So I have 4 cases where Cold critical is better, 20 where Impact critical is better
Yes, but that is in "brute", splitting criticals results in single damages, an example:
Cold, 86-90 | down, s2 (f-70)
Impact, 86-90 | +10, down3, np3
This is, in brute, 2 better for impact, but grouping all damage in one as it is really, what is better?, it is not clear, if can kill the target in those 3 rounds, impact seems better, but if not the case (you need 2 rounds to prepare the spell for example), the (f-70) is clearly better.
Another example is the C(91-95), I think an instant death is good even if it has delay. Or the C(86-90), if you are fighting alone against that target with a spell user (you use a spell user for sure if you do cold/impact criticals), is really
snp2 better than
6/r+
s9 (this are 54HPs for free with no risk for us, and continue bleeding next rounds), IMO not.
For D(96-99) I have d8 in my tables for impact (so this is a
red for cold), and for E(96-99) is d3 for impact. But in any case, generally I don't care some more or less rounds to death if target can do nothing, is not relevant really, so many
greens or
reds for this is not very important IMO.
I think looking at the whole damage is very different.
This is my opinion for each result using the entire damage (e=equal, c=cold, i=impact):
81-85: A(e) B(c) C(c) D(e) E(i)
86-90: A(c) B(e, because with armor(i) but without(c)) C(c) D(c, is out in any case) E(i)
91-95: A(i) B(i) C(c, because delayed die) D(i, but is kill with no activity in any case) E(e, because impact could allow a spell user to cast a spell before die, but does more damage for non-spell users)
96-99: A(c, is out) B(i, but both are out in any case) C(i) D(e) E(e, impact gives us greater bonus but could allow a spell caster to cast with a SCSM for no hand use).
100: A(c, both are out but cold is death always) B(c, impact allows to cast spells with SCSM) C(i) D(i) E(c, nr+20)
Counting... Cold = 10, Impact = 9.
The problem is that for these high result, all is important damage, so 1 or 2 better results for each one is not very relevant, IMO is more interesting look at medium results to make an average critical damage. I'll look them and then I'll post other comparsion type A() B() C() D() E().