1.10s Strafe question

[Nicator]

I read your article and Zath's, at AB, before starting this thread:

Ruvanal, AB, Jul 22, 2002

Strafe.
I see from some digging at old posts and the guides that there is a misconception on the missiles that are generated with Strafe. The claims are that the extra missiles that are generated by the presence of minions, hirelings, decoys and such are phantom missiles that do not hit. All the missiles from the Strafe volley are real and can hit, but there is another more troubling bug with the skill that has lead to this false belief. I did not see any of the reports on how this testing was done to see where the errors actually came from, since the original posts have been deleted.

Test set up.
Changed zombies to have 150 life, 100% physical resistance (physically immune), 0% resistance for the other attributes, set their damage regeneration to zero. Equipped the test amazon with a charm the delivered 50-fire damage (the only non-physical damage on her). She was already equipped with a charm that gave 100% chance of deadly strike (for the spark effect).

Observation from attacking zombies using Decoy and Valkyrie to get 2 extra shots in. It is possible to kill a single zombie in a three-arrow volley of strafe; all arrows can do damage. But there does seem to be something broken with Strafe none the less. The amazon was showing as having a 95% hit rate (zombie mlvl 1, amazon clvl 32 with an attack rating of 1764 using strafe), but about 1/2 of the strafe arrows were consistently missing stationary targets (engaged with the Valkyrie or Decoy). This last resulted in only about 10% of the tests on a single target actually hitting for all 3 arrows in the volley (which is close to consistent with 0.5^3=0.125). Also it was not a case of every other arrow missing as I could at times have every arrow hitting or only 1 or 2 when testing in large groups that would have 6+ arrow salvos. So while all the arrows can do damage there appears to be a problem with the arrows hitting with the chance that they should be.

What this translates to is that while all the missile are real and can hit, they only have a chance-to-hit that is about half of what is should be. This chance-to-hit being so low is what is making this skill less than what it was before and is giving the illusion that some of the missiles are phantoms.
Strafe.

and Zath, AB, Oct 10, 2003

I know Ruvanal's original test was performed in v1.09d, but I have checked the MPQs and it seems like nothing has changed in this respect.

While I accept your observations there's one thing in your test you didn't mention: Attack speed. Assuming NextDelay 4 frames window, 2 out of a 3 and 4 missile volley can hit if a 4-frames strafe, and only on those two chances-to-hit should be analyzed.

Besides that, in your very first thread, there was this reply:

TrueMuppet, AB, Aug 16, 2002

Hm. Maybe we need to go back to square one with Ruvanal's Strafe testing - I performed a similar experiment and could not duplicate his results or anything that'd prove the existence of a flat-out 'Strafe has lowered / bugged chance to hit' phenomenon.

Test setup: Old strafe testazon I had around: lvl 75 Zon, 300 DEX, roughly 1700 AR; single 6 fire dmg charm to visually verify hits; no skills except 7 points in strafe, 20 in Valk & Decoy. Equipment: 1.) Ballista of molasses (-20 IAS, to achieve as slow as possible strafe speed); 2.) Ballista of molasses w/ ITD (to test whether the lowered CtH, if it existed, was AR dependent); 3.) Ballista of molasses w/ +100 AR / cLvl; +10% AR / cLvl (achieve ITD by hugely blown up AR, to test whether results would differ from 'plain' ITD)

Opponent: Hell Zombie set to 100% physical resist, otherwise unchanged.

CTH according to the lying char screen: 85% with weapon 1 & 2, 95% with weapon 3.

First round of testing immediately revealed that there is no flat CtH reduction - single target strafe (i.e. one arrow) against normal attack; 100 shots each.

Strafe hits: 81
Normal attack: 88

Well within the parameters of what should happen.

Same picture strafe against 3 targets (no valk & decoy), 100 strafe cycles (single click each time):

Average number of hits per cycle: 2.42; '1 out of 3' were few & far between, didn't get a single '3 misses' cycle.

So far, so good. Strafe is working as advertized. Got another single target & added Valk to the mix:

2 arrow strafe against 1 opponent, 100 shots:

Average number of hits per cycle: 1.46

Two blanks this time, with 73% average CtH straddling the line between 'number of shots too low' and 'something's fishy here.

Added decoy, pretty much same picture - 2.15 average hits now against a single.

Where the weapon and ias was established: a 18/5 strafazon. Therefore, no NextDelay 4-frames window is applied and an explanation for the different results on both tests.

The way that Stafe was set up work in v1.09 is not the same as it is set up to work in v1.10.

I accept it, and accepted the 3/4 damage because that bug. However, I'm not aware of anything against NextDelay on 1.09's strafe, and truely a target's 4-frame immunity fits well, both in 1.09 and 1.10s, explaining strafe missiles' behaviour.

Therefore my conclusion regarding target's memory and a single strafe attack (up to 10 missiles):

5 frames and slower strafe: All missiles will be checked by chances to hit;
4 frames strafe: 1 missile will go unchecked after an effective hit;
3 frames strafe: 1 missile will go unchecked after an effective hit; and,
2 frames strafe: 2 missiles will go unchecked after an effective hit.

Another thing I didn't mention: It seems, from my experience, that the angle made by the strafazon and the target(s) has something to do with the chances to hit and/or pierce. I can't explain it but it looks like I get better results when attacking to 0, 180, 90 and 270 degrees.