well yes on one hand i cant deny that my current playtime suggests i am the wrong to judge, but on the other hand i also played empires regularily for about 11 iterations over 5 years (but some only lasted a few days, so its probably 9) with the same perception. even if it was me who told empiresmod-forum about emptys armor-weapon stat sheets, leading to him being unbanned and getting part of the dev team, i still dont believe in raw stats. i think DPS calculations dont hold much ground since unlike in RTSs and MMORPGs you dont automatically hit so its nice theory and they certainly shouldnt be neglected but in the end its handling deciding. as best example take times when we could put dual homing on NF tanks. stat wise (DPS) they where a lot worse then UML but their ability to land >90% hits made them super-imba in comparison.
It would be nice if we could measure accuracy better. You'd have to have the game keep track of his and misses for each weapon and then eventually provide a nice sample to model from.
I dont exactly agree 100% here, but its a fucking thread title. Don't use acronyms unless it makes the title ~100 character long. Especially if you want to use ones that are not common knowledge.
I remember how long i resisted to look up what tl;dr meant. I passionately hated people who used that newfangled acronym.
Yeah. We know it's easier to hit a moving target with a faster projectile, but how much easier? Not being able to easily measure that drives me up a wall.
its not only hitting more important would be hitting what you wanted (i mainly think sides here, but also intentional misses with AoE) which would prove tricky to code ...
I hadn't even thought it intentional misses. It's truly a difficult problem. It might be possible to simplify the problem to get an acceptable approximation. We've got a 2d system with a time variable. A point (or collection of points?) has a coordinate pair for each time interval. The pairs have the same distance from the pair for the time interval before them. You could assign a small probability where the distance is zero (and maybe the subsequent angle is tighter than normal?) The tricky part is how to vary the angle between the lines formed by connecting a points at times x & x+1 (for discrete time) and x+1 & x+2. I initially thought some auto regressive process, but that's too nerdy. You can just manually assign probabilities since the angle is nicely bounded. Then you have a vector that "launches" a point that has a much larger distance between its ordered pairs at subsequent time values. The angle between the lines created by points at three different time values is always zero. Then you just see how often the two points are at the same ordered pair at the same time value. I'm sure you could jury rig some kind of "splash" as well.
now you accidentiallied my whole brain :D am i wrong that this isnt so great if you switch targets often? point being intentional misses to deal damage around a corner or to hit a more damaged side, also how to distinquish if you wanted to hit tank a or b? but then, the more i think about it, maybe its neglectable at all, in the vast majority of times the intention probably is to directly hit. traveling down the offtopic highway ...
You're right that a lot of the stuff that goes into a rich model is negligible. That's why we could probably get away with one "tank" launching something at another"tank" to see how projectile speed affects accuracy.
OK - one of my pet peeves is shit changing unannounced without release notes and we have to lose games before anyone realizes why the research that used to work no longer does. HOW CAN YOU FIX THIS ? Can you list the PLATE counts for each armour ? How about strength of each weapon. Clearly BE Rifleman/Tanks/APC are much stronger than NF. BE seems to majorly unbalanced from NF. Hell push a UGL APC with a second seat gren on BE and you win early game. NF can't do second seat and their heavy tanks are no match for BE heavies. The game is NOT evenly matched by a long shot. All you can hope for is players who like a challenge and the skill to win on NF. How many of the developers actively play? We had 3 vs 2 on the head to head tank match with flags and got our asses handed to us by BE tanks. And the skill of the tank driver was not that disproportionate. And people wonder why so few players are still playing. One match like that and you would never play NF again. Maybe a random unannounced change is called for again - sheesh.
BE Rifleman aren't much stronger than NF. BE APC are better with the 2nd seat. NF jeep are better with 4 seats. Tanks? Yeah, I'd say BE have some advantage, but that advantage is not at the level of fucking the game up. That 3v2 round, are you refering to a Vehicletraining round just days ago, with Me&KyrO vs you&urethra(can't recall) and one other guy? You guys were not even constantly driving Heavy tanks. I saw a lot of Meds/Artys. I partly agree with your points(BE tanks are better, but not outstandingly better. BE APC is better).
HE was made OP last patch, so obviously. 3v2 doesn't really prove anything. Especially not on vehicletraining.
Solution : Make NF heavies have a 3slot mg and a 3slot ML (with second 2slot), rebalance mg's and keep the HE balance that exists. That way no dual nuke(or homing salvo) but still good nf guided+mg boats.
MG's aren't supposed to be a main weapon. Plus, that'd require NF to research both missiles and machine guns to be effective, while BE still only needs HE.
