Thanks to Sylvain posting a "Ju-Ju" <G> processing tutorial over at QCUIAG I finally started to get some better results!
These are 10 second AVI's (320x240 at 30fps) captured with Vega Pro that were registered and stacked using AVI2BMP. I don't any reject images from the stack as I'm a really lazy bugger and going through 300 images looking for more fuzz / less fuzz just doesn't turn me on. Maybe I should but hey...
The trick I learned from Sylvain was to resize your source image to 2x or 3x the original and to un-sharp mask pretty heavily the larger image. When you re-size back to a smaller image you soon find you can get away with a _lot_! These were taken in 320x240 mode but these results are 2x the size of the original having done a first un-sharp mask at 3x the original. The final touch is a light gaussian blur to give the image a smoother effect.