|
|
Post by toto on Jan 4, 2011 2:20:43 GMT -5
Still doesnt explain why the M14 never kicks down, and why its kick is so much more than the FALs. Let's see if I can explain. I'll state that I am assuming that the game calculates centerspeed for the horizontal and vertical axes independently, based on Den's findings in this thread. Firstly, we need our formula I wrote down earlier: (FireRate * Centerspeed) / 5 = Maximum viewKick allowed to be able to fully recover from shot firedAccording to Den's charts, centerspeed is equal to 1400 (for both the M14 and FAL). But what is the Fire Rate? The M14 (and FAL) is listed at 625 RPM. I'm assuming this is the fastest speed at which the M14 can be fired. But in a real game, you probably won't even get close to that - an RPM in the 300's is more likely. Let's assume you can click pretty fast at a rate of 400 RPM. The Fire Rate would then be 0.15. Using this value, we find that: (0.15 * 1400) / 5 = 42 viewKick allowed This means that any roll of 42 or less in a given direction and the gun will fully recenter (from that direction) before the next shot is fired. Since the maximum viewKick in the left, right and down directions is only 40, it means that the M14 will not be able to actually kick in those directions. The viewKick of 80 in the up direction means that it can only kick straight up, and boy will it kick up. Fortunately, adding a grip increases the centerspeed to 1600, which changes things slightly: (0.15 * 1600) / 5 = 48 viewKick allowed The FAL has higher viewKick values in the left and right direction, meaning there is a chance it can veer left, right and up. It also doesn't have the grip, so it will suffer from more recoil. Remember, the FAL and M14 can shoot only as fast as the player's ability to click the mouse. The faster you're able to shoot, the lower the viewKick allowed, and the more recoil that will be felt. However, during actual gameplay, the player is most likely not to fire anywhere close to the 625 RPM it's rated at. Conversely, firing slow enough will allow you to negate any effects of viewKick and recoil. To sum it up: M14 - Heavy upward kick FAL - Small, random kick in all directions |
|
asasa
True Bro
fuck
Posts: 4,255 
|
Post by asasa on Jan 4, 2011 16:39:07 GMT -5
Right, but if a dice rolls between -40 and 80, it makes no sense that it would always be positive...
Why not just make it 0-40 if that -40 is essentially non-existant?
Horizontally guns kick equally left and right if they are.. equal.
|
|
|
|
Post by Ishbane on Jan 4, 2011 17:52:36 GMT -5
|
|
|
|
Post by psijaka on Feb 13, 2011 14:49:20 GMT -5
This is a subject that is of interest to me. I have recently been involved in some heated debate on the Black Ops official forum, about the effect of recenter speed and recoil on the performance of an automatic gun at long range.
I recently made a breakthrough when I was informed that recenter speed/5 = recenter speed per second (thanks to asasa on the BOps forum).
From here it was easy to determine what the recenter would be before the next round was fired, and I am glad to say my calculations match Den's for the Skorpion (still some grey matter working after 49 years).
The next step was to determine what proportion of second shots from a full auto gun would result in the gun being on target. To do this I applied a linear random distribution to the kick spread figures for each gun, and calculated the proportion of times that the gun recovered before the next round. I applied recenter figure to the vertical and horizontal component separately. My methods were a bit ugly, using brute processing power rather than mathematical finesse, but that is what spreadsheets are for!
The results for a selection of full auto guns are below. The figure is the proportion of second shots that are on target; in other words, when the gun has fully recovered from the kick of the previous round.
57.00% Skorpion grip
51.00% Skorpion
27.00% M60 grip
27.00% Galil, Enfield
23.00% AK74 grip post nerf
22.00% Famas
21.00% AK47, Comm.
21.00% HK21
21.00% AK74
20.00% M60
20.00% MP5
17.00% M16
15.00% AK74 RF grip post nerf
15.00% G11
14.00% Aug
13.00% Stoner
13.00% AK74 RF
11.00% MP5 RF
7.00% RPK
I would be happy to post part of the spreadsheet to show my method if anyone is interested, and would welcome constructive criticism.
|
|
|
|
Post by Contrary on Feb 13, 2011 14:59:37 GMT -5
Great work! A really good way to put things in perspective with hard data. I tsk tsk at 74u and Famas. I expected the RPK to be worse honestly but that's still pretty bad.
How exactly do these results work? Is this just firing two shots full auto and seeing how many of them recenter exactly? So can we extend these percentages to subsequent shots? Like would a Skorpion with grip have a 32% chance of having its third shot precisely on target?
|
|
asasa
True Bro
fuck
Posts: 4,255
|
Post by asasa on Feb 13, 2011 15:06:13 GMT -5
Lol @ RPK. Very useful info. Adding more weapons would be great Now if you could figure out the odds of being on target at different ranges considering the max body size[Not crouched, fully visible] that would be awesome  . You'd have to do a lot more than its probably worth, though. |
|
|
|
Post by maradona on Feb 13, 2011 16:30:32 GMT -5
This is a subject that is of interest to me. I have recently been involved in some heated debate on the Black Ops official forum, about the effect of recenter speed and recoil on the performance of an automatic gun at long range. I recently made a breakthrough when I was informed that recenter speed/5 = recenter speed per second (thanks to asasa on the BOps forum). From here it was easy to determine what the recenter would be before the next round was fired, and I am glad to say my calculations match Den's for the Skorpion (still some grey matter working after 49 years). The next step was to determine what proportion of second shots from a full auto gun would result in the gun being on target. To do this I applied a linear random distribution to the kick spread figures for each gun, and calculated the proportion of times that the gun recovered before the next round. I applied recenter figure to the vertical and horizontal component separately. My methods were a bit ugly, using brute processing power rather than mathematical finesse, but that is what spreadsheets are for! The results for a selection of full auto guns are below. The figure is the proportion of second shots that are on target; in other words, when the gun has fully recovered from the kick of the previous round. 57.00% Skorpion grip 51.00% Skorpion 27.00% M60 grip 27.00% Galil, Enfield 23.00% AK74 grip post nerf 22.00% Famas 21.00% AK47, Comm. 21.00% HK21 21.00% AK74 20.00% M60 20.00% MP5 17.00% M16 15.00% AK74 RF grip post nerf 15.00% G11 14.00% Aug 13.00% Stoner 13.00% AK74 RF 11.00% MP5 RF 7.00% RPK I would be happy to post part of the spreadsheet to show my method if anyone is interested, and would welcome constructive criticism. thanks for doing this. isnt it not quite clear though what exactly the recoil numbers mean in relation to centerspeed? see Den's last posts ITT. also, did you use 1 or 2 random numbers each for the vertical and horizontal recoil? there was some discussion about which one was the correct method if i recall correctly. |
|
|
|
Post by Ishbane on Feb 13, 2011 16:50:51 GMT -5
I'd appreciate if you could do a full list for all guns and allow me to add them to the chart.
|
|
|
|
Post by imrlybord7 on Feb 13, 2011 17:35:49 GMT -5
What Ishbane said. I'm curious about the Spectre.
|
|
mannon
True Bro
wordy bastard PSN:mannonc Steam:mannonc XB:BADmannon
Posts: 15,371
|
Post by mannon on Feb 13, 2011 17:55:24 GMT -5
Right, but if a dice rolls between -40 and 80, it makes no sense that it would always be positive... Why not just make it 0-40 if that -40 is essentially non-existant? Horizontally guns kick equally left and right if they are.. equal. Horizontally and Vertically guns behave the same. What's being said is not that a -40 pitch kick has no effect at all, just that the gun can recenter before the next shot is fired. In other words the gun CAN kick downwards, but it will recenter before the next shot if it does and therefore cancel out the kick. Every gun has basically a box (not quite square but close to it according to Den) around the middle. The size of said box is determined entirely by centerspeed. If the viewkick happens to fall within this magic box then the gun will completely recenter before the next shot. The viewkick determines by how much or even whether or not the gun can kick outside of this box and on which sides. (Actually it's even more complicated since apparently weapons can recenter upon each axis independently. This means that a significantly higher proportion of second shots will land exactly in the center due to recentering, but also a higher proportion of second shots will land either directly horizontal or vertical of the center due to recentering in one axis but not quite recentering in the other if I'm reading this right. I think that I am, though as this would account for the high degree of accuracy you can get countering mostly vertical kicks. You can clearly see the weapon still kicking left and right, but don't need to correct horizontally due to the recentering before the next shot is fired.) So when you do get the down kick, basically it becomes purely visual much like the SPAS shotgun and Intervention sniper rifle from MW2. They recenter before the next shot can be fired... but they DO still kick. Removing the visual kick would make the game feel more like Quake than something based on RL weapons, as well as making the weapons much easier to aim. I think this also tells me why I have such a hard time adjusting for recoil... I always over adjust because I'm reacting to where the gun kicks to right after the last shot, but it will automatically correct for a large portion of that on it's own usually causing me to overcompensate and then let the weapon drift back across my target as it recenters. Since you can no longer see where your actual aim point is due to the entire view kicking around I've always had a terrible time with it. Obviously I just needed to make smaller adjustments and only when outside a certain range, but never developed the knack for it. (I fell in love with burst fire and low recoil guns... sue me.) BTW it also occurs to me that viewkick numbers of exactly 50,-50/50,-50 (50 in all directions) will not result in an exact square either, since it does not reflect a number of pixels in each direction, but rather represents the rotation of the view on an axis. I believe that it should map instead like longitudinal lines on a sphere and would thus be subject to increasing perspective distortion the further from center, at least when attempting to map it to your actual view in game. Anyway, very damn cool this was finally worked out! Now I just wish we had the stats for the MW2 guns. heh (Yes I still don't have BO, and I'm not totally sold on it either. I know it's popular to hate on MW2 these days, but frankly, I still like the game. It was a big step up from W@W IMO and most of the things that do bug me about it aren't that big a deal to me. BO... well I get mixed feelings when I read about it, but it does look fun.) |
|
|
|
Post by psijaka on Feb 13, 2011 18:18:21 GMT -5
Great work! A really good way to put things in perspective with hard data. I tsk tsk at 74u and Famas. I expected the RPK to be worse honestly but that's still pretty bad. How exactly do these results work? Is this just firing two shots full auto and seeing how many of them recenter exactly? So can we extend these percentages to subsequent shots? Like would a Skorpion with grip have a 32% chance of having its third shot precisely on target? Thanks for the feedback. It is as you say, what happens to the second shot when you fire an automatic. If the second shot was fired when the gun had fully recovered from the kick of the first shot, then the percentage would apply to the third shot. And so on. But it is not quite as simple as that. What happens to the third shot if the second shot was not on target? Some 3rd shots will still be on target. I need to do more work on this. But more guns first. |
|
|
|
Post by psijaka on Feb 13, 2011 18:27:49 GMT -5
Lol @ RPK. Very useful info. Adding more weapons would be great Now if you could figure out the odds of being on target at different ranges considering the max body size[Not crouched, fully visible] that would be awesome . You'd have to do a lot more than its probably worth, though. Thanks for the feedback. I will be going through all the combinations of full auto very soon. I have been doing some work on calculating the percentage of "near misses", which I arbitrarily define as the gun returning to within an absolute value of 10 of it's starting point. But I don't know what 10 means in terms of shooting an enemy player at a certain range. I would imagine it is way too big for longshotting snipers on the mast in Array; nothing but a perfect shot will do in this case. Any help on what would be a reasonable figure for a near miss would be welcome; I can just plug in the new figure easily. But I am going to concentrate on covering all of the full auto guns first. |
|
|
|
Post by psijaka on Feb 13, 2011 18:35:48 GMT -5
thanks for doing this. isnt it not quite clear though what exactly the recoil numbers mean in relation to centerspeed? see Den's last posts ITT. also, did you use 1 or 2 random numbers each for the vertical and horizontal recoil? there was some discussion about which one was the correct method if i recall correctly. If I have understood Den's figures, the recoil figures are the spread of random kick that you get when you fire a gun. I have assumed a separate linear random distribution for both vertical and horizontal kick. It would just not make sense any other way. I have applied the recenter figure to the vertical and horizontal component separately; I note that there has been some debate about this. But I believe that this is correct; it would take a lot more processing power to apply it any other way, for no real gain. |
|
|
|
Post by psijaka on Feb 13, 2011 18:41:51 GMT -5
I'd appreciate if you could do a full list for all guns and allow me to add them to the chart. I will be posting the full list shortly; feel free to add them to your excellent chart. I would like some more "peer review" of my method before you do so, however. |
|
mannon
True Bro
wordy bastard PSN:mannonc Steam:mannonc XB:BADmannon
Posts: 15,371
|
Post by mannon on Feb 13, 2011 18:44:55 GMT -5
Hmm... I just had another thought.
If centerspeed truly works on each axis completely independently as suggested then this means that the resetting after view kick only ever travels in 8 principle directions. It will go directly horizontal, vertical, or at a 45 degree angle until one or both axis recenter or the next shot is fired. The centerspeed would never, for example, move the view back to center along a trajectory of 30 degrees or any such thing.
More specifically if up is 0 then centerspeed would only ever move the view on a vector along the following trajectories, in degrees: 0, 45, 90, 135, 180, 225, 270, 315.
Additionally during a recentering the view might be moved along one of the diagonal angles until recentering on one axis, at which point it will transition to either horizontal or vertical. But it would never, ever transition from horizontal or vertical movement to diagonal.
Thus given any known viewkick you can predict exactly the path taken to return to center. This may be fairly irrelevant to full auto weapons but knowing this could be useful for semi auto or burst fire weapons that you may wish to fire rapidly before a full recenter. (In other words you could predict a single axis recentering and chose to compensate along the other axis that won't have a chance to recenter.)
This should surely be testable.
mmm... Sorry... Just been wanting to know this information since W@W and it makes me giddy!
|
|
mannon
True Bro
wordy bastard PSN:mannonc Steam:mannonc XB:BADmannon
Posts: 15,371
|
Post by mannon on Feb 13, 2011 18:56:27 GMT -5
Oh wait... I should make a small correction on the recentering pattern I mentioned before. With each axis considered independently the only reason the point in the center is more likely than the horizontal and vertical lines of the cross that I mentions is because those lines cross each other there, and thus both axis recenter. That just makes the center point of the cross twice as likely as any single point on the lines of the cross.
Wow... it's SO much less complicated with centerspeed applied independently to each axis... I hope that really is correct.
|
|
phale
True Bro
Posts: 635
|
Post by phale on Feb 13, 2011 19:22:04 GMT -5
57.00% Skorpion grip 51.00% Skorpion 27.00% M60 grip 27.00% Galil, Enfield 23.00% AK74 grip post nerf 22.00% Famas 21.00% AK47, Comm. 21.00% HK21 21.00% AK74 20.00% M60 20.00% MP5 17.00% M16 15.00% AK74 RF grip post nerf 15.00% G11 14.00% Aug 13.00% Stoner 13.00% AK74 RF 11.00% MP5 RF 7.00% RPK Just a question. How do these numbers relate to the numbers in this thread? Most of them are very similar. |
|
asasa
True Bro
fuck
Posts: 4,255
|
Post by asasa on Feb 13, 2011 19:34:46 GMT -5
Right, but if a dice rolls between -40 and 80, it makes no sense that it would always be positive... Why not just make it 0-40 if that -40 is essentially non-existant? Horizontally guns kick equally left and right if they are.. equal. Horizontally and Vertically guns behave the same. What's being said is not that a -40 pitch kick has no effect at all, just that the gun can recenter before the next shot is fired. In other words the gun CAN kick downwards, but it will recenter before the next shot if it does and therefore cancel out the kick. Every gun has basically a box (not quite square but close to it according to Den) around the middle. The size of said box is determined entirely by centerspeed. If the viewkick happens to fall within this magic box then the gun will completely recenter before the next shot. The viewkick determines by how much or even whether or not the gun can kick outside of this box and on which sides. (Actually it's even more complicated since apparently weapons can recenter upon each axis independently. This means that a significantly higher proportion of second shots will land exactly in the center due to recentering, but also a higher proportion of second shots will land either directly horizontal or vertical of the center due to recentering in one axis but not quite recentering in the other if I'm reading this right. I think that I am, though as this would account for the high degree of accuracy you can get countering mostly vertical kicks. You can clearly see the weapon still kicking left and right, but don't need to correct horizontally due to the recentering before the next shot is fired.) So when you do get the down kick, basically it becomes purely visual much like the SPAS shotgun and Intervention sniper rifle from MW2. They recenter before the next shot can be fired... but they DO still kick. Removing the visual kick would make the game feel more like Quake than something based on RL weapons, as well as making the weapons much easier to aim. I think this also tells me why I have such a hard time adjusting for recoil... I always over adjust because I'm reacting to where the gun kicks to right after the last shot, but it will automatically correct for a large portion of that on it's own usually causing me to overcompensate and then let the weapon drift back across my target as it recenters. Since you can no longer see where your actual aim point is due to the entire view kicking around I've always had a terrible time with it. Obviously I just needed to make smaller adjustments and only when outside a certain range, but never developed the knack for it. (I fell in love with burst fire and low recoil guns... sue me.) BTW it also occurs to me that viewkick numbers of exactly 50,-50/50,-50 (50 in all directions) will not result in an exact square either, since it does not reflect a number of pixels in each direction, but rather represents the rotation of the view on an axis. I believe that it should map instead like longitudinal lines on a sphere and would thus be subject to increasing perspective distortion the further from center, at least when attempting to map it to your actual view in game. Anyway, very gosh darn golly gee whiz cool this was finally worked out! Now I just wish we had the stats for the MW2 guns. heh (Yes I still don't have BO, and I'm not totally sold on it either. I know it's popular to hate on MW2 these days, but frankly, I still like the game. It was a big step up from W@W IMO and most of the things that do bug me about it aren't that big a deal to me. BO... well I get mixed feelings when I read about it, but it does look fun.) Fire an M14 at a wall. It has the largest down number of any gun. Continuously do so until you see it push down a single time. Hint: You will never stop. |
|
mannon
True Bro
wordy bastard PSN:mannonc Steam:mannonc XB:BADmannon
Posts: 15,371
|
Post by mannon on Feb 13, 2011 20:24:01 GMT -5
That depends on if you are waiting for one of your bullets to hit lower than the last one or if you are watching for visual recoil.
Here this video demonstrates it a bit, where he's comparing the FAL and M14.
Just skip to the M14 part, then ignore the bullet decals, those are all after centerspeed, remember. Watch the horizontal boards on the garage. If the view only ever kicked up then you would only see them move down as the view climbed steadily up. But this is not what you see. You see them mostly move down with the view going up, but here and there the boards go the other direction very briefly as the view actually kicks down, then quickly recenters.
It is only ever a very brief bump down and then back to center due to centerspeed, but it does happen. This is also partly responsible for the uneven clumping. Shots tend to climb, but not at even intervals, and although you cannot have a shot fall below the last shot they can fall at the same height. You know that's possible because you can see right after the first reload there are at least two places with 2, 3+ shots on a horizontal line. In all likelihood these shots were kicked down, or only a short up kick and recentered vertically.
The downward kicks are pretty much irrelevant performance-wise since they are rendered visual only, but that doesn't mean they don't exist.
|
|
asasa
True Bro
fuck
Posts: 4,255
|
Post by asasa on Feb 13, 2011 20:30:33 GMT -5
I dont see what you mean Mannon.
Recentering if it had a downward recoil would not make it push up, it would still pull down. [As long as its above the initial point]
I fired the M14 at a wall for like 10 mins. Not once did it ever go down. [Firing single shots, waiting for recenter, and the opposite]
|
|
|
|
Post by maradona on Feb 13, 2011 20:39:48 GMT -5
thanks for doing this. isnt it not quite clear though what exactly the recoil numbers mean in relation to centerspeed? see Den's last posts ITT. also, did you use 1 or 2 random numbers each for the vertical and horizontal recoil? there was some discussion about which one was the correct method if i recall correctly. If I have understood Den's figures, the recoil figures are the spread of random kick that you get when you fire a gun. I have assumed a separate linear random distribution for both vertical and horizontal kick. It would just not make sense any other way. I have applied the recenter figure to the vertical and horizontal component separately; I note that there has been some debate about this. But I believe that this is correct; it would take a lot more processing power to apply it any other way, for no real gain. this is the post i was referring to: I expected the same, but I made sure and tested it beforehand. If both X and Y were reduced to 70, the view recovers well before a full second. But when that difference is distributed between the X and Y, the View re-centers almost exactly on one second. The area of a 1.0 second recenter comes out like more of a skewed octagon than a circle.  Actually... wait. Hold on... I can go even greater, up to 0.95 and the view will still recenter before the next shot. It's not a perfect square, but it's very close. Perhaps CenterSpeed handles X and Y speeds separately. now if i understood your method correctly you first calculated how long it takes the y-value to reach 0, then how long it takes the x-value to reach zero and then you added those two times together. but, using your method and the numbers in the above example (centerspeed of 500 and x- and y-values of 70 each), it would take [(70 + 70) * 5] / 500 = 1.4 seconds to recover, which obviously contradicts what Den says. |
|
mannon
True Bro
wordy bastard PSN:mannonc Steam:mannonc XB:BADmannon
Posts: 15,371
|
Post by mannon on Feb 13, 2011 21:23:20 GMT -5
In the video I linked (http://www.youtube.com/watch?v=fRKhvwnbEew) you can see as he is firing the view does indeed kick steadily up (garage moves down in view as view moves up) but every now and then the reverse happens and the garage moves up in the view for a split second then down again.
We've known about this type of behavior for quite some time, it's just that now we have a much better handle on the actual math behind it.
In other words if the above stated figures are correct (I'm assuming they are) the M14 can viewkick down up to 40 "viewkickunits" but can recenter up to 42 or so of those units. Thus even at a maximum downward kick there is still enough time before the next bullet is fired for the view to return to center. Ergo the view goes down, then up, but very quickly, and the next bullet is never ever ever fired lower than the last one without help from the player.
It's not so mysterious. But don't worry about it. All you really need to care about is where your next bullet will land. Purely visual viewkick can mostly be safely ignored. (Then again some people would still put a grip on the SPAS in MW2 just to reduce the purely visual recoil, even though the gun always recenters before you can fire it again... and it's a shotgun at that with no need for precision.)
I will retract one of my previous statements, though. I should say that the centerspeed is always only working in one of the 8 primary direction, but if it is applied more like an accelerating force rather than purely as a speed then the actual movements wouldn't always bee along those 8 principle directions. (Though they would still be dominant, especially horizontal and vertical.)
I'd have to dig a little more into Den's math, but I'm trying to just generalize so far. I mean... I don't even have BO and can't play MW2 or W@W online right now anyway, but meh. I can't stay away from such an interesting topic.
If anybody else is still confused about the M14 could somebody else explain it to them? I can't think how to make it more clear without drawing a picture... Mebbe I should.
|
|
|
|
Post by imrlybord7 on Feb 13, 2011 23:57:23 GMT -5
Great post mannon, really explained a lot that I hadn't quite understood. Sorry you can't CoD it up at the moment.
|
|
Den
He's That Guy
Posts: 4,294,967,295
|
Post by Den on Feb 14, 2011 2:20:13 GMT -5
There's nothing to dig. Everything is already explained.
It takes five times the CenterSpeed of a ViewKick to neutralize the Kick at exactly one second.
A certain CenterSpeed recovers a certain amount of ViewKick at a certain time.
One-fifth of the CenterSpeed multiplied by the FireTime is how much of the area the CenterSpeed will recover before the next shot.
The "Neutral Number".
1000 x 0.064 = 12.80
1100 x 0.064 = 14.08
1200 x 0.064 = 15.36
1300 x 0.064 = 16.64
1400 x 0.064 = 17.92
1500 x 0.064 = 19.20
1600 x 0.064 = 20.48
1700 x 0.064 = 21.76
1800 x 0.064 = 23.04
1000 x 0.080 = 16.00
1100 x 0.080 = 17.60
1200 x 0.080 = 19.20
1300 x 0.080 = 20.80
1400 x 0.080 = 22.40
1500 x 0.080 = 24.00
1600 x 0.080 = 25.60
1700 x 0.080 = 27.20
1800 x 0.080 = 28.80
1000 x 0.048 = 09.60
1100 x 0.048 = 10.56
1200 x 0.048 = 11.52
1300 x 0.048 = 12.48
1400 x 0.048 = 13.44
1500 x 0.048 = 14.40
1600 x 0.048 = 15.36
1700 x 0.048 = 16.32
1800 x 0.048 = 17.28
1000 x 0.060 = 12.00
1100 x 0.060 = 13.20
1200 x 0.060 = 14.40
1300 x 0.060 = 15.60
1400 x 0.060 = 16.80
1500 x 0.060 = 18.00
1600 x 0.060 = 19.20
1700 x 0.060 = 20.40
1800 x 0.060 = 21.60
---
If the recoil amount randomly chosen on an axis is less than that amount, that axis will have recovered fully before the next shot.
Two examples. The Spectre and the AUG.
The Spectre's left and down kick is completely neutralized in less than one twentieth of a second - barely two frames - and will neutralize out to 21 ViewKick up and right
before the next shot.
23 with a Grip.
16 with Rapid Fire.
The AUG will neutralize out to 19 in any direction.
-
Recoil is in the field of random events.
Specific numbers here don't have that much of a meaning since a 50/50 chance coin toss could still land on Heads a hundred times in a row.
Anyway, the "Neutral Kick Chance". Only applies to shots made from the Center, once that aim goes everywhere, you'll have... leftover ViewKick.
1. Divide the Neutral Number by all four ViewKick amounts.
21 / 60 = 0.35
21 / 10 = 1.0*
21 / 60 = 0.35
21 / 20 = 1.0*
When a whole direction is neutralized, just make it a 1.
19 / 60 = 0.316
19 / 60 = 0.316
19 / 60 = 0.316
19 / 30 = 0.633
2. Grade your tests. Horizontal and Vertical are separate since their axes do not influence one another.
(0.35 + 1) / 2 = 0.675
(0.35 + 1) / 2 = 0.675
The Spectre has a 67.5% chance of having Neutralized Recoil on each axis before the next shot.
0.316 + 0.316 / 2 = 0.316 (oh boy, how redundant. When both sides are equal, there's no need to do this part)
0.316 + 0.633 / 2 = 0.475
The AUG has a 31.6% chance of Neutral horizontal Kick and a 47.5% of Neutral Vertical Kick.
Another case, the Skorpion.
1700 and 1800 Centerspeed. 0.08 and 0.06 Firetime.
28 with the Grip
20 with Rapid Fire
With the Grip, the whole Vertical axis is almost completely Neutral, a 96.7% chance.
The Horizontal axis has a 56% chance.
---
And for a 100% accuracy guarantee, take the highest possible ViewKick number, divide it by the Neutral Number and multiply by the firetime.
Spectre
60 Viewkick divided by it's Neutral Number of 21.76.
Multiplied by its FireTime of 0.064.
0.177 seconds.
|
|
|
|
Post by psijaka on Feb 14, 2011 2:19:52 GMT -5
That depends on if you are waiting for one of your bullets to hit lower than the last one or if you are watching for visual recoil. Here this video demonstrates it a bit, where he's comparing the FAL and M14. Just skip to the M14 part, then ignore the bullet decals, those are all after centerspeed, remember. Watch the horizontal boards on the garage. If the view only ever kicked up then you would only see them move down as the view climbed steadily up. But this is not what you see. You see them mostly move down with the view going up, but here and there the boards go the other direction very briefly as the view actually kicks down, then quickly recenters. It is only ever a very brief bump down and then back to center due to centerspeed, but it does happen. This is also partly responsible for the uneven clumping. Shots tend to climb, but not at even intervals, and although you cannot have a shot fall below the last shot they can fall at the same height. You know that's possible because you can see right after the first reload there are at least two places with 2, 3+ shots on a horizontal line. In all likelihood these shots were kicked down, or only a short up kick and recentered vertically. The downward kicks are pretty much irrelevant performance-wise since they are rendered visual only, but that doesn't mean they don't exist. I have had a look at the M14 If it were fired as fast as possible, with the 2nd round being fired 0.096s after the first (Den's figures), then recentering between rounds is 26.88 (or 30.72 with grip). So it is theoreticaally possible to have the seond shot go downwards, as max downwards kick is 40. Very unlikely to happen in reality though; how many people can consistently loose off 10 shots per second whilst aiming accurately? |
|
|
|
Post by psijaka on Feb 14, 2011 2:27:41 GMT -5
now if i understood your method correctly you first calculated how long it takes the y-value to reach 0, then how long it takes the x-value to reach zero and then you added those two times together. but, using your method and the numbers in the above example (centerspeed of 500 and x- and y-values of 70 each), it would take [(70 + 70) * 5] / 500 = 1.4 seconds to recover, which obviously contradicts what Den says. No I apply the two calculations simultaneously. So the recenter time for a shot is governed by whichever is biggest, the horizontal or the vertical kick. I know I am sticking my neck out a bit here, but it would take up a lot more "in game" processing power for it to work any other way, for very little gain. I could be wrong though! The Squashed Octagon is interesting, but I think Den subsequently says it is almost a square. Again, there would be no reason for this added complexity; what is the gain? I would really like to thrash out issues like this before I publish any more figures. Keep the questions coming! |
|
|
|
Post by psijaka on Feb 14, 2011 2:31:45 GMT -5
57.00% Skorpion grip 51.00% Skorpion 27.00% M60 grip 27.00% Galil, Enfield 23.00% AK74 grip post nerf 22.00% Famas 21.00% AK47, Comm. 21.00% HK21 21.00% AK74 20.00% M60 20.00% MP5 17.00% M16 15.00% AK74 RF grip post nerf 15.00% G11 14.00% Aug 13.00% Stoner 13.00% AK74 RF 11.00% MP5 RF 7.00% RPK Just a question. How do these numbers relate to the numbers in this thread? Most of them are very similar. I hadn't spotted this. Thanks for pointing it out. First impressions are that I don't like the look of the methods, but I will need to look a ot more closely. |
|
|
|
Post by psijaka on Feb 14, 2011 8:45:50 GMT -5
Two examples. The Spectre and the AUG. The Spectre's left and down kick is completely neutralized in less than one twentieth of a second - barely two frames - and will neutralize out to 21 ViewKick up and right before the next shot. 23 with a Grip. 16 with Rapid Fire. The AUG will neutralize out to 19 in any direction. - Recoil is in the field of random events. Specific numbers here don't have that much of a meaning since a 50/50 chance coin toss could still land on Heads a hundred times in a row. Anyway, the "Neutral Kick Chance". Only applies to shots made from the Center, once that aim goes everywhere, you'll have... leftover ViewKick. 1. Divide the Neutral Number by all four ViewKick amounts. 21 / 60 = 0.35 21 / 10 = 1.0* 21 / 60 = 0.35 21 / 20 = 1.0* When a whole direction is neutralized, just make it a 1. 19 / 60 = 0.316 19 / 60 = 0.316 19 / 60 = 0.316 19 / 30 = 0.633 2. Grade your tests. Horizontal and Vertical are separate since their axes do not influence one another. (0.35 + 1) / 2 = 0.675 (0.35 + 1) / 2 = 0.675 The Spectre has a 67.5% chance of having Neutralized Recoil on each axis before the next shot. 0.316 + 0.316 / 2 = 0.316 (oh boy, how redundant. When both sides are equal, there's no need to do this part) 0.316 + 0.633 / 2 = 0.475 The AUG has a 31.6% chance of Neutral horizontal Kick and a 47.5% of Neutral Vertical Kick. The way that you are calculating whether a gun will fully recover from vertical assumes that 50% of the kick is up and 50% is down. I will use the Aug as an example. (19/60+19/30)/2=0.475 (as you state) However, with KVmax (Kick, Vertical max) = 60 and KVmin (Kick, Vertical min) = -30 (I have used a negative for downwards kick to avoid errors) Surely we would expect 67% of the shots to kick upwards and 33% to kick downwards? So we would expect the expression for the likelihood that a gun will recover from vertical kick to be RCeff / KVmax * KVmax / (KVmax - KVmin) + RCeff / KVmin * KVmin) / (KVmax - KVmin) EDIT - this can be simplified to 2 * RCeff/(KVmax-KVmin) in this instance!where RCeff = effective recentering between rounds This gives an answer of 19/60 * 60/90 + 19/30 * 30/90 = 0.422 We can do the same for horizontal kick. To get the overall probablilty that a gun will recentre on both axes, we multiply the vertical and horizontal probabilities together. As you point out, if RCeff > the max kick, the probability of the gun recovering from kick in that direction is 1. We can use the MIN function available on spreadsheets to cover this eventuality. Complications arise with the auto guns that always kick upwards (M60, RPK), and we need a different expression for these. I will eventually publish a table of results for all guns with every combination of attachments (grip, RF, ACOG, IR etc). I would also like to look at what happens to subsequent shots, and also calculate the likelihood of near misses (to be defined). Do forum members think that this would be worthwhile enough to merit a new thread? This would enable me to easily update the opening post and make it easier for forum members to find the table. I ask as I am new to this forum. |
|
|
|
Post by psijaka on Feb 14, 2011 11:44:09 GMT -5
Have done some initial modelling of what happens to the 3rd and 4th shots as well as the 2nd. | 2nd shot | 3rd shot | 4th shot | Weapon | | 56% | 56% | 56% | Skorpion grip | | 27% | 16% | 12% | Galil, Enfield | | 22% | 11% | 7% | Famas | | 21% | 19% | 17% | AK47, Commando | | 14% | 12% | 10% | Aug |
The percentage is when the gun has fully recovered from kick before that shot is fired. |
|
|
|
Post by rubionubio on Feb 14, 2011 14:17:30 GMT -5
That last table is very enlightening, I think the information regarding 3rd and 4th shots to be the most beneficial, from a gameplay perspective. For instance, the Famas' 2nd shot is more likely to still be on target than the Aug's 2nd shot, but by the 3rd shot the Aug is actually more likely to recenter, with subsequent shots widening the gap further and further. I know this is already sort of intuitive based on their recoil patterns, but I find the actual numbers for comparison to be very very useful.
I would definitely say start a new thread and possibly sticky it as it is ongoing work and will make for easy location when it's time to modify. I would also ask that when you do "update" the OP of your new thread as you mentioned, post a new reply so that those of us that have subscribed to the thread will more easily note when you've updated it. Does that make sense?
Appreciate your work, very handy!
|
|
.
