The UKF Base Arti has 4 super accurate shells that will land dead center. It comes from the first fire base. This REQUIRES vision or scatter goes up 1000x. Due to the TINY hit box of 2cm Flak, it might take 2 barrages. |
Hallelujah! |
Yea. I just dont have a strong opinion on it. Size is small, but it is built on a light tank. Its just a casemate light tank, basically a brick of steel. |
Yep, its size 15. It is pretty small. Shows up late, has only 180 armor and 640 HP as a flame tank. Generally, the problem is no one expects the Hetzer. It's on a single REALLY broadcasty build, so just spot and be ready. |
that's some interesting analysis, thanks a lot for sharing!
as hannibal already pointed out we've both been working on very similar scatter and AoE-related projects lately, and i think your approach is a good way to do it.
after reading over your methodology, i'd suggest a few edits to your formulae to get a bit more accurate results though:
i think the approximation you used for the double integral under the AoE curve is quite crude and calculating the precise "volume" is actually not much more complicated than what you've already done.
instead of approximating the area under the curve with rectangles and integrating to get the volumes of the resulting cylinders, you can split the complete area into right trapezoids and calculate the precise volume for the corresponding cone sections via pi*h/3*(r^2+rR+R^2), where r and R are the inner and outer radii and h is the damage at each radius (n, m, f)
with respect to the scatter cone, there is one important varable that you may have missed from what i've read: distance_scatter_max.
this will cap the horizontal scatter you get from distance x dynamic_scatter_ratio (this even caps the FoW multiplier!) and lead to smaller actual scatter areas, especially at max range.
apart from this i think your method works great to get a somewhat reliable impression for raw or "on paper" performance of arty pieces and a handy gauge for comparison. however, relating this to in-game performance is a bit tricky, due to the sheer volume of additional variables factoring in here, as pointed out by others already.
i've been working on a simulation to get actual squad sizes and formations into the equation for quite some time, which i plan to release soon(tm). with a few modifications this would probably work well for artillery and indirect fire, too.
Crunched the numbers. Its a LOT of extra damage. The Priest and Sexton get 22.
4% MORE damage concentration. At Max range (WITHOUT vet 3), the Priest gets a 31.2% scatter region premium. Jesus. Vet makes this all even nastier.
From this, the Sexton AND priest beat out the base-line leIF (AND ANY OTHER ARTILLERY) when fighting at 160 range into FoW. Even the Super Charged rounds are better. And the Priest just bodies it (13% more dps in area). Counter Barrage is still solidly the best artillery in the game, but dang, that max scatter is naaaaaasty.
Super Charged barrages are mildly more effective than leIF and ML/20 into FoW, but worse against spotted. Due to how Super Charged Valentine works, mostly will be in vision (mostly...).
What is the point of this? the Max Distance Scatter makes the mobile artillery precision FoW snipers instead of close support. Mobile howitzers are Back Breakers, not close support (compared to fixed howis).
Jives with my experience. I can snipe enemy rocket artillery reliably with a Sexton at max range (my primary use for it, when not shelling forward bases).
So... Balance Team. What do you want artillery to do? |
that's some interesting analysis, thanks a lot for sharing!
as hannibal already pointed out we've both been working on very similar scatter and AoE-related projects lately, and i think your approach is a good way to do it.
after reading over your methodology, i'd suggest a few edits to your formulae to get a bit more accurate results though:
i think the approximation you used for the double integral under the AoE curve is quite crude and calculating the precise "volume" is actually not much more complicated than what you've already done.
instead of approximating the area under the curve with rectangles and integrating to get the volumes of the resulting cylinders, you can split the complete area into right trapezoids and calculate the precise volume for the corresponding cone sections via pi*h/3*(r^2+rR+R^2), where r and R are the inner and outer radii and h is the damage at each radius (n, m, f)
with respect to the scatter cone, there is one important varable that you may have missed from what i've read: distance_scatter_max.
this will cap the horizontal scatter you get from distance x dynamic_scatter_ratio (this even caps the FoW multiplier!) and lead to smaller actual scatter areas, especially at max range.
apart from this i think your method works great to get a somewhat reliable impression for raw or "on paper" performance of arty pieces and a handy gauge for comparison. however, relating this to in-game performance is a bit tricky, due to the sheer volume of additional variables factoring in here, as pointed out by others already.
i've been working on a simulation to get actual squad sizes and formations into the equation for quite some time, which i plan to release soon(tm). with a few modifications this would probably work well for artillery and indirect fire, too.
MMX! MY MAN! Yep. What you did is a better approximation. Should be almost exactly the same. I got lazy on properly doing the areas. Should be REALLY CLOSE though, if not equal.
Distance Scatter Max is a weird beast that really makes some guns over perform scatter-wise. Both Mobile Howitzers are actually constrained by this at 124 range. Thats 3/4ths range. It REALLY improves long distance scatter, with max scatter for Sexton and Priest should be 20 and 22.5. It's actually fixed at 15.5. That is HUUUUUUGE long range accuracy bonuses.
leIF and ML/20 only get to exploit this once they get Vet 3 and the extra range. Give me a minute, gotta get my spread sheet including Max Scatter. Might do some disturbing things. |
1. Valentine Super Charged Victor Barrage disabled during Perimeter Overwatch. Might be intentional.
2. Concentration Barrage (called out as "Direct Barrage") is NOT disabled during Perimeter Overwatch. It really should be. |
Thread: Panther15 Apr 2020, 18:08 PM
Big team games always result in needing different strategy, ESPECIALLY Red Ball's super laniness. "Just flank" is the poor man's "just use AT guns, you don't need REAL Anti Tank" (that cuts both ways depending on timing).
I think its been discussed over and over, heavy(-er) tanks should have a size increase, Heavy TDs then can have some accuracy cut, and then the Stug would be able to do budget-Jadgpanzer duty (which it kind of did in real life, a nice bonus).
Might need to further improve the Stug against Heavy TDs... And give USF Non-doc Medium TD or balance around Medium Tanks + AT Gun. There's design space. It's "just" a really massive overhaul and a pile of work. |
(just realized this post has become super long, sorry for that)
MMX and I have been working independently on theoretical AoE calculations as well.
To be honest some those numbers don't say too much, that is my conclusion from thinking about it and discussing with him for quite some time. MMX's work has the advantage of also simulating scatter and thereby more "in-game like" scenarios. But let's break it down point by point to be more precise:
First table: Overall AoE damage. Represents the AoE "volume".
This is the core issue since all other calculations rely on this. This is also what I tended to calculate when I set up my Excel sheet. But to be honest this does not say much by far. The issues here are:
- what are we shooting at? All of these units are mostly used against infantry. So if we want to compare against infantry, we need to cap all damage at 80, otherwise overkill damage will be factored in which falsifies the data.
- The damage is not normalized and hard to normalize. Damage in the inner core is WAY more valuable than on the edges, since it can insta-kill models or bring them down to a point where small arms can quickly finish them off. However, the outer third of the circle where the damage is abysemal often contributes a whole lot to the overall damage volume.
- Against somewhat decent players, you usually only shoot at 1-2 squads at most. So most of the (often outer circle) damage will not be applied.
- The damage in the inner core is also more like to be actually applied, of course depending on the scatter.
This leads to the next point:
Normalization vs scatter. This is by itself good. One minor issue you make is that you neglected the scatter_offset (haven't checked, but I assume those units also have that value; Please correct me here if I am wrong). From the current scatter model, this value basically leads to a larger scatter area by moving the scatter center point further back. However, scatter_offset might be the same for all units, so the introduced error is the same or at least similar.
While this value is better in terms of it shows which units are better able to really damage the intended area, it still gets fed "bad" data due to the above reasons.
Afterwards, you normalize vs time etc which is all fine apart from the already mentioned base data.
Now don't get me wrong, I don't want to shit talk your work here. I like this kind of approach and I think for generally comparing those units within the group itself it can show some differences. BUT:
We should keep in mind that nobody knows how this data actually translates into in-game performance. Does 1000 more damage volume help you? Well it at least does not hurt. If it's overkill damage than it does not help against infantry, just against buildings. Maybe it does not help at all because the damage comes from small damage values in a large outer area of the AoE circle. But since the cannon tends to hit well the outer areas are rarely applied, etc etc. Reliability is a huge thing. The B4 can kill a heavy tank, but people don't like it because your investment into it might only pay off every Xth game or so, while it single handedly wins you the next one because two shots are on target. Then there is shock value. The air time of the shell, which is probably one of the most important things. Unit density (mostly depending on the game mode played). Now obviously you did not claim that your numbers would factor all this in and your calculations are the end-all for defining the performance on this unit. But as I said, as much as I like and also trust in theoretical calculations, we should keep in mind that they have some shortcoming that do not allow us to really draw conclusions for the game itself if we do not correct for them.
Now the thing I think is probably be the "best" indicator is what MMX is currently doing: Making a simulation against a "real" squad. Calculate how long a unit takes to kill a whole squad. Obviously this still does not simulate moving of the squad etc, but it's at least something. If you want to put real work into this, I'd recommend getting in touch with MMX.
Dang! Thanks for the reply. Yep, I agree with what you said. My last howitzer analysis was entirely on Death Zone (that 80 number), which REALLY had funky results (Sexton was like...40% or more worse than leIF we know it isn't really). Overkill for arti is actually up to 160 or more, due to green cover and target-damage types. Full health Team Weapons in Green Cover live through DIRECT SZF shots with 20 hp per model (found that out when an MG in a trench lived through it). Really depends on what you are trying to do.
My current hypothesis on all artillery is it should have a pretty tiny Death zone, as the huge zones on some guns are prone to wipe pain and un-stoppable veterancy degradation. Wipes should be because you screwed up or got out played, not cause someone clicked a single ability and waited. having long shelling barrages, large AoE, low per-shell damage would separate Howitzers and Rocket arti. Need more data and good players to verify any of this.
I'll get in touch with MMX. |
They do not.
Actually Ro.A is a very strong commander in large mod and the ammount of arty in bring down is simply ridiculous. It does not need are more buff and it should probably be nerfed.
Did you read the data? the shells do 10% less for no described reason, looks like someone typo'd the AoE number. I'd like to not have another one of these special little game knowledge things.
Ro.A is MASSIVELY muni intensive and Coordinated Barrage call-in is garbage 25 lb shells. The real value is in the 7 sniper shells if you maintain LoS. How are you getting shelled more than usual against a UKF?
Also, which Super Charged rounds? the Valentine ones or the Creeping? Creeping Barrage has some weird cooldown stuff, takes longer to cooldown post barrage than the barrage does.
It's also Sexton rounds. These are not the Random Acts of Death of most artillery, they have a pretty tiny 80 dmg area. I've watched many a player just retreat with minimal losses. |