Originally Posted by
Brusalk
I'm all for dynamic gameplay, but I feel that there is a high-enough level of dynamism introduced in encounters already.
At the least it is way too random right now. When you have phases of fights that have a built in damage check (Lei-shen) and never get a trinket proc during that amount of time, that's just bad.
Failing an encounter's damage checks just because you didn't get a trinket proc is frankly rather stupid. That's not to say that there aren't times when it's caused by sub-optimal play, or other causes. What I'm specifically referring to is when undergearing an encounter that it's entirely possible that more trinket procs would've gotten you past that damage check.
We completely agree on that point.
Originally Posted by
Brusalk
Personally, I think that trinkets should always proc within the first 5ish seconds of a fight, and then have an ICD in the old sense of the word, however rather than it proccing right away, you have ~15 seconds before and after the ICD occurs that you have a potential for a proc. Anything before 15 seconds of the ICD, and any time after 15 seconds of the ICD, and you can be assured that there won't be a proc.
That way you will pretty much can expect a proc to happen, but it won't happen right away and there's still some randomness.
It would indeed make for an interesting gameplay. As I explained in the earlier posts in this thread you can "engineer" any distribution for the time between two procs by tweaking the chance to proc per unit of time (as a function of the time since last proc).
As an example, let's derive the math for this :
Code:
f(t) = 0 on [0, a] U [b, +infinity], 1/(b - a) on [a, b]
h(t) = ?
where dP = h(t) dt
We have :
h(t) = H'(t)
f(t) = - g'(t)
g(t) = exp(- H(t))
From which :
Integral(0,t)[f(u) du] = 1 - exp(- H(t))
H(t) = - ln(1 - Integral(0,t)[f(u) du])
h(t) = f(t) / (1 - Integral(0,t)[f(u) du])
And in this case :
Integral(0,t)[f(u) du] = 0 on [0, a], (x - a) / (b - a) on [a, b], 1 on [b, +infinity]
And using matlab to plot the graphs for a = 50s, b = 70s (guaranteed proc in a 20s interval centered on one minute) :
Which shows the probability density of the time between procs in blue, and the "engineered" chance to proc per second in green.