Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Google's Stadia is the Next Generation of Gaming That Doesn't Require a PC or Console - MMORPG.com

14567810»

Comments

  • QuizzicalQuizzical Member LegendaryPosts: 25,516
    Quizzical said:
    Quizzical said:
    On another site, someone pointed out that judder is also going to be a huge problem.  If your average ping time to Google's servers is 20 ms, but that's bouncing around a lot between 10 ms and 50 ms, then a lot of games will be an awful mess of things speeding up and slowing down erratically

    Anything sensitive to timing like combos in Street Fighter may well be completely unplayable.  If you enter a combo properly, but one key press being delayed a little more than another means that 10% of your combos aren't recognized as such on the server, you basically can't play the game.

    Now, you could work around that in a sense by adding extra latency so that everything seems like consistent latency up to some maximum.  For example, they could pad your latency such that it always acts like 100 ms ping exactly unless it spikes above that.  That would mostly avoid the judder problem, but at the expense of making the latency problem far worse.
    That should already be addressed considering multiplayer games and server side authority. Animations for example run at 60fps and servers/client comms is typically 15 and less so to ensure the proper animations start when they should you send 60 frames worth of commands every 15 with included time stamps and the receiving end handles it. 
    Erratic network latency doesn't cause judder in a traditional MMORPG.  It can cause something to occasionally look like a lag spike, but that's about it.  The problem is that, while there won't be any judder as the game is rendered on a remote server, there sure will be when it gets to you.  You can't re-render the game locally at 60 frames per second because the whole point of this is that you're not rendering the game locally at all.

    There's also minimal judder on input latency because a dedicated USB wire is pretty consistent.  Windows polls USB devices every 8 ms by default, and some will reduce that further.  Online games rarely have sensitive timing-based things at all, and to the extent that they do, there's likely to be more trusting the client on when it says a button was pressed than in other places in the game.
    Queue up a series of commands and send it to the google servers. That's all that needs to happen. Then the server treats those as input in the game in the same order you sent them. Nothing will be missed.
    There are a lot of things that you can do, none of which will work well in all circumstances.  Should a dropped packet mean that all other input for the next second is ignored until the missing packet can be resent?

    Any time you try to queue up commands in any way, you're adding latency.  And remember that latency is your enemy.  If you want a few hundred ms for a full combo to be implemented before you send any of it, then that input latency completely kills you.  If the server trusts the client too much on when you say that you did something, then you open up any multi-player games to all sorts of egregious cheating.  There are no good solutions, but only trade-offs.
  • BobVaBobVa Member UncommonPosts: 125
    Still going on about latency eh?
    The guy , at this point, is clearly trying so hard to tell us all , how he's right and we all, including Google, are wrong. 

    I mean, up until a point, he was 100% sure, that Stadia will depend heavily on the user configuration. After he realized that all you actually get is a.."video" , he's now starting to tell us all, how latency/ping will be so bad, that we couldn't play any game. 

    Assassin's Creed on Chrome would like to have a word with you!

    PS: ..and that was last year, and in Beta! :)
  • ChildoftheShadowsChildoftheShadows Member EpicPosts: 2,193
    Quizzical said:
    Quizzical said:
    Quizzical said:
    On another site, someone pointed out that judder is also going to be a huge problem.  If your average ping time to Google's servers is 20 ms, but that's bouncing around a lot between 10 ms and 50 ms, then a lot of games will be an awful mess of things speeding up and slowing down erratically

    Anything sensitive to timing like combos in Street Fighter may well be completely unplayable.  If you enter a combo properly, but one key press being delayed a little more than another means that 10% of your combos aren't recognized as such on the server, you basically can't play the game.

    Now, you could work around that in a sense by adding extra latency so that everything seems like consistent latency up to some maximum.  For example, they could pad your latency such that it always acts like 100 ms ping exactly unless it spikes above that.  That would mostly avoid the judder problem, but at the expense of making the latency problem far worse.
    That should already be addressed considering multiplayer games and server side authority. Animations for example run at 60fps and servers/client comms is typically 15 and less so to ensure the proper animations start when they should you send 60 frames worth of commands every 15 with included time stamps and the receiving end handles it. 
    Erratic network latency doesn't cause judder in a traditional MMORPG.  It can cause something to occasionally look like a lag spike, but that's about it.  The problem is that, while there won't be any judder as the game is rendered on a remote server, there sure will be when it gets to you.  You can't re-render the game locally at 60 frames per second because the whole point of this is that you're not rendering the game locally at all.

    There's also minimal judder on input latency because a dedicated USB wire is pretty consistent.  Windows polls USB devices every 8 ms by default, and some will reduce that further.  Online games rarely have sensitive timing-based things at all, and to the extent that they do, there's likely to be more trusting the client on when it says a button was pressed than in other places in the game.
    Queue up a series of commands and send it to the google servers. That's all that needs to happen. Then the server treats those as input in the game in the same order you sent them. Nothing will be missed.
    There are a lot of things that you can do, none of which will work well in all circumstances.  Should a dropped packet mean that all other input for the next second is ignored until the missing packet can be resent?

    Any time you try to queue up commands in any way, you're adding latency.  And remember that latency is your enemy.  If you want a few hundred ms for a full combo to be implemented before you send any of it, then that input latency completely kills you.  If the server trusts the client too much on when you say that you did something, then you open up any multi-player games to all sorts of egregious cheating.  There are no good solutions, but only trade-offs.
    If you send the same amount of info, like 60 frames, for every tick you send to the server (remember we're only sending input information) and a packet is dropped you'll pick it up on the next frame. The frames are timestamped and it will be read in order as it was meant to be.

    I'm not familiar with how Google is handling any of this, but like I said this is actually a real problem in today's online games. The example I gave with animations can be thought of like this. Let's say you have the client (current clients, not stadia) performing animations and you are sending what animations you should be playing to the server (this is an antiquated method, because servers should have authority over this as well). You are sitting there idle and on your screen you are idle. On someone else's screen you are also idle. Now you press the jump button. Because the client is only sending animation information 15 times per second if you press that jump button between those intervals you will not be telling the server to play the jump animation. Instead your character will jump while performing the idle animation. This isn't bullshit. I'm not pulling it out of my ass. This can be demonstrated easily. So instead you queue up 60 frames worth of animation commands and send that list every 15. Every frame you are shifting the array and placing in new information. When the server gets the animation commands it looks where it last played an animation and looks at the list consisting of the last 60 frames and grabs the next one that should be played.
  • esc-joconnoresc-joconnor Member RarePosts: 1,097
    klash2def said:
    How they make money and bring it to market is going to very important to note. If it works, considering how most people don't even buy physical games anymore, its going to be a big shift in gaming. Imagine no console, the games are gonna skyrocket to 150-200.00 per game. How will they make up for the loss on the hardware side? Idk if the big three are prepared to take that loss just yet. Not to mention you need the ISPs to help make it work on a global scale.. But we will see. 
    You have it backward. They lose money on the hardware because they know they can make it back on the games. Like how mobile phones are cheaper when you buy them with a plan. With a streaming service they make more money with the hardware because it's not dedicated.
    gervaise1
  • OzmodanOzmodan Member EpicPosts: 9,726
    Quizzical said:
    DMKano said:
    There was an interesting article about this - that basically stated the reason for Stadia is not game streaming as much as it making videos of gameplay and seamlessly putting them on YouTube. 

    So its more of content creation than just pure gameplay as they want to compete with Twitch who owns the live streaming space, but afterwards the videos end up on YouTube. 

    I think that makes a lot more sense - as YouTube streaming will be built in.
    It's unlikely that that's the main reason or even a significant reason for Stadia.  If getting a cut of the Twitch market is the goal, then this is a creatively idiotic way to go about it.  Bandwidth requirements are a major problem, and needing to stream the game in real time makes that bandwidth problem massively worse than when you're not latency sensitive, as is the case with Twitch.
    I don’t think it was their main motivation either, but streaming from stadia will have zero impact on the player because it will stream from googles servers to YouTube. You won’t feel a thing. 
    Nonsense.  Few people have an internet connection that can support game streaming unless you are talking about an extremely simplistic game.  It really cracks me up when people bring up 5G as they are clueless how much such a feature is going to cost.
  • ChildoftheShadowsChildoftheShadows Member EpicPosts: 2,193
    Ozmodan said:
    Quizzical said:
    DMKano said:
    There was an interesting article about this - that basically stated the reason for Stadia is not game streaming as much as it making videos of gameplay and seamlessly putting them on YouTube. 

    So its more of content creation than just pure gameplay as they want to compete with Twitch who owns the live streaming space, but afterwards the videos end up on YouTube. 

    I think that makes a lot more sense - as YouTube streaming will be built in.
    It's unlikely that that's the main reason or even a significant reason for Stadia.  If getting a cut of the Twitch market is the goal, then this is a creatively idiotic way to go about it.  Bandwidth requirements are a major problem, and needing to stream the game in real time makes that bandwidth problem massively worse than when you're not latency sensitive, as is the case with Twitch.
    I don’t think it was their main motivation either, but streaming from stadia will have zero impact on the player because it will stream from googles servers to YouTube. You won’t feel a thing. 
    Nonsense.  Few people have an internet connection that can support game streaming unless you are talking about an extremely simplistic game.  It really cracks me up when people bring up 5G as they are clueless how much such a feature is going to cost.
    If you're streaming the game already, streaming to YouTube will have zero impact because you do not stream it, it is streamed from the data center to YouTube directly.

    That was the topic.
  • ChildoftheShadowsChildoftheShadows Member EpicPosts: 2,193
    klash2def said:
    How they make money and bring it to market is going to very important to note. If it works, considering how most people don't even buy physical games anymore, its going to be a big shift in gaming. Imagine no console, the games are gonna skyrocket to 150-200.00 per game. How will they make up for the loss on the hardware side? Idk if the big three are prepared to take that loss just yet. Not to mention you need the ISPs to help make it work on a global scale.. But we will see. 
    You have it backward. They lose money on the hardware because they know they can make it back on the games. Like how mobile phones are cheaper when you buy them with a plan. With a streaming service they make more money with the hardware because it's not dedicated.
    What plan sells a cheaper phone? I'm genuinely curious. I pay a monthly fee that will total the same as buying outright.
  • adam_noxadam_nox Member UncommonPosts: 2,148
    I take it most of you have not done game streaming, cause yeah this already exists in the form of remote play. And it doesn't work great over 4g currently. Not because of throughput but because of latency, that will likely not go away with 5g.

    Secondly, wired internet from one location to another 3 hours away still has very slight input lag. Luckily, I don't care about twitch games. It works great for anything even a little forgiving. 
  • QuizzicalQuizzical Member LegendaryPosts: 25,516
    Ozmodan said:
    Quizzical said:
    DMKano said:
    There was an interesting article about this - that basically stated the reason for Stadia is not game streaming as much as it making videos of gameplay and seamlessly putting them on YouTube. 

    So its more of content creation than just pure gameplay as they want to compete with Twitch who owns the live streaming space, but afterwards the videos end up on YouTube. 

    I think that makes a lot more sense - as YouTube streaming will be built in.
    It's unlikely that that's the main reason or even a significant reason for Stadia.  If getting a cut of the Twitch market is the goal, then this is a creatively idiotic way to go about it.  Bandwidth requirements are a major problem, and needing to stream the game in real time makes that bandwidth problem massively worse than when you're not latency sensitive, as is the case with Twitch.
    I don’t think it was their main motivation either, but streaming from stadia will have zero impact on the player because it will stream from googles servers to YouTube. You won’t feel a thing. 
    Nonsense.  Few people have an internet connection that can support game streaming unless you are talking about an extremely simplistic game.  It really cracks me up when people bring up 5G as they are clueless how much such a feature is going to cost.
    If you're streaming the game already, streaming to YouTube will have zero impact because you do not stream it, it is streamed from the data center to YouTube directly.

    That was the topic.
    For a lot of games, for the streamer to himself stream the game via Stadia as a way of playing it would be such a crippling disadvantage as to be out of the question.  Even for games that aren't so sensitive to latency, it will still be clunky and unreliable.  That the game is streamed to others wouldn't directly affect the streamer, but trying to play it himself via streaming would be awful.

    If they're going after the Twitch market other than by assuming that all potential customers are idiots, they'll need to go about it a different way.  But a different way presents itself pretty readily.

    If a game is online, then all that Google needs is for the game to make all information necessary to render what the player sees available to one of their data centers.  This requires support on a per-game basis, rather than being able to run arbitrary games.  Still, games that have a spectator mode might already have the support that they need.

    At that point, what could happen is that the streamer renders his game locally as normal, while opting in to allow Google to do likewise.  Meanwhile, Google also renders the game in one of their data centers to show what the streamer sees.  There's no need for Google's rendering to reach the person playing the game or vice versa.  Rather, anyone who wants to watch the streamer gets sent Google's rendering of the game, not what the streamer sees himself.

    There's no need for Google to use the same graphical settings in their render of the game that the streamer is using.  The streamer himself can run the game at a low resolution with a lot of settings turned down or off to maximize his frame rate.  Meanwhile, the Google rendering of the game could be super-max settings, with 4K and some high degree of SSAA, or heavy amounts of ray tracing, or whatever.  If the game is amenable to it, use an entire server full of GPUs to render what a popular streamer sees.  Or an entire rack full of them.

    From the streamer's perspective, he gets all the benefits of running the game locally because that's exactly what he's doing.  Additionally, he gets the load of encoding and uploading the game off of his system.  He may still need to upload some audio, or video of himself to cover a small portion of the screen.  But that's a much lighter load on his system than uploading a full video of the game he's playing.

    People watching the streamer play a game would download the version rendered by Google.  It would have a modest delay of perhaps several seconds, to allow buffering the video a little, compressing images across time, resending dropped packets, or whatever.  Exactly how long the delay should be would depend on the fine technical details.

    From the perspective of the people watching the game stream, the trade-offs would be similar to what you have with YouTube or Netflix.  You can get better image fidelity at the expense of using more bandwidth.  But it would at least be possible to get pretty good image fidelity.  The trade-off curve of what you can do when you're not sensitive to latency is massively better than what can be done if you're really sensitive because you're playing the game yourself.

    While this requires per-game support, that support could end up being pretty common.  If you're a game developer and want your game to get streamed for reasons of marketing, adding this support would not merely lead to more people watching more streamers play your game.  It would also make your game look a lot better visually to the people watching.  And it's probably not that hard to add, anyway.

    Would that be revolutionary to the Twitch market?  I'm not sure.  I really don't get game streaming of that sort, as why watch someone else play a game that you could play yourself?  But it would at least have some real benefits to offer game streamers, which the naive approach of letting the game streamer play a game remotely really doesn't.
  • ChildoftheShadowsChildoftheShadows Member EpicPosts: 2,193
    Quizzical said:
    Ozmodan said:
    Quizzical said:
    DMKano said:
    There was an interesting article about this - that basically stated the reason for Stadia is not game streaming as much as it making videos of gameplay and seamlessly putting them on YouTube. 

    So its more of content creation than just pure gameplay as they want to compete with Twitch who owns the live streaming space, but afterwards the videos end up on YouTube. 

    I think that makes a lot more sense - as YouTube streaming will be built in.
    It's unlikely that that's the main reason or even a significant reason for Stadia.  If getting a cut of the Twitch market is the goal, then this is a creatively idiotic way to go about it.  Bandwidth requirements are a major problem, and needing to stream the game in real time makes that bandwidth problem massively worse than when you're not latency sensitive, as is the case with Twitch.
    I don’t think it was their main motivation either, but streaming from stadia will have zero impact on the player because it will stream from googles servers to YouTube. You won’t feel a thing. 
    Nonsense.  Few people have an internet connection that can support game streaming unless you are talking about an extremely simplistic game.  It really cracks me up when people bring up 5G as they are clueless how much such a feature is going to cost.
    If you're streaming the game already, streaming to YouTube will have zero impact because you do not stream it, it is streamed from the data center to YouTube directly.

    That was the topic.
    For a lot of games, for the streamer to himself stream the game via Stadia as a way of playing it would be such a crippling disadvantage as to be out of the question.  Even for games that aren't so sensitive to latency, it will still be clunky and unreliable.  That the game is streamed to others wouldn't directly affect the streamer, but trying to play it himself via streaming would be awful.

    If they're going after the Twitch market other than by assuming that all potential customers are idiots, they'll need to go about it a different way.  But a different way presents itself pretty readily.

    If a game is online, then all that Google needs is for the game to make all information necessary to render what the player sees available to one of their data centers.  This requires support on a per-game basis, rather than being able to run arbitrary games.  Still, games that have a spectator mode might already have the support that they need.

    At that point, what could happen is that the streamer renders his game locally as normal, while opting in to allow Google to do likewise.  Meanwhile, Google also renders the game in one of their data centers to show what the streamer sees.  There's no need for Google's rendering to reach the person playing the game or vice versa.  Rather, anyone who wants to watch the streamer gets sent Google's rendering of the game, not what the streamer sees himself.

    There's no need for Google to use the same graphical settings in their render of the game that the streamer is using.  The streamer himself can run the game at a low resolution with a lot of settings turned down or off to maximize his frame rate.  Meanwhile, the Google rendering of the game could be super-max settings, with 4K and some high degree of SSAA, or heavy amounts of ray tracing, or whatever.  If the game is amenable to it, use an entire server full of GPUs to render what a popular streamer sees.  Or an entire rack full of them.

    From the streamer's perspective, he gets all the benefits of running the game locally because that's exactly what he's doing.  Additionally, he gets the load of encoding and uploading the game off of his system.  He may still need to upload some audio, or video of himself to cover a small portion of the screen.  But that's a much lighter load on his system than uploading a full video of the game he's playing.

    People watching the streamer play a game would download the version rendered by Google.  It would have a modest delay of perhaps several seconds, to allow buffering the video a little, compressing images across time, resending dropped packets, or whatever.  Exactly how long the delay should be would depend on the fine technical details.

    From the perspective of the people watching the game stream, the trade-offs would be similar to what you have with YouTube or Netflix.  You can get better image fidelity at the expense of using more bandwidth.  But it would at least be possible to get pretty good image fidelity.  The trade-off curve of what you can do when you're not sensitive to latency is massively better than what can be done if you're really sensitive because you're playing the game yourself.

    While this requires per-game support, that support could end up being pretty common.  If you're a game developer and want your game to get streamed for reasons of marketing, adding this support would not merely lead to more people watching more streamers play your game.  It would also make your game look a lot better visually to the people watching.  And it's probably not that hard to add, anyway.

    Would that be revolutionary to the Twitch market?  I'm not sure.  I really don't get game streaming of that sort, as why watch someone else play a game that you could play yourself?  But it would at least have some real benefits to offer game streamers, which the naive approach of letting the game streamer play a game remotely really doesn't.
    Google didn't create Stadia to compete with Twitch, YouTube already does that with their game streaming section. What I'm talking about in this particular post is because players are already playing games that are on the data center, meaning the game is not local to your device, one of the advantages is being able to split the stream so one of them goes to the player, you, and the other goes to YouTube for a live stream or recording. There could be a chosen delay and will likely be a buffer while live streaming on YouTube. No extra support required. That's just one of the advantages to this online gaming setup.
  • MyrdynnMyrdynn Member RarePosts: 2,484
    edited March 2019
    I can't see how this is good for gaming as of right now, my son streams off my Steam account rather than downloading and both our systems are hooked to ethernet and 10 feet apart

    his lags so much more than mine ever does
  • esc-joconnoresc-joconnor Member RarePosts: 1,097
    klash2def said:
    How they make money and bring it to market is going to very important to note. If it works, considering how most people don't even buy physical games anymore, its going to be a big shift in gaming. Imagine no console, the games are gonna skyrocket to 150-200.00 per game. How will they make up for the loss on the hardware side? Idk if the big three are prepared to take that loss just yet. Not to mention you need the ISPs to help make it work on a global scale.. But we will see. 
    You have it backward. They lose money on the hardware because they know they can make it back on the games. Like how mobile phones are cheaper when you buy them with a plan. With a streaming service they make more money with the hardware because it's not dedicated.
    What plan sells a cheaper phone? I'm genuinely curious. I pay a monthly fee that will total the same as buying outright.
    Really? With no discount? The original contract might work that way, but in Tokyo I get a discount for every month I actually am paying out the term. If I were to cancel I'd have to pay more for the phone.
Sign In or Register to comment.