Hi All:
https://youtu.be/uuAIfcAqc-w
For those following the 3d thread.. I thought you might like to see how the algorithms are coming along during my summer break. This is a low res image ( 320x240 ) and took 8 seconds to create in the new code base. An 800x600 takes about 30 seconds. The quality is better than the previous code after 24 hours of run. If you look carefully you can see the face pulls out at a sharper angle from background features. IT much better follows the gradient contours of the original data. Still lots more optimising to do yet though.
The new code breaks an image down into frequencies, and then does a 3d pull from each of the frequencies adding them up as it collapes
a pyramid of mip maps of the original data, this makes it faster, and much more sensitive.
I still have more to add and a couple of ideas of how to turn it into more of an interactive experience for the user to guide it into
the result you may want for a given use.
Anyway, back to playing..
Art