Dexterity, We Have a Problem.

I like the idea of giving evasion a chance to make you take half damage. IMO that would easily solve evasion's primary problem, which is that the damage you take is too erratic.

Evasion isn't horribly broken, but to make it work we need another kind of defense too. At least rangers lend themselves easily to being.... ranged. Should almost always be kiting, never taking direct fire from anything other than spell/ranged. Evasion on its own does seem rather weak.

It seems to work fine for shadow, though, since we get evasion block AND es all together. I can't speak much to a duelist. But having a large ES allows you to dodge out and fill your shield anytime you take a large hit. Most situations you can evade enough to just keep on fighting while your shield refills.

My experience, anyway. I run a high block/high evasion/high ES, crit build. And I did also die most recently by using flicker strike poorly. It's a trap :)

Think prob been mentioned but I think Dex should lead to crit chance, and str crit multiplier.. and intelligence... not sure?

"

Zakaluka wrote:

Evasion isn't horribly broken, but to make it work we need another kind of defense too. At least rangers lend themselves easily to being.... ranged. Should almost always be kiting, never taking direct fire from anything other than spell/ranged. Evasion on its own does seem rather weak.

When I say evasion is a broken mechanic I'm judging it as a primary armor type that is supposed to be equal to both armor and energy shield.

Armor tends (for my characters) to lower incoming damage by about 30% without passives enhancing it. So it's roughly equivalent to evading 30% of the time. But it's deemed to be a very powerful tool because each hit is weaker and you have a strategic chance to drink from a flask, dodge out of battle, run away, or whatever . . .

Energy shield is also strategic. It works like an extra layer of bonus health that regenerates quickly when your not getting hit. If your ranged it soaks the blows that do land, and as you kite away it regenerates and you just keep dishing out the damage. The CI passive node shows how powerful this can be as high levels of energy shield can actually replace health entirely.

Then we get to evasion. High levels of evasion as your ONLY defense currently means that your either immortal or your dead. This isn't strategic at all, everything depends on the roll of the dice.

You will get hit 5% of the time, and when you get hit you will get hit hard. Furthermore nothing is stopping that 5% of the time from happening all at once. You can spend hours not getting hit at all, and then get hit 50 times in one minute. It's unlikely but it's possible.

That's why I say that the basic mechanic is broken. It needs fundamental changes to the way it works in order to be comparable to either armor or energy shield.

Strategy is good, constant prayers to the RNG is not fun. :)

"

JohnChance wrote:

"

Zakaluka wrote:

Evasion isn't horribly broken, but to make it work we need another kind of defense too. At least rangers lend themselves easily to being.... ranged. Should almost always be kiting, never taking direct fire from anything other than spell/ranged. Evasion on its own does seem rather weak.

When I say evasion is a broken mechanic I'm judging it as a primary armor type that is supposed to be equal to both armor and energy shield.

Armor tends (for my characters) to lower incoming damage by about 30% without passives enhancing it. So it's roughly equivalent to evading 30% of the time. But it's deemed to be a very powerful tool because each hit is weaker and you have a strategic chance to drink from a flask, dodge out of battle, run away, or whatever . . .

Energy shield is also strategic. It works like an extra layer of bonus health that regenerates quickly when your not getting hit. If your ranged it soaks the blows that do land, and as you kite away it regenerates and you just keep dishing out the damage. The CI passive node shows how powerful this can be as high levels of energy shield can actually replace health entirely.

Then we get to evasion. High levels of evasion as your ONLY defense currently means that your either immortal or your dead. This isn't strategic at all, everything depends on the roll of the dice.

You will get hit 5% of the time, and when you get hit you will get hit hard. Furthermore nothing is stopping that 5% of the time from happening all at once. You can spend hours not getting hit at all, and then get hit 50 times in one minute. It's unlikely but it's possible.

That's why I say that the basic mechanic is broken. It needs fundamental changes to the way it works in order to be comparable to either armor or energy shield.

Strategy is good, constant prayers to the RNG is not fun. :)

Agreed, this is the fundamental problem. Evasion seems fine at lower difficulties, when damage comes in smaller bursts from a bunch of mobs and you have time to heal or back off before dying. Over the long haul, dodging 30% of attacks is the same as mitigating each one by 30%.

The problem is that late game, you are liable to die in 1 or 2 unmitigated hits. Mitigation means something that was gonna 1 shot you might need 2 hits now. Dodge means it either misses or you're dead. In MMO terms, dodge leaves you very vulnerable to burst DPS.

There's also the problem of synergy. Each point of armor makes your entire hp pool effectively 'better' (because it takes more incoming damage to remove 1 point of hp), and vice versa; it's multiplicative. ES and HP are additive: your ES doesn't make your HP any more effective, but it acts as an additional pool of HP so it still helps you survive bigger bursts of damage without getting 1 shot. Evasion does absolutely nothing to increase the maximum incoming damage you can take without dying; it just gives each enemy attack less chance of dealing damage in the first place. This is really the heart of the problem - not only do dex builds have less access to good hp nodes, they can't even use them as efficiently.

Maybe evasion could cap the maximum damage you can take from a single hit as a percent of your maximum life? That's an arbitrary idea and I'm not sure exactly how the formula would be scaled, but it would make Evasion a unique defensive stat that doesn't rely too much on a huge hp pool. Really low hp evasion builds would be doable, but very vulnerable to AoE/DoT damage sources that can get around the damage cap by dealing multiple ticks of damage, and would still have to watch out for swarms.

Yeah, i have to agree at exactly what @up said ! Thats the main problem i encountered. Evade or die. Not funny in long terms especially as shadow. Not that bad as ranger.

A cap at a percentage of life would be the same as not being able to be killed at all. Eventually you'd have people running around endlessly at 1 health point laughing at any enemy they saw.

We also have to remember that not getting hit via evasion has synergy with both other armor types WAY out of proportion to the way it works alone.

If I get hit less than my energy shield regenerates more often during battle, which means that I can take more hits for the same amount of energy shield.

If I take less hits then there is less damage for armor to mitigate, and my health regeneration or flask use can keep up with each bit of damage I take more easily.

My suggested solution (on a previous page) may or may not be the answer, but any solution the will work has to take both of these factors into account.

Evasion isn't simply broken, it's almost impossible to balance because it's very easy to knock it either into the realm of invincibility or uselessness, and it's easy to amplify the effect it has on other armor types, making them too broken.

A possible option that just came to me is to make it so that evading an attack is reliant entirely upon you doing just that (physically moving your character out of the way). Then instead of having dex increase some arbitrary evasion stat, have it decrease the size of your "hit-box", so that evading (physically) is easier and easier to do, because you are presenting a smaller and smaller target. Additionally, it should reduce the critical hit chance against you.

The current "accuracy" stat would then simply increase the thickness of your attacks (the connect-box). Essentially, they present harder to hit targets and you attack with broader smacks, making the two stats have the same relevance they do now, but without the issues they face in their current implementation.

A combination of a smaller hit box and higher movement speed would make for excellent evasive opportunities (reliant on the player themselves), and the lower crit rate means that when you stop dodging and wade into combat, your smaller hitbox is simulated by a smaller critical hit target being presented to the opponent (since you can't really dodge in melee).

This would also allow for increasing the size of the hit-box as a penalty to some other keystone, for example a 100% increase in hit-box size paired with a 100% movement speed boost (or other interesting ideas). Similarly, you could fiddle with weapon sizes in the design space.

One approach to implementing partial evasion.

For everyone proposing additional mechanics to make evasion "better" - adding glancing blows allows the underlying system to stay exactly as it is, while improving the usefulness of evasion substantially. Look at the illustration below.

Must keep this as short and concise as possible. (seems I failed, 1 whole page XD)

An Approach to Partial Evasion



Steps:
1 - Double the size of the "miss" range, and label this the "partial evasion range". For the 20% evasion character illustrated, 40% chance of partial evasion.
2 - Use an inverse linear plot such that ∫D/I = 0.5 [D: damage multiplier I: interval length] - this means that 50% of incoming damage applies across the whole range.
3 - Preserve a segment where a "full miss" occurs (D=0) so that the feeling of evasion as a system doesn't change too much from gameplay.
4 - Everywhere else on the damage characteristic D = 1.

End result: Restrict the plot to keep ∫D[0..100] = %hit. In the case plotted in blue, ∫D[1..100] = 0.8, or 80% of the damage gets through evasion. As a result, evasion reduces exactly the same amount of damage as before.

Now when applying evasion rolls, can just scale this characteristic in various ways. When evasion <=50%, the actual evasion roll gets telescoped. When evasion >=50%, the evasion roll gets telescoped, along with the output getting multiplied by some offset.

Case 2: high evasion (60%). This character no longer evades all 60% of incoming hits. However, he can never get hit for 100% damage - incoming hits always partially evade by 20% or more. This is just a side effect of this new implementation, no special considerations required. Just an extra ratio is needed to keep area under the curve = %eva.

Caveats:
- the original miss/hit roll is still meaningful. On-hit procs, life on hit, etc still apply to everything that's a "hit", even if partially evaded. Using the un-scaled damage values. Even partially evaded "hits" can crit, and secondary effects must apply. Secondary effects on crit must use the un-scaled damage values.
- partially-evaded misses must be able to crit, but cannot gain a secondary effect from the crit. partially-evaded misses cannot gain benefit of any type of on-hit proc.

Development standpoint: First, it takes some thinking to get past the >=50% case. Can teach the computer to go through this process with a few simple multiplications by using a step-piecewise characteristic. It would be fairly unacceptable to require exponents and/or logs being used in combat.

Still, there is a drawback for optimization. Before, we just rolled against %evasion. Now we have to at the very minimum perform a table lookup and two more multiplications per attack. Can be a difficulty for optimization.

I know it seems a little OCD of me to model that all out, but I wanted to know how cumbersome such a system really would be on the server side.

Answer:
one table lookup or logic branch
edit: three multiplications/one division. Maybe one final operation to eliminate strange cases where high damage can get through the characteristic (because it's piecewise-step, and we're restricting it/scaling it)

Sometimes when everyone asks for a feature, it's simply not feasible to implement. Either because it affects underlying systems too much (and something like this shouldn't), or because the implementation would just take way too much CPU time on the server side (in this case, depends). Or, lastly, if it's just not worth the effort. It is a fairly complex problem after all.