] <burek> and btw, take a look at the credits lie
[11:51] <burek> line*
[11:51] <burek> does it say FFmpeg developers or Libav?
[11:51] <md_5> Libav
[11:52] <burek> then you are not using ffmpeg
[11:52] <md_5> avconv
[11:52] <burek> well that's not ffmpeg
[11:52] <burek> it's a bad fork
[11:52] <md_5> thought they were the same project :|
[11:52] <burek> if you want ffmpeg, you'll need to compile it from source
[11:52] <burek> well, that's to be thanked to ubuntu/debian distro maintainers
[11:52] <burek> who decided to make it look like so
[11:53] <burek> and I guess you too have that stupid line like "ffmpeg is deprecated" something like that
[11:53] <md_5> yep, hence I switched to avconv thinking it was the same thing
[11:53] <md_5> ooh yay git
[11:54] <md_5> damn thats a huge repo *downloads tarball
[11:54] <burek> why not just git clone --depth=1
[11:55] <md_5> you make a point
[12:04] <md_5> burek thanks for that, compiled and working, Ill see how it goes
[12:04] <burek> :beer: :)
[12:05] <md_5> sudo apt-get purge ffmpeg
[12:05] <md_5> :D
[12:05] <burek> or --purge autoremove ffmpeg
[12:05] <burek> :)
[12:05] <md_5> grgh avconv is still installed
[12:06] <burek> autoremove removes all its dependencies too :)
[12:06] <md_5> yeah autoremove didnt remove anything
[12:15] <burek> well remove it manually with apt-get --purge autoremove <pkgname>
[12:15] <burek> dpkg -l | grep -i ffmpeg
[12:15] <burek> to get the list
[12:15] <burek> or grep libav
[13:59] <killown> It's possible to record the internal System Sound?
[14:00] <killown> For example now I am playing a game and I would like to record the sound
[14:01] <zimbatm> probably, depends on your audio system
[14:01] <zimbatm> http://ffmpeg.org/ffmpeg.html#Input-Devices
[14:33] <Olson> heyo.. I am having troubles :(  I make tutorials using x11grab and ffmpeg.. but right now the audio is going out of sync.. it starts off ok, but gradually gets worse, by about 2 minutes it, the audio is getting so far out it's impossible to follow.. i not sure what to do
[14:34] <Olson> I did it like http://pastebin.com/uvz2mGvf
[14:35] <Olson> I believe I'm using more or less the latest FFMPEG from the GIT source, after building a few days ago
[14:38] <Olson> it is strange because it used to work, and I recorded many videos
[14:38] <Olson> now I sort of stuck
[15:16] <Olson> oh woes
[15:25] <Olson> http://pastebin.com/hp9ad27t  this rather long, but it's what terminal said.. I notice the fps fluctuate, 28fps, 29, 30... I recorded at 30.. would this account for the video advancing in front of the audio?
[15:30] <Olson> o/
[15:33] <iive> no, not likely.
[15:33] <iive> if frames are dropped, then the video should lag behind the audio.
[15:34] <iive> capturing from a mic sometimes could be a tricky subject. I'm not aware how ffmpeg handles it, or how alsa does.
[15:35] <iive> in the past with discrete sound cards there was a big problem caused by the sampling clock oscillation.
[15:36] <Olson> if less video frames per second are recorded, and played back at constant fps then surely video is faster
[15:36] <Olson> if i record 20 fps and play back at 30 then..? no?
[15:37] <iive> you'll play 20 frames in 2/3 of second, not 3/3 of a second.
[15:37] <iive> if you think it is regression, then try to find the last working version of ffmpeg.
[15:38] <iive> having a multiple hour recording to trigger the bug sounds like nightmare.
[15:38] <iive> and keep asking, somebody may know more.
[15:39] <Olson> it's going out of sync within a minute or so
[15:39] <iive> oh, this is fast!
[15:39] <Olson> in fact, noticeably so after I saw from the log that fps went to 29 rather than 30, but that could be coindidence
[15:40] <Olson> uhm *coincidence
[15:40] <iive> just for the test, you may try mjpeg, as the fastest encoder.
[15:40] <Olson> i went to 50 seconds in where it say around frame 1533 would be and indeed after that.. it's start to go awry
[15:41] <Olson> rather than libx264 in there?
[15:43] <iive> yep, just in case x264 eats too much cpu.
[15:44] <iive> it is not likely, but just in case.
[15:44] <Olson> I do have the 6 cores
[15:44] <Olson> i will try
[15:44] <Olson> thank you
[15:49] <mparodi> what're the normal parameters of -b:v, -minrate and -maxrate for a video on the web??
[16:55] <dericed> i'm testing ffmpeg's new iec61883 support to possibly replace my use of dvgrab. I have two dv decks attached via 2 ieee1394 cables to an imac running ubuntu with latest ffmpeg from git and am having trouble getting ffmpeg to capture from the second deck.
[16:55] <dericed> When I specify "ffmpeg -f iec61883 -i 0 -an -c copy -y out.dv" I can receive data from the first deck. When I run "ffmpeg -f iec61883 -i 1 -an -c copy -y out2.dv" I get an error "[iec61883 @ 0x246ae80] Selecting IEEE1394 port: 1 / [iec61883 @ 0x246ae80] Failed setting IEEE1394 port."
[16:56] <dericed> I can interact with both decks properly with dvgrab (using -guid) or using dvcont using "dev 0|1"
[19:44] <sgfgdf> hello, guys! i have avi file with this video spec -- "Stream #0:0: Video: mpeg4 (Advanced Simple Profile) (XVID / 0x44495658), yuv420p, 1280x690 [SAR 1:1 DAR 128:69], 23.98 tbr, 23.98 tbn, 23.98 tbc" which is 2.36GB. now i wanted to convert the video to H.264 and don't mess with the audio. so i used -- "ffmpeg -i in.avi -acodec copy -vcodec libx264 out.mp4" and the video become 1.43GB. is this normal and is this the way i can loose quality as low as possible?
[19:53] <JEEB> you can use -preset to control encoding speed vs compression ratio (default is 'medium'), and if it looks good you can try a higher crf value, and if it looks bad you can make it a bit lower (default is 23)
[19:53] <JEEB> list of x264's presets http://mewiki.project357.com/wiki/X264_Settings#preset
[19:59] <sgfgdf> JEEB, what are the ranges of that crf?
[19:59] <JEEB> --crf <float>           Quality-based VBR (0-51)
[20:00] <JEEB> (this is from x264's command line help)
[20:00] <iive> 0 is special case lossless. the rest is logarithmic.
[20:00] <JEEB> yeah, was going to mention that
[20:01] <iive> value of 18 is the same as quant=2 in mpeg1/2. (max quality)
[20:01] <JEEB> is it exactly the same?
[20:01] <JEEB> I mean, crf is not constant quants IIRC
[20:01] <JEEB> anyways, 23 is a pretty good baseline, it compresses nicely in most cases and in most cases for most people it looks "good enough"
[20:01] <iive> no need to be exact... mpeg2 also have option for logarithmic quantizer.
[20:02] <JEEB> yeah, I was just a bit o_O at someone comparing crf with constant quants
[20:02] <JEEB> because it's not exactly the same AFAICS
[20:02] <iive> and bigger values produce lower quality and smaller files.
[20:02] <sgfgdf> JEEB, i did not understand if the crf conrtols the quality or the -preset or both of them. so does veryslow means better quality compared to veryfast
[20:03] <JEEB> preset lets x264 use more features for compression
[20:03] <JEEB> and crf sets the "quality level"
[20:03] <iive> crf and cq control how much data would be lost. presets control what features and how much effort would be spent trying to find better match.
[20:04] <JEEB> yup
[20:04] <sgfgdf> JEEB, so it just makes my file smaller, but very slowly, but i don't loose quality during the process?
[20:04] <JEEB> you always lose quality
[20:04] <JEEB> in lossy encoding
[20:04] <JEEB> whether or not that quality degradation is visible to you is a whole different affair
[20:05] <sgfgdf> JEEB, yea i mean compared to -preset ultrafast and -preset veryslow
[20:05] <JEEB> you can get the same quality if you adjust the crf somewhat, but in most cases veryslow will be quite a bit smaller than ultrafast
[20:05] <JEEB> also ultrafast is pretty much only meant for benchmarks
[20:05] <JEEB> so is placebo
[20:06] <JEEB> ultrafast is meant to basically show how fast x264 can be, and placebo is something to show how well x264 can compress, I don't really recommend using either of those, although if you have the cpu power and time to burn, be my guest with placebo
[20:06] <sgfgdf> JEEB, but that was my question if i don't change anything just the -preset, will the difference be only on the size or quality also?
[20:07] <JEEB> crf values are not static between settings because the way crf values determine the output depend on the various algorithms used, which depend on presets and settings in general
[20:07] <iive> i think placebo does use tranform for ME, not the casual SSqE
[20:07] <sgfgdf> JEEB, btw while using this encoding do i need more CPU or the graphic card is needed?
[20:08] <iive> cpu
[20:08] <sgfgdf> JEEB, okay so in case i use static crf lets say 23 and just change the -preset? so i assume that there should be only size difference. am i right?
[20:09] <iive> most of less.
[20:09] <iive> more or less....
[20:09] <JEEB> yeah, generally crf is the "quality lever"
[20:09] <JEEB> but the result does depend on the presets
[20:10] <iive> i may be wrong but I think that crf calculated some stuff using the motion in the scene
[20:10] <sgfgdf> JEEB, or you mean even if i set staic crf it won't be what i specifyed depending on the -preset (for example)?
[20:10] <sgfgdf> *static
[20:10] <iive> so if motion estimation is done differently, it may tweak something differently.
[20:11] <JEEB> sgfgdf, it will not be stable, yes -- although in many cases the difference won't be really visible :P
[20:11] <JEEB> just wanting to tell you how it is instead of a simplified version
[20:11] <sgfgdf> JEEB, :)
[20:12] <sgfgdf> JEEB, so lower crf is better quality, right?
[20:12] <JEEB> yes
[20:12] <sgfgdf> JEEB, as someone mentioned crf=0 is losless is then there no compression?
[20:12] <JEEB> no, lossless just means that input = output
[20:12] <JEEB> compression will be done
[20:13] <iive> it would compress it, like .rar archive.
[20:14] <sgfgdf> JEEB, but how can i achieve this since all those codecs are named to be lossy? and isn't it be lossless if the source is in lossless format?
[20:14] <JEEB> uhhh
[20:14] <JEEB> I'm not sure what you're asking there :D
[20:14] <cbreak> the name is irrelevant
[20:14] <cbreak> you can call a chicken a chair, it still won't be sittable.
[20:15] <cbreak> Codecs have some steps that are lossy, and others that are not.
[20:15] <iive> sgfgdf: all lossless codecs compress very little. 30% to 50%. all lossy codecs compress more than 10 times.
[20:15] <cbreak> and if you leave away the lossy parts, like not doing any quantification, then you won't lose anything
[20:15] <cbreak> (apart from rounding errors)
[20:20] <sgfgdf> i thought that every time i reencode something it loose some quality, so thats why i asked, how can i achieve no loss. if you want just to change the encoding of a video and use this lossless way (H.264 allows you) will you get bigger file than the original or it can't be predicted?
[20:21] <JEEB> a lossless file in 99% of all cases will be bigger than your lossy source
[20:21] <JEEB> also if you want to play that H.264 stream on a plastic box kind of thing you don't want to use lossless H.264
[20:22] <sgfgdf> JEEB, isn't in becoming just like raw video or something similar?
[20:23] <iive> sgfgdf: lossy codec would encode the image differently. This my definition it always lowers quality. In practise however, you are not going to notice these differences.
[20:23] <sgfgdf> JEEB, from what you say it probably will be a lot bigger than the source.
[20:23] <JEEB> well, yes -- if you compare the lossless output to raw YCbCr or RGB, then it will be smaller, the problem is that if you've got an already compressed source (that will then get decoded and recompressed losslessly), the output file will be bigger than the source file in 99% of all cases
[20:24] <JEEB> *an already lossily compressed source
[20:26] <iive> sgfgdf: if you try to encode this rip with lossless, the encoder would spend a lot of bits trying to preserve data that is actually error/artifacts introduced by the previous encoder, not real signal that have existed in the original original.
[20:26] <sgfgdf> i think i got it.
[20:26] <burek> "how can i achieve no loss. if you want just to change the encoding of a video" - you can't
[20:27] <JEEB> well, you can
[20:27] <burek> I mean you can, but not without size impact
[20:27] <JEEB> yeah
[20:27] <sgfgdf> JEEB, iive, thank you very much for your assistant!
[20:27] <sgfgdf> thanks to the others too!
[20:54] <newl> wow you guys get to have assistants?
[20:54] <newl> i gets to do my own goolgling for answers
[21:13] <iive> newl: you should be asking simpler and general questions. :P
[22:10] <mape> Hi, I'm doing a project which allows a user to drag&drop a video file onto a website, this starts an XHR upload (octet-stream) which is then piped into ffmpeg so that the user can se real time updates of what frames are being processed on the server while the file is being uploaded. On some files I get the error "Could not find codec parameters (Video: h264, 13973 kb/s)" which from some googling seems to be that the meta data ("moov" box) isn't 
[22:10] <mape> Is there any way of solving this without having to alter the input file? Or is the only solution to just buffer the entire file and then use it with ffmpeg?
[22:11] <mape> stderr: pipe:: could not find codec parameters
[22:17] <newl> sounds pretty cool if it works ....  try installing x264
[22:17] <newl> then you will have to recompile ffmpeg to use it
[22:18] <mape> It is compiled with --enable-libx264
[22:19] <mape> And works with other files (.mkv rather then mp4 though) that use "H264 - MPEG-4 AVC (part10) (avc1)
[22:20] <newl> ffmpeg -codecs | grep 264
[22:20] <mape> which is the same codec as the other file uses
[22:20] <mape>  D V D  h264            H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
[22:20] <mape>   EV    libx264         libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
[22:20] <newl> is the ffmpeg used on their system? maybe it doesn't have x264 in it
[22:22] <mape> ? I'm testing both files on my system, on one file it works and one it doesn't. Both seems to be using the same codec. So I'm thinking it can have something to do with the meta data
[22:22] <mape> http://stackoverflow.com/questions/10381695/ffmpeg-pipe-input-error
[22:22] <mape> Seems similar to that, at least that the answer could be the same
[22:23] Action: newl passes mape up to level 2 tech support
[22:25] <rainmaker1> I want to save multicast stream to a file and I am using this command
[22:26] <rainmaker1> /usr/local/bin/ffmpeg -y -i udp://230.0.0.3:5000 -c:v copy -c:a copy -map 0 -f segment -segment_time 60 -segment_format mpegts -segment_list test.m3u8 stream%05d.ts
[22:26] <rainmaker1> now, when I play with vlc it plays fine
[22:26] <rainmaker1> but when I try to stream that file via rtmp I get audio/video out of sync
[22:26] <rainmaker1> any ideas why?
[22:53] <mindfire6482> Hello everyone, if i might trouble someone, does anybody know the FFMpeg command for deleting the original file after it has been converted?
[22:58] <burek> mape, google for qt-faststart
[22:59] <mape> burek: jup, have looked at that, but it wouldn't help when I'm piping in foreign files from users
[22:59] <burek> well in that case you don't show the progress with frames
[22:59] <burek> only %
[23:00] <mindfire6482> does anybody know the ffmpeg command to delete the original file after conversion?
[23:00] <burek> mindfire6482, no
[23:02] <mape> burek: well all I'm doing is making the video into frames, as the service is to do gifs for bug reports, but guess I can just save to disk if it breaks when trying to stream.
[23:02] <mindfire6482> thanks burek, ive read about scripting it but using && rm throws out errors
[23:02] <burek> mape, not all formats are streamable
[23:03] <burek> that's the main problem for you
[23:03] <burek> mindfire6482, try ffmpeg -i bla ... && rm -f bla
[23:03] <mape> Yeah, guess it is the way it is
[23:03] <burek> mindfire6482, if on windows, use del bla
[23:04] <mindfire6482> im on linux burek
[23:04] <burek> then rm should work
[23:04] <burek> if you have write access/ownership over that file
[23:04] <burek> mape, you can also create your own uploader
[23:04] <burek> which will do qt-faststart on chosen file(s)
[23:04] <burek> before upload
[23:05] <mape> would have to do that in js then, or flash, or java applet, all seem a lot more clonky. I reccon it is good enough now, I just have to add some fallback in case it isn't streamable
[23:06] <mindfire6482> all my permissions are set correctly, im using winff as a gui for ffmpeg, there is a tab under options for additional command line options but if i put in & rm i get errors an simply using rm throws out a diffrent error
[23:07] <burek> mindfire6482, this is not winff support channel, unfortunately
[23:07] <burek> you need to go to their user support
[23:07] <burek> and ask for help
[23:09] <mindfire6482> but, stop me if im wrong, winff simply gives you a graphical way of passing arg's to ffmpeg, so... there should be a command i can add to the ffmpeg preset as though i was using it solely command line
[23:09] <burek> no
[23:10] <burek> there is no such thing as an additional command line in ffmpeg
[23:10] <burek> you need to ask there for help, not here
[23:10] <mindfire6482> im not explaing it very well
[23:11] <burek> "there should be a command i can add to the ffmpeg preset as though i was using it solely command line"
[23:11] <burek> that is simply not true
[23:11] <burek> it doesn't exist
[23:11] <burek> got it?
[23:11] <eivanov> Hi! I convert image to video: ffmpeg -f image2 -r 150 -i img%d.jpg a.avi  I expected it to be 6s, but it's less than 1 :/
[23:11] <newl> make a script
[23:11] <eivanov> I have just 1 img, btw.
[23:12] <eivanov> How can I make it 6s?
[23:12] <mindfire6482> cheers burek..
[23:12] <burek> eivanov, you said that ffmpeg should read 150 images in a second
[23:12] <burek> do you have 900 images?
[23:13] <burek> if you have 1 image, you need -loop
[23:13] <eivanov> burek: just 1, I want a 6s movie from it.
[23:13] <burek> take a look at "tutorials"
[23:15] <eivanov> burek: thank you very much!
[23:15] <burek> :ber: :)
[23:15] <burek> :beer:
[23:16] <microchip_> :whiskey:
[23:30] <eivanov> Damn, in trouble again. I try to merge 2 avi files, like explained here: http://ffmpeg.org/faq.html#How-can-I-join-video-files_003f
[23:30] <eivanov> Sound in resulting file is shifted (from second video).
[23:30] <eivanov> There is a delay about 1-2s, sound starts earlier. How can I merge 2 avi files properly?
[23:31] <burek> well, you should use a decent video editor to join media files anyway :)
[23:32] <eivanov> I failed to install kdenlive and avidemux on linux, nothing official for OpenSUSE 11.4 and building from source fails.
[23:33] <eivanov> The only thing I need is to add a title to the video. I already did a video from title image with your help :-)
[23:34] <burek> title?
[23:34] <burek> like rename a file?
[23:35] <eivanov> A picture with some text.
[23:36] <burek> use gimp/photoshop? :)
[23:36] <burek> I don't understand :)
[23:36] <eivanov> The video should start from that picture. I should be embedded/encoded into the video :)
[23:36] <eivanov> s/I/it
[23:38] <burek> well, that really isn't a job for an encoder like ffmpeg :)
[23:38] <burek> try to uncompress your video to raw video
[23:38] <burek> and also create uncompressed video output from an image
[23:39] <burek> and then try concat: protocol on those 2 uncompressed videos, and after that encode it to the output video
[23:39] <burek> ctrl+F concat
[23:40] <eivanov> burek: thanks again :)
[23:40] <burek> :)
[23:40] <burek> np
[23:40] <microchip_> :whiskey:
[00:00] --- Fri Jul 13 2012