Just saw this video, and wanted to post this. See that you posted this 23min ago :D Question, since you have to use other bind methods (alias and cfg execs) in cs2. Could this possibly not create a level of 'lag'/delay with the testing? Would it not have been better to use the same bind method in CSGO to see if it works the same there?
I was just about to correct you on that one, it doesn't really matter though, your theory might be close to what's actually happening though. I'm sure valve's servers are powerful enough to handle the load, BUT we're currently in beta testing which means valve are logging absolutely everything happening all the time which puts a ton of strain on the servers. The subtick system would be amazing for lan or very low ping games 5-10 on all players, higher ping makes it a mess though, I surely hope they have an idea of how to compensate that. Great video.
the only 2 flaws that ive found were about: 1 subtick partially uses 64 tick rate for things that are less demanding live grenade trajectory, that why on 128 tick even on cs2 lineups are diferente. and 2 that the sub tick need to register every movement that ur mouse does, it doenst need to process every thing, just the position of your cursor on the fire.
Valve never said subtick sends server requests dynamically, and it would be pretty hard to implement something like that because the server won't have a good trigger for when it updates the clients, since it has to update them all. It's more likely subtick is just timestamped event data being sent on a fixed tick rate so teh server can theoretically simulate the events more accurately
I thought the client sent the tickrate right away, that's the point, not because the shots were not accurately registered in time but the objective is the faster validation and display back to the client, so we're supposed to see less missed shots that were displayed locally (happens often) or maybe as often but the false impact will disappear faster.
so subtick is only a minor deconfliction improvement but if the mouse pos is not updated as often, it's limited in its positive effects, maybe it shouldn't have been inflated by the valve communications @@cicolas_nage
correction: if i understood correctly, subtick doesn't stress servers any more than normal tickrate as it just registers all inputs between ticks and executes them at that place and moment when the next tick comes instead of waiting for the next tick to execute all of that (for example, if in between ticks 1 and 2, a player was in the center of your crosshair and you fired, but then moves out of your aim when tick 2 arrives, subtick would know that and register it as a hit instead of a miss)
I don't think this is a server stress problem, but instead, a display problem - human baseline reaction time is around 10milisecond, which actually coincides with the 15ms 64 tick rate in CSGO, that is why it feels natural to most players when they flick shot timing, or even spraying because it's only after we see and register in our brain (which doesn't have sub tick lol ;) ) ak-47 start to shoot sideways that we adjust by a counter-motion using a mouse to spray control. But that doesn't explain why the macro this guy used to stop and shoot at the same time will cause the crosshair to not be on the shothole - but you can apply a similar theory, most gaming keyboards have an input latency of 1-2milisecond, and the receiver (PC) + loading into CS2 software will up this to around 5-15 ms depending on your specs. Since this issue doesn't always occur (sometimes the crosshair is on the shothole), in science, we call this unreliable, so there 'must' be a random error that is causing a disconnect between what "you see and what you get", and I suspect that sub-tick is working compartmentalized for single actions such as shooting (aim) or movement keys, but not calibrated to work well when together. You might argue that sub-tick shouldn't cause this issue (which is what Valve is insinuating), but from a practical perspective counting units such as ticks is incredibly important when trying to achieve replicable results (which is a fancy way to say reliable results). For example, if you have a ruler with millimeter units and ask 10 people to measure the length of a 4.25mm string, they are most likely going to give results such as 4mm, or 4.5mm because your measuring equipment is not accurate enough to give more decimal places, so you naturally just round to a nicer number like a half. If we do the experiment again with a ruler with a mix of units such as mm, cm, micrometer, and nanometre (to emulate how sub-tick tries to get every input in between ticks) and then ask 10 people to measure the same string, some people won't try hard to measure and just go with 4mm, some people will use the smaller units and try to get 4.2mm or even more extreme some people will attempt to go 4.2479...mm, now you have a wider array of numbers as results which represents the inconsistency that you feel, and amplify that with so many actions that you are doing in CS2, it's more likely than not that errors will occur - and reasonable that we have this problem.
Makes so much sense, I feel like something has been fundamentally wrong since the beginning, awping feels off with flicks and so does spray control with rifles, good video man! To be fair I think this does make it unplayable to a degree, with peek advantage being so heavy at the moment everyone is trying to wiggle peek or re peek for advantage so most of us are moving, if you sit and try to hold an angle half the time you will die simply from behind a wall without seeing the other player.
literally. i've gone from hitting the majority of my flicks, to the bullet always going to the left/right of a player. It is actually game breaking. Raw accel also doesn't see to work so my sens is fixed unlike in csgo
Awp feels of because in CSGO when you flicked the bullet would slingshot to the position you flicked after your shot, like a string attached to the bullet and cursor, in CS2 the awp shots perfect, to perfect. the bullet will land exactly where you shot and not travel with your flick cursor thats why people missing a lot now. As a awp main i noticed that too that i sometimes shot like silver 1 in CS2
@@ownagebox how can you say in ''cs2 the awp shots perfect, to perfec, the bullet will land exactly where you shot'' When the video, CLEARLY, shows that it doesnt
@@cuppavids you can clearly see hes always barely moving after the shot in CS2 most visible with awp and auto the bullet holes are exactly where he triggers not where the cursor ends. I had some other article about this with a were the shots where scripted points, maybe i can find it
Question is, WHEN do you click? In CS2, when you click, it remembers the exact position/rotation and that's where your bullet starts of from. In CSGO, your click is queued for the next tick and THEN it looks at rotation etc, so there is a random delay between clicking and the click taking effect. Therefore it's possible in CSGO to click->move mouse and the shot fires from the new position and not from where you clicked. In CS2 it will fire from where you clicked, being more precise. What you observe right now, is that the animation is still tick and not subtick dependend so it fires on the next tick, while the actual shot got prepared on the subtick precision and therefore it feels off. The click for the shot probably happens a few moments before the rotation ends. In CSGO everything was on tick so it was executed when the rotation ends. In CS2 it's now "off" because the action is registered more precicely.
This is correct actually. It makes sense. It was actually the animation and the scope that stopped later, rather than the shot being off or registered earlier. So even if it seems off, it’s actually more accurate. It’s kinda confusing :))
This is all because of the subtick system, in csgo there is server compensation but in cs2 it registers 2 things, where shot started, where shot should end, immediately due to subtick. For me personally this feels really bad. After playing cs2 for a week i went on to play csgo again and god the shooting felt like a breeze of cold wind on a sunny day.
I wanted free leetify premium for a day, so I changed my tag within csgo and had to play a game. God damn is it such a crisp game when you put it up next to cs2. I'm really enjoying cs2. I can see the promise in it, but it needs so much work. I think Valve has no choice but to delay or fix this fast.
Exactly, for some reason the shooting in CS2 feels just weird and I could tell that almost immediately. Then I tried to go back to CSGO and man it feels so much better.
@@Kylo_tonny I think it will be many fixes over the next year. I think we get a slight delay while they iron out the worst of it, but this is a new CS game on a new engine. It'll probably take time to fix like csgo did.
Be careful with the assumption that your mouse movement input is sent to the server immediately. The subtick system just receives the input with a timestamp, indicating how far into the tick the action happened. Games can handle high polling rates, but for online games they have to linearly interpolate between frames to keep Internet traffic of inputs at a manageable level. Source: I've been programming 2d and 3d applications for about 6 years now and I'm working on a small game engine of my own. Valve will come up with a solution for this. Since this is not an easy problem to fix, it might take them some time, but they will come up with an acceptable solution
Is this even an problem lol? People got used to press fire a dozen miliseconds before stopping crosshair at the target, and the game registered it at the end of the tick. Simply muscle memory issue. Now it shows off as people's shots appear before the crosshair stop. Edit: Shots are still processed when the tick ends, it's that now they appear accordingly to input time. Maybe they make a line between 2 crosshair positions, and, let's say if you pressed at 0,5s after previous tick, shot will appear between 2 positions.
I am a game developer and what you’re saying it sort of right. There are not thousands of inputs being sent to the server when you move your mouse. As that would be a big security and performance issue, your mouse input is very likely all managed on the client. It’s when you shoot, thats what being sent to the server. The only thing that your mouse replicates on the sever is probably the players look direction like head and body. As for the delayed shot, its one of two reasons. The sub tick rate or the anticheat. Since your input is instantly sent on the server when you shoot without waiting for the tickrate, the server will trigger that shot instantly at where your character is aiming on the SERVER. And as you know the client is always ahead of the server, this can cause your bullet to look delayed. This is just an assumption and i think he should test this on lower ping and higher ping to see if the delay difference, if theres a larger distance on higher ping then my theory is correct. However if not then it could be the anticheat or another technique on the server.
I think that bullet goes exactly where You clicked Your mouse. Animation and console command processing probably is still in ticked system. This way bullet goes where you shot, but the unscope animation and stop rotation are still processed on the end of tick. This is why it's sometimes close to the crosshair (You shoot and send stop rotation command close to the end of tick, so You stop rotation almost immediately) or really far away (you shoot on the beginning of tick and move till the end of it, so You have almost 1 tick of discrepancy (15ms) between shot and stop). This is also why it feels of, we have lagless shooting, where bullet's can go back in time to be resolved, cause they have timestamps, but rest of the game - is still bound to the ticked system. Where in CS:GO - everything was resolved on the end of the tick. And as I always say about those things - in games we can pretty easily accommodate to latency as long as it's stable (that's why we can play games in 30fps as long as it's locked, and we get new frames in stable manner) - but when latency starts to fluctuate - it fucks our gameplay. And that's why game feels bad when it dropps frames from 300 to 150 (where on stable 150 it would be playable AF for anyone after some accommodation period) and this is why shooting feels of as You can't accommodate for the lag that fluctuates between 0 and 15 ms.
yeah this is what it is and why spraying feels a bit off. the shot is perfectly on your crosshair at the time you shoot because that’s the whole point of sub tick the animations are just a tiny bit delayed. still feels jank because it’s out of sync but i feel like videos like this should have almost any amount of research behind them lmao
If you have so much knowledge why aren't you a game developer? What you're saying sounds like a bunch of word salad, he tested on CS:GO first for a reason to show that 64 tick is more stable using this option.
@@scambaitspidey3470 he is literally right, 64 tick system wasnt more stable, it just sent everything a the end of the tick, making it so that if you pressed in the middle of a tick, the bullet would go where the crosshair was at the end of the tick, because this state was the one sent to the server.
Cs2 is a mess and valve doesnt have any competent developers working on anything but dota and steam. Theyll kill this game if they keep going like this
The reason this feels so wrong is because Sub-Tick is working properly, you DID actually click at those locations but the kicker is your gun didnt shoot till the actual tick update making your shot look delayed to where your bullets are actually going.
no, the point is that the animation of the shot and tracer happens on the next tick. wheres your shot calculation is sub-tick. So it will look like you're shooting later than you actually are, and later than when the server calculates your shot. hence, the actual shot is more accurate than the animation and tracer. @@PultsMoizer
@@PultsMoizer its not, there are other videos that perfectly show that where you shoot is pretty much where the bullet lands, the animation is just very delayed
Actually the bullets hit right where you click them thanks to subtick but the visual representation of it is only rendered at the next tick. That is the reason why they are in sync in CS:GO but in fact the shot in CS:GO is delayed until the next tick and NOT hitting the spot where you clicked. This is also the reason 128 tick in CSGO feels so much better than 64 because ticks are only half as long and your shots are then hitting closer to where you actually clicked. With subtick though it registers at that exact same moment of the click but the shooting animation comes 1-16ms after depending where you shot between the ticks and the bullet hole is represented where it is according to the subtick/where you clicked. This means it would still yield you a kill if you clicked on the enemys head even though at the time of the tick the enemy might already be away from your crosshair. When you understand how the engine works in CS2 this is actually a good thing. They just need to adjust the visual part of it to match better with the client side actions.
The red and blue are where the client and server thought the shots hit because he is shooting and stopping instantly thats where he was aiming at the time and yet the shots are not going in the same place so if it is just a visual representation and not shots lagging behind why when he is stopping spinning are the shots still behind?
I think your theory is right. Assume you have 0 ping and say you shoot at exactly the start of second 1 of the server's time, the mouse updates 1000 times a second, therefore, the server receives the shot at 1/1000 of second 1 but cannot send it back until at least 16/1000 of second 1 because the server is still running on 64 tick. Introduce 30+ ping and you're now waiting at least 30+30+16=76ms until your shot is sent back from the server. Do this same test with 100 or 200 ping and the distance from the shot on the wall to the crosshair should be even greater. If someone did that then it would confirm this theory. In CS:GO this shouldn't be a problem since the client and server both run on 64 tick, so they know exactly when the shot happened and if they agree, then that's it, it just gets received the same way it was sent out.
@@Tricker1266 Actually in CSGO the shot will just completely miss solely due to the delay of 64 tick when ever the click has happened just after a tick.
Quick fyi on how source/S2 networking works- the game is “simulated” at whatever tickrate it has (which is why nade lineups were different in GO). The game also sends packets every time it simulates. The current working theory for subtick is that every input is timestamped as best as possible relative to whatever frames it happens between. Additionally, chances are, they’re not recording each louse movement. They just check your mouse movement every frame and send that to the server, and then also log your mouse movement every time a shot happens, so that the shot isn’t misaligned.
It occurred to me that in cs2 i miss a lot of flicks, most of the time when it happened on cs:go I could recognize that I simply missed it, while on cs2 it happens VERY often that I genuinely say "wtf?" Since I'm pretty sure I should've hit that But I'm always too unsure to say "It's cause of new bs valve made" and most of the time I simply justify it with "I'm bad bruh"
@@StarComet7 From my understanding it's not exactly like that but feel free to correct me if I'm wrong; don't want to go around spreading misinformation From what I understood if we were to take 1 singular tick and divide it in 100 fragments, no matter if you shot it in the 64th or 99th of a tick, it will be registered as if you shot it at the very start of a tick (1/100) So (impossibile example but exaggerating things help out understanding) if I were to move my crosshair 500 pixel away from the registered position where my crosshair previously was, then my shot would be registered as 500 pixel before my current position Again, I'm not 100% sure I understood correctly so if I'm wrong please correct me
@@whitinnn what you described is actually how csgo calculates. In cs2 the tick would carry the timestamp of you click and the server would register a better guess of you crosshair position inside the tick.
This video is great! I do have one comment. I could be wrong but I think your reasoning is a bit off. Cs2 servers are still 64 tick. Originally in csgo your inputs would be registered every tick. The subtick system in cs2 now LOCALLY saves your inputs between ticks with timestamps. When your computer communicates with the server on the next tick, it will register your shot to where you ACTUALLY clicked the mouse because it has a timestamp associated with the input. The important thing about this system is this: MOVEMENT IS NOT INCLUDED IN THE SUBTICK SYSTEM. The reason it's not included is simple. Visuals are updated locally before they are updated to the server. If it wasn't, it would be the equivalent of playing the game in 64 fps. If visuals were included in the subtick system, your crosshair would flick back to where your bullet was shot on the next tick. This is a glaring fundamental flaw with the subtick system. I think it is the cause of this problem along with other issues in CS2 including the "Peek advantage" issues. Overall, I'm not certain of this theory but I am pretty confident. If this is correct, the subtick system is working as planned... it was just a bad plan.
I saw a video that was posted on Reddit recently where someone made a macro that would move the mouse to one side and then in less than 1 frame would flick to a target, fire, then move off the target (Sub-frame mouse input, or why flicks feel different in CS2 by MrPyber). In CSGO the shot would be fired at the same location as the next rendered frame, but in CS2 the shot was fired exactly where the mouse was when the input was received. I can't claim to know what's going on here, but it could be that while shooting is recorded at the exact ms the input is received, the visual updates like unscoping don't occur until the next frame. As for the spin test I don't have a guess. As far as how many mouse inputs CS2 needs to handle, USB mice can only send inputs to Windows at up to 1000hz independent of the DPI. Some USB mice can "pretend" to be multiple devices in Windows and feature 4000hz polling or more, but the highest I've seen is 8000hz. Maybe now that CS2's mouse inputs aren't capped at your frame rate, ultra high polling mice will explode in popularity.
I think this is arguably the best explanation as to why this is a thing. The subtick system works in-between frames - and as a result even with the most precise of slow motion recording - itll still seem like it didnt hit the same frame granted it tracks the time you hit your mouse without account for frames. If your mouse's polling rate is moved down to your monitor refresh rate (144hz, 60hz, etc) you'd likely see the bullets hitting in their exact positions similar to CSGO. Could be wrong though.
I keep my mouse at 2000 MHz polling rate, but shooting still feels off...not to mention spraying is the worst thing in the game atm. Will try 4k polling rate to see if it feels better
This is the real explanation. It isn't that the shooting uses the mouse position form the last tick but that the unscoping/movement stopping only happens on the next tick. I.e. the movement stops late which makes it look like the shot hit too early
He probably sends mouse events using native code to the window. Instead of moving the mouse along a spline which sends a mouse event for every coordinate on the spline he hops the mouse from coordinate to coordinate. hop(500, 500), leftclick(), hop(1000, 1000) for instance. Doing this you can basically do all the 3 actions in a time frame of 1ms. I thought CS2 used timestamps which normally uses nanosecond precision. Or maybe Valve is cheap on the data traffic and uses millisecond precision since 64 tick is 15,625ms/tick .
The worse your ping the more you see and feel this effect. Even now. You can still see this problem in community servers that you have 60 or higher ping in, joining a closer one feels drastically different
wow thx, this really explain how my awp shot somehow missed with me flicking right on the body. I was so confused since I missed like 2 shot that feels like should have hit, to the point I just stopped picking up the awp
I guess it makes sense. I'm one of those guys that when I pick up the awp I flick crazy fast like I'm trying to break my wrist, and I've struggled adjusting to cs2. My other friends who use a more passive / slow style rarely flick fast and have much more easily adjusted to it.
Same, I'm not flicking all the time, but I do flick, I need it for no scopes for example or if the enemy wasn't in the side I thought he was then I flick to shot him.
well the reality was that while subtick registers your mouse click at the exact moment you clicked and sends it to the server, it has to wait until the next tick arrives (yeah servers still run at 64 tick) and server confirms that the shot was made. because the other person you shot may have shot you before you click. if this confirmation was removed, you would see that the animation happens and player gets killed then the next tick arrives and server denies your click request because other player clicked before you did and all of a sudden you die even when it looks like the other person should i still think my guess was really great, even though *its not correct:* about how the game works, actually tickrate is still 64, but on each packet sent to the server, client attaches a timestamp with the packet. when server has to do a tick, server reads each packet and merges them in order accordingly to the timestamp they have. after this, server sends the data to all player on that tick. i agree on you that this is about subtick system because as you said, the distance from the real shot and the correct shot is random, its probably not random and probably about how far you are to a game tick. my guess is that server merges packets in the wrong order even though i make games i don't work at valve and just making an educated guess*
bro! the fact that this is your first youtube video, it is doing so good! the content you provide is better then most. keep up the good work and let this channel grow huge! and thank you for making this video, it explains why so many of my awp flicks miss😂
I've seen someone that explained that in csgo the bullet goes to where your crosshair is when the animation happens. But, in cs2 the bullet goes to where was your crosshair at the moment you clicked your mouse. Im not a game developer either and no absolutely nothing but i think its a fair summary.
the DPI part is wrong to an extend. what you're saying is right but its not the DPI that determines how many inputs/sec your mouse sends. Its the hz of the mouse. The DPI determines how many counts (also called counts per inch) your sensor does when you move your mouse. the updaterate which sends this data to your computer is hz. so for a 1000hz mouse that is 1000 updates/second and it doesn't matter whether you move your mouse or not. the data will get sent anyways. i personally use a 8k hz mouse and 8k hz in cs2 feels a lot more reliable than 1000hz is all i can say. this is most likely because the 8k hz sends the data earlier (because sending more often/sec = sending earlier to the actual position of the mouse). To sum it up: DPI determines how finely-grained the mouse's sensitivity to movement is, the polling rate measures how often the mouse checks its sensors for changes.
no he's as right as you in a way, the mouse do not sent an move event when not moving, the maximum rate of change for the mouse is when the sensor detect any movement at each mouse refresh tick, if some tick don't contain any movement, there is no event. So the refresh rate is the maximum event rate, and the actual rate depends partially on the DPI and movement speed of the mouse as well as the resolution, I'm sure cs or the mouse driver don't consider a mouse position change until it's bigger than a pixel. What is wrong here is the scale of the argument, imagine you move half of the screen in one sec (approx 1000 px for full hd) you already max out the fastest mouse out there (1000 Hz), and a half screen flick is not uncommon and often way faster than 1 sec, so most of the quick movements will absolutely max out the refresh rate. If you play 800 600 and you have the speed of a snail, then the mouse events will be a lot less frequent.
I was a pro player in a former lifetime (meaning 20 years ago) but I still dabble once in a while. My only way of awp'ing was flicking, and I could usually pwn a DM server easily in GO after a little warmup. Now I cannot hit with awp/scout AT ALL. I hope this get fixed so I don't have to change my playstyle at an age of 41.
Easily my fav part of the video is the humility and reminding everyone that this isn’t some kind of confirmation of fact, and just a test. Love to see that
hello, game developer here. Just a quick note: what you see on your screen, and what the server gets told, processes, and then sends to other players isn't always the same thing. Also, tick dependent shooting has it's flaws. in CS:GO what is on your crosshair when you shoot and what is at your crosshair on the next tick isn't always the same, so if you shoot and move the crosshair the server may think you missed when you actually hit. While with sub-tick stuff it notes where you looked when you shot so it knows exactly where you were aiming. The bullet hole being off the crosshair is weird and should be looked into, but it may also just be client-side (what you see) and it is correct server-side (what the server gets told and then others see) What I believe is most likely is that player's muscle memory got used to the slight delay in CS:GO and are now missing shots in CS2 because the delay isn't there. Though it is completely possible that valve messed up. PS: It doesn't send the click immediately, it just stores the current angle, position, timestamp, etc and that just gets sent to the server on the next tick. It's still sub-tick but it doesn't send for example 100 packets if you click 100 times between 2 ticks, it just sends it once (or maybe a few ones depending on the max size of the packet) PPS: it doesn't send every movement you do, it sends only some, the current position on each tick. Then the game for other players just interpolate between you last position and your current one. (though it does send more position stuff if you shoot, etc, as mentioned above)
for anyone wondering this still happens, the amount of times i shoot and see the bullet tracer go off to whatever angle i just flicked from is unbelievable.
as a gamedev i have to say, that the mouse dpi doenst affect the tickrate. in games you just get the mouse axis, x and y and then you rotate the player based on that value. (0f-1f)
The problem is simply that sub tick is a tick that happens in-between the slow 64 tick rate. It's never gonna be compatible because if you land a shot immediately after a server tick is gonna be 15ms delayed whilst "registering accurately" but you're already way past that in your motions. It would work more fluently if they just gave us 128 tick. At this point it's a huge step backwards
Subtick is a clientside feature. The server still operates at 64 tick. The only thing that subtick means in the end is, that it has certain actions that send updates to the server at any time, not bound to a tick. Regardless, you have a very interesting find there. I hope valve sees this video. The reason why this is happening is hard to figure out from the outside.
its the opposite, the one validating your shots with the perfect timestamps is the server, not your pc, if not, hacker would validade shots however they wanted
@@nexuhs. yes but the client will send the stamp to the server and the server will decide. but it isnt working properly. lag, performance, latency and other issues are favouring better systems and ping
It doesn't even send updates to the server at any time. It is still sent in packets every 1/64 second. It's just that some actions are now time stamped, so the server can more accurately calculate when something happened and in which order.
People have been over compensating for the tick rate. Being able to flick, click, then rest on the target before the bullet actually fires. The reason your spin and shot don’t line up is the buffer process of the code since you can’t dual bind a key.
CS2 still runs on ticks, you can verify this because the game will send 2 data packets to the server 64 times per second. That makes 128 packets which they need to send because the subtick system really just adds metadata to each event, the metadata makes the packet too large for a single packet so two are sent. It could still be related to the subtick system but not in the way you suggest.
Awesome video. You showcased the bullet inaccuracy really well, and I hope Valve looks into it. I do want to mention however that the part where you mention 160,000 updates isn't quite right. The server will only update 64 times per second. If you move your mouse 1 pixel, or 1,000 pixels, it doesn't matter. After 15.625 milliseconds the server will see your new mouse position, and your character model's view angles will be sent to all other players on that single tick. I don't know exactly how sub-tick updates work, but if I had to guess I would say all commands (or inputs) are marked by a clock independent of actual game ticks. The clock can be used to see when, within a single tick, a command was issued. For example, if the game is on let's say tick 1000 (since the map started) then the games clock would be 15.625 seconds (yes, seconds), which means tick 1001 will occur at 15.640625 seconds. Now if you send a command such as +attack in the middle of tick 1000, the clock would read 15.6328125 seconds. By marking the player's command +attack with that specific time value, when tick 1001 finally occurs, the command will go through, but not before taking that clock value into account. For example, if two players fire their weapons on tick 1000, we can determine who actually shot first by examining both of their clock values for each of their +attack commands. That would be one use case for a sub-tick clock assuming my guess is accurate. It could be used further to determine all sorts of data such as view angles between ticks, when a command occurred. It's hard to say exactly what's going on at the moment, but take my theory with a grain of salt.
Bullet inaccuracy? What? No. It's you who is inaccurate. Gone are the times when you guys rely on luck to hit flicks that doesn't make sense. Imagine in real life you pull the trigger and then you flick and that's only when it fires. Doesn't make sense right? This is the accurate way.
in fact locally on your client, the render will show the last local mouse position, if you get 300 fps and your mouse is 500 Hz, the view angle will be updated 300 times a second using a position which is max 3/5 of a frame old (I did the math at random), so the test he's doing is comparing this local position to what ever position gets recorded by the server prior to validating the shot and then displayed at the next tick when the server response gets used locally. It means that what he is showing is comparing the shift between server shots and local shots, for instance in a game, some shots seems straight, display impact, then disappear and completely missed, because the client must display something temporarily before updating with the latest info from the server. The servers send an updates every bunch of frames at the 64 tick rate, the diff between csgo and cs2 is how much the position changes. Subtick means that between ticks, the clients send more often update potentially, because it sends right away key press. Mouse events while rotating are at the max FPS (I don't think cs registers position multiple times between frames, all the code is frame based). If the subtick was triggered even for mouse events, the number of events received will be enormous (I'm all for it even then) so the cost for the servers will rise, it could improve the shift but I don't think it's even that, because csgo does better at 64 ticks without subticks, so I'm sure it's related to another difference, like the server deconfliction policies and even it could be caused by some random bugs in the new code used in cs2 (I'm sure they finally rewrote core stuff)
@@raph151515 "It means that what he is showing is comparing the shift between server shots and local shots" The problem is that he is using sv_showimpacts 1 which shows local bullet prediction as well as server prediction. If there was a shift between client and server we would see the blue and red boxes misaligned greatly every time. "Subtick means that between ticks, the clients send more often update potentially, because it sends right away key press." This doesn't appear to be the case. A TH-camr named MrMaxim performed a Wireshark analysis of game packets months ago which showed that there were no additional packets being sent. Something worth considering is that Valve cannot afford to congest the network in a fast-paced game like CS2. Fewer smaller packets is the best way to ensure low bandwidth. It's also the reason why they use UDP packets for speed. 160,000 packets per second has the potential to literally knock people offline. In MrMaxim's video, Valve's official 64-tick server transferred 64 packets per second. This is why I believe that sub-tick doesn't literally mean sending data during world update deltas.
I played a lot CS2 and wondered why my AWP flicks aren't hitting anymore. Ich changed my dpi and aim speed several times but it's still a pain to hit shots right. THEN yesterday I played CS:GO with a friend without access to cs2 and instantly my shots were accurate again. I got really nice frags again and was unhappy about cs2
brooooooo i face same prblmm everdayyy i think at everynight when i go for sleep that i am became noob then i saw ur comment and now i feel that i am wrong bcz i eliminatedd many pro playerss in cs goo cs 2 is very baddd
The more I've been playing CS2, the more I feel like subtick is a solution to a problem we didn't actually have. If they spent whatever $ it cost to develop this 'tech' on 128 tick servers everyone would be having a much better time lol
When you combine this with the input lag being 2x and the other netcode issues that are also present it does become "literally unplayable" as far as I'm concerned. The entire game feels completely off when you play it and I personally wont bother till these issues are ironed out.
idk why everybody discussing 'new cs2 problems' but i actually gets perfect glock shots VS compare to csgo where i was complete noob on first round as Terr, i still have questions for scout fast stooting but meanwhile awp is just fine for me.
@@BriJeeTVGameDev Yes tap firing is more accurate than ever in CS2 but with that also comes the horrendous downside of spraying being completely screwed up and flick shots also being screwed up. The core of the issue right now is the lag between subtick and server tick. When you fire a shot subtick instantly registers it but then the server, which is 64tick, updates like 15ms later and since the weapon animations are tied to the server tickrate it means the bullets you're shooting happen way earlier (15ms is a lot for these things) then the actual animation of the gun. Search on TH-cam "MrMaxim Is this why shooting feels off in CS2?" and you'll understand, it has a visual demonstration.
I grew up on cs 1.6, was lvl 10faceit, Radiant in Valorant, and honest to god, ive never felt like something was so off about a game like i do in cs2. I like to 1tap, it keeps me focused and stops me from going for bad sprays, and ever since CSGO came out, it was possible to do it consistently. Valorant obviously made it even easier with their upgrades to hit reg and servers, so you could legit never miss if you had the aim. But then CS2 came out, and I was so hyped to get back into the faster paced gunfights of CS, now with fixed hit regs and servers, but holy shit, the game is not just buggy, its unplayable. There is no way you can ever get a 1tap, the same shit happens to rifles as it does to the awp, I tested it out on some bot games where id flick onto their heads and shoot 1 bullet, and the bullet literally ended up where my mouse was a millisecond ago. Spraying has never been this bad, even cs 1.6 had more consisten spraying it feels like. Just disappointing how the core of the entire game will never be on the level as Valorants, cause I really do miss CS, its simpler and more fast paced, which is honestly more fun if done well, its just the fact that your shots dont go to where your aiming, or your opponents body isnt actually where its showing to you, making the bullets miss again. Its been an issue for 15years we har CS:GO, very weird how they still havent come up with a real solution whilst their competitors have.
It definitely makes sense based on my experience. And the bit at the end about your crosshair rarely moving while you shoot, I’m not so sure. A lot of times you are making microadjustments to move to your opponent’s head as you click, in a 1v1 fight
there's this "flickshot" thing... It's just stupid that the speed of your mouse matters now, esp because csgo had some of the best flickshots/qs mechanics
I'm not an expert either but the dpi I think it relates to the mouse sensor capability to scan the surface and the rate of transmitting these info relates to the polling rate of the mouse.
People probably thinking that 'follow crosshair' is to help new players with the spray. But after seeing this video i'm thinking it's because of the Sub-ticks update 😂 'follow crosshair' exists to compensate for those effects, or drawbacks for some people.
This is subtick working as intended, your shot is going where your crosshair was the instant the mouse button was pressed instead of being teleported to where your crosshair is on the next tick. This is an improvement over csgo that’s simply going to take some getting used to since it is different
It's an improvement in theory because its more "accurate" but its worse gameplay wise and noone asked for the change. People have practiced for 15 years with it the other way we dont want it different.
@@tomklie98 it’s random because him pressing the fire button is not synced with the tickrate, which is the entire point of subtick. The shot lands exactly where it’s supposed to, which it doesn’t in csgo, and then a random amount of time passes between then until the next tick, at which point he stops rotating.
@@NiceEvils "its too old now dont bother improving it" is exactly the type of thinking that prevents improvement, just get used to the new system and let the meta shift, if you cant adapt its your fault. csgo players are all about getting what you see until it doesnt benefit them
it does feel like ur right, on csgo the ssg 08 feels much snappier when moving ur crosshair around, but on cs2 you have to stop right before shooting to actually hit something
I saw something about jump-throw binds being inconsistent, but setting up an alias for them first and then binding that alias made it perfectly consistent. Do you think it's possible something like that is coming into play here? (i.e. Create an alias for shooting, then binding mouse1 to that alias)
This is actually a problem with csgo. There the info of your bullet is only registered at the next tick with the firing animation, so they are in sync. Therefore the bullet, firing animation and crosshair are all a bit too far. In cs2 the bullet info is saved immidiately and sent to the server on the next tick at the same time as the animation. This sub-tick info is then applied, so the bullet shows where it should, but the crosshair placement and animation dont have subtick so they travel too far. IMO this is better than your bullet going too far as well
You should have used the exec cfg method in csgo as well instead of aliasing. Probably doesn't matter but there could be some minor overhead on executing multiple commands via a file instead of with aliases, it's good practice to eliminate unnecessary variables.
I love how valve released a whole new game to try to fix a tiny issue that a small part of the community has been bitching about but ended up making things much worse. What was wrong with the 64 tick servers again? What did subticks improve exactly?
Subtick doesn't immediately talk to the server it sends a packet to the server just as often as the csgo servers, but that packet CONTAINS timestamps logging exactly when each action happened. This may all be a symptom of subtick, but I think people are misunderstanding that whole concept. I'm in software dev., not networking, so I could be mistaken too, but I know FOR SURE that it's not sending a packet every 1/800th of a mouse movement!
The theory about why it's happening is way off. Subtick works by having timestamp for events (like mouse click). Not by sending info to the server as soon as event happens, that would not be technically possible.
@@samsharma3216 5028 hrs deep. Now im quiting because of this stupid chit. Movement is a big deal now im realising aim is affected too. They killed cs:go
@@altaccount2179 I couldn't agree more, 3k hours been playing from 8 years. When I started playing CS2 it's like I forgot how to play at all, sure some shots connect but many don't. They've made everything worst that can be no wonder 75% of the player base has moved to something better.
this finally explains it the Awp flicks never hit and it feels so bad i only played csgo for 64hrs and still got so used to its tick rate i can't play at all in cs2 and its very frustrating
The reason why spraying in cs2 feels bad is that it is way harder than it was in csgo. In csgo every shot was processed every 128th/64th of a second at the end of a tick. That means you had 8 or even 16 miliseconds to correct your aim during spraying to land your next shot. In cs2 every shot is still processed every 64th of a second but due to subtick the shot is calculated to happen on the exact moment your gun fired. That means, you have to correct your aim every single milisecond to land your next shot. What they could do to "fix" this issue is simulating the old 128 tick, meaning every shot and it's timestamp would be pushed to either the middle or end of a tick (1/64). In other words reducing the sampling rate from 1000 to 128. Which will make spraying easier again.
Unless you almost immediately headshot your opponent or 1 tap them you’ll be constantly moving your crosshair to adjust for spray, this means that by your logic it is constantly falling behind, which could definitely cause players to miss/hit shots they would’ve or wouldn’t have hit, which in my opinion is kind of game breaking. I’ve also noticed this myself, gotten shot through a wall you can’t shoot through and shot an opponent multiple times yet i only hit them once supposedly, could just be cause i’m bad, but not that bad. Don’t know if this is just me or if this is happening to other people as well.
The fix you proposed sounds just like what sub-ticks are supposed to be. Also, we should remember that if something is simple, it would already be fixed in the first place (not really in this case since it's a beta). Also the "they did us dirty with the 64 ticks" doesn't make sense since Valve's server (not communities') is 64 ticks in csgo. I'd say that nobody in the community can make an accurate idea of what is the problem since here it's about sending/receiving data (except for the one with a degree in it I guess). It's not like the issue with spraying which is visible (the tracers aren't accurate to the actual bullets). Here it requires a deeper knowledge and we do not have the source code to understand what's happening behind. Remember that it is a beta so if there is an issue, send an email to them explaining the issue with the most precision you can.
Subtick registers the exact moment you click. Where you click is where you hit, not where your crosshair is when the next tick is loaded. So when you spin in csgo, you click and on the next tick it fires and hits where your crosshair is In cs2 when you click it remembers where and when it was, and when next tick loads it shoots where you had your aim the exact moment you clicked
I think it is sth like, when you click, the game remembers the position of your crosshair, and will shoot the bullet to that position when the server refresh(new tick), so when you click, before the new tick(the bullet is fired), you might have already moved your crosshair to a new position, and the shot is fired on the position the game remembered, so while the shot is actually perfectly accurate(fired on where it should land), your crosshair is already at a new position, making the shot not on your crosshair position, I guess you can say it as the animation can't keep up to the actual game
While it does seem that you're witnessing subtick, I dont think you're seeing what you think you are. One of the issues with subtick is that it can make the animations etc feel delayed while the bullets arent. Subtick doesn't include mouse movements unless someone can point to a reference to say otherwise. The bullets are likely going in the right place now, but because the calculated time was within the current tick where the alias is being processed, its "behind" where you're aiming. It's the same with the weird gun feelings. The shots are registered between ticks, but animation still has to happen on a tick so occasionally feels a slight delay. In general, your shots are *technically* more accurate if you put server lag aside. When the servers run well subtick is absolutely superior.
I don't think this is a server stress problem, but instead, a display problem - human baseline reaction time is around 10milisecond, which actually coincides with the 15ms 64 tick rate in CSGO, that is why it feels natural to most players when they flick shot timing, or even spraying because it's only after we see and register in our brain (which doesn't have sub tick lol ;) ) ak-47 start to shoot sideways that we adjust by a counter-motion using a mouse to spray control. But that doesn't explain why the macro this guy used to stop and shoot at the same time will cause the crosshair to not be on the shothole - but you can apply a similar theory, most gaming keyboards have an input latency of 1-2milisecond, and the receiver (PC) + loading into CS2 software will up this to around 5-15 ms depending on your specs. Since this issue doesn't always occur (sometimes the crosshair is on the shothole), in science, we call this unreliable, so there 'must' be a random error that is causing a disconnect between what "you see and what you get", and I suspect that sub-tick is working compartmentalized for single actions such as shooting (aim) or movement keys, but not calibrated to work well when together. You might argue that sub-tick shouldn't cause this issue (which is what Valve is insinuating), but from a practical perspective counting units such as ticks is incredibly important when trying to achieve replicable results (which is a fancy way to say reliable results). For example, if you have a ruler with millimeter units and ask 10 people to measure the length of a 4.25mm string, they are most likely going to give results such as 4mm, or 4.5mm because your measuring equipment is not accurate enough to give more decimal places, so you naturally just round to a nicer number like a half. If we do the experiment again with a ruler with a mix of units such as mm, cm, micrometer, and nanometre (to emulate how sub-tick tries to get every input in between ticks) and then ask 10 people to measure the same string, some people won't try hard to measure and just go with 4mm, some people will use the smaller units and try to get 4.2mm or even more extreme some people will attempt to go 4.2479...mm, now you have a wider array of numbers as results which represents the inconsistency that you feel, and amplify that with so many actions that you are doing in CS2, it's more likely than not that errors will occur - and reasonable that we have this problem.
The server does not update each time your mouse updates. It still updates at the same rate, 64 times per second. The issue is that in CSGO the game would update everything at once. So if in that single tick you move your mouse 5 pixels to the left, clicked once, and moved 5 pixels up, the game would register5 left + 5 up + fire, and the firing would always be the last thing to account for. If you were moving, fired and stopped, in the same tick, you would fire as if you stopped first. But in CS2 the server knows EXACTLY the order of events and follows it to a T. So if you move 5 pixels to the left, fire and then 5 pixels up, this is exactly what the server will calculate. In your head the bullet SHOULD go to where the crosshair is, because for the human brain everything that happened in 15ms is basically the simultaneously. But the server will still fire that bullet where you clicked. Because CS2 is still hitscan (unlike Battlefield or Warzone), the shot is fired and accounted for instantly, but the animation lags behind. In CSGO that wasn't a problem because EVERYTHING was lagging behind, but now it's uncanny. Similarly, if you walk, fire and then stop, in CSGO that was fine, the server would think you stopped before clicking, now it knows you didn't and your shot won't be accurate.
Sub-tick is basically interpolation of an action in between two ticks - specifically shooting and WASD-movement (as stated in the valve video) which, unlike mouse movements, don't have thousands of inputs per second. Let's imagine a very minuscule moment in a game where the server tick-rate is on tick#30 and proceeding to tick#31, a span of 15.625 milliseconds. The player, while quickly swiping their mouse, points at direction XYZ and clicks their mouse button 10.000ms after the instant tick#30 was processed. The game records that the player shot their weapon at that point in time, and tells the server "This player shot at direction XYZ 10.000ms after tick#30". The server will then use that information to create the effects of such an action upon the rest of the game upon the arrival of tick#31. However, there are still 5.625ms remaining until tick#31 arrives, and because your mouse is capable of inputting to your game thousands of times per second, it has moved to direction ABC during the 5.625ms remaining until tick#31 begins. So once the final results of what happened during tick#30 are processed and presented by the server, your shot will have arrived at direction XYZ but your mouse will now be facing direction ABC, which will create the exact same discrepancy showcased in this video. What this means is that in CS2, all of your shots will go in the exact direction you shot them at regardless of tick-rate, although that exact direction is so precise to the point where it's happening in-between what is perceivable by the server's tick-rate. This is different from CSGO where your shots will only go where your mouse is pointing at upon the moment a tick arrives, which means your mouse's direction is only recorded at tick#30 and tick#31 and never anywhere in-between. So CS2's sub-tick system is definitely far more precise than anything a higher tick-rate could achieve in the most literal sense, but it also means that what you're being shown on your screen is ever so slightly different from what you're inputting into the server because you now have the freedom to input actions in between what you can perceive. In fact, it's almost like CS2 has removed the tiny input lag that is present between you inputting an action and the server's tick-rate trying to follow up with your input, but everyone's used to being dependent on the tick-rate which makes the new tick-rate-independent system feel awkward as a result.
Just to add on about the proposed solution at the end of the video, making the mouse position be sent to the point of fire will probably make things feel even worse because that's essentially preventing your mouse from moving for the remainder of that tick, which means you won't be allowed to move your mouse for upwards of ~15ms in the worst case scenario and ~8ms on average. This probably won't have too much of an impact on tap-firing, but it's gonna be super noticeable during spray-downs if your mouse is being frozen in place after every shot.
Bro, thank you! I knew something was wrong. I was telling it to everyone, but even I was like, maybe I was just bad. Since the release I haven't really played because it just felt wrong.
I have given up awping for now because I had great flicks in CSGO but now nothing hits... I went back to CSGO and enjoyed awping so much. Now I understand why.
UPDATE (please read before you comment):
There is a lot of mistakes in this video. I’ve made a new one, where I correct them
Just saw this video, and wanted to post this.
See that you posted this 23min ago :D
Question, since you have to use other bind methods (alias and cfg execs) in cs2.
Could this possibly not create a level of 'lag'/delay with the testing?
Would it not have been better to use the same bind method in CSGO to see if it works the same there?
This makes a lot more sense now that you added that, because with your testing your mouse technically isnt moving.
I was just about to correct you on that one, it doesn't really matter though, your theory might be close to what's actually happening though. I'm sure valve's servers are powerful enough to handle the load, BUT we're currently in beta testing which means valve are logging absolutely everything happening all the time which puts a ton of strain on the servers. The subtick system would be amazing for lan or very low ping games 5-10 on all players, higher ping makes it a mess though, I surely hope they have an idea of how to compensate that. Great video.
Respekt kárečku.
the only 2 flaws that ive found were about: 1 subtick partially uses 64 tick rate for things that are less demanding live grenade trajectory, that why on 128 tick even on cs2 lineups are diferente. and 2 that the sub tick need to register every movement that ur mouse does, it doenst need to process every thing, just the position of your cursor on the fire.
ngl this could explain why spraying feels weird as well. Interested to see if valve do something to fix this
Yeah, I think things like this will be fixed with/before the offical release.
it's not actually a bug in CS2 its actually supposed to be like this.
Read my comment for more.
@@Renos44no way lmao, ig the game keeps like this forever its dead
@@Renos44 Your high if you think it suppose to be like this...
@@Renos44 which comment?
I KNEW I WASNT CRAZY
same. people will flame you for noticing what they're incapable of. visually i couldn't notice but i could feel something strange.
You are, wake up.
This is all your imagination.
You can See it clearly in smokes😢 bring csgo back
Valve never said subtick sends server requests dynamically, and it would be pretty hard to implement something like that because the server won't have a good trigger for when it updates the clients, since it has to update them all. It's more likely subtick is just timestamped event data being sent on a fixed tick rate so teh server can theoretically simulate the events more accurately
I thought the client sent the tickrate right away, that's the point, not because the shots were not accurately registered in time but the objective is the faster validation and display back to the client, so we're supposed to see less missed shots that were displayed locally (happens often) or maybe as often but the false impact will disappear faster.
so subtick is only a minor deconfliction improvement but if the mouse pos is not updated as often, it's limited in its positive effects, maybe it shouldn't have been inflated by the valve communications @@cicolas_nage
if that was the case it would be even more data-intensive than just having 128 tick servers@@raph151515
@@raph151515if the client sent the update right away then we'd be on an events driven update for game networking which would be ass for FPS games
Yeah, thanks for adding that. Thought the same.
correction: if i understood correctly, subtick doesn't stress servers any more than normal tickrate as it just registers all inputs between ticks and executes them at that place and moment when the next tick comes instead of waiting for the next tick to execute all of that (for example, if in between ticks 1 and 2, a player was in the center of your crosshair and you fired, but then moves out of your aim when tick 2 arrives, subtick would know that and register it as a hit instead of a miss)
Exactly that
I don't think this is a server stress problem, but instead, a display problem - human baseline reaction time is around 10milisecond, which actually coincides with the 15ms 64 tick rate in CSGO, that is why it feels natural to most players when they flick shot timing, or even spraying because it's only after we see and register in our brain (which doesn't have sub tick lol ;) ) ak-47 start to shoot sideways that we adjust by a counter-motion using a mouse to spray control. But that doesn't explain why the macro this guy used to stop and shoot at the same time will cause the crosshair to not be on the shothole - but you can apply a similar theory, most gaming keyboards have an input latency of 1-2milisecond, and the receiver (PC) + loading into CS2 software will up this to around 5-15 ms depending on your specs. Since this issue doesn't always occur (sometimes the crosshair is on the shothole), in science, we call this unreliable, so there 'must' be a random error that is causing a disconnect between what "you see and what you get", and I suspect that sub-tick is working compartmentalized for single actions such as shooting (aim) or movement keys, but not calibrated to work well when together. You might argue that sub-tick shouldn't cause this issue (which is what Valve is insinuating), but from a practical perspective counting units such as ticks is incredibly important when trying to achieve replicable results (which is a fancy way to say reliable results). For example, if you have a ruler with millimeter units and ask 10 people to measure the length of a 4.25mm string, they are most likely going to give results such as 4mm, or 4.5mm because your measuring equipment is not accurate enough to give more decimal places, so you naturally just round to a nicer number like a half. If we do the experiment again with a ruler with a mix of units such as mm, cm, micrometer, and nanometre (to emulate how sub-tick tries to get every input in between ticks) and then ask 10 people to measure the same string, some people won't try hard to measure and just go with 4mm, some people will use the smaller units and try to get 4.2mm or even more extreme some people will attempt to go 4.2479...mm, now you have a wider array of numbers as results which represents the inconsistency that you feel, and amplify that with so many actions that you are doing in CS2, it's more likely than not that errors will occur - and reasonable that we have this problem.
Makes so much sense, I feel like something has been fundamentally wrong since the beginning, awping feels off with flicks and so does spray control with rifles, good video man!
To be fair I think this does make it unplayable to a degree, with peek advantage being so heavy at the moment everyone is trying to wiggle peek or re peek for advantage so most of us are moving, if you sit and try to hold an angle half the time you will die simply from behind a wall without seeing the other player.
literally. i've gone from hitting the majority of my flicks, to the bullet always going to the left/right of a player. It is actually game breaking. Raw accel also doesn't see to work so my sens is fixed unlike in csgo
The spray difference is due to the AK having higher recoil in the first 9 bullets than in csgo
Awp feels of because in CSGO when you flicked the bullet would slingshot to the position you flicked after your shot, like a string attached to the bullet and cursor, in CS2 the awp shots perfect, to perfect. the bullet will land exactly where you shot and not travel with your flick cursor thats why people missing a lot now. As a awp main i noticed that too that i sometimes shot like silver 1 in CS2
@@ownagebox how can you say in ''cs2 the awp shots perfect, to perfec, the bullet will land exactly where you shot'' When the video, CLEARLY, shows that it doesnt
@@cuppavids you can clearly see hes always barely moving after the shot in CS2 most visible with awp and auto the bullet holes are exactly where he triggers not where the cursor ends. I had some other article about this with a were the shots where scripted points, maybe i can find it
Question is, WHEN do you click?
In CS2, when you click, it remembers the exact position/rotation and that's where your bullet starts of from. In CSGO, your click is queued for the next tick and THEN it looks at rotation etc, so there is a random delay between clicking and the click taking effect.
Therefore it's possible in CSGO to click->move mouse and the shot fires from the new position and not from where you clicked.
In CS2 it will fire from where you clicked, being more precise.
What you observe right now, is that the animation is still tick and not subtick dependend so it fires on the next tick, while the actual shot got prepared on the subtick precision and therefore it feels off. The click for the shot probably happens a few moments before the rotation ends. In CSGO everything was on tick so it was executed when the rotation ends. In CS2 it's now "off" because the action is registered more precicely.
This is correct actually. It makes sense. It was actually the animation and the scope that stopped later, rather than the shot being off or registered earlier. So even if it seems off, it’s actually more accurate. It’s kinda confusing :))
now add imput delay 7-25ms and you get fully random shots all the time. with qued ticks you can acually understand where bullet will go.
@@airbl4d3HDexactly, same shit that happens in valorant most likely. Its horrible.
@@WorldKeepsSpinninvalorant has 128 tick servers, subtick is unique to cs2, what are you on about
@@airbl4d3HDif you'd actually read the comment properly you'd understand the guy you're replying to is saying literally the opposite
This is all because of the subtick system, in csgo there is server compensation but in cs2 it registers 2 things, where shot started, where shot should end, immediately due to subtick. For me personally this feels really bad. After playing cs2 for a week i went on to play csgo again and god the shooting felt like a breeze of cold wind on a sunny day.
I wanted free leetify premium for a day, so I changed my tag within csgo and had to play a game. God damn is it such a crisp game when you put it up next to cs2. I'm really enjoying cs2. I can see the promise in it, but it needs so much work. I think Valve has no choice but to delay or fix this fast.
Exactly, for some reason the shooting in CS2 feels just weird and I could tell that almost immediately. Then I tried to go back to CSGO and man it feels so much better.
Same
this will prob get fixed at early release
@@Kylo_tonny I think it will be many fixes over the next year. I think we get a slight delay while they iron out the worst of it, but this is a new CS game on a new engine. It'll probably take time to fix like csgo did.
you put so much effort into this. thank you for the insights.
Be careful with the assumption that your mouse movement input is sent to the server immediately. The subtick system just receives the input with a timestamp, indicating how far into the tick the action happened. Games can handle high polling rates, but for online games they have to linearly interpolate between frames to keep Internet traffic of inputs at a manageable level. Source: I've been programming 2d and 3d applications for about 6 years now and I'm working on a small game engine of my own. Valve will come up with a solution for this. Since this is not an easy problem to fix, it might take them some time, but they will come up with an acceptable solution
Is this even an problem lol? People got used to press fire a dozen miliseconds before stopping crosshair at the target, and the game registered it at the end of the tick. Simply muscle memory issue. Now it shows off as people's shots appear before the crosshair stop.
Edit: Shots are still processed when the tick ends, it's that now they appear accordingly to input time. Maybe they make a line between 2 crosshair positions, and, let's say if you pressed at 0,5s after previous tick, shot will appear between 2 positions.
@fantastikboom1094 now I gotta do math while gaming
Such a small content creator but amazing video quality. Subbed and liked
yes same here! well done :D VALVE PLS FIX
I am a game developer and what you’re saying it sort of right. There are not thousands of inputs being sent to the server when you move your mouse. As that would be a big security and performance issue, your mouse input is very likely all managed on the client. It’s when you shoot, thats what being sent to the server. The only thing that your mouse replicates on the sever is probably the players look direction like head and body.
As for the delayed shot, its one of two reasons. The sub tick rate or the anticheat. Since your input is instantly sent on the server when you shoot without waiting for the tickrate, the server will trigger that shot instantly at where your character is aiming on the SERVER. And as you know the client is always ahead of the server, this can cause your bullet to look delayed. This is just an assumption and i think he should test this on lower ping and higher ping to see if the delay difference, if theres a larger distance on higher ping then my theory is correct. However if not then it could be the anticheat or another technique on the server.
Production is crazy, appreciate all the work u put into this
I think that bullet goes exactly where You clicked Your mouse. Animation and console command processing probably is still in ticked system. This way bullet goes where you shot, but the unscope animation and stop rotation are still processed on the end of tick. This is why it's sometimes close to the crosshair (You shoot and send stop rotation command close to the end of tick, so You stop rotation almost immediately) or really far away (you shoot on the beginning of tick and move till the end of it, so You have almost 1 tick of discrepancy (15ms) between shot and stop).
This is also why it feels of, we have lagless shooting, where bullet's can go back in time to be resolved, cause they have timestamps, but rest of the game - is still bound to the ticked system. Where in CS:GO - everything was resolved on the end of the tick. And as I always say about those things - in games we can pretty easily accommodate to latency as long as it's stable (that's why we can play games in 30fps as long as it's locked, and we get new frames in stable manner) - but when latency starts to fluctuate - it fucks our gameplay. And that's why game feels bad when it dropps frames from 300 to 150 (where on stable 150 it would be playable AF for anyone after some accommodation period) and this is why shooting feels of as You can't accommodate for the lag that fluctuates between 0 and 15 ms.
yeah this is what it is and why spraying feels a bit off. the shot is perfectly on your crosshair at the time you shoot because that’s the whole point of sub tick the animations are just a tiny bit delayed. still feels jank because it’s out of sync but i feel like videos like this should have almost any amount of research behind them lmao
If you have so much knowledge why aren't you a game developer? What you're saying sounds like a bunch of word salad, he tested on CS:GO first for a reason to show that 64 tick is more stable using this option.
@@someoneelse564 Why, where?
@@scambaitspidey3470 he is literally right, 64 tick system wasnt more stable, it just sent everything a the end of the tick, making it so that if you pressed in the middle of a tick, the bullet would go where the crosshair was at the end of the tick, because this state was the one sent to the server.
@@ssame9722 everyone has to be on the same page. Without a tickrate it is impossible as there will always be input lag. Unless everyone is on LAN.
Cs2 is a mess and valve doesnt have any competent developers working on anything but dota and steam. Theyll kill this game if they keep going like this
The reason this feels so wrong is because Sub-Tick is working properly, you DID actually click at those locations but the kicker is your gun didnt shoot till the actual tick update making your shot look delayed to where your bullets are actually going.
which is pretty stupid if this is the result no?
Thats wrong. If you slow down you can see where youre shooting and its not on the location. That showed at the beginning
no, the point is that the animation of the shot and tracer happens on the next tick. wheres your shot calculation is sub-tick. So it will look like you're shooting later than you actually are, and later than when the server calculates your shot. hence, the actual shot is more accurate than the animation and tracer. @@PultsMoizer
@@PultsMoizer its not, there are other videos that perfectly show that where you shoot is pretty much where the bullet lands, the animation is just very delayed
So in conclusion, the game is dogshit
Actually the bullets hit right where you click them thanks to subtick but the visual representation of it is only rendered at the next tick. That is the reason why they are in sync in CS:GO but in fact the shot in CS:GO is delayed until the next tick and NOT hitting the spot where you clicked. This is also the reason 128 tick in CSGO feels so much better than 64 because ticks are only half as long and your shots are then hitting closer to where you actually clicked. With subtick though it registers at that exact same moment of the click but the shooting animation comes 1-16ms after depending where you shot between the ticks and the bullet hole is represented where it is according to the subtick/where you clicked. This means it would still yield you a kill if you clicked on the enemys head even though at the time of the tick the enemy might already be away from your crosshair.
When you understand how the engine works in CS2 this is actually a good thing. They just need to adjust the visual part of it to match better with the client side actions.
The red and blue are where the client and server thought the shots hit because he is shooting and stopping instantly thats where he was aiming at the time and yet the shots are not going in the same place so if it is just a visual representation and not shots lagging behind why when he is stopping spinning are the shots still behind?
yes this, the shots will happen in the server in the exact point in time when you click and it won't wait till the next tick, like in csgo.
@@Tommo_ yes but the moment he clicks on client side is the moment he stops aiming and yet the shots lag behind
I think your theory is right. Assume you have 0 ping and say you shoot at exactly the start of second 1 of the server's time, the mouse updates 1000 times a second, therefore, the server receives the shot at 1/1000 of second 1 but cannot send it back until at least 16/1000 of second 1 because the server is still running on 64 tick. Introduce 30+ ping and you're now waiting at least 30+30+16=76ms until your shot is sent back from the server. Do this same test with 100 or 200 ping and the distance from the shot on the wall to the crosshair should be even greater. If someone did that then it would confirm this theory. In CS:GO this shouldn't be a problem since the client and server both run on 64 tick, so they know exactly when the shot happened and if they agree, then that's it, it just gets received the same way it was sent out.
@@Tricker1266 Actually in CSGO the shot will just completely miss solely due to the delay of 64 tick when ever the click has happened just after a tick.
Quick fyi on how source/S2 networking works- the game is “simulated” at whatever tickrate it has (which is why nade lineups were different in GO). The game also sends packets every time it simulates. The current working theory for subtick is that every input is timestamped as best as possible relative to whatever frames it happens between. Additionally, chances are, they’re not recording each louse movement. They just check your mouse movement every frame and send that to the server, and then also log your mouse movement every time a shot happens, so that the shot isn’t misaligned.
It occurred to me that in cs2 i miss a lot of flicks, most of the time when it happened on cs:go I could recognize that I simply missed it, while on cs2 it happens VERY often that I genuinely say "wtf?" Since I'm pretty sure I should've hit that
But I'm always too unsure to say "It's cause of new bs valve made" and most of the time I simply justify it with "I'm bad bruh"
That's why things like shadowplay or obs replay are a blessing, sometimes the 60FPS recording is enough to tell game's bs
You shouldn't have hit it. Shots go off position when you exactly pressed the button, but the animations plays with tick
And since skill no longer works on cs2, its fair to say all those hardcore top fraggers are actually hackers
@@StarComet7 From my understanding it's not exactly like that but feel free to correct me if I'm wrong; don't want to go around spreading misinformation
From what I understood if we were to take 1 singular tick and divide it in 100 fragments, no matter if you shot it in the 64th or 99th of a tick, it will be registered as if you shot it at the very start of a tick (1/100)
So (impossibile example but exaggerating things help out understanding) if I were to move my crosshair 500 pixel away from the registered position where my crosshair previously was, then my shot would be registered as 500 pixel before my current position
Again, I'm not 100% sure I understood correctly so if I'm wrong please correct me
@@whitinnn what you described is actually how csgo calculates. In cs2 the tick would carry the timestamp of you click and the server would register a better guess of you crosshair position inside the tick.
This video is great! I do have one comment. I could be wrong but I think your reasoning is a bit off. Cs2 servers are still 64 tick. Originally in csgo your inputs would be registered every tick. The subtick system in cs2 now LOCALLY saves your inputs between ticks with timestamps. When your computer communicates with the server on the next tick, it will register your shot to where you ACTUALLY clicked the mouse because it has a timestamp associated with the input. The important thing about this system is this: MOVEMENT IS NOT INCLUDED IN THE SUBTICK SYSTEM. The reason it's not included is simple. Visuals are updated locally before they are updated to the server. If it wasn't, it would be the equivalent of playing the game in 64 fps. If visuals were included in the subtick system, your crosshair would flick back to where your bullet was shot on the next tick.
This is a glaring fundamental flaw with the subtick system. I think it is the cause of this problem along with other issues in CS2 including the "Peek advantage" issues. Overall, I'm not certain of this theory but I am pretty confident. If this is correct, the subtick system is working as planned... it was just a bad plan.
I saw a video that was posted on Reddit recently where someone made a macro that would move the mouse to one side and then in less than 1 frame would flick to a target, fire, then move off the target (Sub-frame mouse input, or why flicks feel different in CS2 by MrPyber). In CSGO the shot would be fired at the same location as the next rendered frame, but in CS2 the shot was fired exactly where the mouse was when the input was received. I can't claim to know what's going on here, but it could be that while shooting is recorded at the exact ms the input is received, the visual updates like unscoping don't occur until the next frame. As for the spin test I don't have a guess.
As far as how many mouse inputs CS2 needs to handle, USB mice can only send inputs to Windows at up to 1000hz independent of the DPI. Some USB mice can "pretend" to be multiple devices in Windows and feature 4000hz polling or more, but the highest I've seen is 8000hz. Maybe now that CS2's mouse inputs aren't capped at your frame rate, ultra high polling mice will explode in popularity.
I think this is arguably the best explanation as to why this is a thing. The subtick system works in-between frames - and as a result even with the most precise of slow motion recording - itll still seem like it didnt hit the same frame granted it tracks the time you hit your mouse without account for frames. If your mouse's polling rate is moved down to your monitor refresh rate (144hz, 60hz, etc) you'd likely see the bullets hitting in their exact positions similar to CSGO. Could be wrong though.
I keep my mouse at 2000 MHz polling rate, but shooting still feels off...not to mention spraying is the worst thing in the game atm. Will try 4k polling rate to see if it feels better
I think the -right command is only sent on the next tick, resulting in a delay
This is the real explanation. It isn't that the shooting uses the mouse position form the last tick but that the unscoping/movement stopping only happens on the next tick. I.e. the movement stops late which makes it look like the shot hit too early
He probably sends mouse events using native code to the window. Instead of moving the mouse along a spline which sends a mouse event for every coordinate on the spline he hops the mouse from coordinate to coordinate. hop(500, 500), leftclick(), hop(1000, 1000) for instance. Doing this you can basically do all the 3 actions in a time frame of 1ms. I thought CS2 used timestamps which normally uses nanosecond precision. Or maybe Valve is cheap on the data traffic and uses millisecond precision since 64 tick is 15,625ms/tick .
This is what niko was saying if u miss your first one tap just spray,because the one tap wont register in the right spot after the first one
Still broken game
Still i am having fun
The worse your ping the more you see and feel this effect. Even now. You can still see this problem in community servers that you have 60 or higher ping in, joining a closer one feels drastically different
this game is not esports rdy
wow thx, this really explain how my awp shot somehow missed with me flicking right on the body. I was so confused since I missed like 2 shot that feels like should have hit, to the point I just stopped picking up the awp
I guess it makes sense. I'm one of those guys that when I pick up the awp I flick crazy fast like I'm trying to break my wrist, and I've struggled adjusting to cs2. My other friends who use a more passive / slow style rarely flick fast and have much more easily adjusted to it.
Yeah, the AWP was the only weapon I was noticing the bug with in game.
Same, I'm not flicking all the time, but I do flick, I need it for no scopes for example or if the enemy wasn't in the side I thought he was then I flick to shot him.
well the reality was that while subtick registers your mouse click at the exact moment you clicked and sends it to the server, it has to wait until the next tick arrives (yeah servers still run at 64 tick) and server confirms that the shot was made. because the other person you shot may have shot you before you click. if this confirmation was removed, you would see that the animation happens and player gets killed then the next tick arrives and server denies your click request because other player clicked before you did and all of a sudden you die even when it looks like the other person should
i still think my guess was really great, even though *its not correct:*
about how the game works, actually tickrate is still 64, but on each packet sent to the server, client attaches a timestamp with the packet. when server has to do a tick, server reads each packet and merges them in order accordingly to the timestamp they have. after this, server sends the data to all player on that tick. i agree on you that this is about subtick system because as you said, the distance from the real shot and the correct shot is random, its probably not random and probably about how far you are to a game tick. my guess is that server merges packets in the wrong order
even though i make games i don't work at valve and just making an educated guess*
bro! the fact that this is your first youtube video, it is doing so good! the content you provide is better then most. keep up the good work and let this channel grow huge! and thank you for making this video, it explains why so many of my awp flicks miss😂
I've seen someone that explained that in csgo the bullet goes to where your crosshair is when the animation happens. But, in cs2 the bullet goes to where was your crosshair at the moment you clicked your mouse. Im not a game developer either and no absolutely nothing but i think its a fair summary.
the DPI part is wrong to an extend. what you're saying is right but its not the DPI that determines how many inputs/sec your mouse sends. Its the hz of the mouse. The DPI determines how many counts (also called counts per inch) your sensor does when you move your mouse. the updaterate which sends this data to your computer is hz. so for a 1000hz mouse that is 1000 updates/second and it doesn't matter whether you move your mouse or not. the data will get sent anyways. i personally use a 8k hz mouse and 8k hz in cs2 feels a lot more reliable than 1000hz is all i can say. this is most likely because the 8k hz sends the data earlier (because sending more often/sec = sending earlier to the actual position of the mouse). To sum it up: DPI determines how finely-grained the mouse's sensitivity to movement is, the polling rate measures how often the mouse checks its sensors for changes.
no he's as right as you in a way, the mouse do not sent an move event when not moving, the maximum rate of change for the mouse is when the sensor detect any movement at each mouse refresh tick, if some tick don't contain any movement, there is no event. So the refresh rate is the maximum event rate, and the actual rate depends partially on the DPI and movement speed of the mouse as well as the resolution, I'm sure cs or the mouse driver don't consider a mouse position change until it's bigger than a pixel. What is wrong here is the scale of the argument, imagine you move half of the screen in one sec (approx 1000 px for full hd) you already max out the fastest mouse out there (1000 Hz), and a half screen flick is not uncommon and often way faster than 1 sec, so most of the quick movements will absolutely max out the refresh rate. If you play 800 600 and you have the speed of a snail, then the mouse events will be a lot less frequent.
its because what we do is registered in subtic, but what we see on our screens is still 64 ticks. Visuals on CS2 are still on 64 ticks
I was a pro player in a former lifetime (meaning 20 years ago) but I still dabble once in a while. My only way of awp'ing was flicking, and I could usually pwn a DM server easily in GO after a little warmup. Now I cannot hit with awp/scout AT ALL. I hope this get fixed so I don't have to change my playstyle at an age of 41.
yepppp im thinkinh about quiting, awp main here 5000 hours
now i know why i get more headshots, cause i always flick past the head
This video needs millions of views. Share this to everybody so valve can understand! Great work!
Easily my fav part of the video is the humility and reminding everyone that this isn’t some kind of confirmation of fact, and just a test. Love to see that
Thank you for making this.
Thank YOU for this comment. It feels great having positive feedback like this.
hello, game developer here. Just a quick note: what you see on your screen, and what the server gets told, processes, and then sends to other players isn't always the same thing.
Also, tick dependent shooting has it's flaws. in CS:GO what is on your crosshair when you shoot and what is at your crosshair on the next tick isn't always the same, so if you shoot and move the crosshair the server may think you missed when you actually hit. While with sub-tick stuff it notes where you looked when you shot so it knows exactly where you were aiming.
The bullet hole being off the crosshair is weird and should be looked into, but it may also just be client-side (what you see) and it is correct server-side (what the server gets told and then others see)
What I believe is most likely is that player's muscle memory got used to the slight delay in CS:GO and are now missing shots in CS2 because the delay isn't there. Though it is completely possible that valve messed up.
PS:
It doesn't send the click immediately, it just stores the current angle, position, timestamp, etc and that just gets sent to the server on the next tick.
It's still sub-tick but it doesn't send for example 100 packets if you click 100 times between 2 ticks, it just sends it once (or maybe a few ones depending on the max size of the packet)
PPS:
it doesn't send every movement you do, it sends only some, the current position on each tick. Then the game for other players just interpolate between you last position and your current one. (though it does send more position stuff if you shoot, etc, as mentioned above)
This needs more views!! Been noticing this with spraying as well. Awesome video!
for anyone wondering this still happens, the amount of times i shoot and see the bullet tracer go off to whatever angle i just flicked from is unbelievable.
10 months after and nothing has changed😭😭😭 I uninstalled the game they don't deserve my time and my money neither
as a gamedev i have to say, that the mouse dpi doenst affect the tickrate. in games you just get the mouse axis, x and y and then you rotate the player based on that value. (0f-1f)
They pre-nerfed the game because of KennyS yet again.
The problem is simply that sub tick is a tick that happens in-between the slow 64 tick rate. It's never gonna be compatible because if you land a shot immediately after a server tick is gonna be 15ms delayed whilst "registering accurately" but you're already way past that in your motions. It would work more fluently if they just gave us 128 tick. At this point it's a huge step backwards
Subtick is a clientside feature. The server still operates at 64 tick. The only thing that subtick means in the end is, that it has certain actions that send updates to the server at any time, not bound to a tick. Regardless, you have a very interesting find there. I hope valve sees this video. The reason why this is happening is hard to figure out from the outside.
Do you know this for sure? I thought sub tick would mean the server uses an event-based queue with timestamp instead of polling ticks.
its the opposite, the one validating your shots with the perfect timestamps is the server, not your pc, if not, hacker would validade shots however they wanted
@@nexuhs. yes but the client will send the stamp to the server and the server will decide. but it isnt working properly. lag, performance, latency and other issues are favouring better systems and ping
@@bay0r exactly. Its also worth mentioning that there are still certain activities that are strictly tick bound.
It doesn't even send updates to the server at any time. It is still sent in packets every 1/64 second. It's just that some actions are now time stamped, so the server can more accurately calculate when something happened and in which order.
People have been over compensating for the tick rate. Being able to flick, click, then rest on the target before the bullet actually fires. The reason your spin and shot don’t line up is the buffer process of the code since you can’t dual bind a key.
still not fixed. game is unplayable
And it makes sense more because they intentionally hardcoded their tick, no chance to change to 64 or 128
CS2 still runs on ticks, you can verify this because the game will send 2 data packets to the server 64 times per second. That makes 128 packets which they need to send because the subtick system really just adds metadata to each event, the metadata makes the packet too large for a single packet so two are sent. It could still be related to the subtick system but not in the way you suggest.
That could explain my whole experience with shooting in CS2. Told a friend, aiming feels like playing roulette every single shot.
Awesome video. You showcased the bullet inaccuracy really well, and I hope Valve looks into it. I do want to mention however that the part where you mention 160,000 updates isn't quite right. The server will only update 64 times per second. If you move your mouse 1 pixel, or 1,000 pixels, it doesn't matter. After 15.625 milliseconds the server will see your new mouse position, and your character model's view angles will be sent to all other players on that single tick. I don't know exactly how sub-tick updates work, but if I had to guess I would say all commands (or inputs) are marked by a clock independent of actual game ticks. The clock can be used to see when, within a single tick, a command was issued. For example, if the game is on let's say tick 1000 (since the map started) then the games clock would be 15.625 seconds (yes, seconds), which means tick 1001 will occur at 15.640625 seconds. Now if you send a command such as +attack in the middle of tick 1000, the clock would read 15.6328125 seconds. By marking the player's command +attack with that specific time value, when tick 1001 finally occurs, the command will go through, but not before taking that clock value into account. For example, if two players fire their weapons on tick 1000, we can determine who actually shot first by examining both of their clock values for each of their +attack commands. That would be one use case for a sub-tick clock assuming my guess is accurate. It could be used further to determine all sorts of data such as view angles between ticks, when a command occurred. It's hard to say exactly what's going on at the moment, but take my theory with a grain of salt.
Bullet inaccuracy? What? No. It's you who is inaccurate. Gone are the times when you guys rely on luck to hit flicks that doesn't make sense. Imagine in real life you pull the trigger and then you flick and that's only when it fires. Doesn't make sense right? This is the accurate way.
in fact locally on your client, the render will show the last local mouse position, if you get 300 fps and your mouse is 500 Hz, the view angle will be updated 300 times a second using a position which is max 3/5 of a frame old (I did the math at random), so the test he's doing is comparing this local position to what ever position gets recorded by the server prior to validating the shot and then displayed at the next tick when the server response gets used locally. It means that what he is showing is comparing the shift between server shots and local shots, for instance in a game, some shots seems straight, display impact, then disappear and completely missed, because the client must display something temporarily before updating with the latest info from the server. The servers send an updates every bunch of frames at the 64 tick rate, the diff between csgo and cs2 is how much the position changes. Subtick means that between ticks, the clients send more often update potentially, because it sends right away key press. Mouse events while rotating are at the max FPS (I don't think cs registers position multiple times between frames, all the code is frame based). If the subtick was triggered even for mouse events, the number of events received will be enormous (I'm all for it even then) so the cost for the servers will rise, it could improve the shift but I don't think it's even that, because csgo does better at 64 ticks without subticks, so I'm sure it's related to another difference, like the server deconfliction policies and even it could be caused by some random bugs in the new code used in cs2 (I'm sure they finally rewrote core stuff)
@@justcrap3703 I'm talking about the inaccuracy with +attack;-right as shown in the video compared to CSGO.
@@raph151515
"It means that what he is showing is comparing the shift between server shots and local shots"
The problem is that he is using sv_showimpacts 1 which shows local bullet prediction as well as server prediction. If there was a shift between client and server we would see the blue and red boxes misaligned greatly every time.
"Subtick means that between ticks, the clients send more often update potentially, because it sends right away key press."
This doesn't appear to be the case. A TH-camr named MrMaxim performed a Wireshark analysis of game packets months ago which showed that there were no additional packets being sent.
Something worth considering is that Valve cannot afford to congest the network in a fast-paced game like CS2. Fewer smaller packets is the best way to ensure low bandwidth. It's also the reason why they use UDP packets for speed. 160,000 packets per second has the potential to literally knock people offline. In MrMaxim's video, Valve's official 64-tick server transferred 64 packets per second. This is why I believe that sub-tick doesn't literally mean sending data during world update deltas.
underrated asf, not only for the explanation but also for the editing!
I played a lot CS2 and wondered why my AWP flicks aren't hitting anymore. Ich changed my dpi and aim speed several times but it's still a pain to hit shots right.
THEN yesterday I played CS:GO with a friend without access to cs2 and instantly my shots were accurate again. I got really nice frags again and was unhappy about cs2
same this made me quit cs2 for now. its just a stupid feeling missing shots and be left wondering if its game or your fault.
My shots are hitting with CS2, because CS2 actually does it accurately :)
brooooooo i face same prblmm everdayyy i think at everynight when i go for sleep that i am became noob then i saw ur comment and now i feel that i am wrong bcz i eliminatedd many pro playerss in cs goo cs 2 is very baddd
@@Ayan-nz3or meanwhile I learned how to play with the new system. Doesn't work as good as before but it's getting better. Good luck for you bro
The more I've been playing CS2, the more I feel like subtick is a solution to a problem we didn't actually have. If they spent whatever $ it cost to develop this 'tech' on 128 tick servers everyone would be having a much better time lol
When you combine this with the input lag being 2x and the other netcode issues that are also present it does become "literally unplayable" as far as I'm concerned. The entire game feels completely off when you play it and I personally wont bother till these issues are ironed out.
input lag is not 2x that's just misinformation
@@keatonwastakenIf you have a 60ms ping it kind is.
idk why everybody discussing 'new cs2 problems' but i actually gets perfect glock shots VS compare to csgo where i was complete noob on first round as Terr, i still have questions for scout fast stooting but meanwhile awp is just fine for me.
@@BriJeeTVGameDev Yes tap firing is more accurate than ever in CS2 but with that also comes the horrendous downside of spraying being completely screwed up and flick shots also being screwed up.
The core of the issue right now is the lag between subtick and server tick.
When you fire a shot subtick instantly registers it but then the server, which is 64tick, updates like 15ms later and since the weapon animations are tied to the server tickrate it means the bullets you're shooting happen way earlier (15ms is a lot for these things) then the actual animation of the gun.
Search on TH-cam "MrMaxim Is this why shooting feels off in CS2?" and you'll understand, it has a visual demonstration.
I grew up on cs 1.6, was lvl 10faceit, Radiant in Valorant, and honest to god, ive never felt like something was so off about a game like i do in cs2. I like to 1tap, it keeps me focused and stops me from going for bad sprays, and ever since CSGO came out, it was possible to do it consistently. Valorant obviously made it even easier with their upgrades to hit reg and servers, so you could legit never miss if you had the aim. But then CS2 came out, and I was so hyped to get back into the faster paced gunfights of CS, now with fixed hit regs and servers, but holy shit, the game is not just buggy, its unplayable. There is no way you can ever get a 1tap, the same shit happens to rifles as it does to the awp, I tested it out on some bot games where id flick onto their heads and shoot 1 bullet, and the bullet literally ended up where my mouse was a millisecond ago. Spraying has never been this bad, even cs 1.6 had more consisten spraying it feels like.
Just disappointing how the core of the entire game will never be on the level as Valorants, cause I really do miss CS, its simpler and more fast paced, which is honestly more fun if done well, its just the fact that your shots dont go to where your aiming, or your opponents body isnt actually where its showing to you, making the bullets miss again. Its been an issue for 15years we har CS:GO, very weird how they still havent come up with a real solution whilst their competitors have.
so this is why i feel like stupido ....
This explains a LOT, thank you for doing Valve's work for them. "what you see is what you get"... my a$$!
It definitely makes sense based on my experience. And the bit at the end about your crosshair rarely moving while you shoot, I’m not so sure. A lot of times you are making microadjustments to move to your opponent’s head as you click, in a 1v1 fight
there's this "flickshot" thing... It's just stupid that the speed of your mouse matters now, esp because csgo had some of the best flickshots/qs mechanics
You’re constantly moving your crosshair as to try and spray control
best video that makes sense. you did a lot of investigation. just because you are not a developer doesnt mean you havent figured out what is wrong
But I was wrong in the video, thanks to youtube comment I’ve been able to make another, “more correct,” video
Okay will watch it @@vnc_cs thanks
I'm not an expert either but the dpi I think it relates to the mouse sensor capability to scan the surface and the rate of transmitting these info relates to the polling rate of the mouse.
People probably thinking that 'follow crosshair' is to help new players with the spray.
But after seeing this video i'm thinking it's because of the Sub-ticks update 😂
'follow crosshair' exists to compensate for those effects, or drawbacks for some people.
CS2 server sucks too much packet loss, CSGO was way much better.
I literally stopped playing because I thought I just somehow turned garbage at the game but this explained everything
This is subtick working as intended, your shot is going where your crosshair was the instant the mouse button was pressed instead of being teleported to where your crosshair is on the next tick. This is an improvement over csgo that’s simply going to take some getting used to since it is different
It's an improvement in theory because its more "accurate" but its worse gameplay wise and noone asked for the change. People have practiced for 15 years with it the other way we dont want it different.
People did ask for 128 tick servers for well over 8 years though.
While it's not what people asked for it is the same idea
its obviously not as u see in the vid its pretty random, in csgo its at least consistent
@@tomklie98 it’s random because him pressing the fire button is not synced with the tickrate, which is the entire point of subtick. The shot lands exactly where it’s supposed to, which it doesn’t in csgo, and then a random amount of time passes between then until the next tick, at which point he stops rotating.
@@NiceEvils "its too old now dont bother improving it" is exactly the type of thinking that prevents improvement, just get used to the new system and let the meta shift, if you cant adapt its your fault. csgo players are all about getting what you see until it doesnt benefit them
i like it being behind your crosshair because it can stop the run and gunners.
Great video! Really hope this gets fixed, cs2 doesnt feel nearly as responsive when compared to csgo.
it does feel like ur right, on csgo the ssg 08 feels much snappier when moving ur crosshair around, but on cs2 you have to stop right before shooting to actually hit something
I saw something about jump-throw binds being inconsistent, but setting up an alias for them first and then binding that alias made it perfectly consistent. Do you think it's possible something like that is coming into play here? (i.e. Create an alias for shooting, then binding mouse1 to that alias)
Good idea
This explains why I am doing so much better in CS2, my shots are actually landing where I put them.
This is actually a problem with csgo. There the info of your bullet is only registered at the next tick with the firing animation, so they are in sync. Therefore the bullet, firing animation and crosshair are all a bit too far.
In cs2 the bullet info is saved immidiately and sent to the server on the next tick at the same time as the animation. This sub-tick info is then applied, so the bullet shows where it should, but the crosshair placement and animation dont have subtick so they travel too far. IMO this is better than your bullet going too far as well
This explains why I felt like my shot would not register on target when I turn quickly to aim at someone
I doubt it is due to dpi, however this is a really solid showcase and experiment showing that there is a massive problem. Thanks for making this video
This explains why I've suddenly been horrible after getting cs2
You should have used the exec cfg method in csgo as well instead of aliasing. Probably doesn't matter but there could be some minor overhead on executing multiple commands via a file instead of with aliases, it's good practice to eliminate unnecessary variables.
I love how valve released a whole new game to try to fix a tiny issue that a small part of the community has been bitching about but ended up making things much worse. What was wrong with the 64 tick servers again? What did subticks improve exactly?
still not fixed trash
Subtick doesn't immediately talk to the server it sends a packet to the server just as often as the csgo servers, but that packet CONTAINS timestamps logging exactly when each action happened.
This may all be a symptom of subtick, but I think people are misunderstanding that whole concept.
I'm in software dev., not networking, so I could be mistaken too, but I know FOR SURE that it's not sending a packet every 1/800th of a mouse movement!
Still not fixed.
Keep the fancy graphics + change the tick behaviour back as it was in cs:go = perfect game
Very well made video! Have you send a bug report? With this amount of evidence it should be replicable and fixable for the devs :)
The theory about why it's happening is way off. Subtick works by having timestamp for events (like mouse click). Not by sending info to the server as soon as event happens, that would not be technically possible.
Is it fixed?
Lmfao, give it 10 more years.
@@samsharma3216 5028 hrs deep. Now im quiting because of this stupid chit. Movement is a big deal now im realising aim is affected too.
They killed cs:go
@@altaccount2179 I couldn't agree more, 3k hours been playing from 8 years. When I started playing CS2 it's like I forgot how to play at all, sure some shots connect but many don't. They've made everything worst that can be no wonder 75% of the player base has moved to something better.
No
this finally explains it the Awp flicks never hit and it feels so bad i only played csgo for 64hrs and still got so used to its tick rate i can't play at all in cs2 and its very frustrating
CS2 IS TRASH this is the worst shooting game in history
Exactly bring back CSGO 1 man 😢
The reason why spraying in cs2 feels bad is that it is way harder than it was in csgo.
In csgo every shot was processed every 128th/64th of a second at the end of a tick. That means you had 8 or even 16 miliseconds to correct your aim during spraying to land your next shot.
In cs2 every shot is still processed every 64th of a second but due to subtick the shot is calculated to happen on the exact moment your gun fired. That means, you have to correct your aim every single milisecond to land your next shot.
What they could do to "fix" this issue is simulating the old 128 tick, meaning every shot and it's timestamp would be pushed to either the middle or end of a tick (1/64). In other words reducing the sampling rate from 1000 to 128. Which will make spraying easier again.
cs2 is trash
They fked it up soo bad, make me cry.
It ran smoothly on my laptop like 60+ fps and no it's barely over 30....
++++
Unless you almost immediately headshot your opponent or 1 tap them you’ll be constantly moving your crosshair to adjust for spray, this means that by your logic it is constantly falling behind, which could definitely cause players to miss/hit shots they would’ve or wouldn’t have hit, which in my opinion is kind of game breaking.
I’ve also noticed this myself, gotten shot through a wall you can’t shoot through and shot an opponent multiple times yet i only hit them once supposedly, could just be cause i’m bad, but not that bad.
Don’t know if this is just me or if this is happening to other people as well.
its people like you, that i'm thankful for, to deal with problems like this. ngl - thank you very much.
cs:go - you cant walk if u want shoot
cs2 - you cant aim if you want to shoot
cs3 - you cant hold a weopon if you want to shoot
The fix you proposed sounds just like what sub-ticks are supposed to be. Also, we should remember that if something is simple, it would already be fixed in the first place (not really in this case since it's a beta). Also the "they did us dirty with the 64 ticks" doesn't make sense since Valve's server (not communities') is 64 ticks in csgo. I'd say that nobody in the community can make an accurate idea of what is the problem since here it's about sending/receiving data (except for the one with a degree in it I guess). It's not like the issue with spraying which is visible (the tracers aren't accurate to the actual bullets). Here it requires a deeper knowledge and we do not have the source code to understand what's happening behind. Remember that it is a beta so if there is an issue, send an email to them explaining the issue with the most precision you can.
Subtick registers the exact moment you click. Where you click is where you hit, not where your crosshair is when the next tick is loaded.
So when you spin in csgo, you click and on the next tick it fires and hits where your crosshair is
In cs2 when you click it remembers where and when it was, and when next tick loads it shoots where you had your aim the exact moment you clicked
I think it is sth like, when you click, the game remembers the position of your crosshair, and will shoot the bullet to that position when the server refresh(new tick), so when you click, before the new tick(the bullet is fired), you might have already moved your crosshair to a new position, and the shot is fired on the position the game remembered, so while the shot is actually perfectly accurate(fired on where it should land), your crosshair is already at a new position, making the shot not on your crosshair position, I guess you can say it as the animation can't keep up to the actual game
I thought of exactly the same solution after your explanation. I hope it's not too complicated to implement it.
While it does seem that you're witnessing subtick, I dont think you're seeing what you think you are. One of the issues with subtick is that it can make the animations etc feel delayed while the bullets arent. Subtick doesn't include mouse movements unless someone can point to a reference to say otherwise.
The bullets are likely going in the right place now, but because the calculated time was within the current tick where the alias is being processed, its "behind" where you're aiming. It's the same with the weird gun feelings. The shots are registered between ticks, but animation still has to happen on a tick so occasionally feels a slight delay. In general, your shots are *technically* more accurate if you put server lag aside. When the servers run well subtick is absolutely superior.
I don't think this is a server stress problem, but instead, a display problem - human baseline reaction time is around 10milisecond, which actually coincides with the 15ms 64 tick rate in CSGO, that is why it feels natural to most players when they flick shot timing, or even spraying because it's only after we see and register in our brain (which doesn't have sub tick lol ;) ) ak-47 start to shoot sideways that we adjust by a counter-motion using a mouse to spray control. But that doesn't explain why the macro this guy used to stop and shoot at the same time will cause the crosshair to not be on the shothole - but you can apply a similar theory, most gaming keyboards have an input latency of 1-2milisecond, and the receiver (PC) + loading into CS2 software will up this to around 5-15 ms depending on your specs. Since this issue doesn't always occur (sometimes the crosshair is on the shothole), in science, we call this unreliable, so there 'must' be a random error that is causing a disconnect between what "you see and what you get", and I suspect that sub-tick is working compartmentalized for single actions such as shooting (aim) or movement keys, but not calibrated to work well when together. You might argue that sub-tick shouldn't cause this issue (which is what Valve is insinuating), but from a practical perspective counting units such as ticks is incredibly important when trying to achieve replicable results (which is a fancy way to say reliable results). For example, if you have a ruler with millimeter units and ask 10 people to measure the length of a 4.25mm string, they are most likely going to give results such as 4mm, or 4.5mm because your measuring equipment is not accurate enough to give more decimal places, so you naturally just round to a nicer number like a half. If we do the experiment again with a ruler with a mix of units such as mm, cm, micrometer, and nanometre (to emulate how sub-tick tries to get every input in between ticks) and then ask 10 people to measure the same string, some people won't try hard to measure and just go with 4mm, some people will use the smaller units and try to get 4.2mm or even more extreme some people will attempt to go 4.2479...mm, now you have a wider array of numbers as results which represents the inconsistency that you feel, and amplify that with so many actions that you are doing in CS2, it's more likely than not that errors will occur - and reasonable that we have this problem.
The server does not update each time your mouse updates. It still updates at the same rate, 64 times per second.
The issue is that in CSGO the game would update everything at once. So if in that single tick you move your mouse 5 pixels to the left, clicked once, and moved 5 pixels up, the game would register5 left + 5 up + fire, and the firing would always be the last thing to account for. If you were moving, fired and stopped, in the same tick, you would fire as if you stopped first.
But in CS2 the server knows EXACTLY the order of events and follows it to a T. So if you move 5 pixels to the left, fire and then 5 pixels up, this is exactly what the server will calculate. In your head the bullet SHOULD go to where the crosshair is, because for the human brain everything that happened in 15ms is basically the simultaneously. But the server will still fire that bullet where you clicked. Because CS2 is still hitscan (unlike Battlefield or Warzone), the shot is fired and accounted for instantly, but the animation lags behind. In CSGO that wasn't a problem because EVERYTHING was lagging behind, but now it's uncanny.
Similarly, if you walk, fire and then stop, in CSGO that was fine, the server would think you stopped before clicking, now it knows you didn't and your shot won't be accurate.
Sub-tick is basically interpolation of an action in between two ticks - specifically shooting and WASD-movement (as stated in the valve video) which, unlike mouse movements, don't have thousands of inputs per second.
Let's imagine a very minuscule moment in a game where the server tick-rate is on tick#30 and proceeding to tick#31, a span of 15.625 milliseconds. The player, while quickly swiping their mouse, points at direction XYZ and clicks their mouse button 10.000ms after the instant tick#30 was processed. The game records that the player shot their weapon at that point in time, and tells the server "This player shot at direction XYZ 10.000ms after tick#30". The server will then use that information to create the effects of such an action upon the rest of the game upon the arrival of tick#31. However, there are still 5.625ms remaining until tick#31 arrives, and because your mouse is capable of inputting to your game thousands of times per second, it has moved to direction ABC during the 5.625ms remaining until tick#31 begins. So once the final results of what happened during tick#30 are processed and presented by the server, your shot will have arrived at direction XYZ but your mouse will now be facing direction ABC, which will create the exact same discrepancy showcased in this video.
What this means is that in CS2, all of your shots will go in the exact direction you shot them at regardless of tick-rate, although that exact direction is so precise to the point where it's happening in-between what is perceivable by the server's tick-rate. This is different from CSGO where your shots will only go where your mouse is pointing at upon the moment a tick arrives, which means your mouse's direction is only recorded at tick#30 and tick#31 and never anywhere in-between.
So CS2's sub-tick system is definitely far more precise than anything a higher tick-rate could achieve in the most literal sense, but it also means that what you're being shown on your screen is ever so slightly different from what you're inputting into the server because you now have the freedom to input actions in between what you can perceive. In fact, it's almost like CS2 has removed the tiny input lag that is present between you inputting an action and the server's tick-rate trying to follow up with your input, but everyone's used to being dependent on the tick-rate which makes the new tick-rate-independent system feel awkward as a result.
Just to add on about the proposed solution at the end of the video, making the mouse position be sent to the point of fire will probably make things feel even worse because that's essentially preventing your mouse from moving for the remainder of that tick, which means you won't be allowed to move your mouse for upwards of ~15ms in the worst case scenario and ~8ms on average. This probably won't have too much of an impact on tap-firing, but it's gonna be super noticeable during spray-downs if your mouse is being frozen in place after every shot.
Bro, thank you! I knew something was wrong. I was telling it to everyone, but even I was like, maybe I was just bad. Since the release I haven't really played because it just felt wrong.
i was flick god in csgo ... i try few minutes ago on aim bots and it feels so bad :(
I have given up awping for now because I had great flicks in CSGO but now nothing hits...
I went back to CSGO and enjoyed awping so much.
Now I understand why.