Namco on Tekken 7's input lag: "Unreal Engine 4 eats 3 - 4 frames"

As far as I can tell IJ2 has the same amount of lag as current SFV, which would suggest that there is no inherent issue with the engine that can't be overcome by developers.

Injustice 2 runs on the same engine as MKX - a customized version of UE3, not UE4. Guilty Gear XRD also runs on UE3 and has about ~5 frames of input lag on console, 2 full frames lower than SFV. The numbers are better on PC for everything, but the gap is actually bigger between the UE4 games and the UE3 games.

This is the fear with Dragon Ball FighterZ. Tekken and Injustice input lag matters a lot less because the faster moves in those games are between 8-10 frames. Street Fighter and GG have 3-4 frame moves, which means DBFZ is likely to have some 3-4 frame moves.

It's the difference between something being impossible to react and being difficult to react.
 
Injustice 2 runs on the same engine as MKX - a customized version of UE3, not UE4. Guilty Gear XRD also runs on UE3 and has about ~5 frames of input lag on console, 2 full frames lower than SFV. On PC, SF is better but the gap is worse between GG/SF.

This is the fear with Dragon Ball FighterZ. Tekken and Injustice input lag matters a lot less because the faster moves in those games are between 8-10 frames. Street Fighter and GG have 3-4 frame moves, which means DBFZ is likely to have some 3-4 frame moves.

It's the difference between something being impossible to react to and difficult to react to.
You could see the difference in SFV play immediately when the input lag patch hit- Ken and Nash specifically were hurt huge as people started to be able to react to things they were doing.
 
I would be delighted if this leads them to turn off all the gaudy post-processing effects to help reduce input lag, or at least put in the option for competitive players. It would kill two birds with one stone.
Either that or (the unlikely scenario) ASW lets Capcom & Bandai Namco use their post-processing stack.
 
Injustice 2 runs on the same engine as MKX - a customized version of UE3, not UE4. Guilty Gear XRD also runs on UE3 and has about ~5 frames of input lag on console, 2 full frames lower than SFV. The numbers are better on PC for everything, but the gap is actually bigger between the UE4 games and the UE3 games.

This is the fear with Dragon Ball FighterZ. Tekken and Injustice input lag matters a lot less because the faster moves in those games are between 8-10 frames. Street Fighter and GG have 3-4 frame moves, which means DBFZ is likely to have some 3-4 frame moves.

It's the difference between something being impossible to react to and difficult to react to.

Tekken has sidestepping, which have around 3 frames of startup. The lag makes reacting with movement far less reliable.

It's just as big of a problem in Tekken as in SF.
 
That would be the best.

Fighters going forward ditch Unreal Engine and use your own engine instead
That's a lot easier said than done, man.

"Hey just ditch this super well documented engine and easy to use software to spend multiple millions of dollars to make your own that you might only use a time or two the entire generation"
 
How much input lag did Tekken 2 on the PSX have? Now that was a sluggish game coming from the super fast and responsive Virtua Fighter 2.
 
He himself has said his #s are not to be trusted alone & should be cross referenced with display lags.

That's not exactly what I said. I strongly recommend comparing my results against my results. They tend to differ from displaylag's by 0.5-1F. i.e. if you compare my results for PC T7 versus their results for PS4 T7 you end up with an extra frame between them.

http://forums.shoryuken.com/discuss...hod-games-controllers-etc-20-4-pc-games-added

If you check the posts from 20 days ago you can see where I go through the differences between our two sets of results.
 
Let's wait and hope that the imminent patch fixes the input lag, and it would be cool if they improve the online.
 
Maybe I'm missing something but considering fps isn't an issue. why would they pick the laggiest input version to be the tournament standard?
 
This doesn't explain the difference between XB1 and PS4. Something weird is definitely going on with the game on all platforms lag-wise, even offline, but PS4 version is apparently really bad on the Pro.
Last lag thing I saw on Kor's said that the Pro had better lag than the base PS4.
 
UE4 is great for developer productivity, but the technical performance on consoles this gen is very poor.

They definitely need to work on the default console rendering pipeline with an eye of minimizing input lag on the PS4 especially. UE4 is a fantastic tool for devs, but if you're making a game requiring fast reaction times like a fighting game it looks like atm you basically have to rewrite the default post-processing pipeline.
 
They definitely need to work on the default console rendering pipeline with an eye of minimizing input lag on the PS4 especially. UE4 is a fantastic tool for devs, but if you're making a game requiring fast reaction times like a fighting game it looks like atm you basically have to rewrite the default post-processing pipeline.
I think a beyond3D dev mentioned that UE4's lack of console optimization particularly to GCN means devs by default leave 20% performance on the table.
 
I think a beyond3D dev mentioned that UE4's lack of console optimization particularly to GCN means devs by default leave 20% performance on the table.

Yowch, that's pretty painful. UE4 is such a fantastic engine in terms of versatility, much like Unity, that it's a shame to see it running poorly on PS4. At least in general it seems to run better than Unity but there's definitely work to be done.

How big is Epic's engine team? They seem to be pretty good at pushing out new updates, but there's also a lot of features I see on their roadmap that end up being backlogged.
 
Is this also a problem in the arcade T7 edition? I figure that that is also UE4.

Tekken 7 arcade had lag that some pro players talked about in some interview, but they probably thought it was due to the networked cabinets rather than game engine quirks.
 
I think a beyond3D dev mentioned that UE4's lack of console optimization particularly to GCN means devs by default leave 20% performance on the table.
bs1lQN.gif


Very interesting, I was thinking UE4 does not fully optimise to each specific console hardware much that's why you see UE4 games with more parity on games between each consoles like Snake Pass etc not 100% optimised to each hardware just optimising some hardware more than the other


He is probably correct. UE4 tends to heavily favour Nvidia architectures.
Nintendo Switch uses Nvidia architecture.

So that's why UE4 engine is very well optimised on Switch compared to PS4/XB1 not as optimised to their strengths fully... Hmm
 
Nintendo Switch uses Nvidia architecture.

So that's why UE4 engine is very well optimised on Switch compared to PS4/XB1 not as optimised to their strengths fully... Hmm
Now I really hope that DBFZ gets on the Switch (& MvCI, but I doubt that'll come to the Switch) so we can compare the input delay between the systems.
 
Modern game engines all tend to exhibit input latency to a degree just due to the complexity of the renderers.

We're never getting back to the retro days of virtually no lag - it's just not possible.
 
Now I really hope that DBFZ gets on the Switch (& MvCI, but I doubt that'll come to the Switch) so we can compare the input delay between the systems.
Yeah that would be interesting and also I expect Switch version overall to hold up much more
https://twitter.com/noodalls/status/882568216588963840
The patch shaves off 7 ms of input lag on the PS4. That's like 0.4 frames.

Even after the patch, Tekken 7 is still the fighting game with the worst amount of input delay out in the market today.
Lol well
 
Yeah that would be interesting and also I expect Switch version overall to hold up much more

Lol well
Why, because of Nvidia's optimizations for UE4? I'd imagine that the game would run at 720p/60fps on the Switch if (maybe 900p, seeing as the PS3 is able to run GGXrd at 720p/60fps).
 
Why, because of Nvidia's optimizations for UE4? I'd imagine that the game would run at 720p/60fps on the Switch if (maybe 900p, seeing as the PS3 is able to run GGXrd at 720p/60fps).
Maybe not just resolution i imagine the Ram constraints for that one but graphical settings/effects should be on par or very close though shouldn't be too hard with it just being a 2d fighter
 
He himself has said his #s are not to be trusted alone & should be cross referenced with display lags.

That means DisplayLag's numbers aren't supposed to be trusted alone either. And for what it's worth, Noodalls is a lot more forthcoming about his process than DisplayLag is.
 
When I get this, it will be on PC. But considering the majority of sales for a fighting game will be on consoles, I am surprised at some of the decision making with this game.
 
Top Bottom