[FFmpeg-devel] [PATCH] DCA floating point output

FB2000 flybird2k
Sun May 16 04:32:51 CEST 2010

On Sun, May 16, 2010 at 4:44 AM, Benjamin Larsson <banan at ludd.ltu.se> wrote:

> On 15/05/10 21:20, M?ns Rullg?rd wrote:
>> NACK.  See email just sent and discussion on IRC.  Will you ever listen?
> Well, the patch is ok, I just wont apply it until we have something that
> doesn't incur a speed penalty on major platforms. And I sent this before I
> brought it up on irc as I know you complained about this before.
Glad to know that. Thanks Benjamin.

I can see that Mans (and probably other developers) favor speed more than
quality. Just to explain it clearer why floating point output is important
to people like me, who use ffmpeg mainly to convert between audio formats,
we do care quality more than speed. It is totally disaster to convert DCA
sample to 16 bit internally, because that way we could never get 20 or 24
bits output that the DCA format is supposed to give.

But please don't take this the wrong way. I am not trying to argue more
about anything here. I can compile float version for my own use anyway,
thanks to great work from you guys. Please apply the patch at anytime you
feel comfortable, just understand there is a need, and that might means a
lot to some people.


More information about the ffmpeg-devel mailing list