Talk:Accuracy (weapon)

Real world angles?
There is an entire (rather large) section in the article regarding the corresponding real life angles, and solid angles and everything. This was written by an anon (http://borderlands.wikia.com/index.php?title=Accuracy&diff=201426&oldid=190605), and I'm doubting the accuracy of what is in this section. Does *anybody* even *have* a way of measuring an angle in borderlands?

AFAIK, the section is pure speculation. I started it as speculation, but at least it read as speculation... happypal (talk &bull; contribs) 13:24, August 22, 2012 (UTC)


 * The whole "solid angle" thing (which I revised into using actual units (square degrees)) is correct so long as the original accuracy equation is correct (the one for calculating the angle of spread).
 * As far as the original accuracy equation goes, I did a little looking at the numbers and figured that they were probably an approximation of nicer numbers. In particular, every point of accuracy corresponded to almost exactly 0.02% of 360&deg;.  Since the original equation is apparently just a rough estimation, I didn't figure there was much purpose in having a bunch of icky decimal places if they're not terribly accurate anyway.
 * If I had access to the game's data files, I'd take a peek at the code to see if I could figure out how exactly the angle is calculated; for now, I'm just working with what others have written.  23:58, August 22, 2012 (UTC)
 * TY for the explanation. It'd be great if you could get that figure. On my hand, I'll try to verify the 35.0 vs 37.7. happypal (talk &bull; contribs) 06:04, August 23, 2012 (UTC)
 * Hi there. I did some research with the above shotguns. My 37.7 is more accurate than my 35.0. Does *anybody* know where this data comes from, or have *anything* that could even remotely support this claim? happypal (talk &bull; contribs) 08:38, August 24, 2012 (UTC)
 * I honestly have no idea where the data came from. I'm procuring Borderlands as of this writing to see if I can do some research.   18:28, August 24, 2012 (UTC)
 * Update: I got debug mode working and did some angle calculations. Essentially, the weapon value "spread" is 1/3 of the diameter of the circle (in, say, inches) made by shooting a million rounds into a target 100 inches away.  For example, 2.2 spread means that the target circle 100" away will have a diameter of 6.6".  The calculations I've written on the page now reflect this.  As noted on the page, I haven't tested this with highly inaccurate weapons yet, and they might deviate from this standard formula.   21:02, August 24, 2012 (UTC)

Holyshit, that is some fine research. I always wondered if "spread" wasn't a the "tan" value of a the firing angle, but never had the guts to actually pull out a ruler and measure this stuff! Thanks. For "0.0" weapons, keep in mind you can get their "true" accuracy/spread with gearcalc: Besides, even with gearcalc, you can extract the accuracy manually after reading the spread from the debug info.
 * Sledge's Shotgun: -67,9%, 20.15
 * Boom Stick: -27.3%, 15.28
 * Jackobs ZZ Barrel 2 Matador: -1.8, 12.22

If you can verify that Sledge's shotgun holds true to your formula (with these numbers), then you'll have proven your formula accurate for all spreads (!)

PS: Could you update my spread formula of "12 &minus; (0.12 &times; Accuracy)"? Your form of "12 &times; (100% &minus; Accuracy)" is more elegant, and only uses one "magic number". I don't want to conflict with you if you are editing. happypal (talk &bull; contribs) 21:17, August 24, 2012 (UTC)
 * This might be silly, but are you doing your tests with Roland? Does he have any points invested in Scattershot? happypal (talk &bull; contribs) 08:14, August 25, 2012 (UTC)
 * Conveniently, the debug menu takes accuracy bonuses into consideration when displaying the current weapon's spread (including when using sights), so it actually doesn't matter. Even so, I did re-spec to make sure nothing impacted the performance of Sledge's Shotgun, which does indeed follow the shotgun formula correctly.  At some point I'll scrounge up each weapon type and make sure it checks out.   08:28, August 25, 2012 (UTC)
 * I know what is happening! I found a formula which works for all weapons, and all spreads. Gonna post it now, give me a sec. happypal (talk &bull; contribs) 15:32, August 25, 2012 (UTC)

Here's the thing: AccuracyPool needs to be taken into account. I did a test with a battery of weapons (from Sledge's to Hyperion Lances), writing down the spread, the AccMin, and taking a screenshot, and then measuring the size of the reticule. This gives more consistent results, as weapon with/without stocks will have a different handling, and it will be hard to consistently measure their "pools".

The results is that the size of the reticule is always EXACTLY (give or take 1%) equal to "25 &times; ( spread + AccPool )". I got this result regardless of the weapon type.

My guess is that the results you saw came from shotguns having particularly high AccPools, in particular, even sighted, remain un-accurate.

So know, we've proven there is a linear correlation of "25&times;spread" to "2&times;tan(aperture/2)", how do we actually get the ratio between the two?

Here is my thought: I did the tests with a fixed 85° FOV, and my screen is 1920 pixels large. So, a 42.5° weapon would have a reticule of 1920 pixels. Also, a weapon with 1 Spread+Acc has a reticule of 25 spread. This gives us the system: a &times; 2 &times; tan(42.5°) = 1920 a &times; 2 &times; tan(Aperture / 2) = Spread &times; 25

a = 960/tan(42.5) = 960 / 0.916 = 1047 Aperture = 2 &times; arctan(Spread &times 12.5 / a)

Aperture = 2 &times; arctan(Spread &times 0.0122)

This is actually incredibly close to your (original) formula, which was indeed accurate for all weapons!

If you want the input data, please find it here in HTML comments, in tabulated text format.

I think the last thing to do though, might be to move the section "up" back into the normal Accuracy page?
 * Doesn't check out&mdash;but first, some background with regard to FOV. One of the first things I tried to do when testing accuracy in general was get a correlation between pixels and aperture.  Unfortunately, despite everywhere listing the game as having a FOV of 85&deg;, that's not quite accurate.  Once I got the debug menu working, though, I used the player rotation, pointing at the two sides of the crosshair and noting the difference (rotators in Unreal Engine are integers, with 360&deg; = 216 = 65 536 ).
 * Secondarily, just to clarify: the aperture is the total angle from side to side (diameter, if you will), not the angle from center to side ("radius").
 * Now, I've nailed down the base Sledge's Shotgun to 22.40&deg;–22.85&deg; (depending on if the reticle is on the inside or outside of the aperture). Using 20.15 with my shotgun formula calculates 22.79&deg;, which is right on the money.  Adding in the 2.4 AccMin bumps it up to some 25.4&deg; (and using your 0.0122 would get an even less accurate figure).  In addition, testing with a certain pistol yielded ~3.55&deg;; my &times;1.6 formula calculated 3.57&deg;, your addition formula calculated 3.94&deg;.  Testing with a certain other shotgun yielded ~3.50&deg;; my &times;1 formula calculated 3.47&deg;, your addition formula calculated 4.31&deg;.
 * So there is unfortunately no cigar there. Even so, I can't shake that the accuracy pool must impact the reticle, since shooting makes the reticle larger directly with the pool.  Not sure how to go about this...   18:36, August 25, 2012 (UTC)
 * Note: when employing your formula I ditched the 0.0122 because even using 0.01 resulted in overestimates, and 0.01 is a much nicer number.
 * About aperture: Well, yeah? But if you want to get the diameter of the projected cone from the apperture, you have to apply a tan to the angle from the center of the circle, to the edge, ergo, the half apperture. This will get you the radius of the circle, so the base formula. "2&times;a&times;tan(Apperture/2)" is accurate.
 * About FOV: I didn't know that. Thankyou. There is indeed a "rotation" field. I calculated that my "half" FOV span a rotation of 6362 units, giving me a total FOV of "360&times;2&times;6362/65536=70°"
 * Recalculating my above formula with this new number, I get:
 * FiringCone = 2 &times; arctan((Spread+AccPool) &times; 0.0093)
 * Does this correlate better with your measures?
 * Were all your "other" weapon tests with accurate weapons? Because for an unsighted Pistol (or sniper rifle), the Pool will probably be just as big as (or bigger) than the spread. Do you have an Anarchy lying around? Test it.
 * Regarding Sledge's Shotgun: What stock? Because the stock has an effect on the accuracy pool, hence the bullet pattern. It is accounted for in my formula. happypal (talk &bull; contribs) 19:48, August 25, 2012 (UTC)
 * Edit: In original formula, I meant "Spread+AccPool" when I wrote "Spread", as well as "FiringCone" when I wrote Apperture, I corrected this in my updated formula. happypal (talk &bull; contribs) 19:57, August 25, 2012 (UTC)

Random Break
Hi Proton, I did some more "number crunshing", and I noticed something: The calculated angles are seriously close to the spead/AccPool themselves. Then I thought to myself: Given the low angles of the values we have been calculating, "tan(x) &asymp; x". Also 0.0093 is suspiciously close to &pi;/360 (which is 0.0087) ...

I tried to do some "extreme" testing, and the initial formula remains the most accurate one, but I think I found something interesting if you bear with me: If we express the formula in degrees, then ... FiringCone = 2&times;arctan((Spread+AccPool)/2) So how do we extract "Deviation"(Player Accuracy) and "Apperture"(Gun Accuracy) from this? Technically, we can't, but we can give a pretty damn good approximate: Deviation = 2&times;arctan(AccPool/2) Apperture = 2&times;arctan(Spread/2) See where I'm going with this? This can be approximated to: Deviation = AccPool Apperture = Spread Which brings us back to the original formula: FiringCone = Spread+AccPool

I feel Derp :/

I re-tried all my calculations with the above "simplified" formula, and there are some deviations (10% for Sledge's, 5% for a 7% Matador). Still, as a "layman's" rule of thumb, I think our readers will appreciate the effort.

PS: I'll state my final "accurate" formula as: FiringCone = 2&times;arctan((Spread+AccPool)/2 * 2&pi;/360) happypal (talk &bull; contribs) 07:46, August 26, 2012 (UTC)