I use the following tools: K3CCDTools, Registax and Photoshop.

1. FIRST I start with loading the AVI into K3CCDTools and after opening the AVI I invoke the Planetary Wizard and follow the instructions that are given per step.
   When the Planetary Wizard asks me to specify the Reference Frame I usually select the one suggested by K3CCDTools.
   Depending on the quality of my raw frames I use 50-70% for further processing, as my experiment The impact of the stack size on the quality of the result image has convinced me that larger stacks usually give a better final result.
   The 50-70% should match the overal quality of your AVI of course: make sure that you do not include unwanted low quality frames!

   When I have a huge AVI - e.g. 15 minutes @ 10 fps of Saturn giving 9000 frames - I usually start with aligning/stacking a huge stack, say 5000 frames.
   Yes, this takes quite some time, but the nice thing is you have to align such a big stack only ONCE!
   SAVE the stacked result e.g. as xxxx_k3stacknnnn.png (png16) where xxxx=AVIname and nnnn=stacksize.
   Next select a smaller stack - by unchecking in the frame list - WITHOUT ALIGNING AGAIN then only do the summing, SAVE this result as xxxx_k3stacknnnn.png etc.
   Note: when I use the 2x upsample feature in K3CCDTools I save the result as xxxx_k3x2stacknnnn.png

2. NEXT I invoke Registax and I check the Colour and the LRGB box.
   Select and open the PNG16 file that was the output from K3CCDTools in the previous step and I automatically jump to the Wavelet Settings page.
   All I need to do now is the Wavelet processing and finalise: see the links to some Registax Tutorials on my Software page.
   Recommended technique: "Registax Trapping Noise Technique"
   

   When done I save as 16-bit TIFF file: xxxx_k3stacknnnn_wav.tif where 'wav' means 'wavelets have been applied'.
   Note: if you have multiple results coming from the same AVI but with different stack sizes you should play these TIFFs in Irfanview-Slideshow and select the best for final processing.

3. The FINAL processing procedure - which was worked out in close cooperation with Rob Kantelberg - is done in Photoshop: I load the TIFF file from the previous step and use Curves Tool (CTL-M) and the Hue/Saturation (Ctl-M) to adjust the colours.
   After conversion to 8-bits (via Image|Mode) I make two copies of the image (Duplicate) and apply a Filter|Gaussian Blur of e.g. 75 pixels to one of the copies, so that now I have my own Unsharp Mask. I apply this with the function Apply Image|Subtract.
   [Click here to read more about this method of Unsharp Masking.]
   I repeat this last step several times with other Gaussian Blur settings until I am happy with the result. After that I might use the Curves, Hue/Saturation, Brighness/Contrast, Levels functions for a final adjustment.

   Here is an example of the Unsharp Mask procedure:

   

   And here is what an USM of 90 pixels does to a Mars image (but when properly applying the wavelets in Registax the same result can be reached !):

   

   Finally I put a nice frame around my image: DONE!