Google Stadia (formerly Project Stream)

I hope you are right elvis, but these streaming and game rental platforms are a publishers wet dream. As far as the tech, I agree this will be here sooner than people think and better than many would like to admit.

 

"When we launch, because we’ve made some very significant improvements to our encoder, our streamer, and our compression algorithms, we will get 4K / 60 frames per second in about 30 megabits per second." - Phill Harrison

interview via Polygon

Oh and all the next big news is set to arrive June some time (but not at E3 ^^ )

 

Isn't that trade off going to feel terrible though?

Control to screen latency (Input lag) is currently extremely minimal on pc. This allows the player to fire a shot and see it happen almost instantly, before the server has even confirmed that the player was allowed to shoot. Of course once that action data does the round trip the server may decide that the player wasn't allowed to shoot, so the bullet won't do any damage, and the ammo will be refunded, or whatever... but this is much less jarring for a player, than clicking fire, waiting for the entire round trip + de/compression latency, and then the gun shooting.

Basically, isn't input delay is currently minimised due to networking tricks allowing the client to act instantly, then later have those actions verified? This wouldn't be possible with streaming, since your input has to literally go all the way to the server, be processed and streamed back before you can see it on your screen.

 

Go and measure the lag on something like the latest Assassin's Creed game.

I'm not talking your screen's lag. I'm not talking your controller's lag. I'm talking old school - attach an LED to a digital button, record your screen on a 120FPS camera, and count frames from button press to first frame of animation change.

That lag is already enormous. Why? Mostly because games have changed. Developers have been adding in progressively longer and longer activation delays on things mostly because of animation changes. Jump cycles, run cycles, reload cycles - animation is longer and more realistic than ever. Super Mario Bros had Mario's pixels leaving the ground roughly 2 frames after button press, and with 1 frame of animation change. Today, that's way more.

Here's some testing folks did on consoles. PC is the same (yes, really - again these are baked in, in-game, in-engine lags that can't be compensated for, and not display, control or network lag):
https://displaylag.com/video-game-input-lag-database/

8 frames in some cases. For a 1v1 fighter, where 1 frame lags can mean life or death.

I notice it. I played Fortnite for a minute (kids nagged me, I had a go, hated it). Compared to GLQuake or Doom, the lag in button press to thing happening on screen is enormous, even at 144FPS on a kick arse gaming rig. The engine has put that in for various buffering reasons.

That *already* feels terrible to me. Will Google Stadia make it worse? Full end-to-end perceived lag, I think they can get down to almost identical to what happens now. Again, because they're just substituting one lag for another, and new games are already made with this in mind from an engine/animation point of view.

I full understand and appreciate what it is that "PC Master Race" love about their platform. But the reality is that objectively measured, it's already worse today than it was 10 years ago, and mostly as buffering to limit the effect network play has on games. So you've already been slowly conditioned to this for years. Google Stadia is going to use that to slide in unnoticed.

Beyond that, "PC Master Race" isn't the target audience for this. I'm not going to start another "PC is a minority player" argument here. Whether it is or isn't doesn't matter. Folks have to accept that there's more money in folks playing a game for a few hours than there is in people playing hardcore squads and teams for months/years on end, and Google is going to capitalise on that market. The specific cases of massively multiplayer (not just RPG, but shooter as well) and casual gaming are still rife for milking, and Google's going for it. The entire population of OCAU is now a rounding error in their projections for their new audience numbers.

 

I notice it too... but I feel like that may make the lack of being able to mask the delay with networking even worse. There seems to be a threshold where delays become really noticeable. Under that we don't notice, and over that we feel a game is sluggish/unresponsive, and further beyond that, outright laggy.

My point was that you're substituting a lag that can be masked by networking, with one that can't.

I hope so... it is great if it all works. I just hope poor responsiveness isn't further normalised.

So we'd go from:
1) Input to OS
2) Game Engine processing/Animation-starts
3) Client's state Display + Client to Server
4) Server Processing
5) Server to Client
6) Client's server confirmed display (simultaneous with the next round of step 2 at this point in time)

To:
1) Input to OS
2) (thin)Client to Server
3) Server Processing + Game Engine processing (assuming this would be simultaneous)
4) Server to (thin)Client
5) Client's server confirmed display

But wouldn't the animation windups then have to start at the new step5 instead of step2. It doesn't shorten those windups at all.

I'm not trying to be negative... just trying to understand it all since I'm already really sensitive to dropping frames or packets and annoyed by current netcode implementations.

 

As long as you're playing games where people are on packet-switched, congested networks thousands of kilometers away, this will continue to be a problem.

Game streaming from a data centre versus local games sending packets to clients is splitting hairs for which is better or worse. Both have their pros and cons (one will allow far higher scalability of player count, the other ever so slightly lower perceived lag).

If what you absolutely want is low-lag netplay, the only answer is a LAN where you control all of the variables. Everything after that is a compromise. I'm old enough to remember people throwing massive tantrums about online play and swore they'd never play on the Internet, and only ever play on local networks "because lag". Yet most of them today are playing online games, despite their blood oath promises not to.

 

I ran Cat 5 through the walls when I brought my first house (damn expensive then) and set up a dedicated LAN area, swearing never to be an online gamer (still dial-up in my area to be fair), but now even though I still mainly play single player, I do enjoy some online play (and some console gaming and some retro gaming). It is what it is, I fight trends that don't align with me for as long as possible, but I'm not too stubborn to change when I have to. I'd be the one missing out if I did.

 

So is this all working? How much is it? I don’t need a high spec machine for this aye?

 

that's the hype.

 

The other concern I've seen floated around regarding this is Google's commitment to unusual projects. If things don't quite work out as planned, they have a tendency to forget about it, let it languish and then give up and turn it off. I certainly wouldn't be dropping money on this with how some of their projects have turned out.

 

this isn't an unusual project though, this is the future that publishers have wanted since the dawn of gaming.

 

Poor choice of words from me but my point stands, Google just walk away from things and let them die routinely.

 

they have a threshold for success that most of us don't understand. If this passes it, it'll stay.

 

I have no doubt, their stuff that works works well but I want to wait and see it get to that point first.

If it gets to that point though, I worry about the inetivable "oh it's a Stadia exclusive".

 

Google are already well versed in software as a disposable item.

I've bought several games on Android that were a) exclusive to mobile and b) no longer can be downloaded.

 

mirrors many of my thoughts

 

I feel the same
I’ve got wall to wall Wardrobe in my gaming room with my precious collections inside ready to go whenever I feel the desire to play them.

I do really hope we get backwards compatibility with the PS5 thou because that would make playing older gen games oh so much easier.

 

So you are watching a compressed video right? Surely that will look like ass if compared to actual game side by side. Be interesting to see.

I wouldn't touch this unless I am pinging server constantly under 10ms.

 

a few interesting (but obvious) points by someone who tried it out

I look forward to low latency NBN v2 when it arrives in 10+ years . By then hdmi 2.1 will be common so variable refresh rate (VRR) with ALLM (auto low latency mode) will be standard on $400 10K (its like 8K but wider) TV panels.

Search this post in 2029 for an accuracy check

Also interesting that it consumes about 11GB per hour.

 

Live Stream (in approx 12 hours from when I post this) with Google outlining more launch details, hopefully to include pricing and dates etc




An hour after this comes Bungie (first news post split with activision) detailing Destiny 2 plans - which are said to include cross play with xbox, pc and Stadia.