Support NeoGAF

[SJRB]

I can't stop laughing at the girls just whirlwinding their arms at the start of the video.


..but yeah, this is AWFUL.

 

[Kopite]

From the looks of it at the bottom, I think you have to mimic motions and it's not 1:1.

Well I definitely hope so.

 

[Nozem]

Holy input lag, Batman!

 

[cyberheater]

Huge latency, or am I mistaken? Look at the right player:

VG Leaks state the latency for Kinect 2.0 is:-

~60 ms with processing

That video is more like 500ms. Something is seriously wrong.

 

[Alx]

The game itself doesn't look that bad actually. In fact it's exactly what I was hoping for : not too realistic, fancy martial arts, kicking and dodging,...
Hard to tell about the gameplay though. The two girls didn't seem to know what they were doing. Actually I'm not even sure they were playing the game.

Will try for sure.

 

[nelsonroyale]

I think it would be more fun looking at the people playing this game than actually playing it haha. To be honest, real life low-contact sparring is MUCH more fun than that...there is probably a higher chance of connecting here as well

 

[derFeef]

VG Leaks state the latency for Kinect 2.0 is:-

That video is more like 500ms. Something is seriously wrong.

Your interpreting the video wrong I think, like I said above.

 

[Xater]

I'm going to guess IGN was looking at a raw mapping of depth data, maybe overlaid with a video of themselves.

If there was a video overlay or thumbnail, it would of course look like there's no latency...because the video would be synched to depth map. The depth map itself would be only half the story as far as processing involved for most games goes.

All totally different to a real game using skeleton tracking.

Also if Carmack says the tech is not there I would believe him.

 

[Heretic]

From the looks of it at the bottom, I think you have to mimic motions and it's not 1:1.

good catch. hmm...

 

[GetemMa]

Truly awful.

Ubisoft should be ashamed of themselves.

 

[Law Abiding Stranger]

Aym a faytur hurr.

 

[Arulan]

There are people who believe fighting games are just button mashers, which for anyone who has played them in depth knows is false. This game looks like it was made by one of those people, except he decided buttons were too hard.

 

[NuKERxyz]

WTF!
I must be sleeping...

 

[VlaudTheImpaler]

Yeah, I am a HUGE Kinect supporter but...

 

[Nozem]

Good thing every Xbox One owner has Kinect, otherwise they wouldn't be able to experience this brilliant masterpiece.

 

[BleachAndPepsi]

Is this the "unique" exclusive Microsoft had mentioned?

 

[Monumental]

A page full of "lol lag" comments from people who doesn't know what lag actually is.

Kinect 2.0 can track the body with very little lag. But writing the code on how those movements translate to on screen actions and movements on screen, how the animation plays out, which animations have priority, etc...that's all on Ubisoft.

Basically if you see this gif and think thats a problem with Kinect and not Ubisoft then you're doing it wrong.

This video is not an indication of Kinect Lag. Just like how using controllers movement in one game can seem floaty but responsive in another. See Japanese fighting games vs the "stiff" Mortal Kombat animations. Call of Duty vs. Killzone, etc...

 

[enzo_gt]

Just reading some stuff on IGN and other sites and they say the input lag has almost disappeared. I guess the reality is that it hasn't.
It would explain why no Kinect games were shown at E3.

Input lag can be one cause, the other is recognition.

Compare the demos they showed at E3 where it pretty damn well looked like they synced up 1:1 to this and you can tell that when your making an actual game there are a lot more factors to consider. What I think is happening here is that they're flailing and not even necessarily throwing punches but doing a sort of dance move that Kinect is trying to decipher when it's decidedly a punch and not just a push/throw. It's a design issue that won't really be solved no matter how good your latency is because your physical motion needs to complete, which by then might be too late. Hardware is an easy scapegoat for people who just came in here to rag on Kinect. Much more at play that needs to be considered.

 

[DrM]

Why are they even showing the live demo if they have so much problem with input?

 

[Wynnebeck]

UBISOFT. STAHP.

 

[ThoughtsOfSpeaking]

so much for unnoticeable input lag

.

 

[derFeef]

Why are they even showing the live demo if they have so much problem with input?

Game seems to recognize all inputs - look at the bottom.

 

[medze]

From the looks of it at the bottom, I think you have to mimic motions and it's not 1:1.

The interviewer in the video asks "so its really kinda 1:1, you punch and then you see it punch on the screen?" and the guy responds "yeah, its really 1:1"

 

[traidenk]

ok... This look really bad

 

[gofreak]

VG Leaks state the latency for Kinect 2.0 is:-



That video is more like 500ms. Something is seriously wrong.

VG Leaks is a bit vague about what that 60ms gets you. The doc they published says "End-to-end system latency for Kinect is measured as the time from light hitting the sensor through to the display outputting an update based on that input. " - an update to what, though? The minimum they might be referring to would be a depth map update. That would represent the minimum lag rather than the max - the max depending on what processing a game requires of that depth map to infer intended input. For example, skeleton processing may not be counted in that. So you might have more lag inside Microsoft's library - if you need that type of tracking - plus then your own game's pipeline's lag.

I'm sure Ubisoft's own code is a big contributor here but the VG Leaks number may not be the end of the story as far as base processing goes.

 

[Coiote]

well, it seems gamescom already had a great start.

 

[dab0ne]

says the video is private

 

[enzo_gt]

Video just went private?

 

[Dark_castle]

Can Ubisoft scrap this shitty Kinect crap and fund Child of Light instead? Like, this looks terrible.

 

[AlStrong]

stewie & bertram slapfight.gif

 

[Spongebob]

Kinect has it's place, but this reminds me of playing boxing on the Wii with my brother. I tried to take is serious and be strategic. He did this

and kicked my ass.

Yah, in the garbage bin.

 

[dose]

Sorry, but that looks absolute shit. What the players are doing seems to have very little bearing on what the ingame characters are doing, and when they do copy (crouching for example) the lag seems even worse than Kinect 1. Geez.

 

[Poetic.Injustice]

Video went private, Rehost please.

 

[ElTorro]

says the video is private

I downloaded the video before it went private. Can share it. What's the prefered file hoster on GAF?

 

[nib95]

VG Leaks is a bit vague about what that 60ms gets you. The doc they published says "End-to-end system latency for Kinect is measured as the time from light hitting the sensor through to the display outputting an update based on that input. " - an update to what, though? The minimum they might be referring to would be a depth map update. That would represent the minimum lag rather than the max - the max depending on what processing a game requires of that depth map to infer intended input. For example, skeleton processing may not be counted in that. So you might have more lag inside Microsoft's library - if you need that type of tracking - plus then your own game's pipeline's lag.

I'm sure Ubisoft's own code is a big contributor here but the VG Leaks number may not be the end of the story as far as base processing goes.

I don't know what the hell you're on about half the time, but damn do I like getting your insight. Great posts. Always technically versed.

 

[Hydrargyrus]

Seriously, Kinect is not for games

 

[Billen]

While a potential hit at barcades, this looks really uninteresting.

 

[Mpl90]

The lag. It hurts.

 

[Krilekk]

Look at both players lol. Major lag/delay.

There is an explanation for both the reports of barely no lag vs. this video: Kinect lag depends on the number of players, so one player, 120 ms lag, 2 players, 240 ms lag ...

 

[jabuseika]

That is terrible.

Not even once.

 

[CambriaRising]

That looks terrible. So much lag.


The Fight: Lights Out?

Yeah that one.

 

[~~Hasan~~]

weird for me it says this video is private ^_^

why its not working for me ?. i even tried VPN >_>

 

[aly]

I'm not sure what I just watched. I really want Kinect to be useful and fun for things other than Dance Central, but that looked awful.

 

[cyberheater]

I'm going to guess IGN was looking at a raw mapping of depth data, maybe overlaid with a video of themselves.

If there was a video overlay or thumbnail, it would of course look like there's no latency...because the video would be synched to depth map. The depth map itself would be only half the story as far as processing involved for most games goes.

All totally different to a real game using skeleton tracking.

Input lag can be one cause, the other is recognition.

Compare the demos they showed at E3 where it pretty damn well looked like they synced up 1:1 to this and you can tell that when your making an actual game there are a lot more factors to consider. What I think is happening here is that they're flailing and not even necessarily throwing punches but doing a sort of dance move that Kinect is trying to decipher when it's decidedly a punch and not just a push/throw. It's a design issue that won't really be solved no matter how good your latency is because your physical motion needs to complete, which by then might be too late. Hardware is an easy scapegoat for people who just came in here to rag on Kinect. Much more at play that needs to be considered.

You can see from this video that the latency is indeed almost zero.

http://www.youtube.com/watch?v=bdviGrPaQDQ

So It's really up to the developers to write efficient algorithms to convert that skeletal data to controller input.

At this point. I'm going to say. If I see significant lag in Kinect games. I'm going to blame the developer and not the hardware.

 

[ElTorro]

VG Leaks is a bit vague about what that 60ms gets you. The doc they published says "End-to-end system latency for Kinect is measured as the time from light hitting the sensor through to the display outputting an update based on that input. " - an update to what, though?

For matching patterns in real-time video feeds you usually need more than one frame to extract and detect objects in motion. You need the delta between many images to be sure that a certain shape in frame A represents the same physical object as in frame B. I'd guess that this pre-processing step is included and might even be done in hardware. Newer camera systems in autonomous vehicles work that way.

The extraction of a higher-level model is usually done in software. In the case of the Xbox, that would be done by some operating system service. I am not sure, if they included that in the actual latency statement; that might be a theoretical value and the update of different models (skeletal, depth, force) certainly takes different amounts of time.

 

[FranXico]

Movie spoiler, beware! (as unlikely as it is that you didn't know).

It should be called "Fight Club"

Spoiler

and the player should beat the crap out of himself.

 

[madmackem]

Horribad,wow. Im really expecting the worst from kinect 2 at this point, a slightly better still laggy kinect.

 

[derFeef]

First Kinect game they show (at gamescom) and it's an unoriginal stupid fighter, man...

 

[Monumental]

The interviewer in the video asks "so its really kinda 1:1, you punch and then you see it punch on the screen?" and the guy responds "yeah, its really 1:1"

This game is 1:1 the way Twilight Princess' sword fighting is 1:1.

 

[wapplew]

Lag aside, design a proper kinect 1 to 1 fighting game is really hard I think.
Just too much input data need to fine tune. X,Y and Z axis, power and speed, different angel of the punch, just too much factor. Making a rag doll that react to every kind of punch already harder than anything, let alone the complex physics model.