I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
On the other hand, the experience of every online competitive shooter today will dependent on the latency of the internet infrastructure of the players.
In the Stadia version your could (potentially) have a benefit of having most of the client-to-client communication happening within the same datacenter, or over the (assumably) very fast connection between Google data centers.
0
Options
jungleroomxIt's never too many graves, it's always not enough shovelsRegistered Userregular
I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
On the other hand, the experience of every online competitive shooter today will dependent on the latency of the internet infrastructure of the players.
In the Stadia version your could (potentially) have a benefit of having most of the client-to-client communication happening within the same datacenter, or over the (assumably) very fast connection between Google data centers.
So it's a good thing when games become unresponsive because they're unresponsive for everyone?
+1
Options
ChaosHatHop, hop, hop, HA!Trick of the lightRegistered Userregular
I'm really curious how the cost for the consumer will work out.
You can already buy access to virtual machines in Google cloud, same as aws and azure. However, one month access to a machine with similar specs as the Stadia ones is expensive - like something like 100-200$ per month!
(Someone more tech savvy than me - feel free to correct me if i got the Stadia spec, compared to available instance options from aws/Google, wrong).
Will Stadia offer unlimited access per month to a Stadia instance for a lower price? Then what would prevent bitcoin miner to just release a "game" that gave them access to cheaper computation power?
Will the cost for just the access to a Stadia instance be like 100-200$ per month? If so hahaha no thanks.
Will the cost plan be per hour? Then it could work out cheaper for casuals but still be very expensive for gamers who play a lot.
For me, i doubt i would use Stadia if the monthly cost was more than, say, 2x a month of Netflix.
I mean the virtual machine is probably giving you 100% uptime for that pricing and I'm going to assume Stadia is relying on consumers not using it 100% of the time.
I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
On the other hand, the experience of every online competitive shooter today will dependent on the latency of the internet infrastructure of the players.
In the Stadia version your could (potentially) have a benefit of having most of the client-to-client communication happening within the same datacenter, or over the (assumably) very fast connection between Google data centers.
So it's a good thing when games become unresponsive because they're unresponsive for everyone?
I'm saying that every online game today is subject to the quality of the infrastructure.
It's not clear to me what the difference in player experience will be when you increase the latency between player and client, but decrease the latency between clients and server.
I'm really curious how the cost for the consumer will work out.
You can already buy access to virtual machines in Google cloud, same as aws and azure. However, one month access to a machine with similar specs as the Stadia ones is expensive - like something like 100-200$ per month!
(Someone more tech savvy than me - feel free to correct me if i got the Stadia spec, compared to available instance options from aws/Google, wrong).
Will Stadia offer unlimited access per month to a Stadia instance for a lower price? Then what would prevent bitcoin miner to just release a "game" that gave them access to cheaper computation power?
Will the cost for just the access to a Stadia instance be like 100-200$ per month? If so hahaha no thanks.
Will the cost plan be per hour? Then it could work out cheaper for casuals but still be very expensive for gamers who play a lot.
For me, i doubt i would use Stadia if the monthly cost was more than, say, 2x a month of Netflix.
I mean the virtual machine is probably giving you 100% uptime for that pricing and I'm going to assume Stadia is relying on consumers not using it 100% of the time.
Sure, but then someone could publish a 'game' just to get access to computational power.
0
Options
jungleroomxIt's never too many graves, it's always not enough shovelsRegistered Userregular
I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
On the other hand, the experience of every online competitive shooter today will dependent on the latency of the internet infrastructure of the players.
In the Stadia version your could (potentially) have a benefit of having most of the client-to-client communication happening within the same datacenter, or over the (assumably) very fast connection between Google data centers.
So it's a good thing when games become unresponsive because they're unresponsive for everyone?
I'm saying that every online game today is subject to the quality of the infrastructure.
It's not clear to me what the difference in player experience will be when you increase the latency between player and client, but decrease the latency between clients and server.
When you push a button and the action appears on screen is the general mechanic.
Shooters, mobas, fighting games, shmups, platformers, and even some ARPGs depend on some rock solid timing and low latency to be played correctly.
I am not convinced Stadia will be able to do this because of networking basics and the numbers Digital Foundry got during their test.
The thing about Stadia that’s more troubling to me than latency is how smaller development studios or solo devs are going to get paid for their work by putting their games on this service. I would be shocked if Google had worked out a way to price this competitively while also making sure the devs are compensated fairly.
+2
Options
jungleroomxIt's never too many graves, it's always not enough shovelsRegistered Userregular
I'm really curious how the cost for the consumer will work out.
You can already buy access to virtual machines in Google cloud, same as aws and azure. However, one month access to a machine with similar specs as the Stadia ones is expensive - like something like 100-200$ per month!
(Someone more tech savvy than me - feel free to correct me if i got the Stadia spec, compared to available instance options from aws/Google, wrong).
Will Stadia offer unlimited access per month to a Stadia instance for a lower price? Then what would prevent bitcoin miner to just release a "game" that gave them access to cheaper computation power?
Will the cost for just the access to a Stadia instance be like 100-200$ per month? If so hahaha no thanks.
Will the cost plan be per hour? Then it could work out cheaper for casuals but still be very expensive for gamers who play a lot.
For me, i doubt i would use Stadia if the monthly cost was more than, say, 2x a month of Netflix.
I mean the virtual machine is probably giving you 100% uptime for that pricing and I'm going to assume Stadia is relying on consumers not using it 100% of the time.
I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
On the other hand, the experience of every online competitive shooter today will dependent on the latency of the internet infrastructure of the players.
In the Stadia version your could (potentially) have a benefit of having most of the client-to-client communication happening within the same datacenter, or over the (assumably) very fast connection between Google data centers.
So it's a good thing when games become unresponsive because they're unresponsive for everyone?
I'm saying that every online game today is subject to the quality of the infrastructure.
It's not clear to me what the difference in player experience will be when you increase the latency between player and client, but decrease the latency between clients and server.
When you push a button and the action appears on screen is the general mechanic.
Shooters, mobas, fighting games, shmups, platformers, and even some ARPGs depend on some rock solid timing and low latency to be played correctly.
I am not convinced Stadia will be able to do this because of networking basics and the numbers Digital Foundry got during their test.
Could be. I guess it won't be long until we get to find out ourselves given the release date later this year.
0
Options
jungleroomxIt's never too many graves, it's always not enough shovelsRegistered Userregular
I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
On the other hand, the experience of every online competitive shooter today will dependent on the latency of the internet infrastructure of the players.
In the Stadia version your could (potentially) have a benefit of having most of the client-to-client communication happening within the same datacenter, or over the (assumably) very fast connection between Google data centers.
So it's a good thing when games become unresponsive because they're unresponsive for everyone?
I'm saying that every online game today is subject to the quality of the infrastructure.
It's not clear to me what the difference in player experience will be when you increase the latency between player and client, but decrease the latency between clients and server.
When you push a button and the action appears on screen is the general mechanic.
Shooters, mobas, fighting games, shmups, platformers, and even some ARPGs depend on some rock solid timing and low latency to be played correctly.
I am not convinced Stadia will be able to do this because of networking basics and the numbers Digital Foundry got during their test.
Could be. I guess it won't be long until we get to find out ourselves given the release date later this year.
I mean
There's only so much fancy server software can do when so much of the US's net infrastructure is high-latency copper. There's also just so much you can do from input delay (All devices have it), to server-side travel time, to translation of command put to the game engine, to translating that into graphics that go in the engine, that then render in a video stream, and then are sent back with client-side travel time.
There will have to be geographically local Stadia servers in a ton of Metro areas, and even then it probably won't help people out in the burbs or the sticks.
I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
On the other hand, the experience of every online competitive shooter today will dependent on the latency of the internet infrastructure of the players.
In the Stadia version your could (potentially) have a benefit of having most of the client-to-client communication happening within the same datacenter, or over the (assumably) very fast connection between Google data centers.
So it's a good thing when games become unresponsive because they're unresponsive for everyone?
I'm saying that every online game today is subject to the quality of the infrastructure.
It's not clear to me what the difference in player experience will be when you increase the latency between player and client, but decrease the latency between clients and server.
When you push a button and the action appears on screen is the general mechanic.
Shooters, mobas, fighting games, shmups, platformers, and even some ARPGs depend on some rock solid timing and low latency to be played correctly.
I am not convinced Stadia will be able to do this because of networking basics and the numbers Digital Foundry got during their test.
Could be. I guess it won't be long until we get to find out ourselves given the release date later this year.
I mean
There's only so much fancy server software can do when so much of the US's net infrastructure is high-latency copper. There's also just so much you can do from input delay (All devices have it), to server-side travel time, to translation of command put to the game engine, to translating that into graphics that go in the engine, that then render in a video stream, and then are sent back with client-side travel time.
There will have to be geographically local Stadia servers in a ton of Metro areas, and even then it probably won't help people out in the burbs or the sticks.
Yea, I don't like how hard they pushed the "gaming for everyone" message. While I'm pretty sure Google have a very good idea about potential market, with good enough instructed, this is for sure not "for everyone" anytime soon.
I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
Seeing how my lil cousins are straight up beasts on Fortnite on their damn cellphones, all I have to say is...nah....Stadia is going to be fine on the competitive shooters side of things.
The fact that you can say that shows that you fundamentally don't comprehend the problem at hand and the reason that Stadia is doomed in this regard. To compare the 100ms latency of your cousin's fortnite connection to a 100ms input lag for Stadia is apples to oranges.
Here's how a normal client/server game works. In simple terms, both the clients and the server keep track of their own game states of the match at hand. The client's local version of the gamestate is used to render the game world to the player, and the player's controls are used to immediately move themselves about the game world as smoothly as if they were playing a single player game with no network connection. The client is constantly sending any alterations that the player has made to the gamestate to the server. The server receives these updates from all of the many clients, acts as the arbiter in the event of conflicting events, and then updates the clients with the new official gamestate that they clients can use to render the game to the players. The end result is that while the player's version of the gamestate is always slightly out of sync with the server's master copy, the desync does not matter 98% of the time since the client's local version of the gamestate lets them run around smoothly. The other 1.5% of the time the gameplay experience is kept pleasant due to netcode acting as smoke and mirrors to make things look right to the player. The final 0.5% of the time is the player noticing something has gone awry due to excessive latency causing shenanigans like odd hit registration or visual artifacts like rubber banding. In other words, while the latency is always there it is also generally unnoticed unless you're either very attentive or your latency is very poor. Everything still feels right because the player is playing with a local version of the gamestate.
With stadia, your every input suffers from a delay. You don't have a local version of the gamestate to make things look nice, so literally EVERYTHING will be delayed. You start moving your mouse to turn? It's gonna take a tenth of a second before you see yourself start turning. You want to stop moving your mouse so that you can shoot that guy because your cursor is on his head? Whoops, your cursor overshot because it didn't stop moving when you stopped moving. The latency will be grossly apparent with every action. Worse is that you can't even grow accustomed to the delay since the length of the delay will vary with network conditions. Stop moving your cursor a tenth of a second early to try to account for the lag? Too bad, network conditions have degraded and we've spiked to a 150ms ping to the stadia server. Looks like we still overshoot. On top of all of these input problems, we STILL have the latency issues that normal games experience because the players will most certainly not all be connecting to the same stadia server. You'll have clients connecting to stadia which connects to the game server. Any network hiccups anywhere along the chain are just gonna compound on the player.
This isn't the kind of problem that can be solved by some smart engineers and a lot of money. This is more like Benny walking into the room and saying "sorry kid, the game was rigged from the start. *BANG*"
This just straight up isn't gonna work unless Google is also revealing:
A: Some kinda clairvoyant Minority Report AI that profiles the player and knows what they're gonna do before they do it so that it can show you the results of your actions before you make them.
B: FTL communication.
I'm hoping for the later since I don't trust Google with the former.
More realistically the input delay problem won't be solved, stadia will simply fail in its stated goal of delivering the most popular games that people enjoy on PC/consoles without the need for owning their own hardware, and instead stadia will only bring players games from genres where input delay is not so critical. Again, turn based games spring to mind.
I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
Seeing how my lil cousins are straight up beasts on Fortnite on their damn cellphones, all I have to say is...nah....Stadia is going to be fine on the competitive shooters side of things.
The fact that you can say that shows that you fundamentally don't comprehend the problem at hand and the reason that Stadia is doomed in this regard. To compare the 100ms latency of your cousin's fortnite connection to a 100ms input lag for Stadia is apples to oranges.
Did (s)he say anything about latency on phones? The point here is that enough people love the imprecision of 30 FPS touch controls, and might be fine with the imprecision of streaming latency too.
I don't buy that we can't make tech to give back some of the benefits of client prediction. Like, if you hit fire when your screen shows the guy in your sights, the local client could tell the server that you scored a hit instead of simply shoot the gun. It might take a few frames before you see that visually, but it's probably better than missing. However it doesn't sound like Stadia has done any of this homework, so I'm hoping the xCloud rumors have something to them.
I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
Seeing how my lil cousins are straight up beasts on Fortnite on their damn cellphones, all I have to say is...nah....Stadia is going to be fine on the competitive shooters side of things.
The fact that you can say that shows that you fundamentally don't comprehend the problem at hand and the reason that Stadia is doomed in this regard. To compare the 100ms latency of your cousin's fortnite connection to a 100ms input lag for Stadia is apples to oranges.
Did (s)he say anything about latency on phones? The point here is that enough people love the imprecision of 30 FPS touch controls, and might be fine with the imprecision of streaming latency too.
I don't buy that we can't make tech to give back some of the benefits of client prediction. Like, if you hit fire when your screen shows the guy in your sights, the local client could tell the server that you scored a hit instead of simply shoot the gun. It might take a few frames before you see that visually, but it's probably better than missing. However it doesn't sound like Stadia has done any of this homework, so I'm hoping the xCloud rumors have something to them.
Stadia has no local client though.
It's just streaming the output from the client in the cloud to whatever device, as far as I understand it.
So until the controller sends the input to the cloud client, through the latency, it has no way of predicting that a hit was scored.
Basically Stadia sounds to me like a great way for Google to scrape in game text/voice chat, and not much else.
I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
Seeing how my lil cousins are straight up beasts on Fortnite on their damn cellphones, all I have to say is...nah....Stadia is going to be fine on the competitive shooters side of things.
The fact that you can say that shows that you fundamentally don't comprehend the problem at hand and the reason that Stadia is doomed in this regard. To compare the 100ms latency of your cousin's fortnite connection to a 100ms input lag for Stadia is apples to oranges.
Did (s)he say anything about latency on phones? The point here is that enough people love the imprecision of 30 FPS touch controls, and might be fine with the imprecision of streaming latency too.
I don't buy that we can't make tech to give back some of the benefits of client prediction. Like, if you hit fire when your screen shows the guy in your sights, the local client could tell the server that you scored a hit instead of simply shoot the gun. It might take a few frames before you see that visually, but it's probably better than missing. However it doesn't sound like Stadia has done any of this homework, so I'm hoping the xCloud rumors have something to them.
Stadia has no local client though.
It's just streaming the output from the client in the cloud to whatever device, as far as I understand it.
So until the controller sends the input to the cloud client, through the latency, it has no way of predicting that a hit was scored.
Basically Stadia sounds to me like a great way for Google to scrape in game text/voice chat, and not much else.
Bolded the part that people keep seeming to fail to understand. Stadia is just streaming what's running at the data center. There is no hardware on the client's side to host the local player's perspective of the game state. So every single trick in the net code book to hide the latency and make the player's experience mostly match up with the server is immediately null and void. All of those tricks REQUIRE the player to have a local version of the game state to interact with so that the player's experience looks seamless and smooth.
It will be pervasively jarring and uncomfortable to play any game that requires precision input on stadia.
3DS Friend Code:
Armchair: 4098-3704-2012
+1
Options
HenroidMexican kicked from Immigration ThreadCentrism is Racism :3Registered Userregular
Edit - LMAO the still image of the video embed in the embedded tweet there, Jesus jumped up Christ Google.
Googles really huffing their own farts with this advertising. For fucks sake.
+3
Options
HenroidMexican kicked from Immigration ThreadCentrism is Racism :3Registered Userregular
That video
1) co-opts a lot of minority stuff
2) declares Google is making history on the level of really important shit from history
3) ends up being more of an advertisement for watching streamed broadcasts
4) mentions NOTHING about what Stadia is or does
This isn't the kind of problem that can be solved by some smart engineers and a lot of money.
Repeating for emphasis. I don't think Stadia is doing it right, but that these issues can never be worked around in a streaming environment is untrue. There's no reason we can't move into a space where a client does have some degree of game awareness AND relies on the remote server for the majority of the game's rendering. ID already patented a streaming system where you get sent an array of visual data that the local client choses from based on which direction they are pushing the controller once its time to display. We can also imagine a traditional MMO where the raw 3D game is rendered remotely but the UI renders locally, letting you move your mouse, select targets and queue actions with the freedom of a local client while relying on the server to do the impressive 3D graphics that don't need to be pixel-accurate for enjoying the gameplay.
Even within Stadia limitations, a hitscan game can be written such that when the server renders a frame in which an enemy appears in your sights, (and there's no reason the server shouldn't know that since its doing the rendering), if it receives a "fire" command from the controller within X ms, it counts as a hit, even if it just rendered a frame when they were out of range. X here is based on an estimate of the average round-trip latency of video to the client and controller input back to the server, thus simulating the idea that if you hit fire when you see the target in sights, it will probably hit. No its not as good as traditional client prediction nor fully reliable, but its better than just throwing up your arms and declaring any attempts at innovation pointless.
rahkeesh2000 on
0
Options
jungleroomxIt's never too many graves, it's always not enough shovelsRegistered Userregular
That video
1) co-opts a lot of minority stuff
2) declares Google is making history on the level of really important shit from history
3) ends up being more of an advertisement for watching streamed broadcasts
4) mentions NOTHING about what Stadia is or does
That ad is so bad I almost want this to fail due to it's existence.
That video
1) co-opts a lot of minority stuff
2) declares Google is making history on the level of really important shit from history
3) ends up being more of an advertisement for watching streamed broadcasts
4) mentions NOTHING about what Stadia is or does
That ad is so bad I almost want this to fail due to it's existence.
Well lets not forget the PSP and its incredibly racist ads. Obviously those did so well in pushing the hardware Sony is still making handhelds
I have a podcast now. It's about video games and anime!Find it here.
That video
1) co-opts a lot of minority stuff
2) declares Google is making history on the level of really important shit from history
3) ends up being more of an advertisement for watching streamed broadcasts
4) mentions NOTHING about what Stadia is or does
That ad is so bad I almost want this to fail due to it's existence.
I normally don't give one damn about Twitter, but there is some comedy gold over there:
This isn't the kind of problem that can be solved by some smart engineers and a lot of money.
Repeating for emphasis. I don't think Stadia is doing it right, but that these issues can never be worked around in a streaming environment is untrue. There's no reason we can't move into a space where a client does have some degree of game awareness AND relies on the remote server for the majority of the game's rendering.
This would require a thin client per game hosted on the Stadia device, which is not what Google is suggesting.
Google is just sticking the same thing over every game and sending the output to the device, so that the device needs nothing but input/output at its end.
ID already patented a streaming system where you get sent an array of visual data that the local client choses from based on which direction they are pushing the controller once its time to display.
This is the tech that tries to speed satellite internet up.
It preloads the results of all possible website decisions to the end user's device so that when the user selects a particular hyperlink the next webpage is already loaded.
This requires a thin client per game on the Stadia device to choose between the preloaded data (in the above, this is the web browser), but also becomes impossible in any game with 3D graphics and a non-fixed camera.
We can also imagine a traditional MMO where the raw 3D game is rendered remotely but the UI renders locally, letting you move your mouse, select targets and queue actions with the freedom of a local client while relying on the server to do the impressive 3D graphics that don't need to be pixel-accurate for enjoying the gameplay.
Still requires a thin client to discern what the UI looks like and setup the instant action queue.
You are still suffering device -> Stadia server latency * 2 on each action being triggered or camera being moved, which, in the game you've suggested where actions can be queued up, may not be a big deal because actions are often prepared by the player seconds in advance.
Even within Stadia limitations, a hitscan game can be written such that when the server renders a frame in which an enemy appears in your sights, (and there's no reason the server shouldn't know that since its doing the rendering), if it receives a "fire" command from the controller within X ms, it counts as a hit, even if it just rendered a frame when they were out of range. X here is based on an estimate of the average round-trip latency of video to the client and controller input back to the server, thus simulating the idea that if you hit fire when you see the target in sights, it will probably hit. No its not as good as traditional client prediction nor fully reliable, but its better than just throwing up your arms and declaring any attempts at innovation pointless.
This is a description of normal hit-scan mechanics.
It does not solve the floatiness of having no thin client to track your position for you, nor the increase in latency from having the game server not on Stadia's servers if the game is not hosted inside Google.
Like, I don't know if Quake 3 Arena still has any servers up, but I remember that game felt like the player's movement and firing were all sent to the server and not displayed until the server updated the client.
That would give you some feel of the expected latency issues.
Stadia would be that, but also the same latency on moving the camera around.
HenroidMexican kicked from Immigration ThreadCentrism is Racism :3Registered Userregular
Every company has their own net code, and probably even game. And there's some shitty net codes out there, lemme tell you. Compound the streamed input AND output of the Stadia with multiplayer games and you have a disaster.
The most sophisticated net code system I've seen so far is in Overwatch, where it's taking players' latency / interactions with each other in the game and meeting in the middle to kick back a result to the both of them that should appear the same on their ends. This is with 12 players each round, with any combination of interaction going on, and it's being done in milliseconds. And this is just describing multiplayer.
I dunno how the Stadia is going to resolve single player game latency. Can you somehow program a game to give all its video and audio output ahead of what's happening mechanically to allow for the info to get to the player and receive the player's output in a timely manner that makes sense?
Can you somehow program a game to give all its video and audio output ahead of what's happening mechanically to allow for the info to get to the player and receive the player's output in a timely manner that makes sense?
Within the context of what google is doing here? No. Hell no. HELL GOD DAMN NO.
Let's assume that google isn't clairvoyant and can not predict what the player will do before we do it. We can not send the output in advance of the input since the output is a function of the input. What's being proposed is to hide the input delay by sending the results for all possible inputs and then only display the result that matches what the player ultimately inputs. Popular blockbuster titles that you'd normally play on your computer or console have an absolutely monstrous number of potential inputs such as pressing any button on the user's interface individually, simultaneously, or in rapid sequence for combos, or even the myraid degrees of directional input from analog input sources like a joystick. Remember that stadia's entire premise is that the client doesn't do any heavy lifting. Their datacenter is going to need to render the audio/video for every single possible input and send every single one of those video streams to the client so that the client can display the one singular output that matches the input actually chosen. Consequently the bandwidth consumption just fucking exploded since the required bandwidth is now equivalent to the bandwidth required for a single audio/video stream multiplied by the number of possible inputs. The potential userbase just shrunk from those who can handle streaming in hi-def to those who can handle a biblical legion of simultaneous streams in hi-def. Simply put this approach to hiding our input lag is not viable. If the game was designed to heavily restrict the number of possible inputs that the player can give, such as if the game were just an endless string of quick-time events where we either press the appropriate button or we don't, then our number of possible outputs decreases dramatically and things become a bit more reasonable. However then we're failing in stadia's stated goal of bringing people the latest popular games from PCs/consoles without having to own a gaming PC/console.
So we're right back where we started where Google physically can not provide input delay that's remotely comparable to someone with a local console/PC. Not until they either become clairvoyant or crack FTL communication.
That video 1) co-opts a lot of minority stuff
2) declares Google is making history on the level of really important shit from history
3) ends up being more of an advertisement for watching streamed broadcasts
4) mentions NOTHING about what Stadia is or does
Whoa whoa the video is pretty bad but this isn't why.
Like I'm kinda glad the marketing team didn't think to themselves "ok we want to show gaming since the beginning of time, let's show nothing but white dudes playing games. Let's get some middle ages european chess and 1920s american football and call it a day."
I don't know the appropriate amount of showing minority groups gaming in order to represent diversity without becoming co-opting. Is it better if they ask a representative from that group for permission before including it in the video?
That video 1) co-opts a lot of minority stuff
2) declares Google is making history on the level of really important shit from history
3) ends up being more of an advertisement for watching streamed broadcasts
4) mentions NOTHING about what Stadia is or does
Whoa whoa the video is pretty bad but this isn't why.
Like I'm kinda glad the marketing team didn't think to themselves "ok we want to show gaming since the beginning of time, let's show nothing but white dudes playing games. Let's get some middle ages european chess and 1920s american football and call it a day."
I don't know the appropriate amount of showing minority groups gaming in order to represent diversity without becoming co-opting. Is it better if they ask a representative from that group for permission before including it in the video?
I suppose part of the question is "Who is making it" and "who are you selling it to?"
Google is a major international player with I think something like half or more of its staff external to the US. I think their Hyderabad location is their largest or second largest office. It's almost a default white view to assume that a product produced in the first world can't call on the other cultures undoubtedly heavily involved in its production.
So, yeah, that's not co-opting since the people participating and the markets being sold to are not exclusively white, or even necessarily majority white.
Also, I keep going over it and I think the Stadia makes more sense if you start from the viewpoint of someone at google going, "How do we get people playing Google Play games more often than just on their phone while pooping." Gaming is a big revenue stream, and a HUGE number of gamers are middle-aged women who play a different style of game than first person shooters. Putting a controller in their hands to get them playing the type of game THEY like on their tv as well might make sense. And those games don't require the same precision as a AAA shooter or platformer.
Yeah, but if you look at say Hidden Object games:
- You're either running a high-bandwidth service on a mobile device, which is just going to empty data plans unless it's connected to a wifi router or something, or
- You're running a not very resource intensive single player game from a remote server with associated latency when the desktop device in front of you probably could run it anyway.
The first sounds more probable.
Google could get Stadia onto Androids, and people could play more graphically intensive games whilst pooping, as long as their toilet has good wifi.
I'd like specifics and not tricky-dick private demos on $1000 wifi hotspots.
I'm still trying to figure out who exactly this is for. 100/150+ms lag from their highly controlled, extremely expensive, and extremely localized Digitalfoundry test means competitive shooters and MOBAs, the two biggest streaming game types out there, are a complete no go from jump street. Flat out.
I'm still not convinced most countries have the Internet infrastructure capable of sub 100 ms (or, preferably, sub 50 ms) pings for this. Is this for single player games? Because I could see that. Multiplayer, right now, is out of the question. Any kind of lag gating would stratify the player base badly.
On the other hand, the experience of every online competitive shooter today will dependent on the latency of the internet infrastructure of the players.
In the Stadia version your could (potentially) have a benefit of having most of the client-to-client communication happening within the same datacenter, or over the (assumably) very fast connection between Google data centers.
So it's a good thing when games become unresponsive because they're unresponsive for everyone?
I'm saying that every online game today is subject to the quality of the infrastructure.
It's not clear to me what the difference in player experience will be when you increase the latency between player and client, but decrease the latency between clients and server.
When you push a button and the action appears on screen is the general mechanic.
Shooters, mobas, fighting games, shmups, platformers, and even some ARPGs depend on some rock solid timing and low latency to be played correctly.
I am not convinced Stadia will be able to do this because of networking basics and the numbers Digital Foundry got during their test.
Could be. I guess it won't be long until we get to find out ourselves given the release date later this year.
I mean
There's only so much fancy server software can do when so much of the US's net infrastructure is high-latency copper. There's also just so much you can do from input delay (All devices have it), to server-side travel time, to translation of command put to the game engine, to translating that into graphics that go in the engine, that then render in a video stream, and then are sent back with client-side travel time.
There will have to be geographically local Stadia servers in a ton of Metro areas, and even then it probably won't help people out in the burbs or the sticks.
Yea, I don't like how hard they pushed the "gaming for everyone" message. While I'm pretty sure Google have a very good idea about potential market, with good enough instructed, this is for sure not "for everyone" anytime soon.
Gaming for everyone*.
*"Everyone" does not include people with insufficiently high download speeds, insufficiently reliable service, high latency due to local infrastructure, high latency due to home wifi setup, or anyone with a usage cap.
Completely unrelated to the product on hand: does anyone know where that clip of the elderly woman playing what looks like AC:Black Flag or Skull & Bones is from?
There's something to be said for the schadenfreude of a bunch of otherwise privileged people complaining that, because THEIR internet is not fast enough, that Stadia isn't for anyone.
It's interesting how the tone of the thread seems to have shifted from 'streaming is the future, and google might be able to do it' to 'the concept of streaming is impossible and also awful'. I'm hesitant to think that it can't work after so many people tried that AC demo and said it worked pretty well.
It's interesting how the tone of the thread seems to have shifted from 'streaming is the future, and google might be able to do it' to 'the concept of streaming is impossible and also awful'. I'm hesitant to think that it can't work after so many people tried that AC demo and said it worked pretty well.
I can't speak for others, but before Stadia was even announced, I was busy in the Xbox thread going, "Yeah, I guess if anyone could do it Microsoft could. Except they won't, and it's impossible and also awful."
It's interesting how the tone of the thread seems to have shifted from 'streaming is the future, and google might be able to do it' to 'the concept of streaming is impossible and also awful'. I'm hesitant to think that it can't work after so many people tried that AC demo and said it worked pretty well.
The AC demo, AFAIK, was on a wired connection in a building with commercial/enterprise level internet speeds, at a location near one of Google's datacenters.
I'll be far more interested if it works as well on a WiFi/cellular connection in the middle of North Dakota.
. . .to jump in here, I'm sure it's already been mentioned, but DigitalFoundry was saying that data usage was somewhere around 25 MB/s. That. . .is a lot for metered users if my math is right (something like 90 Gigs?)
Posts
On the other hand, the experience of every online competitive shooter today will dependent on the latency of the internet infrastructure of the players.
In the Stadia version your could (potentially) have a benefit of having most of the client-to-client communication happening within the same datacenter, or over the (assumably) very fast connection between Google data centers.
So it's a good thing when games become unresponsive because they're unresponsive for everyone?
I mean the virtual machine is probably giving you 100% uptime for that pricing and I'm going to assume Stadia is relying on consumers not using it 100% of the time.
I'm saying that every online game today is subject to the quality of the infrastructure.
It's not clear to me what the difference in player experience will be when you increase the latency between player and client, but decrease the latency between clients and server.
Sure, but then someone could publish a 'game' just to get access to computational power.
When you push a button and the action appears on screen is the general mechanic.
Shooters, mobas, fighting games, shmups, platformers, and even some ARPGs depend on some rock solid timing and low latency to be played correctly.
I am not convinced Stadia will be able to do this because of networking basics and the numbers Digital Foundry got during their test.
They must be new to gaming.
Could be. I guess it won't be long until we get to find out ourselves given the release date later this year.
I mean
There's only so much fancy server software can do when so much of the US's net infrastructure is high-latency copper. There's also just so much you can do from input delay (All devices have it), to server-side travel time, to translation of command put to the game engine, to translating that into graphics that go in the engine, that then render in a video stream, and then are sent back with client-side travel time.
There will have to be geographically local Stadia servers in a ton of Metro areas, and even then it probably won't help people out in the burbs or the sticks.
Yea, I don't like how hard they pushed the "gaming for everyone" message. While I'm pretty sure Google have a very good idea about potential market, with good enough instructed, this is for sure not "for everyone" anytime soon.
The fact that you can say that shows that you fundamentally don't comprehend the problem at hand and the reason that Stadia is doomed in this regard. To compare the 100ms latency of your cousin's fortnite connection to a 100ms input lag for Stadia is apples to oranges.
Here's how a normal client/server game works. In simple terms, both the clients and the server keep track of their own game states of the match at hand. The client's local version of the gamestate is used to render the game world to the player, and the player's controls are used to immediately move themselves about the game world as smoothly as if they were playing a single player game with no network connection. The client is constantly sending any alterations that the player has made to the gamestate to the server. The server receives these updates from all of the many clients, acts as the arbiter in the event of conflicting events, and then updates the clients with the new official gamestate that they clients can use to render the game to the players. The end result is that while the player's version of the gamestate is always slightly out of sync with the server's master copy, the desync does not matter 98% of the time since the client's local version of the gamestate lets them run around smoothly. The other 1.5% of the time the gameplay experience is kept pleasant due to netcode acting as smoke and mirrors to make things look right to the player. The final 0.5% of the time is the player noticing something has gone awry due to excessive latency causing shenanigans like odd hit registration or visual artifacts like rubber banding. In other words, while the latency is always there it is also generally unnoticed unless you're either very attentive or your latency is very poor. Everything still feels right because the player is playing with a local version of the gamestate.
With stadia, your every input suffers from a delay. You don't have a local version of the gamestate to make things look nice, so literally EVERYTHING will be delayed. You start moving your mouse to turn? It's gonna take a tenth of a second before you see yourself start turning. You want to stop moving your mouse so that you can shoot that guy because your cursor is on his head? Whoops, your cursor overshot because it didn't stop moving when you stopped moving. The latency will be grossly apparent with every action. Worse is that you can't even grow accustomed to the delay since the length of the delay will vary with network conditions. Stop moving your cursor a tenth of a second early to try to account for the lag? Too bad, network conditions have degraded and we've spiked to a 150ms ping to the stadia server. Looks like we still overshoot. On top of all of these input problems, we STILL have the latency issues that normal games experience because the players will most certainly not all be connecting to the same stadia server. You'll have clients connecting to stadia which connects to the game server. Any network hiccups anywhere along the chain are just gonna compound on the player.
This isn't the kind of problem that can be solved by some smart engineers and a lot of money. This is more like Benny walking into the room and saying "sorry kid, the game was rigged from the start. *BANG*"
This just straight up isn't gonna work unless Google is also revealing:
A: Some kinda clairvoyant Minority Report AI that profiles the player and knows what they're gonna do before they do it so that it can show you the results of your actions before you make them.
B: FTL communication.
I'm hoping for the later since I don't trust Google with the former.
More realistically the input delay problem won't be solved, stadia will simply fail in its stated goal of delivering the most popular games that people enjoy on PC/consoles without the need for owning their own hardware, and instead stadia will only bring players games from genres where input delay is not so critical. Again, turn based games spring to mind.
Armchair: 4098-3704-2012
Worked for Smash Ultimate.
Did (s)he say anything about latency on phones? The point here is that enough people love the imprecision of 30 FPS touch controls, and might be fine with the imprecision of streaming latency too.
I don't buy that we can't make tech to give back some of the benefits of client prediction. Like, if you hit fire when your screen shows the guy in your sights, the local client could tell the server that you scored a hit instead of simply shoot the gun. It might take a few frames before you see that visually, but it's probably better than missing. However it doesn't sound like Stadia has done any of this homework, so I'm hoping the xCloud rumors have something to them.
Stadia has no local client though.
It's just streaming the output from the client in the cloud to whatever device, as far as I understand it.
So until the controller sends the input to the cloud client, through the latency, it has no way of predicting that a hit was scored.
Basically Stadia sounds to me like a great way for Google to scrape in game text/voice chat, and not much else.
it could be pretty cool for an MMO or something though.
Bolded the part that people keep seeming to fail to understand. Stadia is just streaming what's running at the data center. There is no hardware on the client's side to host the local player's perspective of the game state. So every single trick in the net code book to hide the latency and make the player's experience mostly match up with the server is immediately null and void. All of those tricks REQUIRE the player to have a local version of the game state to interact with so that the player's experience looks seamless and smooth.
It will be pervasively jarring and uncomfortable to play any game that requires precision input on stadia.
Armchair: 4098-3704-2012
Edit -
Edit - LMAO the still image of the video embed in the embedded tweet there, Jesus jumped up Christ Google.
Googles really huffing their own farts with this advertising. For fucks sake.
1) co-opts a lot of minority stuff
2) declares Google is making history on the level of really important shit from history
3) ends up being more of an advertisement for watching streamed broadcasts
4) mentions NOTHING about what Stadia is or does
Repeating for emphasis. I don't think Stadia is doing it right, but that these issues can never be worked around in a streaming environment is untrue. There's no reason we can't move into a space where a client does have some degree of game awareness AND relies on the remote server for the majority of the game's rendering. ID already patented a streaming system where you get sent an array of visual data that the local client choses from based on which direction they are pushing the controller once its time to display. We can also imagine a traditional MMO where the raw 3D game is rendered remotely but the UI renders locally, letting you move your mouse, select targets and queue actions with the freedom of a local client while relying on the server to do the impressive 3D graphics that don't need to be pixel-accurate for enjoying the gameplay.
Even within Stadia limitations, a hitscan game can be written such that when the server renders a frame in which an enemy appears in your sights, (and there's no reason the server shouldn't know that since its doing the rendering), if it receives a "fire" command from the controller within X ms, it counts as a hit, even if it just rendered a frame when they were out of range. X here is based on an estimate of the average round-trip latency of video to the client and controller input back to the server, thus simulating the idea that if you hit fire when you see the target in sights, it will probably hit. No its not as good as traditional client prediction nor fully reliable, but its better than just throwing up your arms and declaring any attempts at innovation pointless.
That ad is so bad I almost want this to fail due to it's existence.
Well lets not forget the PSP and its incredibly racist ads. Obviously those did so well in pushing the hardware Sony is still making handhelds
I normally don't give one damn about Twitter, but there is some comedy gold over there:
Armchair: 4098-3704-2012
Google is just sticking the same thing over every game and sending the output to the device, so that the device needs nothing but input/output at its end.
This is the tech that tries to speed satellite internet up.
It preloads the results of all possible website decisions to the end user's device so that when the user selects a particular hyperlink the next webpage is already loaded.
This requires a thin client per game on the Stadia device to choose between the preloaded data (in the above, this is the web browser), but also becomes impossible in any game with 3D graphics and a non-fixed camera.
Still requires a thin client to discern what the UI looks like and setup the instant action queue.
You are still suffering device -> Stadia server latency * 2 on each action being triggered or camera being moved, which, in the game you've suggested where actions can be queued up, may not be a big deal because actions are often prepared by the player seconds in advance.
This is a description of normal hit-scan mechanics.
It does not solve the floatiness of having no thin client to track your position for you, nor the increase in latency from having the game server not on Stadia's servers if the game is not hosted inside Google.
That would give you some feel of the expected latency issues.
Stadia would be that, but also the same latency on moving the camera around.
The most sophisticated net code system I've seen so far is in Overwatch, where it's taking players' latency / interactions with each other in the game and meeting in the middle to kick back a result to the both of them that should appear the same on their ends. This is with 12 players each round, with any combination of interaction going on, and it's being done in milliseconds. And this is just describing multiplayer.
I dunno how the Stadia is going to resolve single player game latency. Can you somehow program a game to give all its video and audio output ahead of what's happening mechanically to allow for the info to get to the player and receive the player's output in a timely manner that makes sense?
Within the context of what google is doing here? No. Hell no. HELL GOD DAMN NO.
Let's assume that google isn't clairvoyant and can not predict what the player will do before we do it. We can not send the output in advance of the input since the output is a function of the input. What's being proposed is to hide the input delay by sending the results for all possible inputs and then only display the result that matches what the player ultimately inputs. Popular blockbuster titles that you'd normally play on your computer or console have an absolutely monstrous number of potential inputs such as pressing any button on the user's interface individually, simultaneously, or in rapid sequence for combos, or even the myraid degrees of directional input from analog input sources like a joystick. Remember that stadia's entire premise is that the client doesn't do any heavy lifting. Their datacenter is going to need to render the audio/video for every single possible input and send every single one of those video streams to the client so that the client can display the one singular output that matches the input actually chosen. Consequently the bandwidth consumption just fucking exploded since the required bandwidth is now equivalent to the bandwidth required for a single audio/video stream multiplied by the number of possible inputs. The potential userbase just shrunk from those who can handle streaming in hi-def to those who can handle a biblical legion of simultaneous streams in hi-def. Simply put this approach to hiding our input lag is not viable. If the game was designed to heavily restrict the number of possible inputs that the player can give, such as if the game were just an endless string of quick-time events where we either press the appropriate button or we don't, then our number of possible outputs decreases dramatically and things become a bit more reasonable. However then we're failing in stadia's stated goal of bringing people the latest popular games from PCs/consoles without having to own a gaming PC/console.
So we're right back where we started where Google physically can not provide input delay that's remotely comparable to someone with a local console/PC. Not until they either become clairvoyant or crack FTL communication.
Armchair: 4098-3704-2012
Whoa whoa the video is pretty bad but this isn't why.
Like I'm kinda glad the marketing team didn't think to themselves "ok we want to show gaming since the beginning of time, let's show nothing but white dudes playing games. Let's get some middle ages european chess and 1920s american football and call it a day."
I don't know the appropriate amount of showing minority groups gaming in order to represent diversity without becoming co-opting. Is it better if they ask a representative from that group for permission before including it in the video?
I suppose part of the question is "Who is making it" and "who are you selling it to?"
Google is a major international player with I think something like half or more of its staff external to the US. I think their Hyderabad location is their largest or second largest office. It's almost a default white view to assume that a product produced in the first world can't call on the other cultures undoubtedly heavily involved in its production.
So, yeah, that's not co-opting since the people participating and the markets being sold to are not exclusively white, or even necessarily majority white.
- You're either running a high-bandwidth service on a mobile device, which is just going to empty data plans unless it's connected to a wifi router or something, or
- You're running a not very resource intensive single player game from a remote server with associated latency when the desktop device in front of you probably could run it anyway.
The first sounds more probable.
Google could get Stadia onto Androids, and people could play more graphically intensive games whilst pooping, as long as their toilet has good wifi.
Gaming for everyone*.
*"Everyone" does not include people with insufficiently high download speeds, insufficiently reliable service, high latency due to local infrastructure, high latency due to home wifi setup, or anyone with a usage cap.
Armchair: 4098-3704-2012
I can't speak for others, but before Stadia was even announced, I was busy in the Xbox thread going, "Yeah, I guess if anyone could do it Microsoft could. Except they won't, and it's impossible and also awful."
The AC demo, AFAIK, was on a wired connection in a building with commercial/enterprise level internet speeds, at a location near one of Google's datacenters.
I'll be far more interested if it works as well on a WiFi/cellular connection in the middle of North Dakota.