(Inverse) Fourier Transform

[ajp_anton]

I don't know if this is in any way practical, just for fun/learning...

Is it possible to take the fourier transform and the inverse separately, so you can view the transformed image and play around with it before going back to the spatial domain?

[StainlessS]

Arrh, my maths is real bad, I'm about (maybe to demonstrate that), but I think that the Fourier Transform,
transforms 2D into 1D and so, having a look at it, it would be a 'line'.
If someone less uninteliigent that I can correct that, I would still be just as stupid, but perhaps better uninformed

[gyth]

This (http://homepages.inf.ed.ac.uk/rbf/HIPR2/fourier.htm) might be what you are looking for?

[Lyris]

I'm sure there was a Photoshop plugin and macro that let you do this - transform and back again. People were using it to remove moire artefacts from scanned images. I never had too much success with it for that purpose.

[jmac698]

You can do it with imagej and a few other softwares, search for image processing software.
http://fiji.sc/wiki/index.php/Fiji

[ajp_anton]

It's for a school project. I'm supposed to do it in Matlab, but I'm too lazy to learn it, and I'm already very familiar with Avisynth =). The things I have to do seem easy enough that some masktools stuff is enough. Neither speed nor precision are really a concern, it's mostly to just see "what happens if we do this".

Oh well, if it's not possible, I'll just have to try Matlab. Not that I think it will be difficult, but I really like Avisynth.

[mastrboy]

this might be of help: http://arxiv.org/abs/1201.2501v1

[librarian]

FFTQuiver (http://avisynth.org/vcmohan/FFTQuiver/FFTQuiver.htm) might be the tool in Avisynth.

[sven_x]

An image tool for FFT and inverse is Image Analyzer http://meesoft.logicnet.dk/

[2Bdecided]

It's for a school project. I'm supposed to do it in Matlab, but I'm too lazy to learn itThat is lazy, given that it has 1D and 2D fft built-in, as well as a function to plot any 2D array (i.e. the output of a 2D fft). Not to mention abs and real (which will come in handy!).

whatever you do, don't use Google. There's no way on earth anyone will ever have done this with MATLAB before ;)

Cheers,
David.

[ajp_anton]

I ended up doing it in Matlab anyway. Yeah I know it's much better and easier than with Avisynth, it was just for fun, to see if it could be done, and I'm already familiar with Avisynth, unlike Matlab.

FFTQuiver, from the homepage, looks like something worth looking at, but I was kind of in a hurry so it was quicker to learn Matlab. I might look into it later if I have some spare time =).

[*.mp4 guy]

Doing an arbitrary fft in avisynth is certainly possible, though not worth the effort without 16 bit colorspace support. I implemented both a dct and an idct in avisynth (using the masktools and average plugins), which are related operations. However, even with only 8x8 blocks, numerical precision becomes problematic. It is possible to work around the colorspace bitdepth limitations, but I have not encountered a compelling reason to revisit the project. Based on what I think would be needed to get useful results, you made the right choice using matlab, as doing the cludging needed in avisynth would have multiplied the workload beyond reasonable limits for a school project.

[redfordxx]

Is this what you want?
My MxN DCTFilter (http://forum.doom9.org/showthread.php?p=1537175#post1537175)

Use parameter mode="showF"