probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Feb 17, 2015 2:30:58 GMT -5
|
|
probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Feb 17, 2015 2:37:48 GMT -5
|
|
|
|
probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Feb 17, 2015 2:46:26 GMT -5
|
|
probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Feb 17, 2015 2:49:36 GMT -5
|
|
|
|
|
|
Post by thegentleman on Feb 17, 2015 6:15:53 GMT -5
Great work, Probaddie. Also a major eye-roll moment again at SHG claiming that the SN6 is the "low recoil" SMG.
|
|
|
|
Post by broth3r on Feb 17, 2015 9:24:15 GMT -5
Why oh why, SN6.
Also, Goliath and Express have individual plots despite also being represented by the base version, and the Speakeasy's base and Grip plots have been swallowed by a repeat of the Scout's.
|
|
probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Feb 17, 2015 9:27:16 GMT -5
Why oh why, SN6. Also, Goliath and Express have individual plots despite also being represented by the base version, and the Speakeasy's base and Grip plots have been swallowed by a repeat of the Scout's. Thanks for point these errors out. I'll fix them post-haste. |
|
|
|
Post by ktownlegend on Feb 17, 2015 15:54:24 GMT -5
Why oh why, SN6. Also, Goliath and Express have individual plots despite also being represented by the base version, and the Speakeasy's base and Grip plots have been swallowed by a repeat of the Scout's. Thanks for point these errors out. I'll fix them post-haste. The deceiver should have it's own chart though. since it has +10% recoil. |
|
probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Feb 17, 2015 19:29:42 GMT -5
Thanks for point these errors out. I'll fix them post-haste. The deceiver should have it's own chart though. since it has +10% recoil. Also fixed. Thanks for your help. |
|
banana
True Banana
Zoro > Law
Posts: 1,577
|
Post by banana on Feb 17, 2015 22:14:54 GMT -5
The speakeasy grip and base don't have an inaccuracy score.
|
|
probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Feb 17, 2015 22:24:54 GMT -5
The speakeasy grip and base don't have an inaccuracy score. Fixed. Thank you. |
|
banana
True Banana
Zoro > Law
Posts: 1,577
|
Post by banana on Feb 23, 2015 18:30:42 GMT -5
where's the mp11 squeaker?
|
|
probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Feb 24, 2015 5:52:26 GMT -5
where's the mp11 squeaker? At your mom's house. Kidding, it's fixed. Thanks again. |
|
banana
True Banana
Zoro > Law
Posts: 1,577
|
Post by banana on Feb 24, 2015 8:20:36 GMT -5
You put the kf5 locomotive/scout in there instead of the mp11 squeaker lol
|
|
|
|
Post by ninopettis on Feb 24, 2015 22:45:53 GMT -5
Okay, don't hate me for the work I might be creating for you here...
The KF5 and MP11 sections are a bit messed up. There's some variants in the wrong sections. The KF5 Single Stack and the MP11 Scout are both missing, and the MP11 Devourer foregrip/rapidfire combo is missing.
Aside from those issues, the only things I notice with the KF5 are the issues with frame rounding mentioned in the other thread, that could just be a result of variance. It's just another instance of that. The Locomotive and the Scout have both their firetime and rapid firetime rounded at 60fps, and therefore aren't very different to the base KF5. They're slightly less accurate, but are slightly more accurate with foregrip, slightly more accurate with rapidfire, and again slightly less accurate with both foregrip + rapidfire.
It's the same with MP11. No issues but another occurence of the above. The Deceiver is slightly more accurate than the Savior and Toxic, which both have a fire rate increase that gets rounded. Yet, the Deceiver is slightly less accurate than them when they all have foregrip. The Savior and Toxic both have a higher rate of fire than the Deceiver when they all have rapid fire attached, so they understandably have worse accuracy there and there's no problems.
How come the foregrip barely does anything to the ASM1 Reckless? Is this correct? I don't have the Reckless so I can't test. The only other thing I notice with the ASM1 is that the Sweeper and the Scout have different plots when they have the same accuracy and rate of fire as each other.
The SN6 Cycled is only slightly less accurate than the base SN6, despite having a higher rate of fire that DOESN'T get rounded. That's seems weird. It's especially weird when you compare the Cycled's rapid firetimes with the base SN6. With rapid fire attached, the base and the Cylced both have the same firetime as each otherat 60fps due to rounding, yet the differences are around the same as when the SN6 doesn't habe rapid fire. I hope it's clear what I mean here. The other issue with the SN6 is with the K, the variant that comes with foregrip. It has the same accuracy and rate of fire as the base SN6, yet it's slightly less accurate in comparison (1.20 vs 1.24 and 1.95 vs 1.97). Is this just variance? Anyway, shouldn't it be grouped with the base SN6 in the same way you did with the AMR9 Banger, only including it in the foregrip examples like the Banger's only included in the foregrip-less examples?
The AMR9 Shorty is another example of having a slightly worse accuracy score without foregrip, but slightly better with foregrip, compared to base variant, and it also involves frame rounding. Finally, the Hipshot and the Dynamo have the same accuracy and firerate. Their only difference is that the Dynamo gets a burst delay decrease from 0.1 to 0.09, but that gets rounded at 60 fps anyway. Their results seem too different to each other considering that this is the only difference between them, especially when rapidfire is attached.
I've already looked at Pytaek and Ameli so I'll post things I've noticed in the heavy weapons thread.
|
|
probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Feb 25, 2015 3:05:23 GMT -5
Okay, don't hate me for the work I might be creating for you here... The KF5 and MP11 sections are a bit messed up. There's some variants in the wrong sections. The KF5 Single Stack and the MP11 Scout are both missing, and the MP11 Devourer foregrip/rapidfire combo is missing. Fixed. Also fixed: there was no multiplier to the ViewKick scale to account for the Grip. In combination with the included Rapid Fire it now gets only +3.5% ViewKick (1.15 * 0.9 = 1.035). The SN6 is a funny case. At exactly 60 fps (0.016666...s between frames) you're right that all versions of the SN6 have their fireTime rounded up to 0.08s. The game only deals with times in integral milliseconds, meaning the true minimum time between frames can only be either 16ms or 17ms. Not knowing exactly which time is correct, I chose the slightly larger time of 17ms, thinking it would better represent reality - how likely is it that 16 ms between frames is preserved for every frame? What this means for the SN6 (and Cycled variant) is that the base weapon (fireTime = 0.075) gets rounded to 0.085s (0.017s * 5 = 0.085s) but the Cycled (fireTime = 0.065s) only gets rounded to 0.068s (0.017s * 4 = 0.068s). With rapid fire, both the base version and the Cycled get a fireTime of 0.068s (0.075s * 0.8 = 0.06s, 0.065s * 0.8 = 0.052s). The Cycled has more recoil because of the ViewKick penalty. That is what accounts for the discrepancies there, I believe. If there was some evidence to suggest that the time between frames is actually capped at 16 ms, I would likely go back and re-do the plots for borderline cases such as these. I understand your point about the SN6 - K. I'm a bit reluctant to go ahead with the suggestion, though, just because it requires a bit of mental gymnastics for the viewer to figure out where the K plot is located. The calculus would have to be, "okay, so the K is the same as the base version (but with a foregrip), so I actually have to go to the base plots and look at just the ones with the grip and convince myself I'm looking at the K." With the AMR9 Banger, its already understood that it has an accuracy profile identical to the base version but with one less attachment option. Ultimately, I think I prefer just keeping variants with technically deviant accuracy profiles separate; there's no need to sacrifice clarity just to be clever about the organization of the plots. It turns out - I will completely admit that this is due to pure luck - that the difference does matter at 17 ms between frames! The actual time required between the end of the last burst and the start of the next one for burst fire weapons is fireTime + burstFireCooldown. The Hipshot requires 0.155s between the last burst and the next, and the Dynamo requires only 0.145s. So with 17 ms between frames, the Hipshot can discharge the next volley after 9 frames (9 * 0.17s = 0.153s) but the Dynamo will require 10 frames (10 * 0.017s = 0.170s). Again, this is a complete coincidence, and at 16 ms between frames you'd be entirely correct about the burst delay not making a difference. (Both would need 10 frames between volleys, since 9 * 0.016 = 0.144s and 10 * 0.016 = 0.160s.) I hope this answers your concerns. Edit: Marvel4 pointed out that, at a perfect 60fps, the SN6 variants with faster fire rate actually would fire faster (0.0166...s * 3 = 0.0666...s > 0.065s). |
|
|
|
Post by ninopettis on Feb 25, 2015 23:19:16 GMT -5
Okay, don't hate me for the work I might be creating for you here... The KF5 and MP11 sections are a bit messed up. There's some variants in the wrong sections. The KF5 Single Stack and the MP11 Scout are both missing, and the MP11 Devourer foregrip/rapidfire combo is missing. Fixed. I think the KF5 Single Stack and the MP11 Scout are still both missing. I have no idea how foregrip works so I don't understand the math but never mind. Now I'm only wondering why the Reckless is so different to the base ASM1. They have the same inaccuracy scores with rapidfire but no foregrip (3.38 each), but very different inaccuracy scores with rapidfire + foregrip (2.2 vs 2.66). Reckless gets frame rounded both WITH and WITHOUT rapid fire to be the same as the base ASM so it should be similar in both instances. This is different to the SN6 example later in that the Cycled has a higher rate of fire than the base without rapidfire attached, but the same rate of fire as the base with rapidfire attached. I think you misunderstood me about the SN6. I wasn't clear enough and should have quoted the numbers I have. The SN6 has 0.075 firetime; the Cycled has 0.065 firetime. I did ROUNDUP(60*firetime)/60 (which I got from this thread) to get rounded firetimes. I know that's different to your method but our conclusion is the same in that we both agree that the Cycled has a higher rate of fire than the base SN6 if both DON'T have rapid fire but they have the same rate of fire as each other if they both DO have rapid fire. The SN6's 0.075 gets rounded to 0.083 and the Cycled's 0.065 gets rounded to 0.067, using the formula above. I never said that the Cycled doesn't see a higher rate of fire. I was saying that the Cycled's higher rate of fire should make it more inaccurate than it does. It's currently only 2.10 vs 2.15 without foregrip and 1.2 vs 1.29 with foregrip. That's a very insignificant difference compared to any other example I've seen in your work. All other fire rate increase examples cause much more inaccuracy than that. I also used the same formula above to round firetime with rapid fire attached. The SN6's 0.06 and the Cycled's 0.052 both get rounded to 0.067, similar to how you say. So, since the Cycled fires faster than the base WITHOUT rapid fire, but the same as the base WITH rapid fire, one would expect the inaccuracy between the two to be greater without rapid fire attached, but that isn't really the case. 3.22 vs 3.28 without foregrip and 1.95 vs 1.93 with foregrip. Yeah, the difference is slightly bigger in the example where neither gun has rapid fire, but it's insignificant. Look at any other example between a base gun and a variant with a higher rate of fire, and you'll see it has a much bigger difference in inaccuracy than that. In other words, the rapid fire doesn't do anything to the SN6 Cycled at 60 fps. I probably should have said that sooner. You said the same thing in the text I bolded. If the Cycled's fire rate is 0.068 both with and without rapid fire, then why is its inaccuracy score so different with and without rapid fire. The K actually has the same issue as this, since it has the same firetime stats as the Cycled. I just checked to confirm, and it's true that these are the only two weapons in my spreadsheet that aren't affected by rapid fire. That includes all assault rifles, all SMGs except the SAC3s, plus the Pytaek and Ameli. Finally, the SN6 and the Cycled have the same viewkick stats so that shouldn't be relevant. I did my working wrong. I didn't realize the time between bursts included firetime as well as burstfirecooldown. I rounded firetime individually and burstfirecooldown individually. I didn't add them up and then round them afterward, which I can see produces different times. Thanks for going through all of this with me, I appreciate it.  |
|
probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Feb 26, 2015 5:44:36 GMT -5
Thanks for going through all of this with me, I appreciate it. No worries. Forgive me for not re-quoting your previous post; things were just starting to get messy. 1. I see the problem with those two variants is still there. I guess I forgot to hit "Save Changes"? I'll fix that soon, I promise  . 2. I finally see what you're saying. You're right that there isn't enough of a difference between the SN6 Base and the Cycled. I double-checked the numbers in my spreadsheet and they appeared to be right - almost. What went wrong is that I inadvertently switched the fireTimes in my spreadsheet so that the "true" fireTime was used first for the first three shots and then the "intro" fireTime after that, where it should have been the other way around. Thus, the SN6 in all cases was simulated with a fireTime of 0.068s between shots after the first 3 shots. (Rapid Fire does not affect the "intro" fireTime.) For reference, here is how the SN6 Base/Cycled look after fixing the bug (with 10,000 trials each):   I think there should be no problems here, but let me know what you think. After I talk to Marvel4 sometime today, I'll update the plots with variable firerates. 3. Regarding the burst weapons: it turns out that the expiration of fireTime is also frame-bound - I just learned this myself recently. So in those examples I gave for the AMR9, there would likely be yet an extra frame between volleys. (And yeah, I'll update the burst-fire plots, too.) I hope you don't feel guilty about hounding me like this: you shouldn't. I never have a problem with people correcting me so long as they're civil and reasonable about it, and you've been both throughout. |
|
|
|
Post by ninopettis on Feb 27, 2015 4:34:27 GMT -5
Yeah, the SN6 seems right now. ALMOST everything is good now. I adjusted the formula in my spreadsheet to replace 60 with (1/0.017) so there's a few guns that were rounded before but now aren't and weren't rounded before but now are. The only one that doesn't add up with the plots now is the AMR Pro Pipe and the AMR Outlaw. They both get rounded to the same firetimes and have the same viewkick stats but the Outlaw has a worse inaccuracy score. Aside from that, there's still the ASM1 Reckless that I mentioned above. It has the same viewkick stats to the base ASM1, and its firetime with rapid fire get rounded to the same number as the base ASM1 with rapid fire (Reckless's 0.055 gets rounded to 0.068 and base ASM's 0.063 gets rounded to 0.068 too). Despite that, they only have similar results when there's no foregrip attached. The only other things are some guns with the same stats should be grouped, like the ASM Scout and ASM Sweeper, and the MP11 Scout and MP11 Devourer. Edit: I forgot about this, but shouldn't the SN6 Cycled and also the SN6 K both not be affected by rapid fire? Due to frame rounding rapid fire should have no effect on these guns, unless I'm mistaken. 0.065 gets rounded to 0.068 0.065 * 0.8 = 0.052 0.052 gets rounded to 0.068 Also, I know I said earlier that these are the only guns affected by this on my spreadsheet, but somehow I missed the fact that both the MP11 and AMR9 are affected by this, with the exceptions of a few variants for each. Another thing is, if assault rifles could use rapid fire, the ARX and IMR would also be examples of this. Is my math wrong and I applied the rapid fire multiplier in the wrong order? I assumed it'd be multiplied by the game code first, and then frame rounded afterwards. I tested the MP11 myself and the one with rapid fire does finish firing a magazine a little faster, but that might be expected due to what I spoke about in the Fire rates and Frame rates thread. |
|
probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Feb 27, 2015 8:12:55 GMT -5
Yeah, the SN6 seems right now. ALMOST everything is good now. I adjusted the formula in my spreadsheet to replace 60 with (1/0.017) so there's a few guns that were rounded before but now aren't and weren't rounded before but now are. The only one that doesn't add up with the plots now is the AMR Pro Pipe and the AMR Outlaw. They both get rounded to the same firetimes and have the same viewkick stats but the Outlaw has a worse inaccuracy score. Aside from that, there's still the ASM1 Reckless that I mentioned above. It has the same viewkick stats to the base ASM1, and its firetime with rapid fire get rounded to the same number as the base ASM1 with rapid fire (Reckless's 0.055 gets rounded to 0.068 and base ASM's 0.063 gets rounded to 0.068 too). Despite that, they only have similar results when there's no foregrip attached. The only other things are some guns with the same stats should be grouped, like the ASM Scout and ASM Sweeper, and the MP11 Scout and MP11 Devourer. 1. I accidentally gave the Pro Pipe a Grip (+10% CenterSpeed, -10% ViewKick). That is now fixed. 2. I think there is a bug with the Reckless plots. My plotter has a single variable which controls which version of Call of Duty it is making plots for (to select which spreadsheet to use and to account for dvars and other exceptions), and one more mode for "test" plots whenever I want to do a small subset of plots - this is how I do my revisions. In Advanced Warfare, the Rapid Fire attachment only modifies the "main" fireTime and not the intro firetime. So, if I set the plotter to do Advanced Warfare plots en masse, the proper exception to Rapid Fire is applied. However, the plotter was not applying that exception for the test plots. This means that any plots that I revised with variable rates are wrong. Unfortunately, I have no record of which plots are affected by this, but it seems likely this is the case with the Reckless. So, I'll be re-uploading all the variable firerate SMGs again. For now, here's a comparison of the Reckless with Grip to the Base with Grip and Rapid Fire:   True, but the Rapid Fire attachment in and of itself confers a +15% ViewKick penalty, regardless of what actually happens to the weapon's firerate. That probably accounts for the difference you see. The ASM1 Scout and Sweeper will be merged and updated; I want to keep the MP11 Scout and MP11 Devourer separate, though. The game will report that the two weapons fire at different rates in the Create-a-Class menu - they are, in fact, coded to have different firerates - so I think it's informative to indicate that, at a typical framerate, the two will behave identically accuracy-wise. |
|
|
|
Post by ninopettis on Feb 27, 2015 13:26:03 GMT -5
Cool, thanks for all your work. I forgot that rapid fire did that. Also, my mistake with the MP11 Scout and Devourer. I knew different coded weapons had different plots but just forgot.
|
|
|
|
|
|
|
|
|
|
Post by kylet357 on Apr 24, 2015 1:14:32 GMT -5
"Post-Rapid Fire Nerf"? Did they do something to the recoil increase on Rapid Fire? Or did you mean SN6 post-buff?
|
|
probaddie
True Bro
You're triggering my intelligence
Posts: 11,043
|
Post by probaddie on Apr 24, 2015 1:20:36 GMT -5
"Post-Rapid Fire Nerf"? Did they do something to the recoil increase on Rapid Fire? Or did you mean SN6 post-buff? SHG stealth nerfed Rapid Fire so that it now confers a 20% increase to ViewKick, up from 15% Also, I have to re-do these plots: I found a bug in my code :/ |
|
|
|
Post by kylet357 on Apr 24, 2015 12:14:13 GMT -5
"Post-Rapid Fire Nerf"? Did they do something to the recoil increase on Rapid Fire? Or did you mean SN6 post-buff? SHG stealth nerfed Rapid Fire so that it now confers a 20% increase to ViewKick, up from 15% Also, I have to re-do these plots: I found a bug in my code :/ SHG really hates Rapid Fire. |
|
|
|
Post by Megaqwerty on Apr 24, 2015 17:56:47 GMT -5
What. If they keep going down this road, Rapid Fire will eventually need to give you class points.
|
|
.
