Thanks everyone. I looked at the video and seems like child's play
However, there is some ambiguity in the video. I think the steps are:
1. Get the stacked RGB images.
2. DBE the images... if I ChannelCombine these looks RED
3. LinearFit to the least noisy one... green in my case.
4. ChannelCombination
5. STF linked
6. HTF
End result dioes look better, especially after SCNR to get out green....
Is this the path?