GPU-based image processing under Python

This pages presents an experimental image processing library written on top of the PyFX effect framework. It aims to facilitate ease of programming of GPU-based computational frameworks.

The demo can be downloaded from here and it requires version 0.2.0 of the PyFX framework. It comes with a simple demo application called demo.py in which various filters can be tried out in real time. For a small list of different filters which can be applied please see the enclosed script.txt file.

Screenshots

This is the initial image you will see when starting the demo application. You should also get an interactive prompt > on which you can enter your commands.
Applying and edge-detection filter to this image is as simple as
> show(edges(lena)) 
You can of course assign intermediate results to variables
> tmp = edges(lena)
> show(tmp)
Showing a different image is simply
> show(spectrum)
Now we can take the spectrum image above and do per-pixel and per-channel multiplication with the blurred edge detected version of lena by simply
> show(spectrum*blur(edges(lena)))

Last modified: Thu Sep 23 11:29:20 CEST 2004