The minimum FFT size is a function of the GPU memory block size (ughh).
The minimum FFT size is a function of the GPU memory block size (ughh). Need a better way to capture that. XXX
Take FFT of a complex, centered filter
producing a Fourier transform
of shape shape
.
Take FFT of a complex, centered filter
producing a Fourier transform
of shape shape
.
The shape of the output Fourier transform. Each dimension must be a power of 2.
The filter to be transformed. This must be of odd size and centered.
The Fourier transform of the filter.
Take FFT of a real, centered filter
producing a Fourier transform
of shape shape
.
Take FFT of a real, centered filter
producing a Fourier transform
of shape shape
.
The shape of the output Fourier transform. Each dimension must be a power of 2.
The filter to be transformed. This must be of odd size and centered.
The Fourier transform of the filter.
Take inverse FFT of a complex frequency domain filter
producing the
spatial version of the filter.
Take inverse FFT of a complex frequency domain filter
producing the
spatial version of the filter. This undoes shifting and expansion that
was done in the apply methods.
The shape of the output spatial filter. Each dimension must be less than or equal to the corresponding dimension in filter
The spatial domain version of the frequency filter.
Takes the FFT of a convolution filter. This does not take into account boundary effects--that must be done before calling by using, for example, PadFFT.