Multishot For Shotguns Rework

[Argoms]

"50% of the time a shotgun fires 100% more pellets" and "100% of the time a shotgun fires 50% pellets" average out to the same number of pellets over infinite trials, but they're not identical behaviours and on any single trial can differ by +/- 50%.

 

 

Hm. What I got from reading this part was that there's no randomness. I interpreted 50% of the time it'd fire a single pellet and the other 50% 2 pellets as exact numbers (assuming an even number of shots are fired) and not statistical averages based on this-

 

float fractionalProjectileRollover = 0.0f(f is a typo?);

 

void fire(Weapon weapon, float multishotMultiplier) {

 

float fractionalProjectiles = weapon.projectilesPerShot*multishotMultiplier + fractionalProjectileRollover;

int wholeProjectiles = floor(fractionalProjectiles);

fractionalProjectileRollover = fractionalProjectiles-wholeProjectiles;

}

 

According to that, every other shot would shoot two pellets (exactly a 50/50 split assuming you shot an even number of times).

[Qeveren]

Oh, well without randomness they converge much more quickly. :)

 

If you use a shotgun that fires two pellets, then it's a case of 2-4-2-4-2-4 vs. 3-3-3-3-3-3. 

[Argoms]

Oh, well without randomness they converge much more quickly. :)

 

If you use a shotgun that fires two pellets, then it's a case of 2-4-2-4-2-4 vs. 3-3-3-3-3-3. 

If you have two pellets, and have +50%, you shoot 3-3-3-3. 50% of 2 is exactly 1.

 

One pellet only goes 1-2-1-2 because half of one isn't a whole number. Every shot does 1.5, but the .5 is carried over to the next shot.

[MJ12]

Dude. Its not complicated. nomatter what previous version made whatever apparent or had bugs. etc.

 

Pistol and rifle multishot is a chance multiplier.

 

Shotgun multishot is a static % of extra bullets per shot.

 

 

 

Shotguns:: + percent extra bullets per shot.

 

Pistol/rifle:: Percent chance to get extra bullets per shot.

 

125% chance is always fire 2 bullets with a 25% chance to fire an extra 3rd. Its a percent chance. Chance being a key word.

 

 

All shotgun multishot percentages are not a chance, they are 100% every time. its instead a measure of just how many you get every time.

 

The fact that you ignore the wiki completely shows me that you're just argueing to argue when this is completely true and not a matter of opinion, it is something completely measured and tangible to experiment in game for solid reassurance. So just stop. If you really believe that multishot works the exact same, then please. Please show me from evidence in screenshots in game where you fired one and a half bullets with a rifle or pistol. Because i'd certainly would love to see it. Or show me you shooting two and a quarter bullets. (125%)

 

Here's a simple way for you to figure out that you're wrong.

 

Put a 20% multishot mod on the Bronco. Do you fire 14 pellets 20% of the time and 7 pellets 80% of the time (you don't)? Or do you fire 9 pellets 40% of the time and 8 pellets 60% of the time? (you do). Use a 20% multishot mod on the Hek, which also fires 7 pellets. What happens? Same thing. Shotgun multishot does not work differently from rifle/pistol multishot, it's just that multishot attempts to be as deterministic as possible, so there's only ever a percentage chance of firing 1 extra bullet/pellet.

 

The wiki is wrong about how shotgun multishot works differently. It's not the first time a user-made database was wrong.

[Argoms]

Does anyone know if this applies to the new thunderbolt mod too? Or is that purely random?

 

EDIT: Okay, the additive chance idea (10% = 1 extra bullet every 10 shots exactly) seems to be false, judging by shooting my boltor at a distant wall with 90% multishot- there was not a visible pattern in whether or not a multishot would proc.

Edited May 6, 2013 by Argoms

[TehJumpingJawa]

Does anyone know if this applies to the new thunderbolt mod too? Or is that purely random?

 

EDIT: Okay, the additive chance idea (10% = 1 extra bullet every 10 shots exactly) seems to be false, judging by shooting my boltor at a distant wall with 90% multishot- there was not a visible pattern in whether or not a multishot would proc.

 

Just tested this myself with an akbolto, and you appear to be correct - multishot does not proc in the predictable manner a rollover mechanic would require.

 

This doesn't invalidate the rest of my explanation regarding how pistols/rifles & shotguns are all handled in precisely the same way.

e.g.

HEK 8 pellets + 80% multishot = 8 * 1.8 = 14.4 = 14 pellets + 0.4 (40%) chance of a 15th pellet.

Bolto 1 bullet + 90% multishot = 1 * 1.9 = 1.9 = 1 bullet + 0.9 (90%) chance of a 2nd bullet.

 

Pseudo code:

 

 

void fire(Weapon weapon, float multishotMultiplier) {

 

float projectiles = weapon.projectilesPerShot*multishotMultiplier;

int wholeProjectiles = floor(projectiles);

 

float partProjectile = projectiles - wholeProjectiles;

if(Math.random() < partProjectile) wholeProjectiles++;

}

Edited May 7, 2013 by TehJumpingJawa

[Argoms]

Does anyone else think it'd be better if "random" effects worked like the originally described effect (additive chances- 25% proc will activate every 4 times)? Crits, thunderbolt, multishot, etc. Becoming consistent instead of just luck?

[TehJumpingJawa]

Does anyone else think it'd be better if "random" effects worked like the originally described effect (additive chances- 25% proc will activate every 4 times)? Crits, thunderbolt, multishot, etc. Becoming consistent instead of just luck?

 

Although in the grand scheme of things such a change would probably go unnoticed by most, I agree it would be an improvement.

[Argoms]

Although in the grand scheme of things such a change would probably go unnoticed by most, I agree it would be an improvement.

That's the nice thing about an update like this. They could stealth patch it and unless someone was testing for it (or something that happened to involve it, like crit chance), they might never figure it out. But for the people who do care, it matters a lot.

 

Little tricks that emerge could be quite interesting too.

Edited May 7, 2013 by Argoms

[smithf]

If they released all the changes clearly, our "weapon testing department/experts" will be out of jobs :)

Anyway, I wish to thank all those researchers who took time and effort to find out and teach us how the (warframe) world behaved. It really helps us in mod planning.