Ok, time to eat a big hunk o' crow.
I'll be the first to admit that the celltime recalculations always made my brains hurt, but working with them today I realized that my math was completely wrong.
For Case 1 video (the most common) you multiply by a factor of 1.25. I had wrongly assumed that this was true for most of the other cases. After playing around with some files today it turns out Case 1 is the exception. Cases 2-6 need a factor of 1.2 (I think).
I've never noticed my error before now because by and large cases 2-6 occur in trailers and stuff where there are few if any chapters and the content is short enough that it isn't a large enough discrepancy.
Now I may still be completely wrong and if anyone has a firmer grasp on this please chime in. For now I recommend you check that the chapters are in the right locations after conversion.
Edit: Or not. I'm really up against it here and I'm hoping the Doom9 guys can thrash it out.
The problem is I'm seeing a several second discrepancy now in the chapter locations and runtime (even though the audio syncs).
Something very odd is going on that I mean to find out about.
THIS ONLY applies to cases where you are using 25 to 29.97 pulldown.
Edit 2: Ok, or NOT. Here's the deal, case 1 which you are essentially using 23.976 to 29.97 pulldown is a multiplier of 1.25. ALL OTHER cases require a multiplier of 1.1988.
The guide and spreadsheets have been updated accordingly.