[Ffmpeg-devel-irc] ffmpeg-devel.log.20130609

[burek]

[03:04] <cone-404> ffmpeg.git 03Carl Eugen Hoyos 07master:c153ea4756ce: Support decoding G.722 in aiff.
[09:35] <burek> do we have a cross-fade filter, to cross-fade two video streams?
[11:13] <cone-244> ffmpeg.git 03Anton Khirnov 07release/1.1:042b8c2f06b4: apetag: use int64_t for filesize
[11:13] <cone-244> ffmpeg.git 03Luca Barbato 07release/1.1:8eb7c2566ca2: tiff: do not overread the source buffer
[11:13] <cone-244> ffmpeg.git 03Michael Niedermayer 07release/1.1:01580c0955f9: Merge remote-tracking branch 'qatar/release/9' into release/1.1
[11:29] <burek> ok, it works :)
[11:34] <ubitux> <+burek> do we have a cross-fade filter, to cross-fade two video streams? // only a concat and a fade filter
[11:36] <burek> ubitux, great :) do we have any example command line by any chance?
[11:36] <ubitux> not that i know of
[11:37] <ubitux> might be tricky to do
[11:39] <burek> i guess i could create a "random idea" request on the tracker :)
[11:40] <ubitux> i remember discussing that issue with saste
[11:40] <ubitux> not a simple problem
[11:41] <ubitux> i think it should come with something larger, like and advanced edl support or something
[11:41] <burek> i believe it's not
[11:41] <ubitux> to make something useful and relevant it does
[11:42] <burek> i noticed people are thinking of ffmpeg more like a swiss-army knife tool for everything you can imagine doing with your media files, rather than just an encoder
[11:42] <ubitux> because it is?
[11:43] <burek> it wouldn't surprize me to see in the near future a scripting support for ffmpeg, like in virtualdub, where people would practically play with their media files using their own algorithms and stuff
[11:43] <ubitux> saste is working on this
[11:43] <ubitux> but it will take some time
[11:43] <burek> yeah, he told me a few days ago
[11:43] <ubitux> because it's definitely not trivial
[11:43] <burek> i could help with lua i guess
[11:44] <ubitux> it's mostly an integration issue i'd say
[11:44] <burek> it would really be cool to make something like that
[11:45] <ubitux> there is also the problem of the first pass/indexing
[11:45] <burek> if it proves to be viable to add support for some common scripting languages, like lua, php, javascript, etc. I can only imagine the burst of tools and ideas for ffmpeg usage
[11:45] <ubitux> which is likely necessary for such feature
[11:46] <burek> yeah, i see.. something like sound forge was doing creating those .sfk files for stats and stuff
[11:46] <ubitux> IMO the scripting is not the most important limitation of ffmpeg/libavfilter
[11:46] <ubitux> it's just that avs users are comfortable with this and can't think of something else
[11:47] <ubitux> the real design problem would be that we are in a single-pass stream model for filters
[11:47] <burek> yes, i noticed that as a big limitation several times so far
[11:47] <ubitux> which limits indexing, and thus seeking, and thus a bunch of useful features
[11:48] <ubitux> that can be workarounded, but it's often a pain to do that with the cli
[11:48] <ubitux> and not optimal
[11:49] <burek> im not sure how big of a work that is, but i can imagine it will be another painful bump in the api/abi or so
[11:49] <burek> btw, tbh, im surprised how far ffmpeg has reached, being an open-source/free product
[11:50] <burek> i believe the huge interest in using multimedia on the web has played an important role in all that
[12:05] <cone-244> ffmpeg.git 03Michael Niedermayer 07master:6d3bcbb2114d: avcodec/g2meet: check framebuf has been allocated before use
[12:05] <cone-244> ffmpeg.git 03Michael Niedermayer 07master:ada497e61660: g2meet: Check cursor parameters before writing them in the context
[12:26] <ubitux> Daemon404: are you going to update your fate instances?
[12:29] <burek> ubitux, should this work on windows too:  -vf 'select=not(mod(n\,1000)),scale=320:240,tile=2x3'
[12:29] <burek> because it throws an error to me.. do you see anything wrong in it?
[12:29] <burek> [AVFilterGraph @ 029992e0] No such filter: 'select=not(mod(n,1000)),scale=320:240,tile=2x3'
[12:29] <burek> Error opening filters!
[12:30] <ubitux> ;)
[12:30] <burek> :D
[12:30] <burek> ok, #ffmpeg then :D
[12:57] <burek> err.. an example for "Compose output by putting two input videos side to side" at this page http://www.ffmpeg.org/ffmpeg-filters.html#Examples-16 produces an error "Output pad "default" with type video of the filter instance "Parsed_overlay_6" of overlay not connected to any destination"
[13:02] <burek> ok, removing "[left+right]" makes it work
[13:02] <burek> docs might be updated i guess
[14:59] <ubitux> burek: you should avoid transparent backgrounds for your images
[15:00] <ubitux> burek: because for ppl who override css it doesn't look good (http://ubitux.fr/pub/pics/_transp-css.png)
[15:01] <burek> well, it looks as intended
[15:01] <burek> when i created such images, those pixels were non-existent
[15:01] <burek> i can fill it with white or black, but what's the point
[15:02] <ubitux> look at the screenshot
[15:02] <ubitux> typically the text part
[15:03] <ubitux> or just make sure the text has a background color
[15:03] <ubitux> but it can cause problems for arrows too (or any thin form)
[15:03] <ubitux> that's not really important though :)
[15:03] <burek> does this one look better http://s24.postimg.org/4axi3yzj9/mosaic2a.png
[15:03] <ubitux> no
[15:05] <burek> can you show the screenshot
[15:05] <ubitux> 15:00:19 <@ubitux> burek: because for ppl who override css it doesn't look good (http://ubitux.fr/pub/pics/_transp-css.png)
[15:05] <burek> with 2a.png
[15:05] <ubitux> it's the same
[15:05] <burek> shouldn't be
[15:06] <ubitux> http://ubitux.fr/pub/pics/_trust-me-burek.png
[15:09] <ubitux> burek: btw, on a more important topic, what happened to fate?
[15:10] <burek> what do you mean what happened
[15:11] <ubitux> i'm curious if we're going to have a fate v2
[15:11] <ubitux> or if you let it rot
[15:12] <burek> well i didn't let it rot
[15:12] <ubitux> "it rots by itself!"
[15:12] <burek> :)
[15:13] <burek> we just need to find a way to test it in paralel with existing fate
[15:15] <ubitux> i don't know if it's possible to update the fate_recv of the clients to send to 2 servers
[15:16] <ubitux> but i guess you're changing the report format?
[15:25] <burek> no no, that's not what i mean
[15:25] <burek> there are 2 parts in fate "system"
[15:25] <burek> one that is receiving reports, and the other that's just a web script triggered when someone accesses that web page
[15:26] <burek> so, we don't need to change anything
[15:26] <burek> we can just setup another web page/script, which i wrote for v2, and it'll be located in a different url comparing to the original one
[15:26] <burek> so both can exist at the same time
[15:26] <burek> using the same report data
[15:27] <burek> btw, my web script doesn't write anything (afai remember) to the disk
[15:27] <burek> so it should be safe to run without consequences (i hope) ^^
[16:09] -:#ffmpeg-devel- [freenode-info] channel flooding and no channel staff around to help? Please check with freenode support: http://freenode.net/faq.shtml#gettinghelp
[17:16] <cone-432> ffmpeg.git 03Michael Niedermayer 07master:2d8f880a9b5d: g2meet: fix typo in height comparission
[17:36] <cone-432> ffmpeg.git 03Michael Niedermayer 07master:6241e8a3821d: id2v2: check the return value of decode_str()
[17:53] <wm4> is there any reason why these files "pollute" the global namespace with functions not prefixed with "av"? http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavutil/atomic_gcc.h
[17:54] <wm4> nevermind, the header is "private" in the first place
[22:24] <cone-266> ffmpeg.git 03Carl Eugen Hoyos 07master:052838f765fb: Insert a newline before showing embedded svq1 messages.
[00:00] --- Mon Jun 10 2013