Spaced Ranger said:
What is the advantage, in this case, that changed your approach?
Median1 and TooT both guarantee a rot-free result (under the assumption that if a pixel at [x, y] is rot in source #1, then it is not rot at [x, y] in sources #2 and #3).
g-force made a convincing argument that a TooT is likely to be noisier than a Median1 and I'm pretty sure I can provide a mathematical proof if that level of rigor is needed.
Spaced Ranger said:
Also, would these individual sources also be available in the project (for those wishing to try alternate methods)?
Yes. The first thing I plan to do is post the raws to Usenet.