Mouse movement


(trickykungfu) #41

[QUOTE=SRS-Kap;495851]Hi all,

For those of you experiencing these issues, I would be very interested to hear your feedback if you try the following command in the console:

RawInputUpdateTest true

For those of you that try this, please reply even if it is just to say that this makes no difference (or even if it makes things worse!).

You can also disable this test behavior so you can compare having it enabled or disabled using:

RawInputUpdateTest false

Thanks for your help![/QUOTE]

tried it but don’t feel much different.

I always used nearly the same settings Pixel posted in here. Those really make a big different for me. Maybe i should compare std. settings with the rawinputupdatetest. Or maybe SD should just fore those settings Pixel posted.


(fzl) #42

Pixels settings are not bad…

btw when you set max delta time to 250 and the client rates to 25 or 40000 the wall jumps are easyer than before,…no idea why


(k0k0nat) #43

10x Twitch, but:

For dummies, can I simply copy paste your commands into the cfg?


(trickykungfu) #44

no u can’t. They have to be in the right position.

fzl best command is press shift and set rate to 1 so no1 can hit u while running XD They really have to limit the config :slight_smile:


(fzl) #45

has anyone spoken of cake that you crumble you want to report?


(trickykungfu) #46

pls fix the mouse XD I started playing sniper again and now i know why i quit it many months ago ^^


(Kl3ppy) #47

Wont happen, many ask for some mouse change/improvement and nothing happend.


(dIaBol0s) #48

before the last patch the mouse feeling was ok for me and the servers were better attended. See now …


(sunshinefats) #49

Having tried a couple of the suggestions here I can say it didn’t help my situation much. It still feels off and as a result I find it more often difficult to kill enemies. Moreover difficult to enjoy the game…really thinking at this point you should have gone Id for your engine, but what’s done is done, so let’s hope this gets worked out soon.


(Mustang) #50

Apparently Oculus managed to shave upto 20ms off their input latency with time warping.

//youtu.be/WvtEXMlQQtI


(spookify) #51

Played last night and mouse lag was horrible…

It also seemed like movement was delayed too…

Crazy frustrating!!


(DJswirlyAlien) #52

What is the deal with mouse input at the moment? What fixes and hardware tweaks can I do for a better experience?


(Smooth) #53

[QUOTE=Mustang;498151]Apparently Oculus managed to shave upto 20ms off their input latency with time warping.

//youtu.be/WvtEXMlQQtI
[/QUOTE]

I believe RawInputUpdateTest true updates the camera position as late as possible, at the start of the render-frame rather than the start of the game-frame. This does shave off a few milliseconds.

Implementing ‘Time warping’ so this is updated during the render-frame is possible for us but there are a few issues:

[ol]
[li]Your ‘real’ aiming crosshair wouldn’t actually be there if you fired a bullet (it would be back where is was during the game-frame)[/li][li]The edge of the screen would flicker black (unless we rendered a little more around the edges or performed some trickery but that costs framerate)[/li][li]Instead of a fixed 60/75hz, we try to render as at high a framerate as possible, any extra calculations like this will also lower framerate[/li][/ol]
The first point is probably acceptable as it would only be a tiny imperceptible offset and the screen motion would feel more responsive. The second point is potentially solvable (at a small cost) and the third may also only have a small impact.

That being said, having a higher framerate in itself will always reduce input latency (as the frame is rendered and then displayed sooner - some the OR doesn’t do) so this is our main goal.

As we continue with development you should be able to see gradual performance improvements and much more importantly a more consistent frame rate with fewer dips.


(PixelTwitch) #54

[QUOTE=Smooth;498355]I believe RawInputUpdateTest true updates the camera position as late as possible, at the start of the render-frame rather than the start of the game-frame. This does shave off a few milliseconds.

Implementing ‘Time warping’ so this is updated during the render-frame is possible for us but there are a few issues:

[ol]
[li]Your ‘real’ aiming crosshair wouldn’t actually be there if you fired a bullet (it would be back where is was during the game-frame)[/li][li]The edge of the screen would flicker black (unless we rendered a little more around the edges or performed some trickery but that costs framerate)[/li][li]Instead of a fixed 60/75hz, we try to render as at high a framerate as possible, any extra calculations like this will also lower framerate[/li][/ol]
The first point is probably acceptable as it would only be a tiny imperceptible offset and the screen motion would feel more responsive. The second point is potentially solvable (at a small cost) and the third may also only have a small impact.

That being said, having a higher framerate in itself will always reduce input latency (as the frame is rendered and then displayed sooner - some the OR doesn’t do) so this is our main goal.

As we continue with development you should be able to see gradual performance improvements and much more importantly a more consistent frame rate with fewer dips.[/QUOTE]

What I do not understand (and maybe you can shed some light on the matter.) is why you need to have 250-300fps in Dirty Bomb to have the mouse feel as responsive as other games on other engines running at 90-100 fps (my magic FPS in most games is actually 92)…

Now I am even talking about with Rawinputtest = True and OneThreadLag = 0 (so we should not have double frame latency if I understand correctly).

I do understand that the errors caused by rounding are causing us to over compensate and that could add to the feeling of “lag” however I am dubious that something else is not going on behind the scenes here…

Possible it is something to do with the number of counts per second unreal receives from the mouse?

EDIT///
Normally when a game recieves too few counts you would notice “microstutter” like symptoms at very high frame rates (most noticable if playing a game like UT2k4 with a 125hz mouse at over 125fps) . I am about to test if setting my mouse to 125hz with 300fps results in the same effect… IF NOT… then the Unreal Engine is storing movement between frames and could explain the floaty feeling.

EDIT 2///
The stutter is still there at 356fps when using 125hz so it is likely that the number of times unreal counts mouse movement being a major issue.


(Smooth) #55

My gut feeling is that spikey frame-rate and rounding errors are the current main culprits but I’m not a programmer and SRS-Kap would know more.

He’s the Lead Programmer on DB and very passionate about things like this so it will be improved over time.

Unfortunately the unreal engine does seem to have several elements that introduce input latency as standard and things need to be tweaked and moved around to reduce it.


(PixelTwitch) #56

[QUOTE=Smooth;498365]My gut feeling is that spikey frame-rate and rounding errors are the current main culprits but I’m not a programmer and SRS-Kap would know more.

He’s the Lead Programmer on DB and very passionate about things like this so it will be improved over time.

Unfortunately the unreal engine does seem to have several elements that introduce input latency as standard and things need to be tweaked and moved around to reduce it.[/QUOTE]

Well I am happy to say that you guys have already amazed me when it comes to breaking down the unreal engines limitations…
I know a lot of people likely do not realise the work you guys put in to make these features and fix things but it really shows…

Demo Recording, Adding back First Person Spectator, Reduced mouse latency, Removing frame caps and actually being one of the ONLY developers able to decently optimise the game performance wise…

Possibly the best unreal engine since games like Mass Effect 3 performance wise and BL:R shooting wise.

Keep it up :slight_smile:


(trickykungfu) #57

off Topic or not…

For me in a 5on5 match my mouse feels a lot better. I think because of better FPS. But why does the playercount effect those FPS that much? On a empty server i allways have 120fps but 6on6 its between 30-80…


(PixelTwitch) #58

[QUOTE=trickykungfu;498377]off Topic or not…

For me in a 5on5 match my mouse feels a lot better. I think because of better FPS. But why does the playercount effect those FPS that much? On a empty server i allways have 120fps but 6on6 its between 30-80…[/QUOTE]

The low FPS with lots of players comes from multiple reasons…

First off, the game runs at 100tick and due to the spectator mode I believe the client is actually working with the data that comes from 11 other players 100 times a second… Also, code may still be a little sloppy for certain mercs and abilities. I would expect this to improve a fair amount in future but we will see :slight_smile:


(Smooth) #59

Game-code is currently undergoing an optimization pass, in the next update things should start to improve :slight_smile:


(davidemo89) #60

Whoa! Sometimes next week? =D