I haven't deeply explored each of the heroes yet, but I'm trying to up my map knowledge. It's fun to get a feel for the opposing team, regarding how well the know the maps. Every now and then you get a set of 6 clueless randos.
I ran Pharah in Volskaya on attack, and did the little jet maneuver over the water. All 6 enemies were hogging the main chokepoint, so it caused a major distraction. I was halfway to claiming the objective before they retreated and took me out. It took 4 times, but I kept doing it, and it kept disrupting their chokepoint hold. The team broke through on that 4th one, so I left early and rushed Objective B. I sat in the corner and we immediately started claiming it. As soon as 2 enemies appeared, I unleashed the Pharah special and went back to sitting pretty.
I noticed that since the patch, there have been much more PotGs featuring big damage prevention. You might only get 2 kills in a PotG, but if you headshot a Genji right as he starts his special, or you also pick off a riptire before it explodes, it's a nice play.
I noticed that since the patch, there have been much more PotGs featuring big damage prevention. You might only get 2 kills in a PotG, but if you headshot a Genji right as he starts his special, or you also pick off a riptire before it explodes, it's a nice play.
They seem to be focusing more on shutdowns and plays that lead to caps. I got PotG as Dad76 for shutting down a Bastion and a Pharia to set up for a winning cap on Hanamura B.
This is how I would do it. They presumably have (or could collect) game state history for every game.
From that, you could estimate win probability for each team at any point in a game. For a first cut, maybe consider time left, map, payload location, team health, and team distance from objective.
Given a win probability function for the whole game, calculate the 5 second window with the biggest change. Whichever player moved it the most, that's your POTG.
I would use a hierarchy of relatively simple heuristic trees. Very broad-brush rules at the top (most damage in a short timespan, heavily leveraged ultimate, etc...), and a series of more specific ones the deeper you descend down any tree. At the bottom of the tree, you'd have highly customized (by humans) rules looking for specific events.
So at the bottom of a given tree, there are things like "used an ultimate within 30 seconds of match victory, spent the last 20s of the match ON the objective, spend at least half of that time ALONE on the objective, and spend at least half of THAT time in contested overtime." Very specific, but shit if that isn't an amazing and not-unheard-of play.
Evaluate all the trees. If anything beyond a certain depth occurs, pick the deepest. Otherwise, pick a random tree and do a more focused pass against a wider set of the more customized heuristics. If time runs out, randomly pick among a set of survivor heuristics.
That's how I'd do it, roughly, if it were my project.
I noticed that since the patch, there have been much more PotGs featuring big damage prevention. You might only get 2 kills in a PotG, but if you headshot a Genji right as he starts his special, or you also pick off a riptire before it explodes, it's a nice play.
And it's great because you get people who think it's just looking for the most kills and the same tired response for everyone on the fucking team saying "LOL PRO PLAY".
Ideally, you want it to catch those clutch moments like when an overtime was extended by a D.Va being forced out of her mech but holding on inside the zone until the cavalry arrived.
Even thought it does say during the footage what type of highlight it is I wish they pushed that more in your face so people understand what they're seeing. I've just had a few too many dummies ironically exclaim how cool a play is when they don't understand what they're even watching.
The only text I've seen is "shutdown" when an ulti is cancelled. I think so far the game really only considers total damage and shutdowns as plays of the game. I haven't really seen anything else different.
Was kind of bummed that my play of the game where I was tracer - constantly teasing the enemy team and dashing in and out of the control point 1 against 5 to extend the overtime for 15 seconds - probably wasn't even a recognizable play under the current system even though it clinched the game for my team.
Ideally, you want it to catch those clutch moments like when an overtime was extended by a D.Va being forced out of her mech but holding on inside the zone until the cavalry arrived.
I seem to get all my kills as D.Va after the mech blows up :-p
My D.Va game on offense is often to Scott to a far side flank and constantly fire. Anyone who isn't reasonably seasoned will flinch under the fire despite little actual damage, making them less effective.
Is there a Reinhardt? Just hammer that shield constantly.
My D.Va game on offense is often to Scott to a far side flank and constantly fire. Anyone who isn't reasonably seasoned will flinch under the fire despite little actual damage, making them less effective.
Is there a Reinhardt? Just hammer that shield constantly.
Too many people stop firing when that sheid's up. No, bitches. Break that fucker!
Also, don't underestimate D.Va's boosters. A lot of people panic when they see a MEKA right in their faces, or coming from weird angles (you can fly with that shit, kids!)
My D.Va game on offense is often to Scott to a far side flank and constantly fire. Anyone who isn't reasonably seasoned will flinch under the fire despite little actual damage, making them less effective.
Is there a Reinhardt? Just hammer that shield constantly.
With some time and experience that should stop working and people will just ignore you.
But it's funny how sometimes things that just should not work end up working. And sometimes people just don't seem to get how some aspects of the game work.
Few days ago had a match where we were escorting a payload and the enemy team otherwise did fine, but instead of stopping the payload they walked in front of it, so while the payload moved forward they moved back. And we talked among our team that they could have just moved in and stopped the payload and had the fight there. At least then they would have slowed us down a bit, but they just didn't go for it.
If you haven't yet played as Reinhardt, you may not realize how his shield works (2000 HP). Before I tried him, I wasn't sure if the thing had its own health, or whether it was on a timer and just broke when it was out of energy.
I just assumed you could break Reinhardt shield and would wonder why other players wouldn't focus their fire on his shield. And then I would end up second guessing myself.
I think it's more that people are looking for angles to shoot around the shield and get useful kills. Since I Pharah a lot, I generally try to go over. That said, I also know someone who until recently thought you could break Zarya's barriers and so would happily shoot her full of power.
Comments
オーバーウォッチ&!
I ran Pharah in Volskaya on attack, and did the little jet maneuver over the water. All 6 enemies were hogging the main chokepoint, so it caused a major distraction. I was halfway to claiming the objective before they retreated and took me out. It took 4 times, but I kept doing it, and it kept disrupting their chokepoint hold. The team broke through on that 4th one, so I left early and rushed Objective B. I sat in the corner and we immediately started claiming it. As soon as 2 enemies appeared, I unleashed the Pharah special and went back to sitting pretty.
From that, you could estimate win probability for each team at any point in a game. For a first cut, maybe consider time left, map, payload location, team health, and team distance from objective.
Given a win probability function for the whole game, calculate the 5 second window with the biggest change. Whichever player moved it the most, that's your POTG.
So at the bottom of a given tree, there are things like "used an ultimate within 30 seconds of match victory, spent the last 20s of the match ON the objective, spend at least half of that time ALONE on the objective, and spend at least half of THAT time in contested overtime." Very specific, but shit if that isn't an amazing and not-unheard-of play.
Evaluate all the trees. If anything beyond a certain depth occurs, pick the deepest. Otherwise, pick a random tree and do a more focused pass against a wider set of the more customized heuristics. If time runs out, randomly pick among a set of survivor heuristics.
That's how I'd do it, roughly, if it were my project.
Was kind of bummed that my play of the game where I was tracer - constantly teasing the enemy team and dashing in and out of the control point 1 against 5 to extend the overtime for 15 seconds - probably wasn't even a recognizable play under the current system even though it clinched the game for my team.
Is there a Reinhardt? Just hammer that shield constantly.
Also, don't underestimate D.Va's boosters. A lot of people panic when they see a MEKA right in their faces, or coming from weird angles (you can fly with that shit, kids!)
But it's funny how sometimes things that just should not work end up working. And sometimes people just don't seem to get how some aspects of the game work.
Few days ago had a match where we were escorting a payload and the enemy team otherwise did fine, but instead of stopping the payload they walked in front of it, so while the payload moved forward they moved back. And we talked among our team that they could have just moved in and stopped the payload and had the fight there. At least then they would have slowed us down a bit, but they just didn't go for it.
That said, I also know someone who until recently thought you could break Zarya's barriers and so would happily shoot her full of power.