Comments for Breaking Eggs And Making Omelettes

Comment on Libav/FFmpeg and Google Summer of Code 2012 by Fruit

Fruit — Sun, 06 May 2012 18:52:47 +0000

Opus is also sort of a rewrite, because there is a liberally-licensed refernce code by Xiph…

I hope the HEVC project goes well and provides something useful (hopefully not just unfinished code that will bitrot in a separate repository for years before being scrapped due to some rewrite of something).

Comment on Ode to the Gravis Ultrasound by ABC

ABC — Sun, 06 May 2012 12:01:27 +0000

Im trying to find a song from the experience CD… it something similar to Engima songs and has a Shakuhachi (Japanese flute) playing cropping up once in a while…. just cant find it on that ISO… (I know its in the audio section.. but thats not included)….

Comment on YUV4MPEG2 Origin by compn

compn — Sat, 05 May 2012 11:31:32 +0000

when it violates standard but adds more colorspace or speed or compression, just do what ffmpeg did with huffyuv and make a new codec/format called ffyuv4mpeg :P

Comment on Libav/FFmpeg and Google Summer of Code 2012 by compn

compn — Wed, 02 May 2012 03:00:08 +0000

i guess ffmpeg has rewritten a lot of libs e.g. libopencore liba52 libdca so why am i complaining?

Comment on Libav/FFmpeg and Google Summer of Code 2012 by compn

compn — Wed, 02 May 2012 02:47:07 +0000

3 rewrites (librtmp / ffserver / libutvideo ) and 2 codecs no one uses.

Comment on How To Write An Oscilloscope by Dithermaster

Dithermaster — Sun, 29 Apr 2012 13:07:11 +0000

You can show just 512 of the 1470 samples, or you could graph all 1470 and scale it to fit in 512. I’ve seen it done both ways, it just changes the look a little. Most jukeboxes only give partial data to the viz plug-ins so they can’t be used to store a copy of the song.

The A/V sync could be off because the audio and video engines are running one ahead of the other (due to different processing delays). If you don’t have presentation timestamps, you could play a sync test (one frame of “beep” audio with a flash frame of video, then some frames of silent black, in a loop), and use a fast video camera (60 fields per second at least) to shoot the result and calculate your offset, then build that into your display algorithm.

Comment on How To Write An Oscilloscope by Reimar

Reimar — Sun, 29 Apr 2012 07:09:42 +0000

I am not sure what your purpose is, but nobody lets an oscilloscope just display a run-on data stream. Instead you’d have a trigger to decide when to start “redrawing”.
In the case of audio you can even go further: use auto-correlation like in those speed adjustment filters (e.g. MPlayer’s scaletempo) to replace the data from the current 1/30 s with the most similar one from around the next 1/30th.
What concerns A-V sync I would expect the delay from the audio driver/hardware to more than compensate for starting to draw 1/30th second too late.

Comment on Libav/FFmpeg and Google Summer of Code 2012 by Multimedia Mike

Multimedia Mike — Fri, 27 Apr 2012 15:32:43 +0000

@Jonathan: There’s only one way to find out for certain.

Comment on Libav/FFmpeg and Google Summer of Code 2012 by Jonathan Wilson

Jonathan Wilson — Fri, 27 Apr 2012 12:49:33 +0000

Wont supporting RTMPS/RTMPE/RTMPTE violate the DMCA?

Comment on Libav/FFmpeg and Google Summer of Code 2012 by Kostya

Kostya — Thu, 26 Apr 2012 07:22:33 +0000

Ambitious project was VP7 decoder (just imagine — an opensource On2 VP7 decoder funded by Google), too bad there was no good student to pick it up.