[Ffmpeg-devel] [RFC] Feature request

[Michael Niedermayer]

Hi

On Sat, Aug 19, 2006 at 12:47:23AM +0200, Benjamin Larsson wrote:
> Hi, the ff_float_to_int16 function doesn't take a step argument would it
> be ok to add a step argument to the function? The cook and wma decoder
> interleaves the output directly from float so it's not possible to use
> ff_float_to_int16 without reordering the floatdata first.

codecs should output data in a representation closest to the internal
one, thats probably planar audio (=non interleaved) and possibly floats
unless a fixedpoint decoder is used ... the API should be changed to use
AVFrame for audio too, so that .data[] and .linesize[] would be used to 
represent how audio is stored

format conversation doesnt belong into the codecs, so someone should
1.  write a generic audio format converter for lavc (fairly trivial)
1b. use the various optimized pieces of code in it
2.  switch API to AVFrame or something similar (search the ML archives for
    the relevant disscussions)
3.  ensure that the format conversation is used if the format from a decoder
    cant be used by the encoder (AVCodec should contain a list of support ones
    for each encoder)
4.  make codecs output stuff in a format closest to their internal one




[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

In the past you could go to a library and read, borrow or copy any book
Today you'd get arrested for mere telling someone where the library is