Digital Light & Color

This is probably a naive question: on the page below there is an example of blending an apple and an orange in two different frequencies.

http://www.cambridgeincolour.com/tutori ... ftware.htm

The Composite transformation is already capable of splitting an input image and an overlay into low and high pass components. Would it be possible to use those two components to stitch two images, using mask, so that high pass blends with high pass and low pass with low pass separately?

I imagine that the blending radii of low and high passes should probably differ too - and it is the mask transition that determines the blending distance during normal blend.

And I guess the question is, even if it's technically possible, would it actually work the way I would like it to work (which is realistically stitching the pictures)?

Cheers!

Interesting idea...

The example in the link actually uses different masks to blend the high and low pass images which would not be possible with composite using its current interface. In addition, the technique described would probably have produced halos if the example images had not been carefully selected to have the similar backgrounds. Without trying it out, my intuition is that the technique is more useful for panorama stitching than for general compositing. I can imagine a new transformation that spits an image into high and low pass components and another one that recombines a high and low pass into a single image. Then one could split the image, perform different operations on the two parts, and recombine the two versions but I'm not sure it would be that useful.