[FFmpeg-devel] [PATCH] RDFT for Bink and QDM2

[Michael Niedermayer]

On Wed, Jan 21, 2009 at 01:12:30PM -0500, Alex Converse wrote:
> On Wed, Jan 21, 2009 at 8:36 AM, Michael Niedermayer <michaelni at gmx.at> wrote:
> >
> > On Tue, Jan 20, 2009 at 03:19:46PM -0500, Alex Converse wrote:
> > > On Tue, Jan 20, 2009 at 1:50 PM, Michael Niedermayer <michaelni at gmx.at> wrote:
> > > > On Tue, Jan 20, 2009 at 12:24:06PM -0500, Alex Converse wrote:
> > > >> On Tue, Jan 20, 2009 at 11:52 AM, Michael Niedermayer <michaelni at gmx.at> wrote:
> > > >> > On Mon, Jan 19, 2009 at 07:44:38PM -0500, Alex Converse wrote:
> > > > [...]
> > > >> >
> > > >> > [...]
> > > >> >
> > > >> >> +/**
> > > >> >> + * Sets up a real FFT.
> > > >> >> + * @param nbits           Log2 of the length of the input array
> > > >> >
> > > >> >> + * @param inverse         If TRUE, perform the inverse of the transform
> > > >> >
> > > >> > i suggest if 0 perform the forward transform, if 1 perform the inverse
> > > >>
> > > >> I believe these are equivalent.
> > > >
> > > > I searched for TRUE in the iso C standard and found no match.
> > > > But lets assume it where defined as 1, this doesnt document any
> > > > other value, is inverse=2 or 0 invalid? the forward transform?
> > > > 99% of the people will guess correct but why write something ambigous if
> > > > it can be written unambigous ...
> > > >
> > > >
> > > >>
> > > >> >
> > > >> >
> > > >> >> + * @param sign_convention The sign of j of the forward FFT.
> > > >> >
> > > >> > i do not understand this, j is a variable that has no clear relation to a FFT
> > > >> >
> > > >>
> > > >> j is the unit vector in the vertical direction on the complex plane. j
> > > >> = sqrt(-1).
> > > >
> > > > wasnt that i, i for imaginary?
> > > >
> > > >
> > > >>
> > > >> The forward DFT can be defined as:
> > > >> X_k = \sum_{n=0}^{N-1} x_n e^{-{2\pi j \over N} nk }
> > > >>
> > > >> It also can be defined as
> > > >> X_k = \sum_{n=0}^{N-1} x_n e^{{2\pi j \over N} nk }
> > > >
> > > > you can, but why would we need it?
> > > >
> > >
> > > Bink uses it. As per the comments, the inverse of this transform is
> > > not simply the transform with the opposite sign convention.
> >
> > maybe iam too tired but i see just one transform
> >
> > X[  k] = sum(n=0..N-1) x[n] e^(-2pi*i*n*k / N)
> >        | k= N-m
> > X[N-m] = sum(n=0..N-1) x[n] e^(-2pi*i*n*(N-m) / N)
> > X[N-m] = sum(n=0..N-1) x[n] e^(-2pi*i*n*N / N - 2pi*i*n*-m / N)
> > X[N-m] = sum(n=0..N-1) x[n] e^(-2pi*i*n + 2pi*i*n*m / N)
> > X[N-m] = sum(n=0..N-1) x[n] e^( 2pi*i*n*m / N)e^(-2pi*i*n)
> > X[N-m] = sum(n=0..N-1) x[n] e^( 2pi*i*n*m / N)
> 
> The twiddle has to happen in the frequency domain so we have 4 cases:

i have no clue what you talk about, but after a little testing i have
some doubt that your code works at all.
for example

    ff_rdft_init(&one, 7, 1, 0);

    for(i=0; i<N; i++)
        src[i] = dst1[i] = 0;
    dst1[65]= 1;

    ff_rdft_calc(&one, dst1);

gives me a all 0 array, this cant be, and this isnt the only oddity, just
printing the values of the transforms for a single 1 element with all others
0 does not look like sine/cosines in some cases, rather some elements seem
at the wrong spot.

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

Its not that you shouldnt use gotos but rather that you should write
readable code and code with gotos often but not always is less readable
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20090121/0dcecce1/attachment.pgp>