When will finally get realistic hitboxes, Sup Forums? The hitboxes in Overwatch/Battlefield are fucking terrible

When will finally get realistic hitboxes, Sup Forums? The hitboxes in Overwatch/Battlefield are fucking terrible.

What game has the best ones?

That's not entirely hitbox issues, but also an issue of types of hit registration, tickrates, and data priorities.

Nothing is going to ever be perfect, but BF4 got really fucking close.

MHFU

>a army

arma 3. just hope to god you both have good pings, hitboxes aside.

Because he's stupid and standing in the open.

It's also an SQL issue but I won't go into the details.

Hitboxes should be shaped for gameplay and balance not realism, especially if the game is meant to be competitive.

How "realistic" a hitbox is has nothing to do with its quality.

That is more of an internet connection issue than hitboxes.
It has to due with latency, client/server communication, and sometimes server predictions.
Your character is not properly synced to the server.
It takes a while for your computer/console(client) to tell the server when you input a direction to move your character.
Sometimes server predictions are put in place to make this less noticeable but players can be unpredictable so that can mess up and visually correct itself if the predictions are incorrect or too far ahead.

Its a big problem that probably won't be fixed for a long time if ever. It also has to do with pings and basically geographical location to how far away you are from the server.

I haven't played any games that handled it very well.

this comic is kinda funny

Yeah there's also potential of packet loss, jitter, and bad routes. Heck, there's even possible problems with physical assets like the router or even the hub in the neighbourhood.

you stop raging about it once you realize that faster tick rate and connection would make you just see your death earlier instead of giving you a chance to save yourself.

If you're talking about your picture then that has nothing to do with hitboxes and everything to do with latency, netcode, server tickrate and game design. It's really fucking bad design to be checking if the bullets hit the player on the receiving player's end. Most FPSs do it for the shooters benefit, if he can see you then he can shoot you even due to latency and tickrate you were no longer actually there.

If we're talking about ACTUAL hitboxes then yes, both projectiles and players have absolutely retarded hitboxes on Overwatch.

Don't mind me, just having objectively the best netcode and input in existence.

Gameplay programmer here

Latency always exists. This will always affect players to some extent.

This is completely incorrect and you sound completely inexperienced.

Think of it this way OP, due to network latency people will always see each other in different times.

for example if you hit the button to duck, it might take 200ms for that command to reach the server, and another 200ms to reach that sniper that currently aims for your head. When he finally gets the fact that you ducked, 400ms after you really did, that's when the ducking animation begins which is another 300-500ms.

So you see, in your PC you are already completely under cover (because you have instant feedback on yourself) but the guy aiming on you will still see you exposed for at least 700-900ms, which is a lot.

Now throw in complexity of accurate hitting during animations (for example hitting legs during running animation) and you have a real issue.

There are some ways to estimate hits correctly like roll time back and forth to check collision (it's a really interesting subject really, I think valve made a huge article about it) but at the end of the day it will never be perfect.

Some games favor the attacker (eg those edge cases will hit) and some favor the target (eg will not hit). But at the end of the day there will always be someone who gets pissed off and feels it's unfair.

This is why rollback is a thing. We're now seeing games set a ping threshold were you're punished for having a shit connection instead of rewarded.

>This is completely incorrect

Unless you're directly linked to the other person with a cable, latency is unavoidable.
The moment you plug yourself into a switch or a router, you are already introducing the latency these things have for routing packets to their correct destination.

Stop playing overwatch

Holy shit stop posting.

It's is because most games use clientside prediction to make games feel more responsive and as a result what you experience and what the server calculates is not 100% the same.

The only game I know of that doesn't do this is Dota 2 and it makes people endlessly bitch and moan because lag more pronounced instead of pretending to be 0ms on your side of the game.

My side, the brand slaves are out. Massah dindu nuthing de know bettah you racist nazi.

>what is propogation speed
Cat5 has a maximum velocity factor of 64% of the speed of light.

Now do the math for distances around the globe and youll quickly see that it is impossible to complete remove latency. And this is under IDEAL THEORETICAL conditions, not including all the other chokepoints on the network

So no, hes right.

also if youre too lazy to do the math, it takes 4.76ns at a VF of .7 to travel 1m. There are 3945km from NY to LA. Thats 18.7ms absolute minimum latency one way.

>arena shooter
Dead genre

You're a idiot

I mean seriously; a small amount of interference can ruin a perfectly good packet. How can I be expected to game in these conditions?

>it might take 200ms
No, it may not. If you're okay with playing on more than 100ms, neck yourself.

what is preventing the hitbox from being the model itself? why does there have to be this separate thing

Because the server is not 1:1 synced with your game.

what does the server have to do with hitboxes?

So bottom line there is always unavoidably going to be latency

Glad you agree

kys

...

???

If you are talking about what is shown in your pic, it has nothing to do with hitboxes - this is a latency problem.
- When you press the "forward" button it take X milliseconds for that message to reach the server.
- It also take Y milliseconds for the server's messages to give you the ennemy's last position.
As a results there will always be a X+Y difference between where (you) see yourself & the ennemy and where the server see you and your ennemy.
Shitty coding can exacerbate that a ton, but until the internet's cables are replace with quantum entanglement and shiet, it will happen even in the best case scenario.

This too

there is an answer to this somewhere when valve talked about updating the csgo hitboxes

i was responding to myself so of course i do

Why are you even posting when you obviously have no knowledge of the subject being discussed? Neither of them said anything about any specific games, you fucking retard.

Physics calculations are getting absurdly expansive when you have anything else than basic geometric shapes (and even then, it's not cheap).
Your choice is to either have a massively simplified version of the model for physic purposes, or have the entire server freeze each time someone unload a clip into someone else general direction.

thanks that's what i wanted to know about. surprised its so taxing to calculate complex hitboxes, that seems unintuitive to me, as someone who's ignorant about that stuff.

This is why South Korea is the ultimate gaming nation. As long as the infrastructure is good enough you can have an entire densely populated country within a very small latency overall for players.

I've only ever found the netcode and hitboxes in overwatch to be absolutely fine.

hit detection basically has to check whether a shot intersects with any of the thousands of polygons that make up a player model
the more polygons you have on the model, the more computational effort is required for this
thats why people use simplified hitboxes that are just a bunch of basic shapes like cubes and pyramids, to make hit detection less expensive.

when you dont use hitscan weapons and actually have to simulate projectiles and do collision detection on them, shit becomes even more taxing.

>surprised its so taxing to calculate complex hitboxes
its not. computers can easily do full bodied hit detection. they do it for gameplay reasons

latency always exists you utter mong

its also worth noting that the best actual latency you tend to see from ny to la is around 40ms. so not that far from theoreticals

But it's also a gameplay thing because proper collision models feel like crap to shoot.

Space partitioning makes it plenty fast, but it sucks.

128tic servers on valve games are almost perfect, csgo and tf2 both have extremely well done netcode, the problem is the only 128tic servers are all private, official valve servers are shit

also, a lot of people don't realize there are a lot of client side network settings you have to do yourself for a good experience, and the source engine is one of the only ones that lets the user tweak these settings, so that contributes a lot to the problems
the vast majority of what people call rubber banding is because 99% of games don't let you change your interp setting and force 200-300ms lerp, whereas in tf2 and csgo you can get that down to 10ms with a simple cvar in the console, there are several other things you have to set yourself but that's just a quick example

client side network tweakers are nearly as bad as audiophiles

Oh please, TF2 lags all over the place on my 80/20mbps connection and has done ever since the orange box came out. And yes, I really have tried everything.

>computers can easily do full bodied hit detection

yeah they can
however when you got 10+ people on a server, and they use automatic weapons that go through half a magazine in under a second, then your server has to compute a LOT of shit in a very short amount of time, which could either cause server lag or force the server to lower the tick rate to free up resources. both options negatively impact gameplay.

using simplified cube-stacks for hit detection instead of doing it on a model made of thousands and thousands of polygons frees up an immense amount of computational power, which is absolutely necessary to run a server with large amounts of players on a high tick rate.

no you're vastly overestimating how expensive it is, a computer could do it with a hundred players without even breaking a sweat

You want to minimize useless bullshit if you do it 128 times per second.

Google how new CS:GO hitcapsules look.

there are a million things that could be causing that that have nothing to do with tf2s netcode

the one weakness the source engine has is that certain wifi connections have horrible packet loss, other than that it's near perfect if you have the proper client settings and are on a 128tic server

a lot of the settings are extremely small improvements that most won't notice, but the difference between 10ms lerp and 200ms lerp is insane, after a decade of 10ms lerp anything over 100 feels absolutely unplayable to me and I notice it instantly

You act like nothing can be done to reduce the impact of latency you're fucking retarded.

yeah you do, but dont start talking shit about it being absolutely necessary and freeing up an immense amount of computational power
it'd be easy to make a game use real hit meshes, but there's little reason to apart from appeasing autists complaining about hitboxes, I mean latency is always going to be a thing anyway you'll never have full accurate collisions

you can literally shoot people around corners with projectiles because of how giant the hitboxes are

>tfw you live 2 hours south of dallas so you ping under 15 to most servers on pc games

what is being lerped that you can control on the client exactly? enemy positions?

You think I'm using wifi when sat at the end of a 80/20 pipe?

Anyway, I notice literally no difference versus when I was on a 1mbps connection. I can set rate to whatever and I still get some amount of choke. I can take the cmdrate and updaterate to the full extents of their range and still get choke.

TF2 lags just like CS:S, except it's worse yet because there is latency before projectiles appear in the world (which is necessarily inherited from Quake). Flawed netcode.

then you're a filthy normie that has never played a game competitively in your life
play csgo on a 128tic server with 12 ping then play a game of overwatch and tell me overwatch feels "absolutely fine"

>using csgo as an example of good netcode
aaaaaaaaaaaaaaaaaaaaaaaaaaa

How is he holding it.

By the hilt, obviously

Is rubberbanding still word used in gaming world?
Modern Online games ruin whole meduim(thats just my opinion).
As far as hitboxes goes, old Unreal Tournaments & Quake3 Arena were flawless(considering 56k was still a thing).

each game has its own netcode so I can't say for everything, but 95% of my shooter experience on PC has been half life engine and source engine games, and in those almost everything is lerped, normally you can only change the interp cvar on private servers, valve matchmaking servers force it to 1 I think, whatever setting gives you 100ms lerp

on private servers it depends on the server settings, but any server used for competitive play will let players set it to the minimum, which is 10ms for the source engine

as for other games like overwatch and console shooters, I'm pretty sure they just force it a universal setting for everyone and users don't even have the option to change it

csgo on a 128tic server with the right settings is amazing, if you have a problem with it then it's on your end

csgo is legendary for not having hitboxes reflect the actions of the players. it could be fixed now, but i don't care.

the only time I see rubber banding these days is shitty console shooter servers, because they're like 16tic and throttle everyone's connection to match the slowest brazilian because console shooter devs can't write netcode to save their lives

they fixed that before the game even released, that was only an issue in beta

He shot his gauntlet for a ghost gauntlet. Geez user this is basic ghost fighting techniques 101

we arent talking about reducing the impact with predictive measurements that are often inaccurate. we are talking about actual latency existing, which it does. reread the original claim.

What the fuck is your stupid ass even doing?

> he thinks SQL transactions are involved in game server netcode
hilarious, do you have a demonstration of how you believe this works?