Is equivalent conflict detection simplification in LL(1) ?

[Borneq]

LL parser table.
For LL(1), the following rules apply for deciding whether there 
is a conflict:
for each pair of different productions of the same non-terminal 
symbol (let's call it A):
1. strings beginning with the same terminal symbol cannot be 
output.
2. two productions cannot simultaneously allow the derivation of 
an empty (epsilon) string
3. if an empty string can be output from one production, no 
string starting with any terminal symbol belonging to FOLLOW(A) 
can be output from the other production
This is caused by the fact that the terminal symbol is used to 
select the production; while the first point is clear, points 2 
and 3 result from the fact that when outputting the entire 
string, when the string generated by the productions ends, there 
is a return to the overriding (in the dynamic sense) production 
that generates further symbols, you can imagine it as a return to 
the procedure calling the current procedure. Point 2 follows from 
the fact that otherwise, for both selection paths, we would have 
tokens with FOLLOW(A) (in addition to those specific only to 
those productions)
It follows from conditions 1 and 2 that the FIRST sets should be 
disjoint for both productions. The third condition is equivalent 
to the requirement that if an epsilon belongs to the FIRST of one 
production then the FIRST sets of the other and FOLLOW(A) are 
disjoint.
I have a question: if there is an epsilon, can tokens from 
FOLLOW(A) be added to the set of productions and then only 
compare the sets with each other, is this equivalent or so 
enriched