That wouldn't work with on map arty or even basic engagements where hits can become rng and not kill/kill unit that it isn't supposed to.
They save the game seed so that every RNG roll has the same outcome. That's something that is easily doable.
But maybe someone else can elaborate on what exactly is in this file. |
I didn't mean replay files. But I guess that would be very accurate. I doubt relic would let the extraction of their proprietary play back format even though it's probably a unit positions along with at what frame or however in game time is calculated is attached to it.
Some people extract info like victory points and wins etc from those files. As far as I know it is mostly a file containing orders, and the game then simulates what is happening if you watch a replay.
That's what I use now.
As weird as it sounds, the stats themselves are somewhat inaccurate. As I said, give me a week or two to work out the model properly, I'll then publish it for criticism. |
I fully support that.
From your second post though you were debating if Vipper had a point in general by referring to "He [Katitof] included exactly what you say". But he did not.
I agree though that the far values paint the picture too bright, since you realistically want to close in to a range of 5-20 meters (depending how much you want to risk your tank for that) to get the best shots off. Still even at those values, pen chance against heaavies has gone up ~20-30%. |
Can anyone tell me how to extract info from the game it self. Then even I could just make an app that gives you more accurate dps graphs.
Do you mean replay files?
To be honest, no idea about that. But I also doubt that it is possible. As far as I know (I don't have much knowledge about replay files though), no unit positions are saved. So even if you'd get some info on HP decline, there is no way to get the range. |
Did you read all that was posted? Clearly not. Check your reading before posting. He included exactly what you say.
You're posting stats in the best possible cases. Please post n/m/f too.
I don't want to participate in the discussion what exactly is supposed to be a "flanker" unit since I find it tedious, but Vipper's general point is right. Katitof claimed:
Not a single axis unit in game lost any relevant amount of armor that would even remotely impact result vs T34 main gun at any range, be it frontal or rear armor. Maybe some fringe cases with 5-10% bounce chance.
He fails to back it up though.
Especially heavy tanks across all factions lost mostly rear armor to reward flanking. And regarding the special example of the T34 as discussed here, you are right, the exact numbers obviously depend on the range. Katitof though even claimed "at any range", so he should be able to deal with values calculated for "any range". But even in the "worst" case, Katitof's claim does not hold any truth. |
One thing I find suspicious is the extremely short reaction time.
Even given (Rosbone's point) many of these cases where something happens at the edge of the FOV are inconclusive due to desyncs, you still need some time to react.
However, how exactly does this hack work?
From the accusations it is assumed that he at least sees some ghost units in the FOV. Or does he just get straight FOV? Does the old FOV indication stay? If he sees everything, why does he sometimes react so "late"? In the example of the assgrens sprinting to support a flanked MG42 against Conscripts, he could have sprinted earlier with similar effect.
Why does all of this happen at the absolute edge of the FOV where we are dealing with synchronization issues, and not 5-10 meters out?
The only explanation is that he actively waits until the edge of the FOV to not look suspicious. But in this case he must have a god-like feeling of the exact FOV of his units and always nail it with accuracy of those 1-2 meters.
It would be good to have the actual replay files or at least full length games as video. What matters is not only when he DID do suspicious things, but also the amount of times when he did not in the same game. And nobody can judge that from mostly 1 minute clips. |
When you say "true" you mean from in game testing?
From a model that I made after in game testing and modding rifle stats.
I also made a test with the live Gren K98 and it seems to work well. To be fair though I so far focused on getting the timings right, I have not checked for the actual damage and assumed those are correct so far. That formula is in line though with what we see from all other weapons in the game. |
Thread: KV111 Mar 2021, 09:49 AM
It's only strength is it's HP though, if you give it 800HP it's at the same durability level as a T-34-85 but without a decent gun.
800 HP is maybe indeed a bit too little, but as others already said, it has very good armor. A P4 cannot break through against it, 880 HP would ensure that you can't double snare it. At the time it comes out the only real counter is an ATG, so it will still be a decent addition for your mid game and the late game is all about synergy with the SU85 anyway. |
grenadier K98
Weapon profile has at range 0 at 5.76 DPS Cereal has it at 5.988
at range 30 2.6 vs 2.6
so they seem pretty close.
But all these is theoretical. One could test in game timing the TTK of modified model of lets say 1600 HP.
True DPS is more between 5-5,3ish though, meaning both 'official' values overestimate it by 10-15%. |
Wait, are you saying the DPS calculations for various weapons on Seralia aren't accurate? That's rather unfortunate, I've been looking at it a lot recently to try and gauge some things.
At the moment I can say for sure that the formula is wrong.
However it will still be a decent indication for the time being. But if you want to compare units that are only a couple % apart, then bear in mind that you can't draw any conclusions. |