Fixing Evasion without touching Evasion? Revisiting Acrobatics/Phase Acrobatics

Okay, cyborg, dumped the summary into 2nd post. Charan can fix up the first post if he wants to, but I won't bug him about it.

To be totally honest, it's that juicy life node that would get me excited. But if the A/PA revised cluster does 'pretty much the same thing' as before, with some slight reduction at low evasion, slight improvement at high, I'll not complain at all.

Interesting, though, because what start out as a grand proposal has, in fact, been reduced to adding an existing node from elsewhere (Discipline and Training) and tweaking the existing A/PA cluster with one word: additive rather than multiplicative.

So even if all GGG did was add that D+T notable, I'd be oddly content.


Which is not at all what I came here to argue in the first place.

...I feel a bit like a dude who charges into a room with both barrels blazing and then realises he's only firing blanks.

This is okay, I don't really like guns anyway.

On the scaling of evasion rating, with this idea.

A 10% increased rating node becomes worth a little more than it used to be. Each evade node is still worth less than an 8% life node, though. In terms of EH. Still, avoidance affects healing more than the size of your life globe, and that lends some more weight to rating passives.

Suppose that you have 4,000 base evasion rating. That's 1200 from grace, 2800 from gear. Most people should reach that amount after spending a while in endgame. Let's say also that our character has 140% incr. life from passives.

Minimum rating value always seems to happen at about 57%, so adding another 8% incr life node adds 8% EH. Just a coincidence. (248/240-1)/(1-.57) = 8%.

Against level 60 mobs, minimum rating value is 56 rating per 1% incr EH. For that same character, a 10% rating node is worth 400 rating. 10% rating ~~ 7% EH.
Against level 65, min rating value 77 / 1%. 10% rating ~~ 5% EH.
Against level 70, min rating value 98 / 1%. 10% rating ~~ 4% EH.
It's one very specific narrow view. With amazing gear, it should be possible to improve the relative value of a rating node.

I had actually hoped to make rating nodes mathematically superior to life nodes, by at least a little bit. From here, though, adding more dodge isn't the right way to go about that. Base gear values would have to go up again, i think. Of course, there are other ways to do it.

I wonder if focusing it just on the A/PA cluster doesn't leave someone behind - the Duelist. While it isn't a pure evasion class usually, It uses evasion as well, or should be able to use evasion without IR as well.

Taking A/PA with a Duelist means running pretty far off and more or less making your build almost entirely dex based and not hybrid. I suppose the idea is, then, that the duelist takes heavily into the STR/Marauder part of his tree for STR+Life nodes to compensate for the times where his evasion fails.

But i think that takes the Duelist back to square one - where evasion isn't really all it should be for him as an armor/evasion hybrid character.

Thoughts?

I have a few ideas about the duelist, but it's kind of on a completely different track. Duelists won't be reaching for 20k evasion rating, and that's what this whole thread is about.

When rating values are lower, evasion has never exhibited poor returns. I have a bit much on my plate at the moment - trying a frenzy/viper A/PA melee shadow in hc mode, finally - still planning an IB build - can only do so many things :). But again I suspect there are two extremely efficient ways to gear a duelist, and neither fit in this topic.

There was more here, then I decided to just say this: that new leather and steel circle should go a very long way. It needs some time to settle in. The only issue remaining is both simple and complicated at the same time: the way grace interacts with IR, compared to the (smaller) benefit it provides over hybrid gear without IR.

It's a discussion for a different thread, sadly.

I forgot to account for something up above. 10% incr rating is still usually worth less than 8% incr life. Have another look above.

Adding more dodge isn't really the right answer to that. I'd still be satisfied with this change.

Thought so, but wasn't sure. I won't budge in then. I might open a new thread about it, But i'm not sure if the interest is really out there, So for now, I'll just try on my own to achieve the same with a duelist.

Thanks for the input, though.

Thought up something that could save dedicated evaders.

How about a keystone that grants a percentage of physical damage reduction for each attack that you avoid up to a certain point.
Suppose it lasts for five seconds and gives 5% damage protection the first time you are missed, and each miss thereafter credits you 3% as well as resetting the timer. Cap it at the 50-70% territory for truly evasive monsters perhaps?

'Roll with the punches'

Maybe something glued into intdex hybrid territories that grant energy shield per evaded attack? maybe 5% or a static number per whiff?

'Kinetic Siphon'

Problem with that is there's no easy way to count how many attacks have been avoided, because when you avoid an attack, all it means is the damage packet isn't delivered.

We've tried to brainstorm quite a few ways of making the actual 'evasion' moment stronger (thereby not inadvertently buffing IR users), but it's tricky.

The moment you reduce damage it's no longer evasion. That's primary problem #1.

Primary problem #2 is that for any evasion % that is below a threshold, there is some value of HP that is required to survive the maximum streak size, and probably double or triple that.

It depends how they do their non-streakiness calculation and if they calculate it in batches, but overall you cannot fix evasion without addressing hp recovery.

At best you can do what they did here, which is make sure that each evasion point matters, and that there is enough hp to allow HUMAN ERROR to cause a death, rather than a damage streak.

Particularly in the realm of batch processing of streakiness, I wonder how they do this in conjunction with client side prediction... since client side prediction is based on two aspects typically: Showing other's states in the past, and rendering packets in batches... how they process streakiness when presented as a batch of attacks that either hit or miss, is a major concern, and heavily relies on at what point the player is informed of their current status.

Sending 3 hits 1 miss then 3 hits 1 miss, allows a player to interact between. Sending 3 hits 1 miss 3 hits as a batch doesn't, and can spell major trouble for an evasion player EVEN IF the system prevented a 3+ hit streak.