Conclusion
It felt easier to massage the parameters using the sinc pyramid.
It was easier to eliminate the high frequency image from the hybrid image viewed at a distance.
I also felt that the low frequency image was sharper when the hybrid image was viewed at a distance.
The variance of the Gaussian was a parameter that required a bit of tweaking for each image. This 'free' variable was eliminated in the sinc implementation.
Did this really help? Who knows, but it seemed so to me.
In the end, the issue of pyramid engineering seems to be in o(alignment issues).